The Poetic Genesis of Old Icelandic Literature 3110643936, 9783110643930

Citation preview

Mikael Males The Poetic Genesis of Old Icelandic Literature

Ergänzungsbände zum Reallexikon der Germanischen Altertumskunde

Herausgegeben von Sebastian Brather, Wilhelm Heizmann und Steffen Patzold

Band 113

Mikael Males

The Poetic Genesis of Old Icelandic Literature

ISBN 978-3-11-064183-7 e-ISBN (PDF) 978-3-11-064393-0 e-ISBN (EPUB) 978-3-11-064237-7 ISSN 1866-7678 Library of Congress Control Number: 2019952375 Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at http://dnb.dnb.de. © 2020 Walter de Gruyter GmbH, Berlin/Boston Typesetting: Integra Software Services Pvt. Ltd. Printing and binding: CPI books GmbH, Leck www.degruyter.com

To Freyja

Contents A Note on Translations Introduction 1 1.1 1.2 1.3 2 2.1 2.2 2.3 2.3.1 2.3.2 2.3.3 2.3.4 2.4

IX

1

Metre and Rhyme 20 The Eleventh Century 23 Grammatical and Historical Awareness in the Twelfth Century and Beyond 31 Summary 38

2.4.1 2.4.2 2.4.3 2.4.4 2.4.5 2.4.6 2.4.7 2.5 2.6

Diction: Mythology, Wordplay, Metaphor and Tmesis 39 Kennings, Mythology and Poetic Figures 39 The Early Period (up to c. 995) 49 The Intermediate Period (c. 995–c. 1120) 56 Wordplay, Extended Metaphor and Tmesis 57 Mythological References 59 Specific Mythological References: Eiríkr jarl and Others Specific Mythological References: Kings 68 The Late Period up to Snorri and his nephews (c. 1120–c. 1300) 75 Einarr Skúlason 77 Øxarflokkr 78 Geisli 82 Einarr and Ofljóst 88 Hallar-Steinn 89 The Thirteenth Century 91 Concluding Remarks on Diction 93 Mythology in Medieval Europe 94 Summary 100

3 3.1 3.1.1 3.2 3.2.1 3.2.2 3.2.3 3.2.4 3.3 3.3.1

Grammatical Literature 102 Manuscripts and the Scope of Old Icelandic Grammatica Grammatical Manuscripts 107 Háttatal: Models, Sources and Rhetoric 110 Authorship of the Commentary 114 Building Authority and Creating Order 118 Snorri and Horace 121 Snorri and the Grammarians 125 Litla Skálda, Gylfaginning and Skáldskaparmál 129 A and B 130

63

103

VIII

Contents

3.3.2 3.3.3 3.3.4 3.4 3.4.1 3.4.2 3.4.3 3.4.4 3.5 3.6 3.6.1 3.6.2 3.6.3 3.7

Character of Litla Skálda and Skáldskaparmál 133 Intertextual Connections 140 Litla Skálda, Snorri and the Traditionalizing Position 144 Snorri’s Reception of Poetry: Skáldskaparmál 147 Boðn 150 Óðrerir 156 Kvasir 158 Kenning Formation and Narrative Elements 159 Snorri’s Reception of Poetry: Gylfaginning 164 The First, Third and Fifth Grammatical Treatises 175 The First Grammatical Treatise 175 The Third Grammatical Treatise 178 The Fifth Grammatical Treatise 188 Summary 192

4 4.1 4.1.1 4.1.2 4.1.3 4.2 4.3 4.4 4.5 4.6 4.7 4.8

Prosimetrical Narrative 194 Age and Development of the Prosimetrical Form 194 Prose, Poetry and Prosimetra up to the Mid-Thirteenth Century Reykjahólar 1119 201 A Local Background Nonetheless? 205 Corpus and Method 212 Metrical Archaization in Egils saga 219 The Authorship of Egils saga 233 Metrical Archaization in Other Sagas 244 Archaizing Diction 252 Pseudo-Grettir 264 Summary: The Image of the Poet 274

5 5.1 5.2 5.3

Poetry, Language and Snorri’s Edda in the Mid-Fourteenth Century The Wormianus Redactor 279 Monastic Poets and Snorri’s Edda 290 Summary 296

6

Conclusions

Appendix

303

Bibliography Index

345

321

297

195

277

A Note on Translations Unless otherwise stated, translations are my own. I have opted to translate Old Norse and Latin text myself in order to calibrate the level of exactness in accordance with the requirements of the argument and to fully account for my interpretation of the text at hand, as well as to achieve stylistic coherence. Unlike the new edition of skaldic poetry, Skaldic Poetry of the Scandinavian Middle Ages, ed. Margaret Clunies Ross, Kari Ellen Gade, Diana Whaley et al. (Turnhout: Brepols, 2007–), I translate long kennings with genitival ’s rather than of-constructions, although the former is often less idiomatic, since the genitival ’s-solution makes it possible to proceed in the direction of reading. Thus, for instance, ‘the spears’ path’s [shield’s] noise’s [battle’s] Njǫrðr (name of a god) [warrior]’ rather than ‘Njǫrðr of the noise of the path of spears [shield>battle>man]’, which requires reading both forwards and backwards.

https://doi.org/10.1515/9783110643930-203

Introduction When a river has several tributaries, it sometimes happens that the waters from different valleys flow side by side for long stretches, clearly distinguished by their colour, apparently without mixing. Old Norse literature is reminiscent of such a river. It has two sources. One of these is the ecclesiastical and literary life among the races of the West and South, with whom there were close contacts; the other is the spiritual life of the people itself, where for the longest time impressions of experiences at home and away, memories and reflections had come together and taken shape in poetry, in narrative, in legal and customary regulations. And the two streams run side by side, not without mutual influence, yet each with its own peculiar characteristics, almost from the beginnings of written literature until it dries up and disappears. Marius Nygaard, ‘Den lærde stil i den norrøne prosa’, 1896.1

Drawing on the image of a glacial tributary entering a river, the philologist Marius Nygaard with these words expressed his views on the essential difference between Latin and Old Norse culture and literature; they run the same course, but they never really blend. Today, we see things otherwise. The sharp dichotomy between European Latin and national vernacular literature has been largely rejected. The rivers that we project onto the past have many tributaries, but none of them is glacial and their waters are like those of any river. And yet it remains a common sentiment that there really is something about, for instance, Icelandic sagas which sets them apart from their larger European context. If their distinctness is not due to the local and pristine nature of Old Norse literature, how shall we explain it? This book sets out to address that question through an analysis of the complex relationship between Latin and the vernacular in the golden age of Old Icelandic literature c. 1150–1350. The analysis presupposes that the difference which Nygaard perceived is real in the sense that portions of Old Icelandic literature differ markedly from what one may find in Latin literature and in all or most vernaculars. One may for instance point to the remarkable blend of poetic handbook and mythological overview that is Snorri’s Edda (c. 1220–1240). In its combination of these two features, this work has no real counterparts in either Latin or other vernaculars (though some Latin texts come pretty close; see below). In another regard, however, it is deeply embedded in its wider European context. Literary writing arrived with Latin learning, and all Old Icelandic literature, including such works as Snorri’s Edda, is predicated on this background. Although the question-and-answer format of Snorri’s Edda may be inspired by some of the poems on which he drew, he developed this form into a complex treatise which could hardly have been accommodated within the oral context of previous centuries. Unique and native, yet also distinctly European and learned, Snorri’s Edda is a prime example of the

1 Marius Nygaard, ‘Den lærde stil i den norrøne prosa’, in Sproglig-historiske studier tillegnede professor C. R. Unger (Kristiania: Aschehoug, 1896), p. 153. https://doi.org/10.1515/9783110643930-001

2

Introduction

mechanisms described in this book. Snorri’s Edda centres completely on poetry, some of which dates to the period before the arrival of Latin learning and the rest of which staunchly adheres to conventions established during that early period. Thus, even though the analytical methods used by Snorri and others were indebted to monastic learning, these necessarily had to be adapted to an object of study that had never previously been accommodated within the Latinate tradition, and the unique features of Snorri’s Edda are a result of the challenges involved. The main argument of this book is that much of what sets Icelandic literature apart is predicated on the dynamics of the meeting of local poetic tradition and Latin learned culture, a meeting which took place in the period under study (c. 1150–1350). This is most obviously true of a work like Snorri’s Edda, whose principal focus is poetry and which has no real counterparts within Latin Christendom. In varying degrees, however, the same observation is valid for considerable portions of Icelandic literature – not only for other treatises on poetry and language, but also for the curious blend of prose and poetry which is found in many sagas and which in crucial ways differs from Latin examples of such a mixed form. In many other cases, however, Old Icelandic literature features more or less the characteristics that one would expect from a Latin or a wider European perspective, and such genres will not be treated in this book. Most obviously, the survey exludes translated literature, if translated into prose only, since the content of these texts remained largely true to its background (thus, for instance, homilies, hagiography, Arthurian romance). Although such texts have undergone some adaptation to local conventions, their background is clear and does not call for an explanation except with regard to details of transmission. The study also excludes texts derived from translated prose genres, such as the fornsögur suðurlanda – the local variety of Arthurian literature. The words Old Icelandic in the title could thus be further specified to read: distinctively Old Icelandic with regard to more than linguistic form. Hence there is much that will fall outside the scope of this book, but what it covers are the works and genres which have since made Old Icelandic literature famous, most notably the mythological works and the most well-known branches of saga literature. Even with this limitation, a focus on poetry cannot come anywhere close to explaining the totality of distinctive features within these genres, and such an aim would hardly be realistic. What a poetic perspective can do, however, is to uncover factors without which these genres would not have developed into the form they have. In cases such as Snorri’s Edda, they would not have developed at all. This is what is meant by the word genesis in the title, which could thus be expanded to read crucial factors for the genesis of the forms we now recognize, to which may again be added of a literature that is distinctively Old Icelandic with regard to more than linguistic form. This, then, is the precise meaning of the title The Poetic Genesis of Old Icelandic Literature. In addition, it will be noted that the title reads Icelandic rather than Norse or Norse-Icelandic. The latter two terms include Norway and the

Introduction

3

North Sea dominions, such as Orkney and Greenland. While other regions than Iceland were important for the developments described in this book, the developments themselves were incomparably more prominent in Iceland than elsewhere. Almost all of the sagas were written in Iceland or at least by Icelanders, and this is true also of the Eddas and the grammatical treatises. Other areas, notably Norway and Orkney, will at times enter the discussion, but there can be little doubt that if it were not for the Icelandic developments, Old Norse would be a field of study of which few would had heard today. The wording of the title is intended to acknowledge this fact. Nygaard’s metaphor which opens this introduction – that of a river with two separate streams – presents Latin and Old Norse as two literatures which remained distinct throughout their history of mutual contact. To illustrate the argument put forward in this book, the analogy of Norwegian stave churches is more apt. These wooden structures are built in imitation of medieval stone churches with which we are all familiar; their pillars, capitals, portals and ground plan bear witness to this. Nonetheless, the building material remains the traditional one of Norwegian building, from farmhouse to royal hall – namely wood – and the qualities of wood condition the makeup of the entire structure, from the carving and decoration to the distribution of weight and the connection of beams. As a result, these churches look like nothing else from medieval Europe, and if a visitor does not already know what to look for, Romanesque architecture is unlikely to spring to mind. In the genres of Old Icelandic literature that are best known today, the local building material was not wood, but poetry. The arrival of Latin learning in Iceland produced a range of vernacular responses. Translation and literary calques are the most obvious ones, but it would also appear that a wish to have a historiography and a literature that was distinctly one’s own arose early on. Latin models were important for this undertaking, but they provided no guidance on how local distinctness should be added to the mix. For this purpose, Icelandic would-be authors turned to their closest counterpart to the Latin intellectual tradition, namely skaldic poetry. They began to study it in a new and systematic way, and its elements – metre, diction, imagery and mythology – became the material out of which they shaped their own, unmistakeably Icelandic literature. In their own way, local poetry and poetics provided many of the elements found in the Latin tradition, but the material was different. The metres, the modes of reference, and the images were no more similar to their Latin counterparts than stone is to wood. The resulting edifice could not but leave a Latin beholder with a feeling of alienation. This metaphor differs from that of Nygaard in two ways. First, it does not posit two distinct literatures, blending only at their margins but running pure in their respective courses. Rather, the metaphor suggests that they did indeed blend, and that one could not have become what it did without the other. Second, the metaphor takes into account why they can nonetheless appear so different, since the characteristics of wood – or local poetry – were at variance with the material of the

4

Introduction

Borgund stave church.

model and called for other solutions. But it is time now to leave metaphors behind and to turn to Old Icelandic poetry: to what it was and to why it could have such a transformative effect on external impulses. From Classical Antiquity through the Middle Ages, the study of literary form largely centred on poetry, and Old Icelandic literature was no exception. If anything, Icelanders were more obsessed with poetry than most of their contemporaries, as attested by the Norwegian historian Theodoricus (c. 1180): [. . .] prout sagaciter perquirere potuimus ab eis, penes quos horum memoria præcipue vigere creditur, quos nos Islendinga vocamus, qui hæc in suis antiquis carminibus percelebrata recolunt.2 [. . .] to investigate as thoroughly as we could with the help of those whom we call Icelanders and among whom the memory of these things [the history of Norway] is thought to flourish in particular. They reflect upon these deeds, which are celebrated in their ancient poems.

2 Monumenta historica Norvegiæ. Latinske kildeskrifter til Norges historie i middelalderen, ed. Gustav Storm (Kristiania: A. W. Brøgger, 1880), p. 3.

Introduction

5

Here Theodoricus stresses that poetry was the medium through which Icelanders reflected upon their and their neighbours’ history. Contemporary Danish historians issued similar statements.3 We shall soon see more and stronger indications to the same effect from within the Icelandic tradition itself, but for now it may suffice to note that the emphasis on poetry was very strong among the Icelandic elite. This tradition was susceptible to external influences and adapted to changes in literary taste, but several fundamental features remained stable and these features formed their literary surroundings at least as much as they were formed by them. This legacy was transmitted within two different branches of poetry. These branches were mostly kept apart within the literary corpus, even if the strict generic division between them is a modern scholarly convention.4 The most prestigious and complex kind of poetry is today referred to as skaldic poetry, in contradistinction to the relatively straightforward eddic type. Skaldic poetry was a strongly individualistic art where the name of the poet was usually transmitted with his or her poetry. It featured very complex diction and rigid metre, including a fixed number of metrical positions, alliteration and rhyme. Eddic poetry, by contrast, was generally anonymous, had much simpler diction, variable line-length and alliteration only.5 The analytical discourses about poetry mainly evolved around skaldic poetry, which will be the primary locus of attention in this book, but eddic poetry also influenced the ways that external impulses were transformed in the presence of local tradition. Thus, for instance, and I shall argue that eddic poetry was of crucial importance for the emergence of the typical saga mixture of prose and verse – socalled prosimetrum. This form is also found in Latin literature and it saw a great rise there in the twelfth century, just when the saga style began to develop. The chronological coincidence of Latin and only decades later Icelandic historiographic prosimetra cannot be overlooked, and Theodoricus, who referred to his Icelandic informants in the quotation given above, cites classical poetry eleven times in his short history of Norway. We can thus point to at least one verifiable case of contact between a historian who adhered to the typical twelfth-century mode of Latin historiography and what was presumably members of the Icelandic elite. It is thus plausible that the Latin tradition suggested this form to Icelandic authors. It will be argued, however, that the function of quotation within skaldic prosimetra – most

3 So Sven Aggesen (c. 1185; Scriptores minores historiæ Danicæ, 1, ed. M. Cl. Gertz (København: Gad, 1917), pp. 96–97), and Saxo Grammaticus (early thirteenth century; Saxo Grammaticus, Gesta Danorum: Danmarkshistorien, ed. Karsten Friis-Jensen and trans. Peter Zeeberg, 2 vols (København: Gads forlag, 2005), 1, pp. 74–77). 4 See, for instance, Margaret Clunies Ross, A history of Old Norse Poetry and Poetics (Cambridge: D. S. Brewer, 2005), pp. 10–11. 5 See Margaret Clunies Ross, A history of Old Norse Poetry and Poetics, pp. 21–28; Martin Chase, ‘Introduction’, in Eddic, Skaldic, and Beyond. Poetic Variety in Medieval Iceland and Norway, ed. Martin Chase (New York: Fordham University Press, 2014), pp. 1–7.

6

Introduction

notably kings’ sagas, sagas of Icelanders and contemporary sagas – was predicated not on Latin, but on previous conventions for quoting eddic poetry, and that the resulting historiographic tradition therefore became decidedly different from any Latin counterpart. In most cases, however, the formative impact came not from eddic, but from skaldic poetry. Again, Snorri’s Edda is an instructive case, but in order to discuss the importance of skaldic poetry for the conception of this treatise, I begin with a few words on the extent to which it is, indeed, unique. In its function as a mythological overview, Snorri’s Edda is not without parallels within Latin literature. The most relevant work for comparison is the so-called Third Vatican Mythographer. Its author has been identified as a magister Albericus lundoniensis (master Alberic of London), who composed this treatise in the decades around Snorri’s birth (1179).6 It is found in at least 43 manuscripts across Europe.7 Other examples could be adduced, but the work of master Alberic is probably the closest parallel to Snorri’s Edda among such Latin texts as enjoyed wide circulation in Snorri’s time. Furthermore, while Alberic’s treatise is variously titled Mythologicon, De idolis et astris (On Idols and Stars) or Liber ymaginum deorum (Book on the Images of the Gods) in manuscripts, Helinand in the beginning of the thirteenth century refers to it as Poetarius (Poetary), and fourteenth-century writers used the same name or called it Scintillarium poetarum (Florilegium [lit. ‘collection of sparks’] of the Poets).8 Both the titles and the work itself suggest that it is a mythological overview, but they do not clarify what function such a treatise would serve. The references to the Poetarius or Scintillarium poetarum, however, tell us that the treatise was used in ways more akin to those of Snorri’s Edda than one might otherwise have thought, namely as a guide to the work of the poets. Internally, however, the texts remain fundamentally different. Snorri’s Edda primarily treats poetry and gives mythology as background, whereas Alberic’s work treats mythology and mentions poetry only by way of reference. There would have

6 The attribution was made by Eleanor Rathbone, ‘Master Alberic of London, “Mythographus tertius vaticanus”’, Medieval and Renaissance Studies, 1 (1943), pp. 35–38. Charles S. F. Burnett, ‘A Note on the Origins of the Third Vatican Mythographer’, Journal of the Warburg and Courtauld Institutes, 44 (1981), pp. 160–66, has suggested that the background of the treatise may be more complex than scholars have realized, but this does not weaken the strength of the several explicit attributions in the sources (pace Burnett, ‘A Note’, p. 163). On Alberic’s possible connections to Chartres, see Henning Sjöström, ‘Magister Albericus lundoniensis, mythographus tertius vaticanus. A Twelfth Century Student of Classical Mythology’, Classica et Mediaevalia, 29 (1968), pp. 249–64. 7 Kathleen O. Elliott and J. P. Elder, ‘A Critical Edition of the Vatican Mythographers’, Transactions and Proceeding of the American Philological Association, 78 (1947), pp. 205–07. 8 Scriptores rerum mythicarum latini Romae nuper reperti, ed. Georgius Henricus Bode (Cellis: 1834), p. 110; Jean Seznec, The Survival of the Pagan Gods. The Mythological Tradition and its Place in Renaissance Humanism and Art (New York: Bollingen, 1953 [1940]), pp. 170–73; for the reference in Helinand, see Rathbone, ‘Master Alberic of London’, p. 36.

Introduction

7

been no certain way of telling that Alberic’s book was received as a poetic handbook if it had not been for later references to it, even if an educated guess might have suggested as much. For learning about matters pertaining to poetry that do not involve mythology, the reader would have to turn to other texts. In Snorri’s Edda, by contrast, all aspects of poetry that Snorri deemed relevant are contained within one and the same work. If Snorri’s Edda thus had partial functional counterparts in Latin, its mythological overview is completely unmatched in other European vernaculars. Mythography served as a tool for understanding classical texts, and to that end, the reader needed only to be acquainted with classical mythology. The vernaculars did not have a corpus of pre-Christian classics that set the standard for all later literature and consequently had no need for treatises of this nature. Remnants of other mythologies crop up here and there in Europe, most notably in historiographical contexts, but no systematic overviews were ever written.9 Only in Iceland did traditionalizing and mythologizing diction hold sway to such a degree that the need for a work of this nature arose. More specifically, it was skaldic poetry which required a full command of the mythology, since its oblique references could often not be understood without it. In essence, then, Snorri’s Edda is a handbook of poetics which for practical reasons became one also of mythology. Its focus lies on the intricacies of skaldic poetry, which were many and complex enough to warrant a fullscale prescriptive analysis. Eddic poetry, by contrast, is relatively simple and did not call for a scholarly exposition. Rather, Snorri used eddic poetry in the first part of the work – Gylfaginning – as part of the narrative background which is meant to explain the recondite references of skaldic poetry, as well as to authenticate his own description of the mythology. In Snorri’s Edda, eddic poetry is thus subservient to the main, skaldic focus. Since Snorri’s Edda is primarily a handbook of poetics, it remains somewhat uncertain whether its conception was at all influenced by the mythological revival of twelfth- and thirteenth-century Europe in general and by works like that of Alberic in particular.10 While this is not in itself unlikely, the primary focus on native poetics and the mythology that came with it meant that Snorri’s Edda

9 Irish literature has a relatively rich body of texts where pre-Christian elements are found (thus, for instance, the Lebor Gabála (book of invasions) and Táin Bó Cúailnge (cattle-raid of Cooley)). The Eastern Slavic Primary Chronicle contains some references to pre-Christian beliefs, and some nonSlavic chroniclers, notably Saxo Grammaticus, provide biased accounts of Western Slavic rites and beliefs. English scholars such as Bede, Daniel of Winchester, Ælfric and others describe Germanic paganism in a manner that is indebted to hagiographic conventions. Medieval comprehensive treatises on pre-Christian mythology, however, exist only for classical and Old Norse mythology. 10 On the high medieval mythological revival, see, for instance, Frank T. Coulson, ‘Ovid’s Metamophoses in the School Tradition of France, 1180–1400: Texts, Manuscript Traditions, Manuscript Settings’, in Ovid in the Middle Ages, ed. James G. Clark, Frank T. Coulson and Kathrym L. McKinley (Cambridge: Cambrdige University Press, 2011), pp. 48–82; Peter Dronke,

8

Introduction

could hardly have turned out as anything but distinctly un-Latin, even if Snorri had tried to conform to that tradition. In the event, he chose the opposite route, and his Edda is almost void even of Latin loan words.11 Under these circumstances it is very difficult to assess the plausibility of influence from Latin mythography in anything but the most general terms. What may be regarded as certain, however, is that the notion of composing a comprehensive and systematic treatise on poetry and language was derived from Latin literature. Yet the result of this impulse was a treatise which in its combination of mythology and poetics lacks any real counterparts either in Latin or in other vernaculars. The power of local poetic traditions to transform external impulses into something new and different is thus abundantly clear, even if the precise origin of these impulses within the larger setting of Latin learning remains somewhat uncertain. Describing Snorri’s Edda as a poetic handbook does justice to its contents, but if we attempt to account for its impact with reference to modern literary priorities, where prose occupies the centre stage, we shall be much mistaken. Icelanders had a keen interest in literary composition, style and delivery – what we might refer to as rhetoric, but which the Middle Ages often categorized as part of grammatica.12 Such matters were discussed not only in grammatical texts like Snorri’s Edda, but also in the sagas and even in the laws. In contrast to both Antiquity and Modernity, however, this entire discourse was centred on poetics. Snorri’s Edda, then, was an authoritative and comprehensive handbook on the intellectual discipline that attracted Icelanders the most. The terse and easily recognizable style of Old Icelandic prose, by contrast, was never a topic of discussion and has only received acclamation in retrospect. The Icelandic literary discourse was poetocentric, and other textual entities revolved around this middle point.

Fabula. Explorations into the Uses of Myth in Medieval Platonism (Leiden: Brill, 1974). In Italy, interest in mythology is mainly evident from Dante onwards. 11 Anthony Faulkes, ‘The Sources of Skáldskaparmál: Snorri’s Intellectual background’, in Snorri Sturluson. Kolloquium anläßlich der 750. Wiederkehr seines Todestages, ed. Alois Wolf (Tübingen: Gunter Narr Verlag, 1993), p. 73. The only real exception is the double set of names in the Troy story. These seem to be connected to the notion that linguistic confusion gave rise to paganism, and to the claim that twelve languages were spoken in Troy. See Snorri Sturluson. Edda. Prologue and Gylfaginning, ed. Anthony Faulkes, 2nd ed. (London: Viking Society for Northern Research, 2005), pp. 4, 6, 13, 55; Snorri Sturluson, Edda. Skáldskaparmál, ed. Anthony Faulkes (London: Viking Society for Northern Research, 1998), pp. 5–6. On linguistic confusion and paganism see below pp.121–24. Note that the claim that there were twelve languages in Troy is attested in all manuscripts, but has been removed in Edda Snorra Sturlusonar, ed. Finnur Jónsson (København: Komissionen for det arnamagnæanske legat, 1931), p. 4. Faulkes claims that there are also almost no linguistic calques in Snorri’s Edda, but this statement should be taken with some caution, since these may be difficult to detect: see, for instance, the discussion of nykrat on pp. 121–24 below. 12 Old Icelandic literature has only one reference to rhetorica, discussed below on p. 181 n. 283.

Introduction

9

Authoring a grammatical work like Snorri’s Edda can be understood as translation in a broad sense of the word, since grammatical studies were of Latin derivation and were brought to the North by the church. With Snorri’s Edda, however, grammatica was subjected to vernacular needs in the strongest possible way. In Rita Copeland’s analysis of Chaucer and Gower’s appropriation of the classical tradition, which turned translation and exegetical practices into a form of rhetorical invention which she terms ‘secondary translation’, we encounter vernacular authors whose dialogue with their sources remains visible.13 Indeed, the very visibility of the dialogue draws attention to the act of appropriation. In Snorri’s Edda and many other Icelandic works, by contrast, the dialogue with Latin literature is completely suppressed. As Anthony Faulkes has remarked about Snorri’s Edda, ‘his [Snorri’s] manner and method seem to me to be totally Icelandic’.14 We shall see in Chapter Three that Snorri appears to have aimed for just such an independent vernacular expression, even when he was actually adding Biblical and classical material to the mix. There is something strange about this picture. In Iceland, one may discern a process of increasing vernacular independence or, in Copeland’s terms, a move from primary to secondary translation. Thus, for instance, translated romances over time gave rise to independent compositions, and hagiographic models were successively adapted to the exigencies of national history.15 At other times, however, this development seems to be turned on its head, and Snorri’s Edda is a case in point. Snorri probably composed the bulk of his Edda in the 1220s, and only in the next generation, with Óláfr Þórðarson’s translation of Donatus’s Barbarismus (c. 1250), do we see the emergence of a dialogue between skaldic and classical poetics – assertive, to be sure, but apparently displaying less vernacular independence than Snorri’s Edda. It is thus easy to see how Faulkes arrived at the conclusion that the character of Snorri’s Edda was fundamentally Icelandic, as yet largely untouched by Latin learning. Faulkes writes: ‘I think it hardly possible that Snorri could have read writers

13 Rita Copeland, Rhetoric, Hermeneutics, and Translation in the Middle Ages. Academic Traditions and Vernacular Texts (Cambridge: Cambridge University Press, 1991), pp. 179–220. 14 Faulkes, ‘The Sources of Skáldskaparmál’, p. 75. 15 The chivalric sagas of original composition are collectively called fornaldarsögur suðurlanda in modern research, and they constitute a rich body of texts. For possible hagiographic influences on Icelandic historiography, see Gabriel Turville-Petre, Origins of Icelandic Literature (Oxford: Clarendon Press, 1953), pp. 109–42; Jónas Kristjánsson, ‘Learned Style or Saga Style?’, in Speculum Norroenum: Norse Studies in Memory of Gabriel Turville-Petre, ed. Ursula Dronke et al. (Odense: Odense University Press, 1981), 260–92; Peter Foote, ‘Saints’ Lives and Sagas’, in Saints and sagas. A Symposium, ed. Hans Bekker-Nielsen and Birte Carlé (Odense: Odense Universitetsforlag, 1994), pp. 5–26; Carl Phelpstead, Holy Vikings. Saints’ Lives in the Icelandic Kings’ Sagas (Tempe: ACMRS, 2007); Sîan Grønlie, ‘Saint’s Life and Saga Narrative’, Saga-Book, 36 (2012), pp. 5–26; eadem, The Saint and the Saga Hero. Hagiography and Early Icelandic Literature (Woodbridge: Boydell and Brewer, 2017); Sverre Bagge, ‘The Old Norse Kings’ Sagas and European Latin Historiography’, Journal of English and Germanic Philology, 115 (2016), pp. 1–38.

10

Introduction

like Peter Comestor, Honorius Augustodunensis and Guillaume de Conches, extracted material from them to use in his writings, and remained totally untouched by them in other ways (stylistically, in attitudes and so on)’.16 This claim is problematic, since Snorri was programmatic not only in his avoidance of loan words, but also in rhetorically suppressing the Biblical parallels that he himself had planted in the text, so as to introduce the reader to a world of undiluted paganism (see below pp. 169–74). His vision was one of a sealed mythological and poetic cosmos of complete Icelandicness or Norseness, but this need not imply that he could not have done differently. Indeed, a work by one of the authors mentioned, Honorius’s Elucidarium, or Elucidarius as it is known in Old Norse scholarship, was translated into Icelandic during Snorri’s childhood if not earlier, and the translator opted for a pure Norse style – surely not out of ignorance of the original.17 Some verbal similarities between Elucidarius and Snorri’s Edda may suggest direct influence.18 The extent of Snorri’s Latin learning may not have been vast, but since he remained surrounded by some of Iceland’s most learned clerics from the age of two and throughout his life, it seems simplistic to ascribe the form of his Edda to a lack of options.19

16 Faulkes, ‘The Sources of Skáldskaparmál’, p. 75. 17 Consider the following translations which, although they are mainly calques, make use of native elements only: skilningarhiminn (heaven of understanding; intellectuale caelum), hugrenningar skrímsl (ghosts of thought; imagines phantasmatum), fyrirætlan (intention; praedestinatio), hǫfuðskepnur (main forms; elementa), andlegr hórdómr (spiritual fornication; spiritualis fornicatio), í áblæstri sínum (in his on-blowing; inspirando) (see Gabriel Turville-Petre, Origins of Icelandic Literature (London: Oxford University Press, 1953), p. 140). A comprehensive list of such translations in Elucidarius is given in The Arnamagnæan Manuscript 674A, 4to. ‘Elucidarius’, ed. facs. Jón Helgason (Copenhagen: Munksgaard, 1957), pp. xxix–xxxi. On the date of the translation, see ibid., pp. ix–xix. On the question of the authorship of the Latin text, see Yves LeFèvre, L’ ‘Elucidarium’ et les Lucidaires. Contribution, par l’histoire d’un texte, à l’histoire des croyances religieuses en France au moyen âge. Bibliothèque des Écoles Françaises d’Athènes et de Rome, 180 (Paris: Boccard, 1954), pp. 209–19. 18 The Arnamagnæan Manuscript 674A, 4to. ‘Elucidarius’, ed. facs. Jón Helgason, p. xxxi; Margaret Clunies Ross, ‘Skáldskaparmál’. Snorri Sturluson’s ‘Ars Poetica’ and Medieval Theories of Language (Odense: Odense University Press, 1987), pp. 55–58. 19 Helgi Þorláksson, ‘Snorri í Odda. Um menntun Snorra Sturlusonar, uppeldi og mótun’, Skírnir (2014), pp. 353–80, contends that Snorri would at most have received training as an acolyte. On the other end of the scale in the debate on Snorri’s level of learning, Matthias Teichert presupposes that he had first-hand acquaintance with much Latin literature, including the classics (‘Das Goldzeitalder: Latinität und Fragmente eines griechisch-römischen Mythos in der Gylfaginning’, in Snorri Sturluson – Historiker, Dichter, Politiker, ed. Heinrich Beck, Wilhelm Heizmann, and Jan Alexander van Nahl (Berlin: de Gruyter, 2013), pp. 129–50; similarly Sverrir Tómasson, ‘“Upp í garð til Sæmundar”. Lærdómssetrið í Odda og Snorri Sturluson’, Goðasteinn, 33 (1997), p. 201). For my argument it is sufficient to note, as will be done repeatedly in this book, that Snorri positioned himself in relation to Latin learning. Whether he had access to it through reading or informants is here of limited interest.

Introduction

11

I believe, therefore, that the appearance of Snorri’s Edda must be explained otherwise. In the generations before Snorri, Latin learning was naturalized among poets who tended to be priest and deacons and who were, in addition, often chieftains in their own right. As we shall see in Chapters One and Two, poetics had begun to take on a learned and systematic character in the hands of these men, but it remained committed to skaldic diction and metre and was, to that extent, traditional. In consequence, if one were to write a work on skaldic poetry, the differences between Norse and classical poetics would have made a direct transfer of classical stylistics a problematic point of departure. Rather, to be useful, such a treatise would have to diverge considerably from what Latin literature had to offer, and we shall see that Snorri’s was not the first attempt in this direction (pp. 133–47). Only in the generation after Snorri, when a useful handbook on skaldic poetry on its own terms existed, did Snorri’s nephew, well aware of what had already been done, set out to address the somewhat less pressing issue of adapting Donatus to skaldic needs. The general progression of vernacular assertiveness has here been only seemingly upset. At a glance, one can easily feel that Icelandic literature developed from more to less self-confidence as time progressed. A closer look, however, reveals that the genres that are most different from their European analogues are also the ones that contain most poetry and that focus most on the poetry contained within them. This is notably the case with poetic treatises and compendia, as well as prosimetrical saga genres such as kings’ sagas, sagas of Icelanders and legendary sagas. Any modification of Copeland’s analysis to local circumstances, then, has everything to do with poetry and with the choice to cling to Norse poetry’s longestablished forms. In other regards, vernacular literature did indeed develop a gradual confidence that allowed for original composition in genres of foreign extraction, and secondary translation and appropriation became increasingly more common as time went by. The focus of this book, however, will be on the parts of Old Icelandic literature where poetic conventions interfered with that process and where the results were least predictable from a European and Latin point of view. The demands of the local poetic tradition ensured that in poetic treatises and prosimetrical sagas, external influences underwent a metamorphosis on a scale that cannot be compared with Icelandic adaptation of other branches of Latin or vernacular literature. While some of the characteristics of this corpus are unique to Iceland, the impact of poetry on the development of vernacular literature is evident in many regions of medieval Europe. Irish and Occitan literature offer some parallels, but Italy and Dante provide the readiest comparison. In the opening of his De vulgari eloquentia (On Vernacular Eloquence), Dante produces various arguments to prove that the vernacular is more noble than the grammatical languages (Latin and Greek): first, because the vernacular was the original language of Eden; second, because everyone uses it; and third, because it is natural, whereas Latin in Dante’s

12

Introduction

view was a constructed language.20 Dante quickly abandons this line of reasoning, however, in favour of a vision of the vernacular which is based on poetic practice, and contrary to his initial argument, he bestows particular praise on poets who refrain from using their own natural language. The foundation on which De vulgari eloquentia stands is thus not so much that of Biblical history as that of the authority of the poets. In the opening of book two, Dante explicitly states that poets should be allowed to set the agenda, and throughout the treatise, he lets them do so.21 In both Snorri’s Edda and De vulgari eloquentia, vernacular assertiveness is founded on a poetic tradition which had taken on a semi-learned character from the twelfth century on (in Dante’s case, Occitan and Italian poetry). This tradition provided the authors with the authority they needed to devote a treatise to vernacular poetry, or to the vernacular via poetry. In both cases, their activities were connected to an extraordinary outburst of vernacular creativity, and preceding poetic developments were instrumental to that outcome. All their differences notwithstanding, the cultural contexts of Snorri and Dante are comparable on several points, and while their literary skills were out of the ordinary, a focus on their individual geniuses leaves our understanding of them incomplete. Their shared traits also depend on their respective cultural settings. By contextualizing Snorri in relation to texts that were used or produced in Iceland in the period in Chapters Three and Four, I hope to shed new light on the literary preconditions of Snorri and his fellow historians, grammarians and poets. Many sources attest to the importance of poetry in Icelandic society, and thus to how it could become such a powerful vector of change. No single text, perhaps, illustrates this better than the seemingly absurd regulation of poetic composition given in the law of the Icelandic free-state, the so-called Grágás (grey goose). There we read (following the main manuscript Konungsbók): Hvártki á maðr at yrkja um mann lǫst né lof. Skalat maðr reiðask við fjórðungi vísu nema lastmæli sé í. Ef maðr yrkir tvau orð, en annarr ǫnnur tvau, ef þeir ráða báðir um, ok varðar skóggang ef lǫstr er í eða háðung. Ef maðr yrkir þá vísu um mann er eigi er háðung í, ok varðar þriggja marka sekð. Ef hann yrkir fleira um mann, ok varðar fjǫrbaugsgarð, þótt eigi sé háðung í. Skóggang varðar ef maðr yrkir um mann hálfa vísu, þá er lǫstr er í eða háðung, eða lof þat er hann yrkir til háðungar. Ef hann kveðr þat eða kennir ǫðrum manni, ok er þat ǫnnur sǫk ok varðar skóggang. Svá varðar ok hverjum er nemr. [. . .] Ef maðr yrkir mansǫng um konu, ok varðar skóggang. Kona á sǫk ef hun er tvítug eða eldri. Ef hon vill eigi sœkja láta, ok á lǫgráðandi hennar sǫkina. Ef maðr kveðr skáldskap til háðungar manni, þótt um annan sé ort, eða snýr maðr á hǫnd honum nokkuru orði í, ok varðar þat skóggang. Skal sœkja sem annan skáldskap. [. . .] Ef maðr yrkir víðáttuskáldskap, ok á maðr kost at dragast undir sá er vill ok

20 Dante Alighieri, Opere, 1: Rime, Vita nova, De vulgari eloquentia, ed. and trans. Claudio Giunta, Guglielmo Gorni, Mirko Tavoni (Milano: Mondadori, 2011), p. 1136. On Dante, language, and vernacular literature, see Albert Russell Ascoli, Dante and the Making of a Modern Author (Cambridge: Cambridge University Press, 2008), pp. 130–74, 388–405. 21 Dante Alighieri, De vulgari eloquentia, ed. and trans. Giunta et al., pp. 1364–68.

Introduction

13

stefna um, þó at kviðr beri at hann hafi eigi ort um þann er sœkir um, en þat beri at hann hafi ort, ok seksk hann þó um víðáttuskáldskap. Þat er víðáttuskáldskapr ef maðr yrkir um engi mann einkum, ok fer þat þó um hérað innan, ok varðar skóggang.22 A man shall not compose either blame or praise about another. A man shall not become angry over a couplet, provided there is no blame in it. If one man composes a couplet, and another one more, and if they agree on doing this, then that results in perpetual exile if there is blame or scorn in it. If a man composes about another and there is no scorn in it, then the fine is three mark. If a man composes more than this, then that results in exile for three years, even if there is no scorn in it. It results in perpetual exile if a man composes half a stanza about another, in which there is blame or scorn, or praise that he composes in order to scorn. If he proclaims it or teaches it to another, then that is a case on its own, and this also results in perpetual exile. This is true also of anyone who learns it. [. . .] If a man composes a love poem about a woman, then that results in perpetual exile. The woman may press charges if she is twenty or older. If she will not press charges, the case belongs to her legal guardian. If a man proclaims poetry to scorn a man, even though it was composed about someone else, or if a man turns it against him by changing some word in it, then that results in perpetual exile. The case shall proceed as with other poetry. [. . .] If a man composes ‘wide poetry’, then whoever so wishes may take the poetry upon himself and open the case, even though witnesses confirm that he did not compose about the one who presses charges, but they do confirm that he composed the poem, then he is condemned for ‘wide poetry’. ‘Wide poetry’ is when a man composes about no one in particular, but the poem still circulates within the county, and it results in perpetual exile.

There is nothing in the text or its context to indicate that it is not intended to be taken seriously.23 Even so, the constant quotation of poetry in the contemporary sagas, by men and women, chieftains and farm-hands alike, indicates that an enforcement of this law would lead to exile for a considerable portion of the Icelandic population.24 The author of the largest of the contemporary sagas, Sturla Þórðarson, would merit perpetual exile for the quotation of a slanderous stanza about his uncle Snorri.25 Nothing of the sort, of course, ever happened.

22 ‘Grágás’. Islændernes Lovbog i Fristatens Tid udgivet efter det Kongelige Bibliotheks Haandskrift og oversat, ed. and trans. Vilhjálmur Finsen, 2 vols (Kjøbenhavn: Berling, 1852–70), 2, pp. 183–85. My normalization. 23 The paragraph is placed differently in the two main witnesses, Konungsbók (GKS 1157 fol, c. 1250; ONP Registre, p. 471) and Staðarhólsbók (AM 334 fol, c. 1250–1300; ONP Registre, p. 441). In Konungsbók, it is placed among miscellaneous chapters before regulations on the tithe, whereas in Staðarhólsbók, it is part of the chapter on mutilation and manslaughter, the so-called Vígslóði (consequences of manslaughter); see ‘Grágás’. Islændernes Lovbog i Fristatens, ed. and trans. Vilhjálmur Finsen, 2, pp. 183–85; ‘Grágás’ efter det Arnamagnæanske Haandskrift, nr. 334 fol., Staðarhólsbók (Kjøbenhavn: Gyldendal, 1879), pp. 392–94. 24 The participation in poetic discourse by broad groups in society has been convincingly argued by Jonathan Grove, ‘Skaldic Verse-Making in Thirteenth-Century Iceland: The Case of the Sauðafellsferðarvísur’, Viking and Medieval Scandinavia, 4 (2008), pp. 85–131. 25 See Sturlunga saga efter membranen Króksfjarðarbók udfyldt efter Reykjarfjarðarbók, ed. Kristian Kålund (København: Gyldendal, 1906–11), pp. 340–41.

14

Introduction

The rationale behind some of the prohibitions is not immediately clear. The phrase ‘eða lof þat er hann yrkir til háðungar’ (or praise that he composes in order to scorn) may suggest why all poetry – and not just slander – should be forbidden, since there was always the possibility that the intent was really to scorn in the guise of praise. The prohibition against love poetry may be predicated on the normal function of panegyrics. When a poet composed a panegyric, he could expect a shield or a sword in return. Presumably, the payment he expected of a woman would be of a more erotic nature, and one of which her husband or father might not approve.26 Even if the regulations on poetry were largely impracticable, they nonetheless bear witness to the centrality of poetry in Icelandic society. The line between love and hate is a thin one, and if the literature at large shows us just how much Icelanders enjoyed the intellectual stimuli of multiple meanings, contorted syntax and phonetic correspondences, Grágás reveals their anxiety about assaults on their honour by the very same means.27 A fundamental premise for this book is that the cultural primacy of the poetic discourse, as expressed in Grágás and elsewhere, invested Icelandic poets with the power to affect the course of literary developments. Further, I argue that when their technical skill was bolstered by the analytical tools of Latin schooling, the poets had not only the mandate, but also the means by which to bring about change and to create a space that was recognizably their own, but which was nonetheless situated within the larger sphere of Latin-derived, learned literacy. Because of the strength of the poetic tradition, they retained the old forms and adapted new learning to them. Much of what appears distinct about parts of Old Icelandic literature can thus be attributed to the blend of two components, one of which remained largely invisible, whereas the other completely dominates the view. Invisible are the methods which

26 For a related interpretation, albeit with greater emphasis on mansǫngr as an affront to male relatives of the woman, see Jenny Jochens, ‘Mansǫngr and the Jómsvíkingar’, in Le secret d’Odin. Mélanges offerts à Régis Boyer, ed. Marc Auchet (Nancy: Presses universitaires de Nancy, 2001), pp. 17–26. On love-poetry (mansǫngr) generally, see Theodor Möbius, ‘Vom isländischen Mansöngr’, Zeitschrift für deutsche Philologie. Ergänzungsband (1874), pp. 42–61; Anne Holtsmark, ‘Kjærlighetsdiktning’, in Kulturhistoriskt lexikon för nordisk medeltid, 22 vols (Malmö: Allhems förlag, 1956–78), 8, cols 438–44; Bjarni Einarsson, Skáldasögur. Um uppruna og eðli ástaskáldasagnanna fornu (Reykjavík: Bókaútgáfa menningarsjóðs, 1961); Theodore Andersson, ‘Skalds and Troubadours’, Mediaeval Scandinavia, 2 (1969), pp. 7–41; Alison Finlay, ‘Skalds, Troubadours and Sagas’, Saga-Book, 24 (1995), pp. 105–53); Edith Marold, ‘Mansǫngr – a Phantom Genre’, in Learning and Understanding in the Old Norse World. Essays in Honour of Margaret Clunies Ross, ed. Judy Quinn, Kate Heslop, and Tarrin Wills (Turnhout: Brepols, 2007), pp. 239–62; Jonas Wellendorf, ‘No Need for Mead. Bjarni Kolbeinsson’s Jómsvíkingadrápa and the Skaldic Tradition’, North-Western European Language Evolution, 69: 2 (2016), pp. 130–54. 27 It should be noted that slander of all kind, and not just in poetic form, was forbidden, but the emphasis on poetry in Grágás is nonetheless remarkably strong. See, for instance, Preben Meulengracht Sørensen, The Unmanly Man. Concepts of Sexual Defamation in Early Northern Society (Odense: Odense University Press, 1983), pp. 29–30, 70–71.

Introduction

15

the authors used, which were often of Latin derivation. The visible component is the poetic form itself and this form, furthermore, presupposed certain kinds of content, such as local mythology. Since the native tradition dictated form and content alike, the Latin element was all but bound to remain invisible. Nonetheless, it was there, and this literature would not have taken shape without it. In the present book, the processes involved are analysed through a progression from the general to the specific. The first two chapters outline the long-term evolution of metre and diction, respectively (c. 850–1350). These developments allow us to see how the poet’s profession gradually became one of systematic studies, modelled on Latin literary analysis. In Chapter One, a breakdown of individual metrical features (line-length, rhyme, etc.) is presented in order to show that metrical innovation remained restricted to a limited number of parameters before the twelfth century. Only then, and among poets who appear to have received some formal schooling, does metrical innovation begin to involve many features. The examination indicates that their method was one of systematic study of the early poets and extraction of rules from the possibilities contained within earlier license. While this analysis has some scholarly precedence, the description of the limited number of parameters in the eleventh century is new and allows for a better understanding of the methods employed in the twelfth century. These methods were used by poets who had no longer simply learned their trade from their elders, but who also actively engaged with the past through close textual analysis. This studied approach to poetry was likely prompted by the treatment of Latin texts in the classroom. Chapter Two, on diction, describes a conscious revival of mythology in the midtwelfth century, and due to a new semantic subdivision of mythological references, this comeback can be shown to have been much more thorough and innovative than has hitherto been assumed. Another point that has previously escaped notice is that other archaic features of diction experienced a corresponding revival, and this elucidates, for one thing, just how much mythology was a matter of diction, and for another, how systematically the archaizers went about their business. These two chapters form the historical background to Chapter Three, on grammatical literature, with particular focus on Snorri’s Edda. In the treatment of that work, the main emphasis is not so much on Snorri’s Edda itself as on its place within the poetic and learned tradition and on Snorri’s use of his sources. The preceding chapter shows that Snorri’s Edda conforms to developments that had been gaining ground for some two or three generations of poets by Snorri’s day. Chapter Three investigates how he took these developments to the next stage by means of creative use of his sources. Furthermore, it argues that there are many indications to the effect that a short treatise on poetic diction – the so-called Litla Skálda – preceded Snorri and was used as a model by him. Snorri’s initiative was, in this analysis, not quite as unprecedented or idiosyncratic as is often assumed. Rather, he synthesized a collective effort, spanning several generations of poets and authors, to stake out an Icelandic literary space.

16

Introduction

After this treatment of Snorri, his sources and his background, the chapter concludes with an analysis of the work of his nephew, Ólafr Þórðarson. He took the poetic project to yet another level by finally bringing Icelandic poetics into an explicit dialogue with the Latin authorities. In his translation of Priscian and Donatus, he claimed that both runes and Norse poetry were as good as and even more pristine than their Latin counterparts. His attempt appears to have been well received during the Middle Ages, since among grammatical texts, only his uncle’s Edda is attested in more witnesses. After treating poetry and grammatical texts, I turn to poetic quotation in the sagas in Chapter Four. This chapter covers a good deal of ground and warrants a somewhat longer introduction than the others. The Icelandic sagas are primarily about events rather than poetry, but their authors wrote about poets and quoted their poetry at every turn, and it would thus appear that they, like Snorri, viewed the local past through a poetic lens. The bulk of the poetry in the sagas of Icelanders and the kings’ sagas may well be authentic or at least from roughly the same time assigned to them in the sagas.28 The textual record, however, suggests that historiography developed from prose only (or in some instances poetry only) into an integrated prosimetrical form in the decades around 1200. The chapter thus begins with a study of how prosimetrical historiography developed, based on internal evidence in Icelandic literature as well as on comparative outlooks to Latin and Irish literature. The combined internal and external evidence, here gathered for the first time, suggests that while there were preceding conventions for quoting eddic poetry, transferring these techniques to skaldic poetry was an innovation which it took some time to develop. Once the prosimetrical ideal had taken root, it would appear that it soon became so strong that demand for poetry sometimes outstripped supply and that certain saga authors could not, or at least did not, resist the temptation to compose portions of the poetry pseudonymously. The remainder of the chapter focuses on how they went about this. Even though only a relatively small portion of the poetic corpus is likely to be pseudonymous, the phenomenon is particularly informative about those aspects of the poetic past the authors thought most important. Furthermore, pseudonymous composition typically aims at some degree of historical plausibility, and if the authors’ techniques for achieving this have left traces behind, we can reconstruct their methods. Finally, a focus on spurious poetry makes it possible to investigate various authorial strategies for text production, from conscientious transmission to pseudonymous intervention, and how these strategies influenced the evolution of Icelandic historiography and saga writing.

28 The main study to date is Klaus Johan Myrvoll, ‘Kronologi i skaldekvæde. Distribusjon av metriske og språklege drag i høve til tradisjonell datering og attribuering’ (PhD diss., University of Oslo, 2014).

Introduction

17

Such an investigation, however, depends on our ability to discern pseudonymous from authentic (or at least old) poetry. This is a contested field without an established methodology. I have opted to be methodologically inclusive, but to follow a clear basic principle: if only one feature in a stanza indicates that it is spurious, then that indication must be very strong for the stanza to be considered pseudonymous. For example, if a linguistic form un-attested before the late thirteenth century is ascribed to an eleventh-century protagonist, and if that form is secured by rhyme and thus not due to scribal carelessness, then little additional confirmation is needed. If, on the other hand, several mutually independent features point in the same direction, less absolute indications may be taken into account, as long as they are indeed independent of each other. Such is the basic principle for the detection of pseudonymous poetry in Chapter Four, and while all scholars may not agree on the diagnostic value of individual features, the premises should at least be clear. By this means the chapter proceeds to identify two principal ways in which saga authors sought to produce historically plausible poetry, namely through archaization of either metre or diction. Sometimes the pseudonymous poets composed faulty imitations of early metre, thus making it possible to detect the efforts of a scholarly impostor. No saga betrays the method of the author-poet more clearly than Egils saga, which is the object of the most detailed analysis. The question of authorship is also addressed, using some new parameters, and like a number of previous scholars I conclude that Snorri is likely to be the author of Egils saga. With regard to diction, a number of cases are presented where poetic expressions occur which are otherwise attested only in the prose of Snorri’s Edda, as opposed to earlier poetry, and where there are additional reasons to suggest that Snorri has added the relevant details himself. Another group of archaizing stanzas can be identified based on the study of diction in Chapter Two. Chapter Four arrives at several new conclusions regarding pseudonymous composition and its relationship to verse quotation at large. First, the method of some of these author-poets can be charted in some detail, which deepens our understanding of the dynamics involved in the creation of this literature. Second, the study of archaizing metre and diction shows that pseudonymous composition was not specific to sagas of Icelanders, as is sometimes assumed, but can also be detected in at least two kings’ sagas (Morkinskinna and excerpts from ‘Styrmir’s book’) as well as fornaldarsögur. The distribution is not only dependent on genre, but also on the date of action (before or after the settlement of Iceland) and on what function the stanzas serve. Third, pseudonymous composition appears to have been a matter of individual predilection, and it would seem that even one and the same author could hold himself up to different standards depending on context. There are, for instance, several indications to support that Snorri was a very reliable transmitter of early poetry in his Edda and in Heimskringla and that he composed none of the poetry there himself, but that he did so in Egils saga, and that his friend and contemporary Styrmir (d. 1245) did so in

18

Introduction

his saga about Saint Óláfr (see pp. 72–75). Fourth, pseudonymous composition was practiced from at least the time of Snorri and Styrmir and through the fourteenth century and should thus not simply be understood as sign of a gradually increasing remove from oral tradition. It appears to have been at least as common in the thirteenth century as in the fourteenth, since the linguistic changes in the decades around 1300 were such that there should otherwise have been more traces of them in poetry that can on other grounds be deemed spurious (see pp. 265–66). Even if it always remained less common than quoting from tradition, pseudonymous composition was thus an integral part of the creation of Old Icelandic literature. It appears not to have been used in Snorri’s Edda and Heimskringla (Kringla-version), but it is important to realize that this was an active choice, not simply a traditional way of doing things. Rather, what we see is a constant negotiation of how best to promote the literary past, indicative of the considerable efforts and personal engagement invested in the project. In the final chapter the book turns to fourteenth-century poetic developments at the Benedictine monastery at Þingeyrar in northern Iceland. Around the middle of that century, an anonymous monk compiled the greatest grammatical manuscript that has come down to us from the Icelandic Middle Ages, the so-called Codex Wormianus (W), adding his own thoughts and observations as he went. Among his additions, one finds an inspired and highly original discourse on language and truth, connected to paganism and phonetics. His texts on language and polytheism are edited and translated in the Appendix. At about the same time, and mainly within the same, northern monastic milieux, three other monks reacted vehemently against what they perceived as the deplorably high status of Snorri’s Edda and traditionalizing diction among monastics. Interestingly, the thoughts and reactions of all four centred on Snorri’s Edda. The compiler of W embraced the obscurity of Snorri’s Edda, but at the same time expressed a longing for final clarity and for a day when he would no longer compose using the language of the pagan gods, but rather sing the praises of the one true God in the one true language, which was Hebrew. The other three saw no place for Snorri’s Edda within devotional literature, but even while they rejected it, they took pains to display their knowledge of it. These poets did not, as is sometimes assumed, question the authority of Snorri’s Edda in a general sense. Rather, what they objected to was its use for devotional composition, and there could be no more eloquent witness to the impact of traditional diction and Snorri’s Edda in the Icelandic Middle Ages than the plea of these monks that one should not have to express one’s devotion to God by refererence to pagan demons. In this they were on the whole successful, since religious poetry of the following centuries mainly followed their stylistic recommendations. Christian religious poetry, however, had never been the main playing ground of the gods, and secular poetry remained largely committed to them. At that point, with the end of the most innovative period of Icelandic literary history, this account of the meeting of Latin methodology and native form ends.

Introduction

19

Never again would Icelandic authors so actively negotiate their position in relation to the learned heritage of Latin Chistendom. The Icelandic poetic tradition remained exceptionally strong and external impulses were accommodated in later centuries as well, but the spirit of innovation-through-tradition was never again nearly as intrepid as during the period under study here. It is the long thirteenth century which has most to offer from a comparative point of view, in a period when other vernaculars were also coming into their own. All of them had their peculiar strategies and solutions, and several drew heavily on poetic precedence. Only in Iceland, however, did this poetic tradition lead the authors to study and emulate works that were as pagan as those of classical antiquity and to insist that this heritage be imported wholesale into the new era. The process by which this came about is interesting in its own right, but it also serves as an important point of comparison to other areas of medieval Europe. There is no law which governs how a literature emerges, and no one could have predicted that the Icelanders would come to realize that their pagan poetry could provide them with a classical heritage of their own. But they did, and such is history. An asset is only an asset if perceived as such, and in this instance, the perspicacity of medieval Icelanders has granted them a place all their own in the history of European literature.

1 Metre and Rhyme This chapter gives an overview of metrical developments from the ninth century down to Snorri (d. 1241) and is one of two chapters to provide a background to the following chapters on grammatical literature and prosimetrical narrative. This longterm view makes it possible to see how gradual developments in the tenth through eleventh centuries gained momentum until poetics were reshaped into a formalized field of study in the twelfth. By the second half of the twelfth century, much of the foundations for the following literary developments had been laid; the next step was that of composing grammatical treatises and sagas, to be treated later in the book. In order to include only changes that may be indicative of the poets’ own perceptions of poetry, the study is restricted to highly marked features that the poets are likely to have been aware of themselves. As we shall see, such features were later discussed by Snorri, and they were also the object of conscious elaboration to create pseudonymous poetry for local historical narrative. They thus have bearing on the formative process of Old Icelandic literature in general. Furthermore, a focus on these features makes it possible to discern clear lines of development which previous research has not comprehensively described. Old Icelandic poetics had no concept that exactly corresponded to metre. The closest equivalent, generally translated as ‘metre’, is háttr, meaning ‘mode’ or ‘way’. As Kristján Árnason has argued, this designation was largely based on the context and function of a given form (‘the court way’, ‘the old way’, ‘the Greenland way’, etc.).1 On a formal level, the term háttr could encompass all systematically recurrent features, such as line length, catalexis, rhyme and even diction. Unlike classical and modern definitions of metre, the Old Icelandic concept was not so much based on measure as on social and literary functions and the systematic recurrence of any given feature. Even though changes occurred and outside influences were felt during the period under study, the basic perception of the háttr was retained. In this book, I use the word ‘metre’ for ease of reference, but the reader should be aware of the conceptual discrepancies lurking under the surface of this familiar term. As noted in the Introduction, Old Icelandic poetry is generally divided into eddic and skaldic poetry, where eddic poetry is generally anonymous and has relatively simple metre and diction, whereas skaldic poetry is typically composed by named poets and exhibits complex metre and diction.2 My analysis, here and with few exceptions throughout the book, will focus on skaldic poetry, partly because this was the medium where poets proved their stylistic mettle, but also because skaldic poetry is more securely datable than eddic poetry and allows for a study of

1 ‘Um Háttatal Snorra Sturlusonar. Bragform og braglýsing’, Gripla, 17 (2006), pp. 82–83. 2 On the distinction between eddic and skaldic poetry, see Margaret Clunies Ross, A history of Old Norse Poetry and Poetics (Cambridge: D. S. Brewer, 2005), pp. 21–28; Chase, ‘Introduction’, pp. 1–7. https://doi.org/10.1515/9783110643930-002

1 Metre and Rhyme

21

diachronic developments. Even more importantly, skaldic poetry was the centre of gravity for many of the intellectual developments described in this book, and thus for the genesis of Old Icelandic literature at large. More precisely, the kind of poetry under study here is so-called dróttkvæðr háttr (courtly metre), or simply dróttkvætt, and dróttkvætt-based metres.3 Dróttkvætt is highly regularized, which makes it possible to follow the development of several features over time. The more flexible eddic metres (fornyrðislag, málaháttr, ljóðaháttr and, for the present purposes, kviðuháttr) lack internal rhyme, and they underwent more restricted changes in the course of the poetic tradition than dróttkvætt-based metres. They do not, therefore, lend themselves nearly as well to the study of conscious changes in the modes of composition, and they will be left out of the metrical survey. I turn now to the constituent features of ‘classical’ dróttkvætt, and this description will form the starting point for the following analysis. In dróttkvætt, the smallest metrically self-contained unit is the couplet, whereas a half stanza is four lines and a full stanza is eight. A line has six metrical positions (which in practice tends to mean six syllables) and always ends in a trochee. The distribution of alliteration and internal rhymes in a classical couplet can be exemplified by the first two lines of the stanza on the Karlevi stone (late tenth century): Folginn liggr hinn’s fylgðu flestr vissi þat – mestar He, whom the greatest [deeds] followed, lies hidden. Most people knew that.

The odd line has two alliterations (f : f), placed freely in two of the three stressed positions.4 The even line has one alliteration (f), binding the lines together, and falling on the first position. The penultimate syllable of each line carries the second member of an internal rhyme pair called a hending (pl. hendingar), whereas the first member of the rhyme can be found in various positions in the line.5 The rhymes are on one syllable only. Odd lines have half rhyme – skothending – where vowels are different but the following consonants (one or more) are identical (olg : ylg). Even

3 For medieval use of this terminology, see Hans Kuhn, Das Dróttkvætt (Heidelberg: Carl Winter, 1983), p. 26. 4 There is one restriction: Both alliterations cannot fall before both hendingar (see Kuhn, Das Dróttkvætt, p. 89; Kari Ellen Gade, The Structure of Old Norse ‘Dróttkvætt’ Poetry. Islandica, 49 (Ithaca: Cornell, 1995), pp. 21, 52–54; Þorgeir Sigurðsson, ‘Þróun dróttkvæða og vísuorðhlutar að hætti Hans Kuhn’, Són, 8 (2010), p. 10). 5 They generally occur in positions carrying primary or secondary stress, but it would seem that they in exceptional cases can be found also in unstressed positions (see Kuhn, Das Dróttkvætt, pp. 85–89). Somewhat confusingly, in medieval usage hending can refer to a pair of hendingar or to only one of them.

22

1 Metre and Rhyme

lines have full rhyme – aðalhending – where both vowels and consonants are identical (est : est).6 Classical dróttkvætt was itself the product of an evolution. In the earliest skaldic poetry, the rules given above were not strictly adhered to. Some odd lines might lack hendingar, and some even lines might have only skothendingar and thus be less metrically marked than normal. There would sometimes be a light syllable or two before the alliterative syllable in even lines (anacrusis or onset). The opening half stanza from Bragi’s Ragnarsdrápa may serve as an example. Hendingar are in italics, onset in bold: Vilið Hrafnketill heyra, hvé hreingróit steini Þrúðar skalk ok þengil þjófs ilja blað leyfa.7 Do you, Hrafnketill, wish to hear how I shall praise Þrúðr’s thief’s [the giant Hrungnir’s] footsoles’ leaf [shield], grown with bright colours, and the ruler.

Here, the first line lacks hending where one would expect skothending, whereas the second has aðalhending, as would normally be expected (ein: ein). The third line again lacks hending, and the last has skothending where one would expect aðalhending. The only line which conforms to the normal hending pattern – the second – is deviant in another respect, since it has onset in hvé. The /h/ in hvé does not count as alliteration, since the word does not carry stress. As a result, each line exhibits licences that later poetry would increasingly disallow. Although the stricter kind of dróttkvætt seems to have become the norm by the turn of the millennium at the latest, early examples of the lax, archaic variety remained familiar through transmission. As we shall see below as well as in Chapter Four, such archaic poetry eventually became important in the shaping of a historical awareness of stylistics. A fundamental precondition for the following discussion is that some poetry can be attributed to a poet or at least to his time with an acceptable degree of certainty, whereas other poetry cannot. The most fundamental criterion here is the distinction between authenticating and situational verses in the sagas, to use Diana Whaley’s

6 The number of consonants which belong to the hending varies according to principles that have for a long time remained obscure, but although some clusters may still call for further explanation, Þorgeir Sigurðsson, Kristján Árnason and Klaus Johan Myrvoll have clarified the main principles at work (Þorgeir Sigurðsson, ‘Rímstuðlar. Um tengsl ríms, atkvæðaskiptingar og stuðla í íslenskum skáldskap’, Íslenskt mál og almenn málfræði, 23 (2001), pp. 215–27; Kristján Árnason, ‘On the Principles of Nordic Rhyme and Alliteration’, Arkiv för nordisk filologi, 122 (2007), pp. 87–93, 97–107; Klaus Johan Myrvoll, Kronologi i skaldekvæde, pp. 53–82). Since the matter is complex and of little import to this book, I do not go into specifics here. 7 SkP 3, p. 28.

1.1 The Eleventh Century

23

now common terminology for a distinction first suggested by Alois Wolf in 1965 and further developed by Bjarni Einarsson in 1974.8 When poetry was used to authenticate the historical narrative, authenticity seems to have been important to medieval Icelandic authors. Such authenticating quotations are typically introduced with the words svá segir N. N (as N. N. says) or the like. When it was used for dramatic, emotional or decorative effect, by contrast, authors seem not to have necessarily asked questions about authenticity. These situational quotations are typically introduced with words like þá kvað N. N vísu (then N. N recited a stanza). Bjarni Einarsson’s distinction is discussed in Chapter Four. I there test it against the body of pseudonymous poetry, and it is found to hold up well as a source-critical tool in texts treating the period after the settlement of Iceland (after c. 870–874). The development outlined below is based on poetry that can be considered reliable according to this distinction, and it should be noted that quotations from Skáldskaparmál figure prominently. Bjarni Einarsson does not discuss Skáldskaparmál, but that text almost exclusively uses authenticating quotation, and there are no indications that Snorri composed any of the poetry himself. As we shall see in Chapter Three, his method appears to have been one of creative interpretation, and this speaks against the idea that he would have composed the poetry that he needed himself: if he had, he would presumably not have had to display such creativity in arriving at his interpretations. This impression is corroborated by Heimskringla, where he clearly added no poetry of his own (see below pp. 241–43). Skáldskaparmál thus appears to be an unusually reliable source even within the authenticating group. When unreliable poetry displays features that are only to be found much later in reliable poetry, the assumption is that the unreliable stanzas in question are not authentic, if no particular circumstances indicate the contrary. Using this rule of thumb reveals clear lines of development, although full justice cannot be done here to each doubtful stanza here.

1.1 The Eleventh Century In the eleventh century, the laxer kind of dróttkvætt was no longer productive but had been replaced by the strict variant. Dróttkvætt was also used as a basis for the development of new metres through alteration of the length of the line. The line could be contracted or extended to four, five or eight positions, rather than the

8 Alois Wolf, ‘Zur Rolle der Vísur in der altnordischen Prosa’, in Festschrift Leonhard C. Frans zum 70. Geburtstag, ed. Osmund Menghin and Hermann M. Ölberg (Innsbruck: Innsbrucker Gesellschaft zur Pflege der Geisteswissenschaften, 1965), pp. 459–84; Bjarni Einarsson, ‘On the Rôle of Verse in Saga-Literature’, Mediaeval Scandinavia, 7 (1974), pp. 118–25; Diana Whaley, ‘Skalds and Situational Verses in Heimskringla’, in Snorri Sturluson. Kolloquium anläßlich der 750. Wiederkehr seines Todestages, ed. Alois Wolf (Tübingen: Gunter Narr Verlag, 1993), p. 252.

24

1 Metre and Rhyme

ordinary six. I would suggest that alteration of the length of the line was the main parameter of conscious metrical innovation before the twelfth century. The contracted form with four positions is called tøgdrápulag or tøglag, named, it seems, after the poem Tøgdrápa (journey-poem (?)) by Þórarinn loftunga (probably the first poet to use this form, c. 1028).9 In the same century, Sighvatr used it in his Knútsdrápa (c. 1038). In the twelfth century, it was used by Einarr Skúlason in his Haraldsdrápa II and by Þórarinn stuttfeldr in Stuttfeldardrápa. The form is otherwise rare.10 The contracted form with five positions is called haðarlag (the metre of Hǫðr (?)) and is found in Þormóðr Trefilsson’s Hrafnsmál (early eleventh century), in a stanza by Sigurðr slembidjákn (c. 1139) and in a few poems of the thirteenth century.11 The extended form with eight positions is called hrynjandi or hrynhenda (flowing rhyme form). It may possibly be attested as early as c. 986 in the fragment of a poem called Hafgerðingadrápa, but several factors indicate that the poem might instead be a twelfth-century composition.12 It is first reliably attested in Arnórr jarlaskáld’s Hrynhenda (1046) and later in Markús Skeggjason’s Eiríksdrápa (c. 1104) and Gamli kanóki’s Jónsdrápa around the middle of the twelfth century. It would have a great future ahead of it, above all in religious poetry from the fourteenth century onwards.13 One contracted form with five positions, hálfhnept (half-curtailed), differs from the other metres with altered line length in that it is catalectic (that is, it lacks the last, unstressed syllable). It is also noteworthy that two additional restrictions seem to have followed, namely that the first hending of each line must fall on the first or second position, and that the alliterating staves of odd lines cannot stand in the first and penultimate position of the same line.14 The form is attested in one early poem: Óttarr svarti’s Óláfsdrápa (c. 1018). Some single stanzas in this metre are attributed to even earlier poets, but they belong to the situational type and are more likely to have been composed in the twelfth and thirteenth century than in the tenth.15 The position of rhymes in hálfhnept leads us to what may be the most curious innovation of the period, but one which, unlike the ones surveyed so far, does not affect the line length. Snorri reports that Þorvaldr veili, who was killed shortly before

9 Snorri Sturluson, Edda. Háttatal, ed. Anthony Faulkes (London: Viking Society for Northern Research, 2007), pp. 87, 156. 10 See Snorri, Háttatal, ed. Faulkes, pp. 87–88. 11 See Snorri, Háttatal, ed. Faulkes, pp. 89. 12 Jakob Benediktsson, ‘Hafgerðingadrápa’, in Speculum Norroenum: Norse Studies in Memory of Gabriel Turville-Petre, ed. Ursula Dronke et al. (Odense: Odense University Press, 1981), pp. 27–32. 13 See Snorri, Háttatal, ed. Faulkes, pp. 85–86. 14 See Ernst Albin Kock, ‘Ett kapitel nordisk metrik och textkritik’, Arkiv för nordisk filologi, 49 (1933), pp. 279, 291. 15 Thus one stanza by Haraldr Hárfagri, one by Bjǫrn breiðvíkingakappi and two couplets in Kormáks saga (stanzas by Brynjolfr ulfaldi, Haraldr harðráði and Magnús góði are more difficult to assess; for occurrences, see Snorri, Háttatal, ed. Faulkes, p. 88).

1.1 The Eleventh Century

25

the year 1000 by the missionary Þangbrandr, was at one point shipwrecked with his men on a skerry, and in the bitter cold he composed the poem Skjálfhenda (with shivering rhymes; perhaps with wordplay on hǫnd, and thus also ‘with shivering hands’), and the metre was named accordingly.16 Þorgeir Sigurðsson has described the various occurrences of this metre in the period up to Snorri.17 Nothing is preserved of Þorvaldr’s poem, but seven half stanzas and one couplet of Hallvarðr háreksblesi’s somewhat younger Knútsdrápa (c. 1029) have been transmitted in contexts that suggest that they are authentic.18 The defining features of this metre are found in the third line of each half stanza: its first three syllables are heavy, alliteration falls on the first and third, and the first hending falls on the first or the second syllable.19 I have put alliteration in bold, hendingar in italics: sigrakkr, Sǫlsa bekkjar20 battle-bold, Sǫlsi’s bench’s [the sea’s]

or heiptsnarr hildar leiptra21 quick to enmity, battle’s lighnings’ [swords’]

As Hans Kuhn notes, the metrical structure made it syntactically difficult to employ such lines before the third line, and this, presumably, accounts for their distribution within the stanzas.22 The number of parameters involved in hálfhnept and skjálfhenda is extraordinary for the period and would rather have been expected in the twelfth century. Hálfhnept is catalectic and has regulation of both hendingar and alliteration, whereas skjálfhenda has additional regulation of stress, hendingar and alliteration. The placing of rhyme is identical in the two, which suggests that they are somehow connected. Skjálfhenda is peculiar not only in its unexpectedly high number of regulations, but also in that the circumstances of its invention are given by Snorri. It would seem that its anomalous character already called for some kind of explanation at an early date, and that Snorri is referring to that tradition. While the connection to

16 Snorri, Háttatal, ed. Faulkes, p.18. 17 ‘Nýjar skjálfhendur á 12. öld’, Són, 12 (2014), pp. 55–67. 18 SkP 3, pp. 230–40. 19 This description differs somewhat from those of Anthony Faulkes and Þorgeir Sigurðsson, who do not provide the rule that the rhyme should be in the first or second position (Snorri, Háttatal, ed. Faulkes, p. 59; ‘Nýjar skjálfhendur á 12. öld’, p. 56). Both scholars adhere to Hans Kuhn’s observation that the tendency in such lines is to have the hending in the second position, which is true enough, but whereas hending in the second position is a tendency, hending in first or second position is a rule (Das Dróttkvætt, pp. 333–34). Snorri notes this, and the corpus of skjálfhenda poetry corroborates it (Snorri, Háttatal, ed. Faulkes, p. 16). 20 SkP 3, p. 231. 21 SkP 3, p. 233. 22 Das Dróttkvætt, pp. 333–34.

26

1 Metre and Rhyme

Þorvaldr raises no particular suspicions, the name of the metre seems to describe its auditive qualities, and one may probably surmise that the story is derived from the name. There are a few random lines corresponding to skjálfhenda in earlier poetry (the earliest is to be found in Þjóðolfr’s Haustlǫng), and its development can thus be explained within the dróttkvætt tradition.23 Hálfhnept seems to combine catalexis with features from skjálfhenda, and it therefore seems likely that skjálfhenda has influenced hálfhnept, rather than the other way around.24 Even though the evolution of these metres can thus, to some extent, be explained organically within the native tradition, the number of features involved is remarkable, and the question remains: What prompted this innovation, seemingly so far ahead of its time? I do not find myself in a position to venture an answer, and I am certainly at a loss to produce one that is more attractive than that poor Þorvaldr shivered so with cold that he defied all poetic precedent, only to then have his precocious literary career stopped short by the heavy hand of God’s chosen apostle. In any event, his metre is a salutary reminder that historical developments need not be entirely linear. Apart from the cluster of features involved in skjálfhenda and hálfhnept, the length of lines seems to have been the main object of systematic innovation during the eleventh century. Another was end-rhyme. End-rhyme’s interest to poetic innovators actually emerged well before the eleventh century, with Egill Skalla-Grímsson’s Hǫfuðlausn (c. 935–50), a poem which displays both monosyllabic and disyllabic rhymes.25 The authenticity of Hǫfuðlausn has been the topic of much debate, but one factor, namely the cultural and historical implications of its rhyming patterns, has not been taken into account.26

23 Þorgeir Sigurðsson, ‘Nýjar skjálfhendur á 12. öld’, p. 63. 24 Einar Ólafur Sveinsson has suggested that the Icelandic catalectic metres (hálfhneppt and similar metres of a later date) derive from Irish tradition, but he does not note the similarities to skjálfhenda (‘An old Irish Verse form wandering in the North’, in Proceedings of the Seventh Viking Congress, Dublin 15–21 August 1973, ed. Bo Almqvist and David Greene (Dublin: Royal Irish Academy, 1976), pp. 141–52). If, as I suggest, catalexis was later added to this cluster of innovations, the matter is probably less straightforward than Einar Ólafur Sveinsson contends. 25 The date given here reflects Finnur Jónsson’s and Sigurður Nordal’s datings (936 and 948, respectively), presupposing authenticity and based on the saga account (Finnur Jónsson, Den oldnorske og oldislandske litteraturs historie, 3 vols (København: G. E. C. Gads forlag, 1920–24), 1, pp. 476, 483; Egils saga, ed. Sigurður Nordal, p. lii). 26 On the question of authenticity, see, for instance, Jón Helgason, ‘Höfuðlausnarhjal’, in Einarsbók. Afmæliskveðja til Einars Ól. Sveinssonar 12. desember 1969, ed. Bjarni Guðnason et al. (Reykjavík, 1969), pp. 156–76; Dietrich Hofmann, ‘Das Reimwort giǫr in Egill Skallagrímssons Hǫfuðlausn’, Mediaeval Scandinavia, 6 (1973), pp. 93–101; Bjarni Einarsson, Litterære forudsætninger for ‘Egils saga’ (Reykjavík: Stofnun Árna Magnússonar, 1975), pp. 195–207; Baldur Hafstað, Die ‘Egils saga’ und ihr Verhältnis zu anderen Werken des nordischen Mittelalters (Reykjavík: Rannsóknarstofnun kennaraháskóla Íslands, 1995), pp. 35–66; Jónas Kristjánsson, ‘Kveðskapur Egils Skallagrímssonar’, Gripla, 17 (2006), pp. 7–35; Haraldur Bernharðsson, ‘Göróttur er drykkurinn. Fornmálsorð í nútímabúningi’, Gripla, 17 (2006), pp. 37–73.

1.1 The Eleventh Century

27

From an Old Norse perspective, end-rhymes are an anomaly in the tenth century. This may be cause for suspicion, but the saga reports that the poem was composed at York, which was a linguistic and cultural melting pot at the time. While end-rhyme was not the norm in Old English, the Riming poem attests to its presence in the tenth century, and in Latin hymns it had certainly been used for a long time already. Apart from its rhymes, Hǫfuðlausn is a conventional praise poem and gives no additional cause for suspicion. The saga author shows some diachronic awareness of, for instance, the interaction of pagans and Christians in tenth-century Northumbria,27 but it nonetheless seems unlikely that a later impostor would have had the diachronic awareness to produce a poem that spoke to the cultural interaction of tenth-century York on a formal level. It is therefore likely that the poem is authentic, and this is further corroborated by authenticating quotations of some of its stanzas in Skáldskaparmál.28 Other instances of end-rhyme in the tenth century belong to situational quotations which instill little confidence, and Hǫfuðlausn may thus have been a lone swallow in its time.29 Although not quoted in Skáldskaparmál and thus not as reliable as Hǫfuðlausn, Gunnlaugr is said to have presented a drápa with end-rhyme to Sigtryggr silkiskegg of Dublin in 1003. The refrain and one and a half stanza, all with monosyllabic endrhymes, are transmitted in Gunnlaugs saga.30 Sigtryggr was the son of Óláfr kváran, at one time king of Dublin and York. The strong connection between Dublin and York, where Egill about half a century before had presented his end-rhymed poem, may indicate that Gunnlaugr’s poem is authentic and that the cultural interaction in Britain and Ireland was important for the development of end-rhyme as a possible, if marginal, option for the composition of skaldic poetry. The authenticity of the supposedly slightly younger Grámagaflím by Bjǫrn Hítdœlakappi is difficult to assess.31 It contains both monosyllabic and disyllabic rhymes. The next reliable evidence is found in a poem by Þjóðolfr Arnórsson addressed to Haraldr harðráði around the middle of the eleventh century. Only one full and three half stanzas have been preserved.32 These stanzas, presumably from different parts of the poem, exhibit monosyllabic end-rhyme only. Next comes the poem by Einarr Skúlason that has been given the name Runhenda (endrhyming poem, c. 1155) by editors, again displaying only monosyllabic end-rhyme.33

27 See, for instance, Egils saga, ed. Sigurður Nordal, pp. 128–29. 28 Skj A 1, pp. 35–38. Note, furthermore, that the debate referred in note 26, on the value of the vowel in one rhyming word (/ǫ/ or /ø/), may be of limited value for dating, since inexact rhymes were very common in Europe at large up to c. 1100. 29 See Snorri, Háttatal, ed. Faulkes, pp. 89–91. 30 Borgfirðinga sögur, ed. Sigurður Nordal and Guðni Jónsson. Íslenzk fornrit, 3 (Reykjavík: Hið íslenzka fornritafélag, 1938), p. 75. 31 Borgfirðinga sögur, ed. Sigurður Nordal and Guðni Jónsson, pp. 168–69. 32 SkP 2, pp. 103–07. 33 This is obviously a panegyric for king Eysteinn, and it is mainly quoted in the kings’ saga Morkinskinna (SkP 2, pp. 551–59).

28

1 Metre and Rhyme

Some additional systematic innovations can be identified in the eleventh century, but they are all tied to the refrains of drápur (the most prestigious kind of long poems, precisely because it featured an elaborate refrain). These innovations cannot be defined as metres without further qualifications, since the features involved were generally not upheld throughout a poem or even, with one noteworthy exception, throughout an entire stanza. It would appear that refrains were a locus of innovation, producing short but metrically interesting specimens on which poets of later centuries would draw. Let us first consider a stanza by Þórðr Særeksson, active in the late tenth and early eleventh century.34 It is composed with end-rhyme, which is itself noteworthy, but it also displays one interesting additional feature: for the stanza to give meaning, the first line has to be connected to the fifth, the second to the sixth, and so on. This arrangement of lines is sufficiently systematic to constitute what in the later metrical lists would be considered a metre (háttr).35 The stanza is transmitted in the main witnesses to Skáldskaparmál.36 In all of the skaldic tradition, we find nothing comparable before the composition of Háttalykill in the 1140s, and the stanza thus serves as a valuable indication of developments within skaldic poetry that preceded the learned compositions of a later period. The uniqueness of this stanza could instill doubts in its authenticity, if it did not have contemporary parallels in the refrain of some longer poems. From the early eleventh century on, some poems with refrain (drápur) exhibit so-called klofastef, that is, only a part of the refrain is given each time, and the recipient of the poem must combine the lines from the different versions of the refrain to be able to extract their meaning. The earliest preserved poem of this kind may be Eyjolfr’s Bandadrápa (c. 1010), and the form is thereafter identifiable in some nine poems from the eleventh through thirteenth centuries.37 On the combined evidence of Þórðr’s stanza and poems with klofastef, we may conclude that an additional innovation of the eleventh century was the systematic splitting of semantic units. This was, however, generally confined to refrains, and Þórðr’s stanza constitutes a curious exception to this rule. Another enigmatic poem of an even earlier date is Kormákr Ǫgmundarson’s Sigurðardrápa (mid-tenth century), which features parenthetical mythological references in the end of the fourth line of most half stanzas (so-called stál). Since Kormákr’s

34 SkP 1, p. 236. 35 Cf., for instance, the metre greppaminni in Háttalykill and Háttatal (Snorri Sturluson, Edda. Háttatal, ed. Faulkes, pp. 20 (40), 78; Háttalykill enn forni, ed. Jón Helgason and Anne Holtsmark. Bibliotheca Arnamagnæana, 1 (København: Munksgaard, 1941), pp. 125–26). 36 SkP 3, p. 478. 37 Theodor Möbius, ‘Vom Stef’, Germania, 18 (1873), pp. 134–37, 140–41; cf. Bjarne Fidjestøl, Det norrøne fyrstediktet (Øvre Ervik: Alvheim & Eide, 1982), p. 183.

1.1 The Eleventh Century

29

poem is referred to as a drápa, and since not all stanzas display this feature, we may conclude that it constituted a refrain, albeit of a rare type.38 Only one full stanza is preserved, but it allows us to understand the structure of the stefjabálkr (the section of the poem containing refrains): the first half stanza ends in a mythological statement ‘véltu goð Þjaza’ (the gods deceived Þjazi), whereas the second ends in a statement about the ruler ‘vá gramr til menja’ (the ruler fought for neck-rings).39 Heimskringla quotes this stanza for its historical content, whereas other half stanzas from the stefjabálkr have been included in Skáldskaparmál for their mythological content, and the half stanzas containing statements about the ruler have therefore not been included. Later the refrain of Sigurðardrápa would serve as the basis for the metre hjástælt in Snorri’s Háttatal, and this is a good example of how refrains from the period before the twelfth century afterwards became a treasure trove for grammarians.40 Finally, a truly remarkable innovation is found in a half stanza attributed to Hofgarða-Refr in the 1030s and transmitted in Skáldskaparmál. It is composed in alhent, meaning that it has two rather than one hending pair in each line, resulting in lines almost completely dominated by rhyme. The analysis of this stanza has been substantially and convincingly altered by its editors in SkP 3, and I follow their reading and interpretation here41: Þér eigum vér veigar Valgautr, salar brautar Fals, hrannvala, fannar, framr valdr, ramar gjalda.42

38 The name – Sigurðardrápa – is given in Heimskringla (Heimskringla, ed. Bjarni Aðalbjarnarson, 3 vols. Íslenzk fornrit, 26–28 (Reykjavík: Hið íslenzka fornritafélag, 1979), 1, p. 168). See further Snorri, Háttatal, ed. Faulkes, pp. 10, 78; Finnur Jónsson, Den oldnorske og oldislandske litteraturs historie, 1, pp. 528–29. SkP claims that Sigurðardrápa is not one, but several, poems, disregarding the fact that the type of refrain used in it is unique in the corpus before the thirteenth century (SkP 3, pp. 272–73). The claim is based on reading the ruler’s heiti yngvi literally as a reference to the Yngling dynasty, which is unlikely to be the case (see LP, s.v.), and on the assumption that Hákon could not have been referred to as mærr (famous) when he was still young. Such is the nature of panegyrics, however. The coherence created by the peculiar type of refrain is strong, and the earlier assumption that Sigurðardrápa is one poem should thus be retained. 39 But cf. SkP 3, pp. 283–85. The editors overlook the fact that the other stanzas refer to mythological, not heroic, lore, and that the statement ‘vá gramr til menja’ underlines the topic of the ruler’s generosity in the stanza. 40 Snorri, Háttatal, ed. Faulkes, pp. 10, 77–79. Snorri does not name Kormákr as his model, but since several of the stanzas in question are found in Skáldskaparmál and exactly match his description, it is highly likely that Kormákr played that role (cf. Snorri, Háttatal, ed. Faulkes, p. 78). 41 SkP 3, pp. 256–58. 42 SkP 3, p. 256. See also Snorri, Háttatal, ed. Faulkes, p. 83; Snorri, Skáldskaparmál, ed. Faulkes, pp. 9–10 (17), 155 (4), 159 (17).

30

1 Metre and Rhyme

Valgautr [Óðinn, here for Gizurr, recipient of the poem and bearing one of Óðinn’s names], we must repay you for the potent snow-drift’s way’s [mountain’s] hall’s [cave’s] dwarf’s [giant’s] drink [poetry], excellent wave-horses’ [ships’] owner [seafarer].43

This is an extremely demanding form, otherwise only known to have been used by bishop Klœngr in the twelfth, by Snorri in the thirteenth, and by two more poets in the fourteenth century (see below p. 278). The other stanzas that are likely to belong to the same poem by Hofgarða-Refr are composed in regular dróttkvætt (one half and one full stanza), and this is also true of all his remaining known poetry.44 The form and content of the alhent stanza – it is an apostrophe to Gizurr gullbrárskáld, by way of a reference to Óðinn, in thanks for the gift of poetry – suggests that this is the refrain of the poem. Snorri would later use the form in the refrain of his panegyric to Skúli.45 This and the previous examples suggest that some of the stylistic varieties that would later be construed as metres developed first in the elaborate displays of poetic skill contained in the refrains. The bulk of the evidence reviewed here supports that the eleventh century saw the earliest phase of systematic alteration of highly marked features, although endrhyme may have developed even earlier. It bears notice that the poets involved seem often to have had a connection to Britain and Ireland. Þórarinn, Óttarr svarti, Hallvarðr háreksblesi and Sighvatr composed in honour of Knútr, and Arnórr to the earls of Orkney. Egill dwelt in Northumbria for a time and Kormákr was in Dublin.46 Cultural interaction is thus likely to have served as a catalyst. In Britain and Ireland, Norse poets may have become acquainted with the rhythms and rhymes of Latin hymns; at least in the case of hrynhenda such a background seems highly probable.47 As we have seen, experimentation was so far restricted to the parameters of line length, end-rhyme, catalexis and the enigmatic nexus of innovations involved in skjálfhenda and hálfhnept. Another set of innovations, involving split lines, parenthetic phrases and hendingar, seems to have been limited to refrains and not yet to have taken on a life of its own, although Þórðr’s stanza remains an exception. The parameters involved in the new metres are few, with additional effects and exhibitions of virtuosity loaded into the refrains. Such a distribution accords well with an oral and performative setting, in contrast to the period of systematic analysis and bookish learning which was to follow.

43 Adapted from SkP 3, p. 256. 44 SkP 3, pp. 243–65. 45 On Snorri’s drápa, see Snorri, Háttatal, ed. Faulkes, p. 83. 46 See Judith Jesch, ‘Skaldic Verse in Scandinavian England’, in Vikings and the Danelaw. Select papers from the proceedings of the Thirteenth Viking Congress, Nottingham and York, 21–30 August 1997, ed. James Graham-Campbell et al. (Oxford: Oxbow Books, 2001), pp. 317–18. 47 Snorri, Háttatal, ed. Faulkes, p. 85; Hans Kuhn, Das Dróttkvætt, pp. 312, 337–41.

1.2 Grammatical and Historical Awareness in the Twelfth Century and Beyond

31

1.2 Grammatical and Historical Awareness in the Twelfth Century and Beyond In the twelfth century, poets of a new kind emerged in Iceland and Orkney. These were no longer warriors in the retinue of a king, learned in the craft of poetry, but lacking formal training in the analysis of literature and language. Rather, the new poets were either clerics themselves or were surrounded by and interacted with them. The result was a new way of viewing and producing poetry. This development is best exemplified by the list of metres called Háttalykill, which, according to Orkneyinga saga, was composed by earl Rǫgnvaldr of Orkney and the otherwise largely unknown Icelander Hallr Þórarinsson in the 1140s. Though only preserved in a fragmentary state in paper manuscripts, several factors indicate that the poem we have is indeed the one mentioned in the saga (see pp. 111–12). Háttalykill is a list of metres with two stanzas of each, and these metres are not only constituted by altering line length and end-rhyme, but also by varying the patterns of hendingar and diction. With regard to hendingar in particular, deviant lines in earlier poetry that were the products of poetic licence at the time of composition have been classified and used as building blocks of new metres in the poem.48 In the preceding tradition, Anthony Faulkes has found only one full stanza in an alternative, hending-based metre.49 The stanza in question is attributed to Bjǫrn krepphendi and dated to the late eleventh century.50 It uses only aðalhendingar, and the form thus remains classical dróttkvætt with rhymes that are more marked than usual: a relatively inconspicuous alteration to the hending scheme. The vast majority of the metres in Háttalykill were not independent metres in the earlier tradition, but were only optional patterns within a flexible system. This is true also of diction, including features such as fixed clause length, parenthetical sentences and structured question-and-answer format. The only known antecedents to such fixed patterns of diction are found in klofastef and in the parenthetical phrases of Kormákr’s Sigurðardrápa, both of which pertain to refrains, not to independent metres. With regard to hendingar, the same can be said of the refrain in alhent by Hofgarða-Refr. Now, however, variant lines due to a flexible system and the occasional showpieces of refrains were both used to construct new metres. Earlier, if a poet wanted to go beyond standard dróttkvætt, he might choose to compose in a

48 Most notably þríhent, munnvǫrp, háttlausa, iðrmælt, dunhent (SkP 3, pp. 1019–20, 1022–24, 1058–60, 1066–67, 1073–75). See also Snorri, Háttatal, ed. Faulkes, pp. 82–85. 49 Snorri, Háttatal, ed. Faulkes, pp. 82–83. The late tenth-century poem Vellekla exhibits some experiments with hendingar, but remains within the fluid, archaic mode. On Vellekla, see Hans Kuhn, ‘Vor Tausend Jahren. Zur Geschichte des skaldischen Innenreims’, in Speculum Norroenum: Norse Studies in Memory of Gabriel Turville-Petre, ed. Ursula Dronke et al. (Odense: Odense University Press, 1981), pp. 298–304. 50 Snorri, Háttatal, ed. Faulkes, p. 82; Heimskringla, ed. Bjarni Aðalbjarnarson, 3, p. 217.

32

1 Metre and Rhyme

metre with altered line length, or he chose a simple metre and added end-rhyme. During the eleventh century, these two features had entered the toolbox of the practical craftsman who was then the poet. The structured and multimodal exposition of Háttalykill, by contrast, indicates that poetry had now become the object of systematic study. Háttalykill’s approach to poetry can be compared to that of the contemporary First Grammarian (c. 1150). He notes that in skaldic poetry, unlike ordinary speech, the word járn (iron) may be scanned éarn, and he uses this observation to support his own orthographic system.51 Here, skaldic poetry has become a tool for phono-graphematic investigation within a learned setting. What, then, had happened? When our knowledge of the persons involved is taken into account, the most likely answer is that skaldic poetry had entered the realm of formal learning, and that these poets had received at least some training in the basic discipline of Latin language and literature: grammatica. Although the identity of the First Grammarian remains unknown, it is evident from his text that he had received a relatively substantial education. The intellectual background of Rǫgnvaldr, the author of Háttalykill, is sketchier. He had commenced the building of Saint Magnus Cathedral in 1137, and one can only surmise that he applied his active intellect and his interest in Old Norse court poetry to such learning as his churchmen could provide. In a stanza attributed to Rǫgnvaldr he boasts, among other things, of his acquaintance with books.52 His knowledge of Latin need not have been profound, and he is far from mapping Donatan categories onto skaldic poetry, as Óláfr Þórðarson would do in the Third Grammatical Treatise a hundred years later. Grammatica must here be understood in a more generalized way, as the idea of systematic textual study. What is grammatical about Háttalykill, furthermore, is not so much the individual sources of inspiration to the different metres, although Latin hymns and troubadour poetry have been suggested as exemplars for some metres.53 Rather, what is truly grammatical is the method by which new metres are arrived at, namely

51 The First Grammatical Treatise, ed. Benediktsson, pp. 224–26. See discussion below pp. 175–77. 52 The stanza is not motivated by the plot or dialogue, but is attached to a general description of Rǫgnvaldr, presumably as a well-known stanza of his. There is thus no obvious reason to suspect that it was composed for the saga (Orkneyinga saga, ed. Finnbogi Guðmundsson. Íslenzk fornrit, 34 (Reykjavík: Hið íslenzka fornritafélag, 1965), p. 130; SkP 2, p. 576). 53 Háttalykill enn forni, ed. Jón Helgason and Holtsmark, pp. 121–34. Rǫgnvaldr had been to the Holy Land, and he stopped by the court of Ermengarde of Narbonne, where troubadour poetry thrived. Some of his poetry both within and outside of Háttalykill seems to betray troubadour influence (Háttalykill enn forni, ed. Jón Helgason and Holtsmark, pp. 127–35; Holtsmark, ‘Kjærlighetsdiktning’, p. 441). It should be noted, though, that Háttalykill was probably composed before this trip. With regard to content, rather than style, the validity of the observations of troubadour influence has been questioned. See Roberta Frank, Old Norse Court poetry. The Dróttkvætt Stanza (Ithaca, N.Y: Cornell University Press, 1978), pp. 168–69; Alison Finlay, ‘Skald Sagas in their Literary Context, 2: Possible European Contexts’, in Skaldsagas. Text, Vocation, and Desire in the Icelandic Sagas of Poets, ed. Russell Poole (Berlin: De Gruyter, 2001), pp. 241–43.

1.2 Grammatical and Historical Awareness in the Twelfth Century and Beyond

33

through structured analysis of earlier texts and the subsequent transformation of these observations into rules. Such descriptive and prescriptive study, based on authoritative poetry, lies at the core of medieval grammatica, and Háttalykill provides the first clear indications that Old Norse poets were beginning to take this approach. In general, these developments are discernible in Iceland and Orkney, but not in Norway (even if Rǫgnvaldr was born and raised in Norway, we hear of his literary activities only after he came to Orkney). It should be noted, however, that there is at least one indication of a connection between a systematic poetics and Latin learning also in Norway, though it belongs to the following century. The runic stick B 145, carved in Bergen in the years prior to 1248, has a dróttkvætt stanza which features only aðalhendingar. After the stanza follows a hexameter line from Vergil’s Eclogues 10.69: Omnia vincit amor et nos cedamus amori (love conquers all, we too should yield to love), without mistakes.54 Since this was a famous line, it need not indicate a great amount of learning, but certainly greater than that of the other hand which tried to imitate the text on the fourth side of the stick and made numerous mistakes. This runic stick thus bears additional testimony to the fact that some degree of Latin learning and a systematic approach to skaldic poetry went hand in hand. Another poet, Hallar-Steinn, also exhibits features that bear witness to the new trends. He seems to have been active in the latter half of the twelfth century.55 Here follows the first half stanza from his poem Rekstefja: Hǫrs gnótt Hrunda sléttum hljóðs kveðk mér at óði randhvéls rennu-Þundi; Rekstefju tekk hefja.56 I bid the linen-string’s Hrundirs’ [women’s] throng for silence for the smooth poem on behalf of me, the shield-rim-wheel’s [shield’s] movement- [battle-] Þundr [Óðinn; warrior].57 I commence Rekstefja.

54 See Aslak Liestøl, Wolfgang Krause, and Jón Helgason, ‘Drottkvætt-vers fra Bryggen i Bergen’, Maal og Minne (1962), pp. 98–108. 55 See SkP 1, p. 894. Later dates have been proposed for Hallar-Steinn, but as he is quoted by name in the major manuscripts of Skáldskaparmál, these suggestions are untenable (Jan de Vries, Altnordische Literaturgeschichte, 2 vols. Grundriss der germanischen Philologie 15–16 (Berlin: De Gruyter, 1964–67), 2, p. 44; Sveinbjörn Rafnsson, Ólafs sögur Tryggvasonar. Um gerðir þeirra, heimildir og höfunda (Reykjavík: Háskólaútgáfan, 2005), pp. 185–88; Snorri, Skáldskaparmál, ed. Faulkes, p. 63). 56 SkP 1, p. 897. 57 This translation differs from that of SkP 1, p. 897, where the dative is translated as ‘about’, that is, ‘the poem about the warrior’. The parallel adduced there in the second half of the stanza is not analoguous, since we are there dealing with an indirect object.

34

1 Metre and Rhyme

This poem is composed in the metre tvískelft (double-shaken) – a variant of skjálfhenda which employs its distinctive form not only in the third, but also in the first line: the first three syllables are heavy, alliteration falls on the first and third, and the first hending falls on the first or the second syllable. This metre is not known from earlier poetry, and the poet himself comments on its rarity.58 As well he might, since this is the only poem in that metre to have been preserved.59 Tvískelft is not produced by regularizing features that are found occasionally in earlier poetry, as is often the case in Háttalykill. Rather, an existing metre has been used as the basis for a much more demanding metre, which imposed considerable syntactic constraints on the poet.60 In this case, emulation is the most conspicuous feature of the choice of metre: we see the poet proudly displaying his ability to outdo his predecessors in metrical dexterity. This need not in itself indicate a learned approach to tradition, but since Hallar-Steinn in another stanza refers to manuscript illumination, and since his choice of metre is so contrived as to impose considerable syntactical restraints on the poet, it seems likely that his metrical experiments are indeed connected to his learning (on syntactical constraints, see above p. 25; on illumination, see below p. 89). A further stage in the development of a studied approach to the poetic tradition may be exemplified by Bjarni Kolbeinsson’s Jómsvíkingadrápa (beginning of the thirteenth century). Bjarni was bishop of Orkney, and we may thus presuppose a solid education, a point further underlined by his Ovidian allusions.61 Here follows the first half stanza: Engan kveðk at óði órum malma rýri (þó gatk hróðr of hugðan) hljóðs (atferðar prýði).62 I ask no metal’s destroyer [warrior] for silence for our poem. Nonetheless I have composed a poem about the beloved adorner of good behaviour [Vagn].

This metre is called munnvǫrp: there are skothendingar for aðalhendingar in even lines and no hendingar in odd lines. A couplet or even a half stanza in some of the earliest skaldic poetry may follow this pattern, but it was never regularized. In Jómsvíkingadrápa it runs through 45 stanzas.

58 Snorri, Háttatal, ed. Faulkes, p. 81. The reference to the rarity of the metre is found in the last stanza (see SkP, 1, pp. 938–39). 59 Snorri, Háttatal, ed. Faulkes, p. 81. 60 Kuhn, Das Dróttkvætt, pp. 333–34. 61 See Wellendorf, ‘No Need for Mead’. Bjarni alludes to Ars amatoria by saying that he has not been inspired by Óðinn, just as Ovid says that he was not inspired by the muses, and there is an additional allusion in the description of love as war. 62 SkP 1, p. 958.

1.2 Grammatical and Historical Awareness in the Twelfth Century and Beyond

35

With Jómsvíkingadrápa, I would suggest that grammatical analysis has combined with historical awareness. While the metre of Rekstefja may have been chosen merely for metrical display, the metre of Jómsvíkingadrápa probably was not. Irregular hendingar are typical of the earliest skaldic poetry, and Jómsvíkingadrápa treats the Jómsvikings, heroes of old and, not least, heroes of an old kind – proud, pagan warriors. It would thus seem that Bjarni has striven to make his poem reminiscent of the earliest poetry, even if the Jómsvikings were believed to have lived some time after the productive period of such poetry. It is curious to note that also the saga about these Vikings – Jómvíkinga saga – exhibits features that are typical of sagas treating a period more than a hundred years earlier than the Jómvikings, the so-called fornaldarsögur or legendary sagas. Fights with trolls, a brotherhood of warriors, a quest for a woman and ‘social advancement by means of marriage and service to a social superior’ are some of the elements which tie this saga closer to the fornaldarsögur than to kings’ sagas or sagas of Icelanders.63 It would thus appear that there was something about the stories about these warriors that made them appear as if they belonged to an era much earlier than the one in which they were actually thought to have lived and that this has left its imprint on narrative style and metrical structure alike. This is a telling indication of how closely associated poetics and other literary endeavours were at the time. To return now to metrics, there is one crucial difference between the style of the early poets and Bjarni’s rendering of it. To them, it was precisely the lack of regularization that gave rise to the occasional occurrence of the pattern found consistently throughout Jómsvíkingadrápa. Similar observations have been made regarding the lexicon used in the poem, which is characterized by ‘overuse’ of archaic words and expressions.64 In sum, Bjarni and the earliest poets occupied opposite ends of the scale of orderliness: they composed with considerable licence, while he actively looked for rules and recurrent features. This new amalgam is best described by Snorri, writing his Háttatal a few decades after Bjarni composed his drápa. In the following quotation, I have included the first couplet of two stanzas, to illustrate which features Snorri thought typical of the early poets after whom he named the individual metres (Egill’s, Fleinn’s, and Bragi’s metre have here been excluded). The stanzas, however, are Snorri’s own. In the translation, I briefly describe the relevant features within brackets: Nú skal rita þá háttu er fornskáld hafa kveðit ok eru nú settir saman, þótt þeir hafi ort sumt með háttafǫllum, ok eru þessir hættir dróttkvæðir kallaðir í fornum kvæðum, en sumir finnask í lausum vísum, svá sem orti Ragnarr konungr loðbrók með þessum hætti:

63 Torfi Tulinius, The Matter of the North. The Rise of Literary Fiction in Thirteenth-Century Iceland, trans. Randi C. Eldevik (Odense: Odense University Press, 2002), p. 216. 64 Anne Holtsmark, ‘Bjarne Kolbeinsson og hans forfatterskap’, Edda (1937), p. 9.

36

1 Metre and Rhyme

Skýtr at Skǫglar veðri (en skjaldagi haldask) [. . .] Nú skal rita Torf-Einars hátt: Hverr sér jǫfra œgi Jarl fjǫlvitrum betra [. . .] Víða er þat í fornskálda verka er í einni vísu eru ýmsir hættir eða háttafǫll, ok má eigi yrkja eptir því, þó at þat þykki eigi spilla í fornkvæðum.65 Now I shall write down the metres in which the early poets composed, and these are collected here,66 even though they composed some with irregular metre. These metres are called dróttkvætt in old long poems, but some of them are to be found in single stanzas, as king Ragnarr loðbrók composed in this metre: Skýtr at Skǫglar veðri (en skjaldagi haldask) [. . .] Shoots in Skǫgul’s storm [battle] (but shields cannot be held) [. . .] [No hendingar in odd lines and aðalhendingar in even. Onset before the first alliteration in even lines] Now follows Torf-Einarr’s metre: Hverr sér jǫfra œgi Jarl fjǫlvitrum betra [. . .] Who sees a jarl who is better than the very wise rulers’ subduer [ruler above rulers] [. . .] [No hendingar in odd lines, skothendingar in even lines and only one light syllable between them] It is common in the works of the early poets that the same stanza displays several metres or irregular metre, and we may not compose in that manner, even though it is not seen as unbefitting in old poems.

65 Snorri, Háttatal, ed. Faulkes, pp. 24–26. 66 Faulkes translates ‘which are now made into consistent verse-forms’ (Snorri, Edda, trans. Anthony Faulkes (London: Everyman, 1987), p. 198; Snorri, Háttatal, ed. Faulkes, p. 138). I believe, however, that setja saman here retains its ordinary meaning ‘collect, compile’, the logic being that Snorri has collected their metres even though, in fact, they composed in an inconstent manner which does not exactly correspond to the metres which Snorri presents.

1.2 Grammatical and Historical Awareness in the Twelfth Century and Beyond

37

The final words in this quotation (‘and we may not compose in that manner, even though it is not seen as unbefitting in old poems’) reveal that Snorri’s attitude towards the early poets is similar to Donatus’s towards the classics: when the old authorities deviate from prescribed patterns, it is fine, but the student should stick to the rules.67 This is the attitude one might expect from anyone with some grammatical training in the Middle Ages. Augustine, for instance, says that if a boy is guilty of barbarism and defends it with reference to Vergil, he deserves to be whipped.68 Snorri’s description of the metres of the early poets accords with this prescriptive approach. He presents rules, not licence, and he has therefore taken features that were common in some of the earliest poetry, classified them and distributed them over the canonical poets, creating one metre for each.69 He could probably not have quoted the poets themselves, since all existing poetry indicates that their stanzas would have been irregular.70 This passage constitutes by far the fullest explicit commentary in Old Norse literature on the diachronic development of metre. It is no concidence that the features Snorri focused on are hendingar, just as in Jómsvíkingadrápa. Irregular hendingar are indeed the most striking formal difference between some of the earliest poetry and strict dróttkvætt. Another typical early feature is onset – one or more light syllables before alliteration in even lines – which Snorri has regularized in Ragnarr’s metre. Even if he misconstrues the dynamics of early poetry by regularizing it, Snorri clearly shows an awareness of the stylistic differences between the early poets and the type of dróttkvætt that had prevailed for more than two centuries by his time. Snorri’s Háttatal is the culmination of grammatical experimentation with variants of skaldic poetry. The poem also shows how systematic observation was beginning to be combined with historical awareness, and this seems also to be the case slightly earlier in Jómsvíkingadrápa. In Chapter Four, we shall see how similar observations of metrical change were used to create an impression of historical plausibility when poetry was composed in the name of the early poets.

67 This is implicit in Donatus’s Barbarismus, where examples of the vices are taken from Vergil’s Æneid – the most elevated of Latin poems. See, for instance, the first examples in Donat et la tradition de l’enseignement grammatical. Étude sur l’‘Ars Donati’ et sa diffusion (ivᵉ–ixᵉ siècle) et édition critique, ed. Louis Holtz (Paris: CNRS éditions, 1981), p. 653. 68 Contra Faustum 22.25 (PL 42, p. 417). 69 Snorri lists 12 kinds of licence, but these are all of a quite restricted nature in comparison to the fluid style of early poetry and really amount to rules in their own right (Snorri, Háttatal, ed. Faulkes, pp. 7–8). 70 No poetry that is likely to be authentic is known to be by Ragnarr, and none at all by Fleinn. Even to Snorri, they may have been no more than authoritative names from antiquity. For the issue of irregular features in Egill’s poetry, see below pp. 220–32.

38

1 Metre and Rhyme

1.3 Summary In this chapter, we have seen the development of dróttkvætt and dróttkvætt-based metres in several stages, from irregular to regular dróttkvætt during the tenth century, then to various alterations of line length in the eleventh. End-rhyme appears to have had a precocious beginning when Egill was in York in the first half of the tenth century, but seems not to have been adopted by other poets until the eleventh. We also saw a number of innovations which were limited to refrains and enabled the poets to put their individual stamp on their work. The twelfth century saw a vast proliferation of metres and these were based on systematic study of earlier poetry. The craftsman had now become a scholar. Finally, in Jómsvíkingadrápa and Háttatal (c. 1200–1220), we see these observations combined with historical awareness, and this coincides with the period when Icelanders (and Orcadians) first began to write their own history on a grand scale.

2 Diction: Mythology, Wordplay, Metaphor and Tmesis This is the second chapter to deal with long-term developments from the ninth to the thirteenth century, before the book shifts to a more focused view on the twelfth to fourteenth centuries. Developments in diction were if anything even more important than those within metre for the literary flowering of the thirteenth century. Snorri’s Edda in particular would have been inconceivable without them, but they also had a formative impact on other grammatical treatises and indeed on saga writing. Unlike metre, diction did not evolve organically over time. Rather, it would appear that the stylistic register was first severely hemmed in under Óláfr Tryggvason’s reign, 995–1000, and certainly from the ascension of Óláfr Haraldsson in 1015 onwards. In the twelfth century, however, tenth-century stylistics saw a remarkable revival. This difference in the evolution of metre and diction notwithstanding, the decisive moment for later developments came at the same time: namely, when skaldic poetry made contact with the schoolroom in the twelfth century. With diction, just as with metre, external influences did not generally result in imported forms, but rather in a new approach to the local poetic tradition and to the poets who had shaped it. Paradoxically, Christian learning was the crucial impetus for reopening the door to the pagan past. Based on a new semantic subdivision of mythological references, I shall argue that nearly all such references which denoted anything other than persons or martial phenomena were absent from poetry produced in the period c. 995–c. 1120, with a few noteworthy exceptions. Several other features followed the same trajectory, as well. These features are wordplay, extended metaphor and tmesis, and for now it may suffice to note that they are features of diction and that their disappearance together with most types of mythological references probably indicates an interconnection between them in poetic practice. The chapter explores these developments and attempts to analyse the dynamics involved. It begins with an introduction to the features in question and then proceeds to a chronological breakdown in three sections. Finally, I give a brief overview of attitudes to mythology in Latin literature, so as to provide a broad cultural contextualization to the revival of myth in the twelfth century.

2.1 Kennings, Mythology and Poetic Figures In skaldic poetry, the figure of the kenning is ubiquitous, and it may be taken as one of its defining features from the emergence of the genre until at least the middle of the fourteenth century. Ordinary kennings are circumlocutions formed by one baseword and one or several definers. Thus, ‘the spears’ path’ means ‘shield’, which may

https://doi.org/10.1515/9783110643930-003

40

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

be expanded to ‘the spears’ path’s noise’, meaning ‘battle’, and again to ‘the spears’ path’s noise’s Njǫrðr (name of a god)’, meaning ‘warrior, man’, and so on.1 As seen from this example, kennings could have a mythological component (for the present purposes, I define myth as pertaining to the gods and the creatures and objects associated with them). The system of mythological references was rich, so that not only could the name of a god be substituted for a person, as in the kenning above, but we also find kennings like ‘Óðinn’s drink’ for ‘poetry’, ‘Freyja’s tears’ for ‘gold’ and a large number of other kennings based on narratives or attributes. The use of mythological kennings was not consistent over time. Jan de Vries and Bjarne Fidjestøl have noted a decrease in the use of mythological kennings in the period c. 1000–c. 1150 (c. 1200 according to Fidjestøl).2 Unsurprisingly, the beginning of the slump coincides with the formal introduction of Christianity. Less predictably, after c. 1150–1200, mythological kennings experience a revival, although both scholars agree that their use was now more restricted than during the peak in the late tenth century. It should be noted that while I here treat de Vries and Fidjestøl together, Fidjestøl was critical of de Vries. Unlike de Vries, Fidjestøl claims that there was no sharp decrease with the introduction of Christianity, but only a gradual one over a century or so, and no rise in the twelfth century, but only in the thirteenth, and that de Vries’s analysis lacks ‘any solid foundation’.3 It may be noted that Fidjestøl excludes much of the twelfth-century poetry that most clearly betrays the new trends from his analysis, such as, for instance, Íslendingadrápa and Øxarflokkr. This casts some doubt over which scholar has the more solid foundation, and my own analysis will indicate that de Vries was closer to the truth. I would suggest, however, that the dates of the slump should rather be set at c. 995–c. 1120. Both de Vries and Fidjestøl have overlooked a fundamental semantic division in the corpus which is of great importance for a plausible analysis: the skaldic transmission strongly suggests that mythological kennings should not be treated as one group, but as two. One of these remained in continued use after c. 995, whereas the other disappeared almost completely. The first group denotes either individual persons, including kennings like ‘the swords’ Freyr [man]’, ‘treasure’s goddess [woman]’, or battle and martial phenomena, including kennings like ‘Valkyrie’s/Óðinn’s storm/clothes [battle/armour]’ and various kennings for ravens (a bird of the battlefield) involving Óðinn or a Valkyrie. The continued use of such kennings accounts for the fact that

1 Based on the kenning geira stígs gný-Njǫrðr (see Meissner, Die Kenningar, p. 262). 2 Jan de Vries, De skaldenkenningen met mytologischen inhoud (Haarlem: Tjeenk Willink, 1934); Bjarne Fidjestøl, ‘Pagan Beliefs and Christian Impact: The Contribution of Skaldic Studies’, in Viking Revaluations: Viking Society Centenary Symposium 14–15 May 1992, ed. Anthony Faulkes and Richard Perkins (London: Viking Society for Northern research, 1993), pp. 100–20; Fidjestøl, The Dating of Eddic Poetry, ed. Odd Einar Haugen (Hafniæ: Reitzel, 1999), pp. 270–93. 3 Fidjestøl, The Dating of Eddic Poetry, ed. Haugen, p. 293.

2.1 Kennings, Mythology and Poetic Figures

41

mythological kennings do not disappear from de Vries’ and Fidjestøl’s tables, although some poets, notably Sighvatr, appear to have avoided them most of the time.4 The second group includes all other mythological kennings and indeed mythological references of any kind. It comprises kennings like ‘Óðinn’s drink [poetry]’, ‘Freyja’s tears [gold]’, ‘Ymir’s blood [the sea]’, ‘Óðinn’s mistress [the earth]’ and many others. This group all but disappears around 995, and the exceptions are few enough that they will be treated on a case-by-case basis below. I refer to the first group as generic mythological references (references generally, but not always, being tantamount to kennings), since they draw only on the anthropomorphism of the gods and the martial associations of Óðinn and the Valkyries. I call the second group specific mythological references, since they presuppose precise knowledge of a mythological world with its own cosmology, topography, history of events and genealogies. I would also include mythological poems in this group, since they too presuppose a detailed mythological world, even if they are texts rather than references. No poems of this kind can with any degree of certainty be dated to the period of the slump in specific mythological references. Unlike de Vries and Fidjestøl, I will not conduct a statistical investigation, since I believe that the material calls for a qualitative approach, and this is true not only of the slump, but also of the twelfth-century revival. The mythological comeback seems to have been the product of relatively restricted circles of individuals in western Iceland and to some extent in Orkney, and the character and complexity of mythological references in their poetry probably carries more explanatory force than sheer numbers, spread thin over the skaldic corpus (or the corpus of panegyrics alone, as in Fidjestøl’s studies).5 My analysis of the twelfth century will therefore focus on two of the poets who were most important for the developments in that period. Although I avoid a full statistical coverage, the reader should have some impression of just how drastic the slump was and how much it differs from the preceding and following periods. I therefore give here some numbers of kennings belonging to the three most common specific referents. In the following, I discuss a total of 179

4 See Erik Noreen, ‘Studier i fornvästnordisk diktning’, 3 vols. Uppsala universitets årsskrift. Filosofi, språkvetenskap och historiska vetenskaper, 3–5 (Uppsala: Akademiska bokhandeln, 1921–23), 2, pp. 18–30. 5 A similar argument is made in Christopher Abram, ‘The Post-Pagan Mythological Kenning’, in Eddic, Skaldic, and Beyond, ed. Chase, pp. 60–61. The source situation is not ideal for the period c. 1070–c. 1120, which has left us with a smaller corpus than the surrounding periods: in Skj, the poetry between Óspakr Glúmsson and Einarr Skúlason occupies some 47 of the total 1279 pages, c. 3,5 %, whereas this time span occupies some 13 % of the productive period of skaldic poetry, if counted between 850 and 1400; some pages could be added for Einarr’s early poetry (Skj B 1, pp. 376–423). This increases the possibility that a rise in certain features from around 1120 could have had antecedents in poetry that is now lost. Such poetry as has been transmitted from the period c. 1070–c. 1120, however, displays no increase in the relevant features.

42

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

individual kennings, and rather than listing them all in footnotes, I refer the reader to the pages in Meissner’s Die Kenningar where they are found. Most common are mythological kennings for ‘poetry’, 102 in all (I have added two that are present in SkP but not in Skj and therefore not listed by Meissner).6 Of these, sixty-three belong to the period before c. 995, thirty to the period after c. 1120, and ten to the intermediate period.7 Of the ten, five belong to the poet Hofgarða-Refr (active in the 1030s), leaving only five kennings to poetry of other poets of the period. The poets in question are Hallfreðr vandræðaskáld (shortly after 1000) with two kennings and Bjǫrn Hítdœlakappi (1020s), Haraldr harðráði (c. 1043–1044) and Arnórr jarlaskáld (1060s) with one kenning each. Meissner lists forty-nine mythological kennings for ‘gold’.8 Of these, seventeen belong to the period before c. 995, twenty-seven to the period after c. 1120, and five to the intermediate period. One belongs to Gunnlaugr ormstunga (c. 1001), two to Hallvarðr háreksblesi (c. 1029)9 and two to Hofgarða-Refr. One kenning in an anonymous stanza cannot be dated with any degree of certainty, except that it is older than c. 1150.10 Finally, Meissner lists thirty mythological kennings for ‘earth’ or ‘land’.11 Of these, thirteen belong to the period before c. 995, thirteen after c. 1120 and four to the period c. 995–1120. One is found in Eyjolfr dáðaskáld’s Bandadrápa (1000–1012) and one in Óttarr svarti’s Óláfsdrapa sænska (c. 1018).12 Two are found in the same stanza in Þjóðolfr Arnórsson’s Sexstefja (c. 1065), praising Haraldr harðráði.13 If the three groups are combined, ninety-three of the total number belong to the period before c. 995, seventy to the period after c. 1120, and nineteen to the intermediate period. Of the nineteen, seven belong to the poet Hofgarða-Refr, leaving twelve to other poets. The contrast to the period up to c. 995 is particularly striking. This poetry occupies the pages up to page 150 in Skj B 1, whereas the poetry of the period c. 995–c. 1120 occupies pages 150–64 and 178–466. Ninety-three specific mythological kennings are thus found on 150 pages of text, against nineteen on 304. They are thus about ten times more common in poetry up to c. 995 than in poetry of c. 995–c. 1120, and slightly more than fifteen times more common if one discounts the single poet Hofgarða-Refr. These are remarkable numbers indeed, and ones that are not visible without the distinction into generic and specific references. 6 These are Hallfreðr, landherðar lýða líð (landshoulder’s [stone’s] people’s [giants’] drink [poetry]) (SkP 1, p. 430), and Kormákr, haptsœni (gods’ reconciliation [poetry]) (SkP 3, p. 281). 7 Meissner, Die Kenningar, pp. 427–30. 8 Meissner, Die Kenningar, pp. 226–28. 9 SkP 3, pp. 231, 238. 10 SkP 3, pp. 521–22. Finnur Jónsson dates it to the tenth century (Skj B 1, p. 174; A 1, p. 184). The form áar, used in the poem, disappeared around 1150 (Myrvoll, Kronologi i skaldekvæde, p. 328). 11 Meissner, Die Kenningar, p. 87. 12 SkP 1, p. 460; SkP 3, p. 339 (the title is editorial). 13 SkP 2, pp. 114–16.

2.1 Kennings, Mythology and Poetic Figures

43

Furthermore, the numbers do not tell the whole story. The majority of the specific mythological references in the period c. 995–c. 1120 are found in poetry composed under circumstances that in various ways may account for their presence. Some are found during the period of what appears to have been a revival of pagan style under Eiríkr jarl, others are composed in Iceland, far away from the royal court, during the early decades of the eleventh century, and yet others seem connected to the political ambitions of the earls of Orkney. Circumstances such as these underline the need for a qualitative rather than a quantitative approach. Some degree of uncertainty must perhaps remain with regard to the dating of a few of the kennings, especially when the poets are protagonists of sagas of Icelanders. Nonetheless, the numbers for the period of the slump are probably largely correct, since uncertainty in most cases pertains to whether a stanza was composed by a tenthcentury saga protagonist or by the later author of the saga, that is, either before or after c. 995–c. 1120 (see pp. 194–218, in particular 218). As noted above, the period after c. 1120 is best studied with a view to individual poets, but suffice it to say that in this period, many poets again display a density of specific mythological references that can only be matched by Hofgarða-Refr during the intermediate period. I turn now to a description of various features of the changes to poetic diction, but before doing so, it is important to appreciate the semantic dynamics of what may be called the ‘basic’ kenning system. If one is acquainted with the common kenning types, most kennings can be decoded step by step, depending on their length. The main difficulty then becomes that the poets used a staggering number of synonyms to avoid repetition of the same kennings. Many kenning types, however, are relatively easy to identify based on their structure, and it is therefore often possible to deduce what a word or a kenning ought to mean without actually knowing all poetic synonyms.14 Thus, for instance, if a word for a tree forms part of a kenning, it is likely to be the base-word in a kenning for a man or a woman, and the rest of the kenning can be interpreted in light of this. This is, on the whole, a valid description of the degree of semantic complexity one is likely to encounter in eleventh-century poetry. In the periods before and after, however, some poets opted for producing semantic and structural complexity beyond ordinary kenning formation. This they achieved in four main ways, namely by the use of specific mythological references, wordplay, extended metaphors and tmesis. With regard to the subdivision of mythological references into generic and specific ones, these differ from one another not only in their chronological distribution, but also in their semantic dynamics. Specific references tend to demand much more skill and knowledge from the audience than generic ones. Generic references (the sword’s Freyr [warrior], Gunnr’s [a Valkyrie’s] storm [battle], etc.)

14 See Elena A. Gurevich, ‘The System of Kennings’, Nordica Bergensia, 3 (1994), pp. 148, 150–52.

44

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

typically do not refer to individual narratives and demand nothing more than a general knowledge of kennings.15 Specific references, by contrast, draw on precise knowledge of the myths and are often realized in technically complicated ways. We shall see many examples of specific references, but I present one here to illustrate the stakes involved: Eyvindr skáldaspillir has the kenning Fenris varra sparri (Fenrir’s lips’ spar [sword]).16 The motif of the sword in the mouth of the mythological wolf Fenrir occurs only twice in the skaldic corpus.17 Kennings like these cannot be decoded by recourse to skaldic convention, but demand detailed mythological knowledge. Most mythical motifs used in kennings occur more often than twice, but some are found only once, and we shall see that mythological kennings for poetry, for instance, are often quite complex in addition to demanding a knowledge that generic references do not. Generic references thus on the whole belong to what I above referred to as the ‘basic’ kenning system, whereas specific references mostly do not. Another feature adding semantic complexity was wordplay, which displays the same chronological distribution as specific mythological references. Wordplay beyond ordinary kenning formation comes in two variants, namely ofljóst and wordplay with proper names. The term ofljóst (too clear) was used by thirteenth-century grammarians to designate a particular kind of wordplay, and they seem to have defined this figure as an expression that demands two rounds of decoding by means of homonymy or paronymy (phonetically identical or similar words).18 It was, presumably, ‘too clear’ because the first meaning arrived at, the one which is ‘clear’, is

15 There are some partial exceptions to this general rule, including kennings like Heðins kván (Heðinn’s woman [Hildr = Valkyrie]), which mentions Heðinn, whose wife Hildr has the name of a Valkyrie and figures in the story about the Hjaðningavíg (Snorri, Skáldskaparmál, ed. Faulkes, p. 72). Ullar kjóll (Ullr’s ship [shield]) is similar, though we do not know the underlying story (it is probably somehow related to Saxo’s information that Ullr could cross waters on a bone and thus using an unconventional means of transportation (Saxo Grammaticus, Gesta Danorum, ed. FriisJensen and trans. Zeeberg, 1, p. 210)). Some narrative, martial kennings have a non-mythological background. Thus, in Þjóðólfr’s Sexstefja, there is a reference to the arrows of the Saami king Gusi (r) (SkP 2, pp. 127–28). It should be noted, however, that kennings like these are sparsely used in the eleventh century. Examples include Gizurr gullbrárskáld, foster-father of Hofgarða-Refr, Arnórr and Þjóðolfr (LP, s.vv. Gusi, Heðinn, kjóll, Ullr). 16 SkP 1, pp. 223–24. 17 The other occurrence, in Einarr Skúlason’s Geisli, will be discussed below. These kennings are extreme examples of the exclusivity already inherent to skaldic poetry; on this topic, see John Lindow, ‘Riddles, Kennings, and the Complexity of Skaldic Poetry’, Scandinavian Studies, 47 (1975), pp. 311–27. One or possibly two occurrences of this motif are also found in rímur (‘Hrólfs rímur Gautrekssonar’, ed. Jóhanna Katrín Friðriksdóttir and Haukur Þorgeirsson. Gripla, 26 (2015), p. 116). 18 Snorri, Skáldskaparmál, ed. Faulkes, p.109; Snorri, Háttatal, ed. Faulkes, pp. 12–14; Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, pp. 66, 89. In modern research, ofljóst is defined with varying degrees of inclusiveness (see, for instance, Frank, Old Norse Court Poetry, p. 69; SkP 1, pp. lxxxiii–lxxxiv. For a more inclusive definition, see Clunies Ross, A history of Old Norse Poetry and Poetics, p. 111).

2.1 Kennings, Mythology and Poetic Figures

45

deceptive.19 It may or may not involve a kenning. An example involving a kenning is found in Grettis saga. When the saga hero Grettir reports how he swam under a waterfall, he finishes the stanza by saying: Heldr kom á herðar skáldi hǫrð fjón Braga kvónar.20 Bragi’s wife’s rather powerful enmity came down upon the shoulders of the poet.

The wife of the god Bragi is Iðunn; Iðunnar in the genitive. Thus, the first decoding of the kenning Braga kvónar (Bragi’s wife’s) gives Iðunnar (Iðunn’s), that is, ‘the powerful enmity of Iðunn’. This makes little sense in the context of swimming under a waterfall, but the definite form of the genitive of the word iða (whirling stream) is also iðunnar, and that is the intended referent. The correct reading, then, is: The rather powerful enmity of the whirling stream came down upon the shoulders of the poet.

A stanza from Egils saga may serve as an example of ofljóst not involving a kenning. The stanza is attributed to the aged Egill, who is complaining about his cold feet: Eigum ekkjur allkaldar tvær, en þær konur þurfu blossa.21 I have two very cold widows [ekkjur (widows) > hælar (widows) > hælar (heels)], and those women need heat.

As these examples show, ofljóst is as a rule not really ambiguous, since only one of the readings makes sense. One exception to this general rule will be treated below. The other kind of wordplay involves proper names. Its use is presumably attributable to the fact that the canonical kenning types could not be used to refer to most proper names.22 Such wordplay is attested from the beginning of the skaldic tradition through Bragi’s (ninth century) foglhildr (bird-Hildr) for Svanhildr (swan-Hildr).23 Apart from the simple substitution of a synonym, a kenning could also be combined

19 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 172. 20 Skj B 2, p. 473; A 2, p. 442; Grettis saga, ed. Guðni Jónsson. Íslenzk fornrit, 7 (Rekjavík: Hið íslenzka fornritafélag, 1936), p. 217. 21 Skj B 1, p. 52–53; A 1, p. 59–60; Two Versions of ‘Snorra Edda’, ed. Anthony Faulkes, 2 vols (Reykjavík: Stofnun Árna Magnússonar, 1977–79), 1, p. 375; Egils saga Skalla-Grímssonar, ed. Sigurður Nordal. Íslenzk fornrit, 2 (Reykjavík: Hið íslenzka fornritafélag, 1933), p. 296. 22 I exclude here the type ‘X’s son’, which denotes the person, not the name. 23 Skj B 1, p. 2; A 1, p. 2.

46

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

with an element of the name. Thus, Eyvindr skáldaspillir has álhimins lendingar (eel-sky’s [ice’s] -landers [Icelanders]).24 We have now considered specific mythological references and wordplay as methods of adding semantic complexity. Another figure that was used for the same purpose with an identical chronological distribution was what is generally translated as extended metaphor (nýgerving).25 In Snorri’s definition in Háttatal: Þat eru nýgjǫrvingar at kalla sverðit orm ok kenna rétt, en slíðrnar gǫtur hans, en fetlana ok umgjǫrð hams hans. Þat heldr til ormsins náttúru at hann skríðr ór hamsi svá at hann skríðr mjǫk til vatns. Hér er svá sett nýgjǫrving at hann ferr leita blóðs bekkjar at þar er hann skríðr hugar stígu, þat er brjóst manna. Þá þykkja nýgjǫrvningar vel kveðnar ef þat mál er upp er tekit haldi um alla vísulengð. En ef sverð er ormr kallaðr, en síðan fiskr eða vǫndr eða annan veg breytt, þat kalla menn nykrat, ok þykkir þat spilla.26 It is extended metaphor to call a sword a snake and use an appropriate determinant, and the scabbard its path, and the straps and fittings its slough. It is in the nature of the snake that it glides out of its slough so that it glides towards water. Here, the metaphor is extended in such a way that it [the snake] goes to seek the brook of blood in the place where it glides on the path of the mind, that is, the breast of men. Extended metaphors are thought to be well performed if the chosen subject is retained throughout the stanza. But if a sword is called a snake, and then a fish or a wand, or changed in some other way, that is called a monstrosity (nykrat), and it is seen as a flaw.

In short, if the metaphor of a snake is chosen, it should remain a snake and its attributes and behaviour should be those of a snake. The extended metaphor thus expands on the imagery inherent in a kenning and thereby provides another possibility of semantic play beyond the relatively restricted number of kenning types.27 The ‘monstrosity’ (nykrat) mentioned in the quotation is essentially the same figure,

24 SkP 1, p. 234–35. 25 While I follow the standard translation here, it should be noted that Snorri may have used the terms nýgerving and nykrat somewhat differently to modern scholarship. He may have thought that the nýgerving itself could either be good or nykruð and that both figures today referred to as nýgerving and nykrat, respectively, were nýgervingar. Snorri seems also to have used nýgerving to denote other phenomena. See Snorri, Skáldskaparmál, ed. Faulkes, pp. 41, 74, 108. 26 Snorri, Háttatal, ed. Faulkes, p. 7. 27 Edith Marold, ‘Nýgerving und nykrat’, North-Western European Language Evolution, 21/22 (1993), pp. 283–302, analyses extended metaphor as restricted to kennings only, not including, for instance, verbs. It should be noted, however, that Snorri implicitly includes the verb ‘glides’ (skríðr), which he uses for the snake, as well as for the sword cutting the breast. Furthermore, the stanzas that will be treated below show that extended metaphor in practice involved the imagery at large. It is true that Snorri, apart from his use of the word skríða, restricts his discussions of extended metaphor to kennings, but this is probably a consequence of his focus on kennings, rather than of an active exclusion of other elements of the stanzas. In skaldic practice, extended metaphor was probably a plastic technique that could involve kenning variation as much as other syntactic components. This is what Hallvard Lie calls setningsmetafor (clause metaphor) (Natur og unatur i skaldekunsten (Oslo: Aschehoug, 1957), pp. 49, 60).

2.1 Kennings, Mythology and Poetic Figures

47

but not sustained throughout a stanza. To Snorri, such inconsistency was a flaw, but to earlier poets it was not. I return to this issue on pp. 52–53. One example of a stanza using extended metaphor is Sturla Þórðarson, Hákonarkviða 8 (after 1263)28: Svalg hvert hús heitum munni viðar hundr Verma bygðar, ok svipkárr selju rakki of garðshlið grenjandi fór.29 The wood’s dog [fire] swallowed every house of the district of the Vermir with its hot mouth, and the willow’s violent hound [fire] rushed howling through the yard-gate.

The image of the violent hound is here retained throughout the stanza, and with its hot mouth, it also interplays with the referent of the kennings, namely fire.30 The final feature of diction to be considered does not add layers of meaning, as the other three do. Rather, it simply serves as an impediment to understanding. This is the figure of tmesis, meaning the splitting of a compound word, as in ó- fyr skǫmmu for fyr óskǫmmu (long ago). In his 1933 article ‘Kenningers led-omstilling og tmesis’, Finnur Jónsson list all examples he can find of tmesis.31 His main objective was to prove to his perennial detractor Ernst Albin Kock, whom he does not deign to mention by name, that the poets sometimes made use of tmesis. It is presumably for this reason that he focuses on its continuous use from the earliest period into the fourteenth century, without pointing to chronological gaps that could potentially raise any doubts in the reader’s mind.32 In fact, however, the material he presents has a conspicuous empty space in the period c. 995–c. 1120, and the distribution thus corresponds to that of complex mythological references, wordplay and extended metaphor. Ernst Albin Kock’s Notationes norroenae constitute an attack in more than 3,000 paragraphs on Finnur Jónsson’s edition of skaldic poetry and it belongs to the more

28 For the date, see SkP, 2, p. 699. 29 SkP, 2, p. 705. 30 Guðmundur Finnbogason notes that interplay between the referential and the verbal level of meaning is common in extended metaphor (‘Nogle bemærkninger om skjaldedigtningens kenningar’, Acta philologica Scandinavica, 9 (1934–35), pp. 69–75; see also Einar Ól. Sveinsson, ‘Dróttkvæða þáttur’, Skírnir, 121 (1947), pp. 28–32). 31 Arkiv för nordisk filologi, 49, pp. 1–23. 32 ‘Kenningers led-omstilling og tmesis’, pp. 9, 23. Finnur Jónsson mentions that some of the major poets, such as Arnórr and Þjóðolfr Arnórsson, do not use tmesis, but he does not connect this observation to chronology.

48

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

colourful portions of Old Norse scholarship. One of Kock’s most important tenets was that of a general opposition to intricate features, including tmesis.33 Although he sometimes disregarded the complicated nature of skaldic poetry, some of his critique is well founded and has been generally accepted, and the new edition is more restrictive with positing tmesis than was Finnur Jónsson. This has some consequences for the following analysis. Nonetheless, scholars agree that the phenomenon was real, and it may indeed have been more common than we think, since in one instance, Snorri insists on a reading with tmesis even when modern scholars, including Finnur Jónsson himself, find it unnecessary.34 Furthermore, many of Finnur Jónsson’s examples are convincing, and his tendency to see tmesis more often than other scholars is an asset in this context, since this means that the chronological gap in the material which he collected is all the more likely to be real. Finnur Jónsson makes a distinction between kennings with rearranged order of the elements (‘kenningers led-omstilling’) and tmesis. Coming up with a good English designation for the first phenomenon is difficult, since Finnur Jónsson’s term would conveniently be translated as ‘inverted kennings’, but SkP uses that label for a class of kennings where no inversion takes place, and the editors do not appear to have a separate term for the phenomenon Finnur Jónsson was describing.35 No matter the terminology, however, I believe that while Finnur Jónsson’s distinction may be useful in some respects, it is also potentially misleading, since we are dealing with two different realizations of tmesis: one taking place in kennings which contain compound words, the other in compound words that are not part of a kenning. There is one circumstantial difference between the two, namely that the removed element of the compound in the case of kennings is reattached to the third component of the kenning, whereas it has nothing to attach to in the case of tmesis outside of kennings (see examples below). Nonetheless, the procedure of splitting up (the literal meaning of tmesis) and rearranging the elements of a compound is similar, and it naturally enough occurs only in kennings containing compounds, rather than the more common genitival constructions. I believe, therefore, that the splitting up of compounds should be analysed as tmesis, irrespective of whether the compound is found in a kenning or not, or if the elements have been re-attached to some other word or not. Here follow some examples of the phenomenon.

33 See, for instance, Kock, Notationes Norroenae, § 4. 34 Snorri says that Víga-Glúmr’s ‘með veðrstǫfum Viðris | vandar mér til landa’ should yield the kenning Viðris veðrvandar stǫfum (Viðrir’s (Óðinn’s) weather- [battle-] staff’s [sword’s] staves [warriors]) (Snorri, Skáldskaparmál, ed. Faulkes, p. 74). This is one of the rare instances when Finnur Jónsson rejects Snorri’s reading (LP, s.v. vǫndr). For a recent discussion of tmesis, see Russell Poole, ‘Scholars and Skalds. The Northward Diffusion of Carolingian Poetic Fashions’, Gripla, 24 (2013), pp. 27–35. 35 SkP 1, p. cxlii.

2.2 The Early Period (up to c. 995)

49

The simple example given above is found in stanza 2 of Þjóðolfr ór Hvini’s Haustlǫng: ó- fyr -skǫmmu which must be read as fyr óskǫmmu (long ago).36 A somewhat more demanding example is found in stanza 10 of the same poem: ‘Þá vas Iðmeð jǫtnum | -unnr nýkomin sunnan’ (then Iðunnr had recently come from the south to the giants).37A famous example of tmesis within a kenning is found in a stanza by Egill Skalla-Grímsson: dalmiskunn fiska (fishes’ valley-mercy), which is meaningless within the kenning system. Dalmiskunn must be split and dal- be connected to -fiska, giving dalfiska miskunn (valley-fishes’ [snakes’] mercy [summer, which is kind to cold-blooded animals]).38 This concludes the description of the features discussed in this chapter: specific mythological references, wordplay, extended metaphor and tmesis. They all either add semantic complexity beyond basic kenning formation or in some other way impede understanding, and they all display a similar chronological distribution. On both of these accounts, they differ from generic mythological references, which will therefore be left out of the study. Their shared functions and distribution make it likely that they were seen as somehow connected to each other in poetic practice, and this question will be explored in the section on the intermediate period below. I turn now to the use of these features in three periods, which I call Early (up to c. 995), Intermediate (c. 995–c. 1120) and Late (c. 1120–c. 1300).

2.2 The Early Period (up to c. 995) Here follows an overview of the uses of these features in the period up to c. 995, beginning with wordplay. The convergence of specific mythological references and other features will be noted along the way. Many scholars have tried to find hitherto unidentified instances of ofljóst and other wordplay in skaldic poetry. Because of the ambiguous nature of wordplay, however, it is difficult to evaluate many of these attempts.39 Ofljóst that can be identified

36 This reading seems an uncontroversial reconstruction based on the variants (Skj B 1, p. 14; A 1, p. 16; cf. SkP 3, p. 434, though note the lack of hyphens and mention of the word tmesis there). 37 SkP 3, p. 446. 38 This stanza is discussed below pp. 220–21. 39 Roberta Frank observes that the kenning type ‘the stone’s goddess [woman]’ is used remarkably often in Kormákr’s poetry about Steingerðr. The first element in her name, stein-, means ‘stone’, and the second, Gerðr, is the name of a mythical person straddling the categories of giantess and goddess, and one which is often equivalent to the name of a goddess in kennings. Frank therefore suggests that Kormákr is playing on Steingerðr’s name, and this seems plausible (‘Onomastic Play in Kormákr’s Verse: The Name Steingerðr’, Mediaeval Scandinavia, 3 (1970), 7–34). A very elegant suggestion of ofljóst, with the added benefit of having some support in the adjacent prose, is that of Jón Helgason (‘Ek bar sauð’, Acta philologica Scandinavica, 23 (1957), pp. 94–96). Its degree of sophistication is such, however, that some doubt might remain as to whether it can realistically have

50

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

with a high degree of plausibility is relatively rare. There is a small group before c. 995. The earliest instance may be Þjóðolfr’s Ynglingatal (late ninth–early tenth century) stanza 18: lagar hjarta (the lake’s heart [stone, ON steinn > place name Steinn]).40 It should be noted that the semantic distance between ‘stone’ and the toponym Steinn, presumably drawn from some prominent stone or rock, is not great, and it may be doubtful whether this reference was intended to demand a second round of decoding at the time of composition. In the tenth century there are a few more, relatively simple instances of ofljóst, such as the example by Egill quoted above and Kormákr, lausavísa 26: híðbyggvir (hide-dweller [bear, ON bjǫrn, but here bersi (male bear) > Bersi]).41 Towards the end of the early period, Eilífr Goðrúnarson produced a more demanding one: Þórsdrápa 2 has gammleið (vulture-path [air, ON lopt > alternative name of the god Loki in the accusative Lopt]).42 In SkP, the editors present the new suggestion that there is ofljóst also in stanza 18, þrǫng (tight spot) being used for þjalfi (enclosure), which is also the name of Þórr’s companion in the poem.43 Previous editors have supposed that the word þrǫng (in the genitive þrǫngvar) was an unattested form of a name for Freyja, and that the words ‘þrǫngvar langvinr’ (þrǫng’s old friend) mean ‘Freyja’s friend’ and refer to Þórr. Þórr is not, however, presented as Freyja’s friend in any source, whereas he and Þjalfi are companions in Þórsdrápa. The new interpretation is thus less speculative than the old one (this stanza is discussed below pp. 256–57). I would also draw attention to a factor not noted by the editors: Loki is mentioned by the name Loptr in stanza 1, which is the name to be decoded through ofljóst in stanza 2. Þjalfi is mentioned by name in stanzas 10, 11 and 22, and this may serve as a similar hint from the poet that this is the name one should look for in stanza 18. For these reasons, it seems likely that the new interpretation is right. In both cases of ofljóst in Þórsdrápa, wordplay and mythology converge. In ofljóst, the second meaning usually cancels the first one out. A curious exception, however, is found in a kenning for Hákon jarl in Vellekla, where he is designated as brúna grundar silkis síma geymir (the eyebrows’ ground’s [forehead’s]

been transmitted orally for more than two centuries. With regard to the suggestions of Kari Ellen Gade and Else Mundal, it is difficult to make an assessment (Kari Ellen Gade, ‘Penile Puns: Personal Names and Phallic Symbols in Skaldic Poetry’, Essays in Medieval Studies, 6 (1989), pp. 57–67; Else Mundal, ‘Heilagmann som sa sex. Lausavise nr. 7 etter Óláfr Haraldsson’, in Fjǫld veit hon frœða. Utvalde arbeid av Else Mundal, ed. Odd Einar Haugen, Bernt Øyvind Thorvaldsen, and Jonas Wellendorf (Oslo: Novus, 2012 [1984]), pp. 97–116). The possible presence of ofljóst in Glymdrápa 2 has been suggested, but poses formal problems (SkP 1, pp. 78, 81). 40 SkP 1, p. 40. 41 Skj B 1, p. 75; A 1, p. 84. Another instance is found in Hallfreðr, see SkP 1, pp. 419–21. 42 SkP 3, p. 77. 43 SkP 3, pp. 115–16.

2.2 The Early Period (up to c. 995)

51

silk laces’ guardian [Hákon]).44 This kenning makes sense as it stands, and it contains one of the few references to the head-bands of rulers.45 Such laces were called hlǫð (sing. hlað), and if the gen. pl. silkis síma is replaced with gen. pl. hlaða, it coincides with the genitive of Hákon’s seat of residence, Hlaðir.46 The first reading means ‘ruler’, whereas the second specifies the identity of the ruler as Hákon jarl, Hlaða geymir (the guardian of Hlaðir). It is noteworthy that this exception to the general rule that the second reading cancels the first one out centres on the person of Hákon jarl. As we shall see presently, poetic inventiveness appears to have thrived in the mileu around Hákon, the last great pagan ruler according to the sagas. Other wordplay than ofljóst is also attested in the early period. Among poets who used wordplay but were not connected to Hákon, the most conspicuous one may be Egill, who repeatedly conceals the name of his friend Arinbjǫrn (fire-placebear). He calls him, for instance, grjótbjǫrn (grovel-bear)47 and birkis ótta bjóð-bjǫrn (the birch’s terror’s [fire’s] table-[fire-place-]bear [Arinbjǫrn]).48

44 SkP 1, pp. 299–300; Heimskringla, ed. Bjarni Aðalbjarnarson, 1, p. 241. 45 The corpus contains some elusive references to such head-bands, though it may be noted that silk is mostly associated with women in skaldic poetry (SkP 1, p. 300; LP, p. 496). In Kormákr’s Sigurðardrápa, head-bands are also connected to the earls of Hlaðir (SkP 3, p. 277–78). For prose references to men wearing hlǫð made of silk, see, for instance, Heimskringla, ed. Bjarni Aðalbjarnarson, 3, p. 318 (‘strauk af hǫfði honum silkihúfu, hlǫðum búna’); Brennu-Njáls saga, ed. Einar Ól. Sveinsson. Íslenzk fornrit, 12 (Reykjavík: Hið íslenzka fornritafélag, 1954), p. 304 (silkihlað). 46 This case of ofljóst was identified by Magnus Olsen, ‘Eldste forekomst av navnet Hlaðir’, Maal og Minne (1942), pp. 154–56. His interpretation is not mentioned in SkP (SkP 1, pp. 297–98). Another instance of ofljóst involving Hákon has been rejected in SkP. In a stanza by Eilífr Goðrúnarson, earlier editors have followed the U manuscript and read mæran kon (glorious descendant (acc.)), giving háan kon or perhaps simply há-kon (high/elevated descendant), that is, Hákon (Skj B 1, p. 139; A 1, p. 148; Kock, Notationes Norroenae, § 249; Snorri, Skáldskaparmál, ed. Faulkes, pp. 13 (36), 163, 360). The editors of Eilífr’s stanza in SkP argue convincingly that this is unlikely to be the right reading even in U (SkP 3, pp. 124–25). Still, some tantalizing hints remain that a degree of ambiguity may be intended. In a stanza attributed to Queen Gunnhildr, há [. . .] kon is to be taken as Hákon, though in this instance the poet uses tmesis only and not wordplay (SkP 1, pp. 150–52; Ágrip af Nóregskonunga sǫgum. Fagrskinna – Nóregs konunga tal, ed. Bjarni Einarsson, p. 75). Wordplay involving konr is found in Rígsþula in the play on konr ungr/konungr (Norrœn fornkvæði, ed. Bugge, p. 148). 47 Skj B 1, p. 40; A 1, p. 46. 48 Skj B 1, p. 40; A 1, p. 46. Another circumlocution for Arinbjǫrn is arnstalls sjǫtulbjǫrn (the eagleseat’s [crag’s, stone’s] seat- [stone- (tautology (?)), i.e. the stone of the fire-place] -bear [Arinbjǫrn]) (Skj B 1, p. 48; A 1, p. 55). The interpretation remains somewhat tentative (see Egils saga Skalla-Grímssonar, ed. Sigurður Nordal, p. 200; Meissner, Die Kenningar, p. 431), but the structure (kenning + bjǫrn) is clear. Though some stanzas in Egils saga were obviously composed for the saga, this stanza has much to commend it as a composition of the tenth century, particularly with regard to the hending structure (see pp. 229–31). There is at least one more instance of wordplay in poetry attributed to Egill (see B 1, p. 45 (14.7); A 1, p. 52; cf. Egils saga Skalla-Grímssonar, ed. Sigurður Nordal, pp. 148–49). Jón Helgason’s suggestion of ofljóst in stanza 42 of Egils saga was mentioned above (p. 49 n. 39).

52

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

The figure of extended metaphor was in use in the early period, notably in Þjóðolfr’s Ynglingatal and other poetry not composed in dróttkvætt, but Snorri’s ideal of consistency was often not upheld. Bergsveinn Birgisson argues that constrasting metaphors were put to use in the early period, whereas Edith Marold sees early use as consistent within single stanzas.49 In order to uphold that claim, however, Marold makes exception for refrains, metonymy, mythological kennings and abstractions.50 These exceptions are many and important, and arguably they amount to pushing the evidence in the desired direction.51 It would in fact appear that poets did both: they extended the imagery inherent in kennings, but often shifted freely when another kenning offered other possibilities. Thus, for instance, Eyvindr skáldaspillir, Hákonarmál 7: Brunnu beneldar í blóðgum undum [. . .] svarraði sárgymir á sverða nesi [. . .]52 Injury-fires [swords] burned in bloody wounds [. . .] the gash-sea roared on the swords’ headland [blades] [. . .]

Here, the opposition of fire and water adds to the martial tenor of the stanza. At times during the early period, poets seem to have aimed at engaging the senses through such colliding images, whereas at other times they chose to retain one image.53 These

Additional examples of wordplay that may belong to the early period include Eyvindr skáldaspillir (SkP 1, p. 234) and Gísli Súrsson (Skj B 1, p. 97 (8.2); A 1, p. 102). One example by Óláfr Haraldsson is likely to be spurious (see SkP 1, pp. 524–25). 49 Bergsveinn Birgisson, ‘Inn i skaldens sinn. Kognitive, estetiske og historiske skatter i den norrøne skaldediktningen’ (PhD diss., University of Bergen, 2007); Marold, ‘Nýgerving und Nykrat’, pp. 291–97. 50 ‘Nýgerving und Nykrat’, pp. 294, 297. 51 SkP 1, pp. lxxx–lxxxi, supports Marold’s claim. See further p. 46 n. 27. 52 SkP 1, p. 182. 53 Hallvard Lie, ‘Skaldestil-studier’, Maal og Minne (1952), pp. 36, 43–51; idem, Natur og unatur, pp. 65–66, 78–80, 88–94, 104–05; Bergsveinn Birgisson, ‘Inn i skaldens sinn’. Similar observations are made by Einar Ól. Sveinsson, ‘Dróttkvæða þáttur’, p. 30; Gabriel Turville-Petre, Scaldic Poetry (Oxford: Clarendon, 1976), p. liii; Margaret Clunies Ross, ‘Skáldskaparmál’. Snorri Sturluson’s ‘Ars Poetica’ and Medieval Theories of Language (Odense: Odense University Press, 1987), p. 77; Anthony Faulkes, Poetical Inspiration in Old Norse and Old English Poetry (London: Viking Society for Northern Research, 1997), pp. 24–25; Frank, Old Norse Court Poetry, pp. 46, 157–58; Marold, ‘Nýgerving und Nykrat’, pp. 291–97. Hallvard Lie identifies one style that he calls the nýgerving-style in the early period. He does not, however, draw attention to the fact that these metaphors are generally inconsistent if compared to Snorri’s description and later relevant poetry (Natur og unatur, pp. 66–78, 82–87, 105; the stanzas from Grettir’s Ævikviða on pp. 60–61 are in all likelihood spurious and belong to the late period; see below pp. 266–68).

2.2 The Early Period (up to c. 995)

53

shifts, then, were not just a product of carelessness, as Snorri implies. Indeed, Snorri can hardly have been unaware of this, since in Skáldskaparmál he includes several stanzas that directly contradict his rule. Thus, for instance, his favourite poet Einarr Skúlason in one stanza uses the base-words brandr (fire), sáð (seed) and dript (snowdrift), all in kennings for ‘gold’.54 It would seem that Snorri here establishes a rule which is predicated on the normative mindset of the grammarian rather than on the skaldic tradition. This is not to say that poets would not occasionally sustain the same image for a given referent throughout a stanza, but that they did so only when they found it expedient. Specific mythological references are richly attested in the early period. As mentioned above, the specific nature of mythological poems makes it relevant to consider them alongside specific mythological references, and they also contain many of these. There are a few poems on mythological themes (Bragi’s Ragnarsdrápa, Þjóðolfr’s Haustlǫng, Eilífr’s Þórsdrápa, Ulfr’s Húsdrápa, fragments). As a rule, the references in these poems are so terse, complex and specific that the recipient has to already know the narrative to be able to extract it. Also in poems that were not on mythological themes, specific mythological references often figure prominently. The five half stanzas which open Einarr skálaglamm’s Vellekla, for instance, constitute a display of how to say ‘poetry’ through recondite references to Óðinn and other mythological entities (see below pp. 149–59). The fifth stanza may serve as an example: Hljóta munk, né hlítir, hertýs, of þat frýju, fyr ǫrþeysi at ausa austr víngnóðar flausta.55 I shall pump the battle-god’s [Óðinn’s] wine-ship’s [barrel’s] bilge water [wine, poetry] for the ships’ swift-mover [captain]. There is no need for incitement.

The kenning ‘the battle-god’s [Óðinn’s] wine-ship’s [barrel’s] bilge water’ is structurally complex and furthermore demands specific knowledge of Óðinn’s drink as poetry, thus challenging the listeners not only to decode the metaphorical language of the stanza, but also to display their acquaintance with traditional lore. Towards the end of the early period, some poets would combine the features of wordplay, extended metaphor and specific mythological references. The three thus covary not only over time, but they also cluster within individual poems.

54 Snorri, Skáldskaparmál, ed. Faulkes, p. 44 (147). Other clear examples include, giving only base-word variation: [shield]: brú (bridge), kleif (cliff), fat (clothing) (Snorri, Skáldskaparmál, ed. Faulkes, p. 30 (92)); [giant]: gylðir (wolf), vagn (whale) (Snorri, Skáldskaparmál, ed. Faulkes, p. 31 (95)); [wave]: brúðr (bride), hlíð (slope) (Snorri, Skáldskaparmál, ed. Faulkes, p. 38 (133)); [men]: bjarkar (birches), miðjungar (giants) (Snorri, Skáldskaparmál, ed. Faulkes, p. 67 (223)). 55 SkP 1, p. 289. I retain the manuscript reading hlítir (cf. Snorri, Skáldskaparmál, ed. Faulkes, p. 312).

54

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

Eilífr Goðrúnarson displays staggering mythological complexity in his Þórsdrápa and, as we saw above, he also employs demanding wordplay. The opening stanzas of Vellekla have extended metaphors on theme of the sea and ships interplaying with the mythological perception of poetry as a liquid (note the words ‘pump’, ‘bilge water’ and ‘ship’ above). The most noteworthy example, where myth, wordplay and extended metaphor are combined, is Hallfreðr vandræðaskáld’s Hákonardrápa. The name of the poem has been given by modern editors, based on a description in Hallfreðar saga, where Hallfreðr presents a drápa to Hákon jarl (ruler of Norway c. 975–995).56 There are, however, some indications that the poem was composed later, after the year 1000, and since chronological discribution is important here, I shall briefly rehearse these. The poem is certainly directed to an earl, since it contains two clear references to an earl rather than a king (grams rúni ‘prince’s confidant’ and konungs spjalli ‘king’s friend’).57 Hallfreðr is listed as Hákon’s son’s Eiríkr’s, but not as Hákon’s poet in both versions of Skáldatal.58 In light of the openly pagan imagery of Hákonardrápa, it is easy to understand why a saga author might suggest that Hallfreðr composed such poetry before his conversion. Moving his pagan poetry in the other direction, from before to after his conversion, seems an unlikely thing to do from a Christian, retrospective viewpoint. These considerations suggest that Skáldatal has it right and that the saga author has attempted to construct a linear religious development. True, the saga portrays him as a lukewarm and somewhat torn Christian, but he does not revert to paganism. Furthermore, the opening of a kvæði to Eiríkr is quoted in Hallfreðar saga, and it cannot be ruled out that this is the beginning of Hákonardrápa, since we do not have the typical bid for a hearing in any of the other preserved stanzas.59 At least, the couplet shows that poetry by Hallfreðr for Eiríkr was known, whereas this cannot be demonstrated in the case of Hákon, and Skáldatal suggests that no such poetry was known. It thus seems plausible that the poem was presented to Eiríkr rather than to Hákon. In order to avoid confusion, however, I follow previous editorial practice and continue to refer to the poem as Hákonardrápa. I analyse the poem here as belonging to the early period, but I also mention it in the following

56 Vatnsdœla saga, ed. Einar Ól. Sveinsson. Íslenzk fornrit, 8 (Reykjavík: Hið íslenzka fornritafélag, 1939), p. 151. 57 SkP 3, pp. 219, 223. On Hákonardrápa, see Bjarne Fidjestøl, Det norrøne fyrstediktet, pp. 102–106. The poets around Hákon have been understudied as a group, although the commentaries on individual poems now edited in SkP are valuable contributions. The main published works are Folke Ström, ‘Poetry as an Instrument of propaganda. Jarl Hákon and his Poets’, in Speculum Norroenum: Norse Studies in Memory of Gabriel Turville-Petre, ed. Ursula Dronke et al. (Odense: Odense University Press, 1981), pp. 440–58, and Christopher Abram, Myths of the Pagan North. The Gods of the Norsemen (London: Continuum, 2011), pp. 127–57. 58 Edda Snorra Sturlusonar. Edda Snorronis Sturlæi, ed. Jón Sigurðsson et al., 3 vols (Hafniæ: Legatum Arnamagnæanum, 1848–87), 3, pp. 256–57, 266. 59 Vatnsdœla saga, ed. Einar Ól. Sveinsson, p. 195.

2.2 The Early Period (up to c. 995)

55

section, since it may chronologically belong there. The poem adheres to the stylistics of the early period, whether it chronologically belongs there or rather to the group of pagan or paganizing poetry around Eiríkr. Hákonardrápa is quoted repeatedly in Skáldskaparmál, and at least the stanzas discussed here must have belonged to it, since they are so closely tied together by diction. In four of the half stanzas in Skáldskaparmál, Eiríkr’s (or Hákon’s) conquest of Norway is framed as a marriage or sexual intercourse with Jǫrð, mistress of Óðinn and personification of the earth, and extended metaphor is inherent to some of this imagery: the earl ‘spenr und sik barrhaddaða Þriðja biðkván’ (stretches out below him Þriði’s (Óðinn’s) pine-needle-haired waiting mistress [Jǫrð > jǫrð, earth]), he is ‘mjǫk trauðr eina at láta ítra systur Auðs’ (very reluctant to leave Auðr’s glorious sister [Jǫrð > jǫrð, earth] alone), he ‘átti eingadóttur Ónars, viði gróna’ (married Ónarr’s only daughter [Jǫrð > jǫrð, earth], overgrown with wood), and he ‘gat teygða at sér breiðleita brúði Báleygs’ (enticed Báleygr’s (Óðinn’s) broad-faced bride [Jǫrð > jǫrð, earth] to come to him).60 The personified earth is here accorded female attributes such as hair, sexual submissiveness and marriageability, and this imagery is extended throughout the four half stanzas (we do not know their arrangement in the original poem, but their content suggests that they belonged in sequence). They serve to show that there were examples of full extended metaphor also in the early period (or its stylistic extension under Eiríkr), even though inconsistent use was probably more common.61 The interplay between the primary and secondary level of referentiality in these stanzas is, furthermore, complex: this is not any beautiful woman, but rather one with the broad face of the earth. Her pine-needle-hair indicates that she is not the entire world, but Norway specifically. On the one hand the earl conquers her, because he gains power over Norway, but on the other, he marries her, since he is a good and responsible ruler. The extended metaphor is also combined with considerable mythological complexity. Wordplay or at least some kind of conceptual play recurs four times, all in kennings for the personified earth. The listener or reader must possess detailed knowledge of mythological genealogy in order to arrive at the female being Jǫrð, whose name he must then substitute with its homonym jǫrð (earth), or he must connect the two concepts through some other, extralinguistic mode of association. For ease of reference, I include also the latter possibility under the label wordplay. Extended metaphor, specific mythological references and wordplay are here densely interwoven.

60 SkP 3, pp. 219–25. 61 Another noteworthy instance of full extended metaphor is Kormákr’s lausavísa 28 (Skj B 1, p. 76; A 1, p. 85; the stanza is discussed in Frank, ‘Onomastic Play’, p. 7). The maritime imagary in the opening stanzas of Vellekla would also fit the bill, except for the kenning Kvasis dreyri (Kvasir’s blood [poetry]) in the first.

56

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

Tmesis is found in several poets in the early period. We have already seen examples from Þjóðolfr ór Hvini and Egill Skallagrímsson in the introduction to the features involved. Several further cases have been called into doubt by Kock and the editors of SkP, but others have not.62 Thus, for instance, Guthormr sindri has bríkar bǫðsœkir (board’s’s battle-attacker) for bǫðbríkar sœkir (battle-board’s [shield’s] attacker [warrior]),63 whereas Eyvindr skáldaspillir has Ála galtar éldraugr (Áli’s [a seaking’s] hog’s storm-log) for Ála élgaltar draugr (Áli’s storm- [battle-] hog’s [helmet’s] log [warrior]),64 and Einarr skálaglamm has myrk- [. . .] markar (dark-forest’s = Myrkviðr, between Jutland and Holstein).65 Eilífr Goðrúnarson’s Þórsdrápa is so complex that many readings remain tentative, but three instances appear to be secure.66 Several poets, such as Þjóðolfr, Egill, Einarr skálaglamm and Eilífr Goðrúnarson, display examples of all four features under discussion, and in particular specific mythological references, wordplay, and extended metaphor are often intertwined. This will have some bearing on the interpretation of their disappearance below.

2.3 The Intermediate Period (c. 995–c. 1120) Up to c. 995, most poets probably retained a pagan outlook, and this was the case also with many of their patrons – most notably Hákon jarl of Norway (c. 975–995), who seems to have promoted pagan cult and poetry as a link in his break with Danish overlordship.67 I have set the end of the early period at c. 995, that is, the year of the fall of Hákon and the rise to power of the missionary king Óláfr Tryggvason. During the intermediate period that followed, the features under study became very rare, but after well over a century they experienced a comeback with a number of twelfth-century poets. Although the precise dates of most relevant poems are uncertain, the return of the features involved probably happened no earlier than c. 1120 and no later than c. 1150. For convenience I have set the dates of the intermediate period to c. 995–c. 1120.

62 Thus, for instance, Finnur Jónsson’s reading of Hallfreðr’s baugs naddfár (the ring’s arrowharm) for naddbaugs fár (the arrow-ring’s [shield’s] harm [battle]) has been rejected (SkP 1, pp. 415–16), and so has Guthormr sindri’s vandar valsendir (the pole’s carcass-sender) for valvandar sendir (the carcass-pole’s [sword’s] sender [warrior]) (SkP 1, pp. 159–61). Kock rejects Bragi’s vazt[. . .] undirkúla (fishing-ground-under-knob [rock]), but SkP does not, perhaps because Snorri vouches for it (Kock, Notationes Norroenae, § 222; SkP 3, pp. 62–63). Several further examples could be added. 63 SkP 1, pp. 166–68. 64 SkP 1, pp. 221–23. 65 SkP 1, pp. 315–16. 66 SkP 3, pp. 91, 117–19. 67 For a brief overview of Hákon’s life and further references, see SkP 1, pp. cxciii–cxcv.

2.3 The Intermediate Period (c. 995–c. 1120)

57

In the early period, advanced diction is often found in panegyrics to rulers. Panegyrics are also by far the most common type of poetry to have been preserved from the intermediate period, and this allows for a study of changing conventions within the same genre between the early and the intermediate period. The observations in this section are based on a comprehensive reading of the poetry dated to c. 995–c. 1120 in Finnur Jónsson’s edition (Skj B 1, pp. 148–64, 178–466; A 1, pp. 156–73, 188–494) and SkP 1–3, as well as relevant portions of Rudolf Meissner, Die Kenningar. Other critical scholarship has been consulted in order to secure the readings, such as Ernst Albin Kock, Notationes Norroenae. Anteckningar till edda- och skaldediktning (Lund: Gleerup, 1923–44), Snorri Sturluson, Edda. Skáldskaparmál, ed. Anthony Faulkes, 2 vols (London: Viking Society for Northern Research, 1998) and editions of Íslenzk fornrit (Reykjavík: Hið íslenzka fornritafélag, 1933–).

2.3.1 Wordplay, Extended Metaphor and Tmesis The authenticity of the only unequivocal example of wordplay for more than a hundred years is doubted even in the sources. The stanza in question is presented with the words that ‘some say’ that Óláfr Haraldsson composed it, and it is present only in the Legendary Saga of Óláfr and in two interpolated manuscripts of his saga.68 It is thus more than a little doubtful that it should be dated to the intermediate period. Extended metaphors are few and far between, with the noteworthy exception of the topic of a ship as an animal. Sighvatr has a stanza completely centred on extended metaphor on the theme of animals and ships. This occurrence seems to be predicated on the names of two ships, Ormr inn langi (the long snake) and Visundr (bison).69 Markús Skeggjason has extended metaphor on a ship as a bear.70 Þórðr Særeksson employs extended metaphor on a ship as a horse throughout a stanza.71 Extended metaphors for ships thus appear as something of a topos in the period. Other interesting examples of extended metaphor are found in Þjóðolfr Arnorsson’s Sexstefja, a poem to Haraldr harðráði which is also noteworthy for its mythological references, as we shall see below. Þjóðolfr uses the kenning Yrsu burðar ǫrð (Yrsa’s offspring’s [Hrólfr kraki’s] grain [gold]) and retains the imagery of grain, sowing, fields and soil throughout the stanza.72 In two half stanzas he uses the same imagery to refer to weapons and corpses.73 The analogous imagery makes

68 SkP 1, p. 524. 69 SkP 1, p. 668. 70 SkP 3, p. 296. 71 SkP 3, p. 476. 72 SkP 2, p. 140. 73 SkP 2, pp. 143–44.

58

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

it tempting to think that the one full and the two half stanzas originally formed a sequence, but this is uncertain.74 Nonetheless, Sexstefja remains conspicuous for its use of extended metaphor, and it is worth noticing that this extended metaphor has a narrative base in the story of Hrólfr kraki who escaped his enemy Aðils by ‘sowing’ gold which Aðils stopped to pick up. From this perspective it is akin to the mytological kennings in the poem, which also presuppose a narrative base, and different from the extended metaphors in the stanzas by Sighvatr, Markús and Þórðr, which allude to the names and functions of ships, but not to stories. It should be noted, however, that Þórðr in one stanza refers to the goddess of the sea and her daughters, and that his and Þjóðolfr’s poetry indicates that at least some poets still combined an interest in extended metaphor and mythology.75 Two stanzas containing extended metaphor comparing a woman to a tree are attributed to Óláfr Haraldsson, but as we shall see below, they are likely to be spurious. Other instances of extended metaphors are few and are generally restricted to a part of a stanza during this period.76 Only one poet makes noteworthy use of tmesis, namely Sighvatr Þórðarson. While Finnur Jónsson found no less than nine occurrences, SkP accepts only three of these, and in most cases the critique is certainly justified.77 Even that, however, is somewhat surprising in a poet who generally opted for a relatively plain style. Since he also employed extended metaphor, it would appear that he wanted to retain some of the more advanced features of diction, even though he used no specific mythological references. Other poets, however, followed the trend towards simplification, so that tmesis in the remaining corpus largely disappears around the beginning of the eleventh century. Finnur Jónsson list only one instance later in the eleventh century, and this, as well as almost all instances in Sighvatr’s older contemporaries, has been plausibly refuted by Kock and the editors of SkP.78

74 SkP 2, p. 111. 75 SkP 1, p. 236; SkP 3, p. 480. 76 Most elaborated examples of extended metaphor are transmitted in Snorri’s Edda, as is the case with Markús’s stanza (thus, for instance, Þorleikr fagri (SkP 3, p. 484), who does not, however, abide by Snorri’s rule of consistency). 77 Finnur Jónsson, ‘Kenningers led-omstilling’, pp. 6, 17–19; accepted in SkP: SkP 1, pp. 569–70, 705–06, 725–26; not accepted: SkP 1, pp. 538–40, 657–60, 686–87. 78 Finnur Jónsson notes one instance each of tmesis in some of Sighvatr’s older contemporaries Hofgarða-Refr, Þormóðr kolbrúnarskáld, and Hárekr í Þjóttu, and two each in Þórarinn loftunga and Halldórr ókristni. That of Refr is plausibly refuted in Kock, Notationes Norroenae, § 784; Hárekr’s stanza is problematic, but there is no reason to assume tmesis (SkP 1, pp. 808–10); Þormóðr’s tmesis is refuted by Kock, but through presupposing an unattested meaning of fjǫrr, and the question is perhaps best considered unresolved (Kock, Notationes Norroenae, §§ 696, 2481 E); one of the instances in Halldórr has been refuted (SkP 1, pp. 479–80), whereas the other is problematic no matter if it is read with or without tmesis (SkP 1, pp. 473–75); Þórarinn has been refuted (SkP

2.3 The Intermediate Period (c. 995–c. 1120)

59

2.3.2 Mythological References After c. 995, specific mythological references become very rare in poetry that can be considered datable with a fair degree of certainty; the few noteworthy exceptions are treated in detail below. Hans Kuhn, in a vehement response to de Vries’ study of mythological kennings, argued that the apparent decline in mythological kennings is due to a skewed distribution of different types of poetry in different sources.79 Since the action in the sagas of Icelanders tends to end around 1030–1050, Icelandic paganizing poetry did not decline, but was, according to Kuhn, only underrepresented in the sources for the period from the early decades of the eleventh century onwards. He even insisted that pagan beliefs in Iceland retained their vigour throughout the intermediate period. Even though scholars today are unlikely to accept the latter claim, Kuhn’s observation about the internal chronology of the sagas of Icelanders is correct and must be taken seriously. The distribution of material in Snorri’s Edda, however, indicates that Kuhn’s explanation is not satisfactory. Snorri’s criteria for selecting poetry were stylistic, and he often provides stanzas of a mythologically elaborate nature by poets whom we would otherwise know as relatively plain and factual. This is true not only for the early period, but also for the intermediate one. Thus, for instance, the specific mythological references that can be found in Arnórr’s poetry (mid-eleventh century) are transmitted in Snorri’s Edda and have only been inserted into Arnórr’s longer poems by modern editors. The fact that the four hundred or so half and full stanzas in Skáldskaparmál conform to the chronological distribution outlined above suggests that Snorri did not find much poetry from the period c. 995–c.1120 that contained such mythological references as were of interest to him. Specific mythological references completely dominate chapters 2–22, and the only poet representing the prolific output of the intermediate period there is Arnórr jarlaskáld, with one half stanza.80 The combined evidence and character of the sources thus indicates that the slump is not illusory. In the early period, the inclusion of mythology within a nexus of advanced diction seems to have been doctrinally unproblematic. This situation appears to have changed under the patronage of the missionary kings Óláfr Tryggvason and Óláfr Haraldsson (995–1000, 1015–1030). The first of these seems not to have favoured skaldic poetry, whereas the second did, but his poets display a very different stylistics from that of previous periods. They avoided not only specific mythological references,

1, 854–55, 860–61). After Sighvatr, the only poet in Finnur Jónsson’s list is Stúfr, and this occurrence has been plausibly refuted (SkP 2, p. 353). The prolific poets of the eleventh century, such as Arnórr and Þjóðolfr Arnórsson, are absent altogether (‘Kenningers led-omstilling’, pp. 5–6, 19). 79 ‘Das nordgermanische Heidentum in den ersten christlichen Jahrhunderten’, Zeitschrift für deutsches Altertum und deutsche Literatur, 79 (1942), pp. 133–66. 80 Snorri, Skáldskaparmál, ed. Faulkes p. 6; note also his two mythological kennings immediately after Chapter 22, on pp. 33–34.

60

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

but also the other features of diction under study here. The approaches of both of these kings can probably be connected to their religious zeal. This would explain why Óláfr Tryggvason was sceptical of skaldic poetry with its rich mythological diction and why specific mythological references are not found in the poetry to Óláfr Haraldsson. It is less obvious, however, why other advanced features of diction should pose any problems to Christian orthodoxy. Since specific mythological references were generally interwoven with other features which increase the semantic complexity, it is likely that the abandonment of such references led to the use of simpler diction.81 We saw above how closely these features are interconnected in Hallfreðr’s Hákonardrápa. In Hallfreðar saga, king Óláfr Tryggvason frowns at Hallfreðr’s mythological diction, and the poet produces a number of stanzas often referred to as his ‘conversion stanzas’. In one stanza, the poet complains that ‘trauðr legg ek á frumver Friggjar fjón, þvít vel hugnaðisk Viðris vald skaldi’ (I am reluctant to be hostile towards Frigg’s first husband [Óðinn], because Viðrir’s [Óðinn’s] power was pleasant to the poet [me]).82 A glance at Hákonardrápa shows that, if the gods had to go, so did also much else of the most artistically elaborate features, and Hallfreðr’s despair at being bereft of the tools of his trade is readily understandable. The saga description cannot be taken at face value, but belongs to the thirteenth century. With regard to the religious problems facing the poet, however, it largely conforms to the content of the stanzas in it. The stanzas themselves have few diagnostic features for dating, although Kari Ellen Gade notes that the metrics of the stanzas in Hallfreðar saga generally suggest that they are authentic.83 None of the scholars writing on the date of these stanzas discuss the aðalhending in a: ǫ in one of them (‘fjǫrð létk af dul Njarðar’). Not only are there no certain examples of this after 1200, but such hendingar are unambiguously described as skothendin-

81 Tarrin Wills sees a general decrease in kenning complexity in the intermediate period. See ‘The Development of Skaldic Language’, in Á austrvega. Saga and East Scandinavia. Preprint Papers of the 14th International Saga Conference. Uppsala 9th – 15th August 2009, ed. Agneta Ney, Henrik Williams, Fredrik Charpentier Ljungqvist (Gävle: Gävle University Press, 2009), pp. 1032–38. 82 Vatnsdœla saga, ed. Einar Ól. Sveinsson, p. 157. I render the quotation in prosaic word order. 83 ‘The Dating and Attribution of Verses in the Skald Sagas’, in Skaldsagas. Text, Vocation, and Desire in the Icelandic Sagas of Poets, ed. Russell Poole (Berlin: De Gruyter, 2001), pp. 71–74. Diana Whaley has investigated their content and concludes that it may plausibly be attributed to the conversion period (‘The “Conversion Verses” in Hallfreðar saga: Authentic Voice of a Reluctant Christian?’, in Old Norse Myths, Literature and Society, ed. Margaret Clunies Ross (Odense: University Press of Southern Denmark, 2003), pp. 234–57). Russell Poole treats them as authentic, but does so mainly with reference to Gade’s study (‘The “Conversion Verses” of Hallfreðr vandræðaskáld’, Maal og Minne (2002), pp. 15–37).

2.3 The Intermediate Period (c. 995–c. 1120)

61

gar by Óláfr Þórðarson c. 1250.84 It is thus unlikely that the stanza was composed for the saga.85 Another factor that has not been taken into account is that Einarr Skúlason paraphrased one of the conversion stanzas as well as the stanza immediately preceding them in the saga, sometime before or around the middle of the twelfth century. Since metrical criteria, phonological change and paraphrase are three independent criteria, the case for the authenticity of Hallfreðr’s stanzas appears to be strong, and the matter is treated more fully in the discussion of Einarr on pp. 80–85, whereas its implications for the emergence of prosimetrical saga writing are discussed on pp. 85 and 211. Both Hallfreðr’s conversion stanzas and Hákonardrápa show which poetic tools lay in the balance if specific mythological references had to go, and the saga description seems to be based on an understanding of this.86 The explanation that a number of features disappeared becaused of their association to specific mythological references must remain tentative, since relevant sources are scant. Nonetheless, the temporal coincidence of the patronage of the missionary kings and the slump cannot be overlooked, and it seems reasonable to assume that the association of mythology to paganism was a central issue in this context. To the poets, however, different uses of mythology were connected to different stylistic registers, and orthodoxy thus in the end amounted to simplicity. This, however, raises the question of why generic mythological references (kennings for persons and battle) were retained. I think that a likely explanation is that they were not seen a presupposing a pagan system of beliefs. Kennings for persons simply substitute the name of a supernatural being for that of a person, thus serving almost as heiti (poetic synomyms). They are based on the generic anthropomorphism of the gods, but draw on no distinctive qualities of the individual god or goddess. Kennings for battle and battle-related phenomena allude only to the martial quality of Óðinn and the Valkyries. Both anthropomorphism and martial qualities could easily be accommodated within even the most

84 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 51. On this phenomenon, see Hreinn Benediktsson, ‘Phonemic Neutralization and Inaccurate Rhymes’, Acta philologica Scandinavica, 26 (1964), pp. 1–18). 85 The manuscripts read ‘fiarð’, but this can easily be explained as scribal correction (Skj A 1, p. 168). All scholars agree in reading fjǫrð here, and there can be little doubt that Einar Ólafur Sveinsson is right in translating ‘last year’, rather than the participle of the unattested verb fjarra as proposed by Finnur Jónsson (Vatnsdœla saga, ed. Einar Ól. Sveinsson, p. 158; LP, p. 138). All suggested interpretations demand the root vowel to be /ǫ/, rather than /a/. This is clearly borne out by cognates in the case of the word meaning ‘last year’ (Harald Bjorvand and Fredrik Otto Lindeman, Våre arveord. Etymologisk ordbok, 2nd ed. (Oslo: Novus forlag, 2007), p. 274). For a discussion of this and other occurrences in skaldic poetry, see SkP 1, pp. 234–35, 475–76, 479. 86 On the author’s perspective on Hallfreðr’s conversion, see Christopher Abram, ‘Modeling Religious Experience in Old Norse Conversion Narratives: the case of Óláfr Tryggvason and Hallfreðr vandræðaskáld’, Speculum, 90 (2015), pp. 114-57.

62

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

rudimentary euhemeristic framework. Specific mythological references, by contrast, presuppose a detailed mythological world. Many of these references evoke particular myths dealing with creation or inspiration, and it is easy to see how they could have been repulsive to an early convert. For these reasons, I believe, specific mythological references had to go, whereas generic references did not. As specific mythological references went, so went the other features associated with them. While these differences between generic and specific mythological references may account for why the generic ones could remain in use, this does not mean that they necessarily had to do so. These kennings do, after all, involve the names of pagan deities, and one might have expected that all mythological references would have been dropped at the same time. Here it is instructive to consider what their removal would have done to the kenning system. Specific mythological references cover more meanings than generic ones, but the latter are vastly more numerous in sheer numbers of kennings: to a large extent, skaldic poetry centres on prominent people waging war. The base-words in kennings for men basically come in three varieties: names of gods, trees (including poles and other tree-like objects) and nomina agentis (including semantically related forms like present participles).87 Women, by contrast, are goddesses, trees and tree-like objects and land (perhaps derived from goddesses in the form of the goddess Jǫrð ‘land, earth’).88 Removing the gods from kennings for people would reduce the number of options from three to two in each case. With regard to kennings relating to battle, Óðinn and the Valkyries in particular had a staggering number of names which could serve as functional synonyms to ‘battle’, far outnumbering other words for the same phenomenon (indeed, several names of Valkyries, such as Gunnr and Hildr, are really common nouns meaning ‘battle’). As a result, the removal of mythological kennings for people and battle would lead to a drastic reduction of the poetic repertoire, and it is important in this context to recall that skaldic poetry was an art form where the number of structural options was limited (cf. the three types of base-words for people), but where the unique expression appears to have been actively sought through the use of new combinations of different synonyms.89 For the most common kenning types, it is sometimes possible to find several occurrences of identical kennings when the entire corpus is collated, but typically only one instance of each verbal realization of a given type is found. Consulting almost any of the lists of kennings in Meissner, Die Kenningar,

87 Meissner, Die Kenningar, pp. 259–350. The words skati (generous man) and ǫ́rr (messenger) appear to fall outside of this grouping, but should, I believe, on functional grounds be connected to the nomina agentis, something which Meissner also suggests in the case of ǫ́rr (Meissner, Die Kenningar, pp. 265–66, 272–73). An occasional giant, dwarf, hero, or king can be found in the place of a god, but such instances are exceptional (Meissner, Die Kenningar, p. 263). 88 Meissner, Die Kenningar, pp. 399–412. 89 See Elena A. Gurevich, ‘The System of Kennings’, Nordica Bergensia, 3 (1994), pp. 140–42.

2.3 The Intermediate Period (c. 995–c. 1120)

63

will prove this point. Removing one of the three semantic options for base-words in kennings for people and the bulk of what in practice functioned as synonyms for ‘battle’ would therefore strike at the heart of the skaldic art, rendering it much more difficult to produce the unique expressions that were so important to the skalds. It may be for this reason that generic mythological references were retained, especially since they were relatively innocuous. This interpretation – that the disappearance of all four features under study was predicated on religious sentiment and thus centred on specific mythological references – can be tested to some extent by considering the few exceptions to their general disappearance after c. 995 and the poets and contexts involved. For a closer analysis of specific mythological references during the eleventh century, I refer the reader to my article ‘The Last Pagan’, JEGP, 116 (2017), pp. 491–514. It should be noted, however, that partly due to the fact that SkP 3 had not been published at the time, I overlooked ten eleventh-century references in the poetry of Hallfreðr vandræðaskáld, Þórðr Særeksson, Óttarr svarti, Hallvarðr háreksblesi, Þjóðolfr Arnórsson, Arnórr jarlaskáld and Stúfr inn blindi in the article (see below).

2.3.3 Specific Mythological References: Eiríkr jarl and Others Now follows an overview where I first treat poets connected to Eiríkr jarl, then other poetry which is not connected to kings, and finally poetry to or by kings. Most specific references of the intermediate period are found in poetry of up to c. 1035 and which has been transmitted in Skáldskaparmál. Some of this poetry seems to convey a genuinely pagan outlook, most notably so that of Eyjolfr dáðaskáld, Skúli Þorsteinsson and Hofgarða-Refr.90 The two former served Eiríkr jarl (r. 1000–1012), son of the notorious pagan Hákon jarl. Eiríkr succeded to Óláfr Tryggvason, who had overthrown his father, and it would not be surprising if Óláfr’s policies had little attraction to Eiríkr. Most saga authors suggest that Christianity fared ill under Eiríks reign.91 In 1012 he was summoned to join king Knútr to fight in England and left the rule of Norway to his son Hákon.92 Some English sources mention Eiríkr, but they

90 See SkP 1, p. 454–68; SkP 3, pp. 366–68, 243–65. Poetry by Grettir is also noteworthy for its many complex mythological references, but this poetry is almost certainly spurious and belongs to the late period (see below, pp. 264–73). 91 Only Heimskringla and Theodoricus do not appear to see a problem (Heimskringla, ed. Bjarni Aðalbjarnarson, 1, pp. 554–57; Monumenta historica Norvegiæ. Latinske kildeskrifter til Norges historie i middelalderen, ed. Gustav Storm (Kristiania: Brøgger, 1880), p. 25). Ágrip, Fagrskinna and Historia Norwegiae suggest that Christianity was in considerable hardship or even close to eradication (Ágrip af Nóregskonunga sǫgum. Fagrskinna – Nóregs konunga tal, ed. Bjarni Einarsson, pp. 24, 165–66; Monumenta historica Norvegiæ, ed. Storm, p. 119). 92 Heimskringla, ed. Bjarni Aðalbjarnarson, 2, pp. lxxxix, 30–32.

64

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

have little bearing on his religious policies or inclinations in the preceding Norwegian period.93 The poetry addressed to Eiríkr is our safest and most informative source on the matter. Eyjolfr apostrophes Eiríkr by saying that he ‘dregr land und sik at mun banda’ (conquers land at the pleasure of the gods).94 Eyjolfr here paraphrases Vellekla, Einarr skálaglamm’s famous poem to Eiríkr’s father, where it is Hákon who takes ‘lǫnd at mun banda’ (lands at the pleasure of the gods).95 Skúli has a stanza on the sun as a goddess and mistress of the god Glenr.96 Another poet who composed in honour of Eiríkr, Þórðr Kolbeinsson, uses the kenning Þorins rekka regn (Þorinn’s [a dwarf’s] men’s [the dwarves’] rain [poetry]), a type of kenning which had passed out of use except in the poets under study here (this kenning is further discussed on pp. 90–91). As argued above, it is likely that Hallfreðr’s Hákonardrápa should be connected to Eiríkr rather than Hákon, and this poem is extremely dense in mythological imagery. Halldórr ókristni (the un-Christian) also served Eiríkr, and although we have no mythologizing poetry by him, his epithet suggests that he may have had pagan leanings. Þórðr Særeksson is also listed among Eiríkr’s poets in Skáldatal, and in one stanza, he refers to the goddess of the sea and her daughters.97 Þórðr is an elusive but interesting poet. On p. 28, I noted his experiments with form. In another stanza, he employs extended metaphor on a ship as a horse throughout.98 To most skaldic scholars, he is known as the poet who produced the longest kenning in the corpus.99 He thus seems to have tested the limits of form and diction in a number of ways. He is also listed as one of Óláfr Haraldsson’s poets, and although we have no explicitly Christian poetry by him, the name of his poem celebrating Óláfr’s battle at Helgeå, Róðudrápa (poem with refrain about the crucifix), suggests that he adapted to the religious preferences of that ruler.100

93 The English sources to Eiríkr are scant, the most important being the Encomium Emmae and the Anglo-Saxon Chronicle (Encomium Emmae reginae, ed. Alistair Campbell. Camden Third Series, 72 (London: Offices of the Royal Historical Society, 1949), pp. 22, 30–32; The Anglo-Saxon Chronicle. MS C, ed. Katherine O’Brien O’Keeffe. The Anglo-Saxon Chronicle. A Collaborative Edition, 5 (Cambridge: D.S. Brewer, 2001), s.a. 1016–1017, pp. 101, 103). These and other sources indicate that he lived considerably past 1013, when Heimskringla places his death. See Heimskringla, ed. Bjarni Aðalbjarnarson, 2, p. 32 n. 1. 94 SkP 1, p. 468. 95 SkP 1, p. 292. 96 SkP 3, p. 366; Snorri, Skáldskaparmál, ed. Faulkes, p. 39. 97 SkP 1, p. 236; SkP 3, p. 480. 98 SkP 3, p. 476. 99 SkP 1, p. 237. 100 SkP 1, pp. 242–44.

2.3 The Intermediate Period (c. 995–c. 1120)

65

Hofgarða-Refr was active somewhat later, in Iceland, and he is the only poet after c. 995 and before the twelfth century to make noteworthy use of mythological kennings for poetry – no less than five times.101 Among other poets of the period, only one such kenning each is found in Arnórr jarlaskáld, Bjǫrn Hítdœlakappi and Haraldr harðráði (see below).102 Arnórr is noteworthy for having three specific mythological references at a late date (1060s), all transmitted in Skáldskaparmál. These belong in the poem Þorfinnsdrápa and they are: ‘und gǫmlum Ymis hausi’ (under the old skull of Ymir [the sky]), ‘þýtr Alfǫður hrosta brim’ (Alfaðir’s [Óðinn’s] mash’s surf [ale > poetry] roars) and ‘brestr erfiði Austra’ (Austri’s [a dwarf’s] toil [the sky] splits).103 The attribution of the second and third stanzas to Þorfinnsdrápa is relatively unproblematic, since an earl is mentioned in the second and the name Þorfinnr in the third. The first does not, however, provide us with any clear indications of where it belongs and is quoted without historical context in Skáldskaparmál. It has with some hesitation been assigned to Magnússdrápa by modern scholars and thus to a memorial poem about the son of Saint Óláfr.104 This attribution is, however, based on weak arguments, the main one being that Arnórr elsewhere uses the words skjǫldungr (ruler) and ungr (young), which are found in this stanza, of Magnús.105 The designation skjǫldungr was used of rulers generally.106 While we may perhaps conclude that Arnórr had a slightly greater propensity than others for using skjǫldungr rather than the many other poetic synonyms for ‘ruler’, it does not follow that he would have restricted its use to any one ruler in particular. With regard to ungr, Arnórr uses its comparative form œri (younger) about Þorfinnr in a stanza in Þorfinnsdrápa, saying that no man under the sky, younger than Þorfinnr, has declared himself ready to defend his realm.107 The stanza thus fits equally well in Þorfinnsdrápa as in Magnússdrápa.

101 SkP 3, pp. 250 (1.1), 251 (2.3), 255 (2.3), 256 (3.1), 264 (5.4). For a focused discussion of Refr, see Mikael Males, ‘The Last Pagan’, JEGP, 116 (2017), pp. 491–514. For sources on Refr, see Eyrbyggja saga, ed. Einar Ól. Sveinsson and Matthías Þórðarson. Íslenzk fornrit, 4 (Reykjavík: Hið íslenzka fornritafélag, 1935), p. 30; cf. Íslendingabók. Landnámabók, ed. Jakob Benediktsson. Íslenzk fornrit, 1 (Reykjavík: Hið íslenzka fornritafélag, 1986), pp. 100, 104; Heimskringla, ed. Bjarni Aðalbjarnarson, 2, p. 358. 102 The one in Sighvatr conjectured by Finnur Jónsson is doubtful with regard both to content and authenticity (Skj B 1, p. 253 (29.2); A 1, p. 274). Occurrences found in the poetry of Ormr Steinþórsson (SkP 3, pp. 325, 329) and Steinþórr (SkP 3, p. 390) should probably both be dated after c. 1150 (SkP 3, pp. 323, 390), and with regard to Ófeigr’s spurious Áms ok Austra sættir, see below pp. 253–54. 103 SkP 2, pp. 229, 231, 258. 104 See The Poetry of Arnórr jarlaskáld. An Edition and Study, ed. Diana Whaley. Westfield Publications in Medieval Studies, 8 (Turnhout: Brepols, 1998), pp. 29, 63, 74; Fidjestøl, Det norrøne fyrstediktet, p. 130. 105 The Poetry of Arnórr jarlaskáld, ed. Whaley, p. 29; SkP 2, p. 207. 106 SkP 3, p. 207; LP, s.v. 107 SkP 2, p. 235.

66

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

What makes Þorfinnsdrápa the more likely candidate is that mythological variations on a theme typically serve to tie a poem together, as in, for instance Kormákr’s Sigurðardrápa, Eyvindr’s Vellekla and Hallfreðr’s Hákonardrápa, and one would therefore expect the cosmological Ymis hauss to tie into erfiði Austra. If the attribution of Arnórr’s stanza to Þorfinnsdrápa is correct, that poem exhibits a conspicuous use of specific mythological references. It also contains prayers directed at the Christian God: one in the stanza where we also find the expression erfiði Austra and one in the following stanza.108 Any connection to paganism is thus ruled out, and we must look for other explanations. I believe that one such may be sought in the keen awareness of tradition within the skaldic art – with its paraphrases of the expressions and rhetorical devices of earlier poets – and that the model in this case is to be found in poems like the ones just mentioned: Sigurðardrápa, Vellekla and Hákonardrápa. Here, mythological variation is a cornerstone of the artistic program, and all three poems were composed for earls: the first for Hákon’s father Sigurðr, the second for Hákon and the third for his son Eiríkr (or possibly Hákon). It would appear, then, that a tradition had formed where mythological variation belonged to poetry for earls, and like the former Norwegian earls, the earls of Orkney had pretentions to independence. With great subtlety, then, the mythological references allude to the political legacy of the mainland earls, who could hold their own and rule as kings in their own right. Arnórr’s use of mythology in Þorfinnsdrápa thus appears to have much to do with political and cultural capital, but little with religion. Another factor serves to corroborate that the references are due to a subtle play with tradition: the stanza containing the words ‘brestr erfiði Austra’ also contains a paraphrase of Vǫluspá. Þorfinnsdrápa stanza 24. 1–4 reads: Bjǫrt verðr sól at svartri søkkr fold í mar døkkvan brestr erfiði Austra allr glymr sær á fjǫllum.109 The bright sun becomes black, the land sinks into the dark sea, Austri’s [a dwarf’s] toil [the sky] splits, all the sea resounds against the mountains.

Vǫluspá R 54/ H 50 reads: Sól tér sortna sígr fold í mar110 The sun becomes black, the land sinks into the sea.

108 SkP 2, pp. 258–60. The expression ‘hjalpi Goð’ (may God help) in stanza 24 could be either singular or plural, but ‘Goð, firr [Þorfinn] meinum’ (God, keep [Þorfinnr] far from harms) in stanza 25 is unequivocal. 109 SkP 2, p. 258. 110 Norrœn fornkvæði, ed. Bugge, pp. 18, 25, 32.

2.3 The Intermediate Period (c. 995–c. 1120)

67

Arnórr has two contrasting adjectives (bjǫrt, døkkvan) and uses the adjectival construction at svartri rather than the verb sortna of the same root. The match is so close that there can be little doubt that this is a paraphrase. The internal dynamics of the skaldic tradition, but also the requirements of dróttkvætt, here help us to determine which of the two poems is the giver. Elaborate paraphrase of this kind belongs in the skaldic tradition and presupposes a famous pre-existent poem in order that the skald’s use of tradition should become evident (see pp. 82, 85–86, 93). The stanza from Vǫluspá is transmitted in R, H and Gylfaginning alike, suggesting that it was thought to belong to the poem’s core also later, in the thirteenth and fourteenth centuries. Metrically, the changes have clearly been undertaken to meet the requirements of dróttkvætt, rather than those of fornyrðislag. The additional adjectives help fill the verse. One could have envisioned their removal to shorten the line, but Arnórr’s use of the verb søkkr rather than sígr shows that this is not what has happened. The choice of the adjective døkkvan explains the use of the rhyming verb søkkr. If the verse was transformed from fornyrðislag to dróttkvætt, this explains why the verb was changed, whereas a transformation in the other direction would not require a different verb, since fornyrðislag does not have rhyme. In quoting this stanza from Vǫluspá, Gylfaginning, like Arnórr, uses the verb søkkr rather than sígr. If one were to analyse søkkr in Gylfaginning as the result of uncontaminated variation within a fornyrðislag tradition, the variant would fulfil no purpose and be difficult to explain on either phonetic or palaeographic grounds. The variant therefore appears to be due to contamination from Þorfinnsdrápa. This is not very surprising since the first half of Þorfinnsdrápa stanza 24 is found in Skáldskaparmál in all four main witnesses to Snorri’s Edda (R, T, W and U). Two parameters, then, the conventions of skaldic paraphrase and metrics, indicate that Vǫluspá is the giver and Þorfinnsdrápa the taker.111 This allows us to see that the mythological references in Þorfinnsdrápa are part of a complex play with the earlier tradition, skaldic and eddic alike (compare the discussion of Illugi Bryndœlaskald below pp. 71–72). Finally, Arnórr has one more specific mythological reference, this one unequivocally belonging to Magnússdrápa and therefore treated below in the discussion of poetry connected to kings. The mythological references treated so far are transmitted in the sourcecritically reassuring context of Skáldskaparmál. Outside of Snorri’s Edda, Bjarnar saga Hítdœlakappa transmits two references that may go back to the saga’s date of

111 John McKinnell comes to the same conclusion based on his claim that three stanzas in Vǫluspá are paraphrased in Arnórr’s stanza (John McKinnell, ‘Vǫluspá and the Feast of Easter’, Alvíssmál, 12 (2008), p. 7). It is somewhat unclear whether all of these similarities are intentional, however.

68

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

action in the 1020s. In one stanza, Bjǫrn calls poetry Hǫ́ars bjórr (Hǫ́rr’s [Óðinn’s] beer).112 In another, he has a curious reference to a Valkyrie: dagleygjar hilmis Ilmr (day-fire’s [sun’s] ruler’s [God’s/Óðinn’s] Ilmr [name of a goddess: Valkyrie]).113 In order to produce a Valkyrie, which is needed in the context, the sun’s ruler must be Óðinn, but typologically, such a kenning would normally designate God. A number of mythological kennings are found in what the sagas present as eleventh-century poetry, but which can, for various reasons, be judged spurious. Most notable is Grettir Ásmundarson, whose historical model lived c. 996–c. 1031; the poetry attributed to him in his saga is massively mythologizing.114 Here, we are in the rare position of having solid linguistic evidence for dating the bulk of his poetry to the late thirteenth or early fourteenth century, and Grettir’s poetry is analyzed below in Chapter Four. Here, it may suffice to note that no specific mythological references are found in the stanzas or lines which are also attested outside of the saga and which thus have a better claim to authenticity.115 Grettir, then, has little bearing on the present argument. Two mythological kennings are found in Ófeigr’s poetry in Bandamanna saga, whose date of action is shortly after the middle of the eleventh century.116 The kennings are Iðja hlátr (Iði’s [a giant’s] laughter [gold]) and Áms ok Austra sáttir (Ámr’s [a giant’s] and Austri’s [a dwarf’s] agreement [mead of poetry]), but various circumstances suggest that this poetry is spurious, although this has not been proven linguistically, as in the case of Grettir’s poetry. This issue is treated on pp. 253–54.

2.3.4 Specific Mythological References: Kings Authentic or not, none of the references reviewed so far is connected to a Norwegian kings, or indeed to any king, after the ascension of Ólafr Tryggvason. The distribution of such references in royal poetry is interesting, and we turn now to the occurrences that have been transmitted: twelve that may well be real and three that I shall argue are illusory. The first three are found in Hallfreðr vandræðaskáld’s memorial poem to Óláfr Tryggvason and thus to one of the two missionary kings. The kennings in question are: Surts ættar sylgr (Surtr’s [a giant’s] family’s [giants’] drink [poetry]), landherðar lýða líð (landshoulder’s [stone’s] people’s

112 Skj B 1, p. 282; A 1, p. 304. 113 Skj B 1, p. 282; A 1, p. 304; Borgfirðinga sögur, ed. Sigurður Nordal and Guðni Jónsson, pp. 196–97. 114 Finnur Jonsson, Den oldnorske og oldislandske litteraturs historie, 1, p. 512. 115 Finnur Jónsson, Den oldnorske og oldislandske litteraturs historie, 1, p. 514; Grettis saga, ed. Guðni Jónsson, pp. xli, 92–93, 207–08, 234–35. 116 Grettis saga, ed. Guðni Jónsson, pp. lxxix–lxxx.

2.3 The Intermediate Period (c. 995–c. 1120)

69

[giants’] drink [poetry]) and Norðra niðbyrðr (Norðri’s [a dwarf’s] kin- [dwarf-] burden [sky]).117 Hallfreðr’s use of these kennings is difficult to interpret. First of all, the Kringla version of the Icelandic list of poets – Skáldatal – lists only two poets under Óláfr’s patronage, Hallfreðr and Bjarni, and no poetry by Bjarni has been preserved.118 Óláfs saga Tryggvasonar in Heimskringla is dominated by poetry to Hákon and his son Eiríkr, and it would thus appear that the author did not have access to much contemporary poetry addressed to Óláfr. We therefore have no other poets with whom to compare Hallfreðr and must rely on his poetry alone. As we have seen, Hallfreðr also composed to Eiríkr jarl (or possibly to his father) and he was remembered as one who had some trouble adjusting to the new faith (hence his nickname vandræðaskáld ‘the troublesome poet’). His poetry to Eiríkr (or Hákon) appears to be completely pagan, whereas his memorial poem to Óláfr contains three specific mythological references. Unfortunately, we have no poetry which was certainly composed in honour of the living Óláfr. Six stanzas are generally held to belong to a drápa about Óláfr when he was still alive and contain no specific mythological references, but it is conceivable that these stanzas belong to the introduction of the memorial poem.119 We therefore have little to nothing that can tell us about the character of the poetry that may have been composed in honour of the living Óláfr. This is in itself remarkable. Hákon before him and Eiríkr, Óláfr Haraldsson and later kings after him had many poets. One may suspect that the lack of poetry connected to Óláfr Tryggvason is due to his misgivings about the pagan associations of this art as a whole. It is important not to read later developments back into Óláfr’s reign. His namesake Óláfr Haraldsson (r. 1015–30) had many poets who had successfully developed a religiously acceptable mode of expression, but this process may just have been getting started in Óláfr Tryggvason’s day. Perhaps we find a trace of these ongoing negotiations in Hallfreðr’s kennings in the memorial poem. Furthermore, the poem is composed in honour of the dead Óláfr. Eiríkr jarl then wielded power over Norway, and in light of the pagan imagery in the poetry to Eiríkr, it seems likely that Hallfreðr was now free to use whichever language he preferred. The interpretation that the lack of poetry to Óláfr Tryggvason indicates that he disapproved of its content is admittedly an argument of silence and therefore not

117 SkP 1, pp. 421, 430, 437. The kenning landherðar lýða líð is based on an interpretation by Jón Helgason and is not found in Skj or Meissner, Die Kenningar (SkP 1, pp. 430–31). 118 Edda Snorra Sturlusonar, ed. Jón Sigurðsson et al., 3, pp. 253; Bjarne Fidjestøl, Det norrøne fyrstediktet, p. 171. The Uppsala version lists four, adding Gizurr gullbrárskáld and Sighvatr, but that appears to be due to confusion with the other Óláfr (Edda Snorra Sturlusonar, ed. Jón Sigurðsson et al., 3, p. 261). Furthermore it adds the nickname gullbráskáld to Bjarni, again apparently out of confusion with Bjarni gullbrárskáld (SkP 1, p. 877). 119 SkP 1, p. 387.

70

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

susceptible to testing. From this perspective, it would have been preferable to set the beginning of the intermediate period at c. 1015, when the large body of poetry around Óláfr Haraldsson allows us to see what the poets are doing. But since Óláfr Tryggvason’s rule was in the middle of a period when rulers of Norway strongly favoured panegyrics, the silence of the skalds is noteworthy, even given his short reign. In light of this and of the description of Óláfr’s religious policies in later sources, I believe that it may be justified to set the beginning of the intermediate period at c. 995. If, however, it were moved to c. 1015, the slump in the features under study would be all the more pronounced. In poetry to non-Norwegian kings, three specific mythological references occur. One mythological kenning for ‘earth’ or ‘land’ is found in Óttarr svarti’s Óláfsdrápa sænska (c. 1018).120 In the large body of praise poetry to Knútr, two specific mythological references are found in a poem by Hallvarðr háreksblesi (c. 1029). One is eitrsvalr naðr (poison-cold serpent [Miðgarðsormr]), the other moldreks orð (soilruler’s [giant’s] words [gold]).121 The poem also contains references to prayer and monks and thus exhibits a remarkable blend of mythological and Christian religious references reminiscent of Arnórr’s Þorfinnsdrápa. The Norwegain king Haraldr harðráði stands alone in both having produced and received specific mythological references. In a famous stanza about his accomplishments he refers to poetry as Yggs líð (Yggr’s [Óðinn’s] drink [poetry]) (c. 1043–1044). The stanza is only transmitted in the Morkinskinna tradition, which may not in itself be very reassuring, but a twelfth-century paraphrase of it suggests that it is authentic.122 Two mythological kennings for ‘earth’ or ‘land’ are found in the same stanza in Þjóðolfr Arnórsson’s Sexstefja (c. 1065), praising Haraldr harðráði.123 As noted above, this poem also makes conspicuous use of extended metaphor. Another of Þjóðólfr’s poems to Haraldr contains the kenning Járnsǫxu veðr (Járnsaxa’s [giantess’s] wind [mind]).124 His situational stanza containing the expression ‘goð vǫ́ru grǫm Dǫnum’ (the gods were angry with the Danes) may well be authentic, and if so bears testimony to the attitudes towards mythology around Haraldr. The matter is complex, however, and is discussed in some detail on

120 SkP 3, p. 339 (the title is editorial). 121 SkP 3, pp. 236, 238. 122 SkP 2, 39–40. The opening line Íþróttir kannk átta was later paraphrased by earl Rǫgnvaldr of Orkney in the unmetrical line Íþróttir kannk níu. Rǫgnvaldr’s stanza has a faulty cadence, and this suggests that he is playing with the listeners’ expectations, that would presumably have been shaped by Haraldr’s stanza (Orkneyinga saga, ed. Finnbogi Guðmundsson, p. 130; SkP 2, p. 576). For Haraldr’s stanza in its saga context, see Morkinskinna, ed. Ármann Jakobsson and Þórður Ingi Guðjónsson, 2 vols. Íslenzk fornrit, 23–24 (Reykjavík: Hið íslenzka fornritafélag, 2011), 1, pp. 114–17. Haraldr’s stanza in turn appears to be a paraphrase of a stanza by Glúmr Geirason, mentioning the twelve íþróttir of Haraldr gráfeldr (SkP 1, pp. 261, 264). 123 SkP 2, pp. 114–16. 124 SkP 2, p. 105.

2.3 The Intermediate Period (c. 995–c. 1120)

71

pp. 259–62. In a poem commemorating Haraldr, Stúfr inn blindi uses the kenning Gríðar byrr (Gríðr’s [giantess’s] wind [mind]).125 This relatively rich record of specific mythological references makes Haraldr stand out among Norwegian kings in the period between Hallfreðr’s memorial poem to Óláfr Tryggvason and c. 1120. The only other specific mythological reference in a poem addressed to a Norwegian king in the period is gífrs veðr (the troll-woman’s wind [mind]) in Arnórr’s Magnússdrápa.126 We saw that Haraldr received two kennings of the same type, and it is possible that these were seen as relatively innocuous since they lacked a clear connection to the divine realm. Whatever the case may be, the eleventh century was the great era of royal panegyrics, and it is therefore more than a little noteworthy that only one specific mythological reference can be found in poetry to other Norwegian kings than Haraldr. This supports the idea that the religious zeal of kings – Norwegian kings in particular – was the main vector in the disappearance of specific mythological references. It therefore seems likely that Haraldr really did represent a different attitude to mytho-poetics than other Norwegian kings of the eleventh century. In this context, it is worth noting that Illugi Bryndœlaskáld in four half stanzas praises Haraldr with reference to the cycle of Sigurðr and the dragon, Brynhildr and Atli.127 This is heroic rather than mythological material, but it is nonetheless an elaborate example of how poets embellished their compositions by allusions to stories about the distant past, stories which the hearer needed to know in order to decode the language. The references to Sigurðr occupy lines two and three of every half stanza, and we thus find here a systematic placement of the references reminiscent of – and possibly influenced by – Kormákr’s Sigurðardrápa (see pp. 28–29). Furthermore, the poet uses the word eiskald n. (heart), otherwise found only in Fáfnismál in the poetic corpus.128 In both instances, it refers to Fáfnir’s heart. Since Fafnismál contains three unambiguous instances of alliteration in vr-, which passed out of use c. 1000, it would appear that Illugi is paraphrasing a famous poem about Sigurðr in his

125 SkP 2, p. 351. 126 SkP 2, p. 211. 127 SkP 2, pp. 282–85. 128 Kommentar zu den Liedern der Edda, 5: Heldenlieder, ed. Klaus von See et al. (Heidelberg: Carl Winter, 2006), p. 460. The two poetic occurrences suggest that this word is neuter and could be used either in the singular or in the plural (eiskǫld) to denote ‘heart’ in the singular. The statement ‘(also eiskǫld f., eiskaldr m.)’ in SkP 2, p. 283, is presumably due to the fact that the scribe of CR seems to have understood the word as a masculine u-stem, as indicated by his use of ‘etiN’ (=etinn, acc. m.), and to the nominative eiskǫld in the prose of Litla Skálda and Skáldskaparmál (see p. 139 n. 133). The adjective in Illugi’s stanza shows unambiguously that the word is neuter, however (‘eiskaldi [. . .] beisku’). The prose occurrences could just as well be neuter plural as feminine singular, and the scribal representation in CR is probably due to misinterpretation of a neuter plural as a masculine singular. The choice of the singular was presumably guided by the fact that the ordinary word for heart, hjarta, is singular (feminine is ruled out by the participle in CR).

72

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

praise of Haraldr-as-Sigurðr.129 Just like Arnórr, with his mythological references and paraphrase of Vǫluspá in Þorfinnsdrápa, Illugi engages in a complex interplay of traditional motifs and poetic quotation, and it is noteworthy that the two poems are roughly contemporary. Arnórr’s and Illugi’s level of intertextual sophistication stand out in the period and are similar to that of the learned twelfth-century poets. Through these two poets, we see that this approach to the poetic tradition did not die out during the intermediate period, but that it led a marginal existence before it resurfaced in the following century. Illugi’s stanzas, albeit heroic rather than mythological, fit well within the context of mythological references in poetry connected to Haraldr. Though of less source-critical value, the later saga tradition supports that Haraldr had an avid interest in poetry.130 This description of Haraldr’s uniqueness among Norwegian kings of the eleventh century, however, presupposes that three mythological references connected to the saint-king Óláfr Haraldsson (r. 1015–1030) are spurious. They are attributed to him and to his chief poet, Sighvatr. I turn now to these. Sighvatr has two specific mythological kennings. The first is Dáins mun- (var. munn-) vágr (Dáinn’s [a dwarf’s] delight- (var. mouth-) wave [poetry]).131 This quotation is found only in two late manuscripts which feature rich and internally related interpolations of a legendary character, namely Flateyjarbók and Tómasskinna (c. 1387–1400). A variant of the stanza lacking the kenning in question is also found in Bæjarbók, which belongs to the same manuscript group.132 The second is hella mildings mál (caves’ lord’s [giant’s] speech [gold]), whereas Óláfr’s kenning is Mardallar grátr (Mardǫll’s [Freyja’s] weeping [gold]).133 Both are found in Flateyjarbók only.134 The transmission of these stanzas in the so-called ‘interpolated’ manuscripts only (in two cases in Flateyjarbók only) of Óláfs saga is noteworthy. The interpolations in this group of manuscripts from the decades around 1400 have been

129 Haukur Þorgeirsson, ‘The Dating of Eddic Poetry – Evidence from Alliteration’, in Approaches to Nordic and Germanic Poetry, ed. Kristján Árnason et al. (Reykjavík: Háskólaútgáfan, 2017), pp. 35, 54. The occurrences are formulaic, providing the alliterative pair of the long ljóðaháttr lines based on the words vreiðr (angry) and vega (fight), and their formulaic nature might lead one to suspect that they could have been imitated also after the loss of initial v- in vr-initial words. As Haukur Þorgeirsson demonstrates, however, this is unlikely, since vr-initial alliteration passed out of use in skaldic poetry c. 1000 and since later scribes did not understand its dynamics. 130 See, for instance, Morkinskinna, ed. Ármann Jakobsson and Þórður Ingi Guðjónsson, 1, p. 286. 131 SkP 1, pp. 733–34. 132 See Flateyjarbók, ed. Guðbrandr Vigfusson and C. R. Unger, 2, pp. 372; ‘Saga Óláfs konungs hins helga’. Den store saga om Olav den hellige efter pergamenthåndskrift i Kungliga Biblioteket i Stockholm nr. 2 4to med varianter fra andre håndskrifter, ed. Oscar Albert Johnsen and Jón Helgason (Oslo: Kjeldeskriftfondet, 1941), pp. 831, 839 (apparatus); Sigurður Nordal, Om Olaf den helliges saga. En kritisk undersøgelse (København: Gad, 1914), pp. 94, 106–07. 133 See SkP 1, pp. 666–68, 529–31. 134 Flateyjarbók, ed. Guðbrandr Vigfusson and C. R. Unger, 2, pp. 341, 68.

2.3 The Intermediate Period (c. 995–c. 1120)

73

thoroughly studied by Sigurður Nordal and Oscar Albert Johnsen and Jón Helgason, and it would appear that a large group of interpolations that cannot be traced to the so-called Legendary Saga comes from what in Flateyjarbók is called ‘Styrmir’s book’.135 For various reasons, the case for tracing these interpolations to Styrmir’s book is strong. First and foremost, Flateyjarbók adds the so-called ‘articles of Styrmir’ to the end of the saga. This is a set of chapters in Flateyjarbók that are explicitly said to ‘standa í sjálfri lífssǫgu hins heilaga Óláfs konungs þeiri sǫmu er Styrmir prestr hinn fróði hefr saman sett þótt sé eigi svá fulliga skrifaðir hér fyrr í bókinni’ (are found the very Life of the holy king Óláfr which the wise priest Styrmir has compiled, though they have not been as fully written down earlier in the book).136 It is clear from this that the following texts come from Styrmir and that his text has also been used earlier in the compilation. The latter is further borne out by other references in Flateyjarbók: ‘Svá segir Styrmir hinn fróði [. . .]’ (so says Styrmir the wise); ‘[. . .] er Styrmir reiknar í sinni bók’ (as Styrmir counts in his book); ‘eptir sǫgn Styrmis hins fróða’ (according to the account of Styrmir the wise).137 Much of the material in the ‘articles of Styrmir’ belongs to the texts that are common to the interpolated manuscripts but are not found elsewhere.138 Based on this observation, Sigurður Nordal has posited that much other material that is unique to this group of manuscripts is also drawn from Styrmir’s book. The argument appears to me to be sound, and it was accepted by Oscar Albert Johnsen and Jón Helgason, who are the two scholars who have studied the transmission of the Greater Saga of Óláfr and the material added to it most thoroughly.139 Sighvatr’s stanza containing the kenning hella mildings mál is found in a chapter with two references to Styrmir.140 The other two stanzas come in chapters now also attributed to Styrmir, although explicit references to him are lacking.141 It is

135 Sigurður Nordal, Om Olaf den helliges saga, pp. 69–96; ‘Saga Óláfs konungs hins helga’, ed. Johnsen and Jón Helgason, pp. 1127–29. 136 Flateyjarbók, ed. Guðbrandr Vigfusson and C. R. Unger, 3, p. 237. 137 Flateyjarbók, ed. Guðbrandr Vigfusson and C. R. Unger, 2, pp. 67, 68, 118. 138 The parallel texts are conveniently gathered in ‘Saga Óláfs konungs hins helga’, ed. Johnsen and Jón Helgason, pp. 683–713. 139 The interpolated texts were edited by Jón Helgason alone, but both scholars have corrected the edition as a whole (‘Saga Óláfs konungs hins helga’, ed. Johnsen and Jón Helgason, p. xvii). On their view of Sigurður Nordal’s work, see pp. 127–29. Jón Helgason has added Sigurður Nordal’s numbers under the titles of the individual interpolations and clearly considers his conclusions to be in all essentials correct (pp. 714–852). 140 Flateyjarbók, ed. Guðbrandr Vigfusson and C. R. Unger, 2, pp. 67, 68. 141 Dáins mun- (var. munn-) vágr in Sigurður Nordal’s no. 96 (Sighvatr’s return from Rome; Om Olaf den helliges saga, pp. 94, 106–07; ‘Saga Óláfs konungs hins helga’, ed. Johnsen and Jón Helgason, pp. 831, 839 (apparatus); Flateyjarbók, ed. Guðbrandr Vigfusson and C. R. Unger, 2, pp. 371–73); Mardallar grátr in Sigurður Nordal’s no. 73 (Óláfr‘s verses on Ingigerðr; Om Olaf den

74

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

remarkable that the specific mythological references all occur in passages that are unique to the interpolated group. This makes one suspect that the group in these instances draws on one and the same source and one, furthermore, which contained a significant amount of pseudonymous poetry. Since there is an explicit reference to Styrmir in one of these chapters, it seems likely that his book is the source of all three. Sigurður Nordal, Oscar Albert Johnsen and Jón Helgason thought the same about these chapters, based on comparative studies of the transmission and content of sagas about Saint Óláfr. I believe that the combined evidence of their studies and my observations suffices to trace all three stanzas to Styrmir’s book. These kennings are much more conspicuous than those of Hallfreðr or even Haraldr harðráði, since the corpus of poetry connected to Óláfr, including that of Sighvatr, is very rich, and nowhere else do we find anything like this. They are also surprising from a religious and political point of view, since we are here dealing with the saint king of Norway and his chief poet. In all likelihood, then, these kennings should be attributed to a later age, when such language was again in vogue, and thus to the late twelfth or early thirteenth century. One indication of a more exact date of composition of these stanzas may be found in Ólafr’s poetry. Óláfr’s stanza containing the kenning Mardallar grátr also employs extended metaphor. Ingigerðr, who is the topic of the stanza, is referred to as a tree which was previously green, but which has now grown pale. Extended metaphor comparing a woman to a tree which is green in the summer and pale in the autumn also occurs in another of Óláfr’s stanzas which is found only in the interpolated manuscripts.142 Because of the strong thematic connection between the stanzas, the first of which can be judged spurious for reasons given above, and because of the rarity of extended metaphor in the intermediate period (apart from extended metaphor on nautical themes), it seems likely both stanzas were composed for a saga account about Óláfr. This is corroborated by a linguistic feature in the second stanza. This stanza has a line which closely resembles the first line of a situational stanza attributed to Bjǫrn Breiðvíkingakappi in Eyrbyggja saga. Bjǫrn’s stanza is also about the loss of a woman, and it mentions trees and refers to the woman as a tree (þella), even though it does not employ extended metaphor. In Óláfr’s stanza, the first line reads ‘Þann myndak við vilja [. . .]’ (I would wish that tree [woman] to [. . .]), whereas Bjǫrn’s first line reads ‘Guls mundum vit vilja [. . .]’ (Of yellow we would wish [. . .]).143 The related topic of the stanzas shows that the similarity of the lines is due to paraphrase rather than coincidence.

helliges saga, pp. 93, 105–06; ‘Saga Óláfs konungs hins helga’, ed. Johnsen and Jón Helgason, p. 820; Flateyjarbók, ed. Guðbrandr Vigfusson and C. R. Unger, 2, p. 341). See also SkP 1, p. 699. 142 SkP 1, p. 519. 143 Eyrbyggja saga, ed. Einar Ól. Sveinsson and Matthías Þórðarson, p. 78; Skj B 1, p. 125; A 1 p. 133.

2.4 The Late Period up to Snorri and his nephews (c. 1120–c. 1300)

75

The dating criterion lies in the words við/vit. In standard normalization, the first is accusative of viðr ‘wood’, whereas the second is normalized as vit and means ‘we’. In the thirteenth century, however, final -t merged with final -ð in unstressed syllables and unstressed words, so that the pronunciation of both became identical.144 This was presumably the situation when the paraphrase was made, since at an earlier stage, við and vit would not have been homonyms. The date of this merger means that the pseudonymous composition probably took place in Styrmir’s day, and that he or a rough contemporary of his is likely to be the poet behind Óláfr’s two stanzas. In Chapter Four, we return to Styrmir’s book and its implications for the evolution of prosimetrical saga writing. With these stanzas by Sighvatr and Óláfr excluded from the eleventh century, specific mythological references in poetry connected to kings turn out to be rare indeed: three in Hallfreðr’s memorial poem to Óláfr Tryggvason, two in a poem to Knútr, one in a poem to Magnús Haraldsson and one in a poem to the Swedish Óláfr. Haraldr harðráði stands out, producing one and receiving four or five. A number of additional references can be found in poetry from contexts not connected to kings, notably in the poets connected to Eiríkr jarl, son of the great pagan Hákon. One poet, Hofgarða-Refr, remains in Iceland and in the 1030s composes as if nothing had happened to the old gods. The distribution thus suggests that the new trends emanated from the royal courts, with the saint king Óláfr Haraldsson and the substantial poetic output connected to him setting an important precedent for later rulers and their poets. We have here seen that all the features with which we are concerned nearly disappeared during the period c. 995–c. 1120 (and certainly c. 1015–1120). Generic mythological references remained in use, but specific ones largely disappeared. It would seem that specific mythological references were the real problem and that specific mythological references, wordplay and tmesis were so intertwined with them in skaldic practice that they disappeared at the same time (except tmesis in Sighvatr). Nonetheless, specific references do crop up, suggesting that the old poetry was transmitted and that the rich artistic impulses embedded within it were just waiting to resurface. This is precisely what happened in the following period.

2.4 The Late Period up to Snorri and his nephews (c. 1120–c. 1300) From around 1120 onwards all relevant features reappear with a vengeance. There is, however, a marked difference in the use of advanced diction in the early and the late period. In the early period, many of the experiments with diction are

144 Scribal norms, however, lagged behind. See Adolf Noreen, Altisländische Grammatik, 4th ed. (Halle: Niemeyer, 1923), p. 183.

76

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

found in panegyrics to rulers which were performed in a court setting. In the late period, by contrast, advanced diction became a preoccupation of the learned. Although the ties to the court were not cut, the prime locus for advanced diction was no longer in panegyrics, but in poetry aimed at fellow scholar-poets. The features involved are found in a number of poets, and the ones who exhibit one of them typically also display several of the others. Wordplay reappears, but just as in the early period, it seems never to have been very common. 145 Examples of extended metaphors and specific mythological references are too numerous to be listed here (though see the overview of kennings on pp. 41–43). I shall not aim here at an exhaustive overview of the twelfth century, but rather at demonstrating how the revival plays out in two representative poets, namely Einarr Skúlason and Hallar-Steinn. While these two may suffice to give an impression of the developments during this period, other poets would merit study equally well in this regard. Most notable among these may be Haukr Valdísarson – a twelfth-century poet who packed the two opening stanzas of his Íslendingadrápa with obscure mythological references, wordplay and extended metaphors, and who also made repeated use of tmesis.146 I also move beyond the twelfth century for a brief discussion of how the legacy was upheld and further developed by Snorri and his nephews, through whom these stylistic ideals entered the centre-stage of Old Icelandic literature.

145 Examples of wordplay not treated here include: Einarr Skúlason (SkP 3, p. 160), Rǫgnvaldr (SkP 3, p. 345); Rǫgnvaldr and Hallr Þórarinsson (SkP 3, pp. 1081–82), Rúnolfr Ketilsson (Skj B 1, pp. 513–14; A 1, pp. 533), Haukr Valdísarson (Skj B 1, pp. 539 (1.8; 2.2; 3.2), 543 (17.4); A 1, pp. 556, 558), Nóregs konunga tal (SkP 2, p. 802), Stríðkeravísur (SkP 3, p. 628), Guðmundr Oddsson (Skj B 2, p. 92 (7.4); A 2, p. 81), Sturla Þórðarson (Skj B 2, p. 136, (4.4; 4.7), Einarr Gilsson (Skj B 2, p. 435–36 (4.4; 8.5); A 2, pp. 408–09; cf. Meissner, Die Kenningar, p. 85), Arngrímr Brandsson (Skj B 2, p. 386 (54.4); A 2, p. 359). 146 For the first two stanzas, see Skj B 1, p. 539; A 1, p. 556. For tmesis, stanza 12 has fannar arm-viðir (the snowdrift’s arm-trees) for arm-fannar viðir (arm-snowdrift’s [gold’s or silver’s] trees [men]), stanza 14 sárjǫkuls geima þrymsvellir (the wound-glacier’s sea’s noise-increaser) for sárgeima jǫkuls þrymsvellir (the wound-sea’s [blood’s] glacier’s [sword’s] noise- [battle-] increaser [warrior]), stanza 23 has unnar hyr- [. . .] tælir (the wave’s fire- [gold-] destroyer [generous man]) (Finnur Jónsson, ‘Kenningers ledomstilling’, pp. 6–7, 20–21; Skj B 1, p. 542, (12, 14), 544 (23); A 1, p. 558–59. Kock accepts all of these instances (Den norsk-isländska skaldediktningen, ed. Ernst Albin Kock, 2 vols (Lund: Gleerup, 1946–49), 1, pp. 263–64). On the dating of Íslendingadrápa, see Jónas Kristjánsson, ‘Íslendingadrápa and Oral Tradition’, Gripla, 1 (1975), pp. 76–91. Later dates have been proposed, but largely without responding to the solid arguments of Jónas Kristjánsson: thus Bjarni Einarsson, ‘Íslendingadrápa’, Tímarit Háskóla Íslands, 4 (1989), pp. 127–31; Ernst Walter, ‘Argumente zur Bestimmung des Alters der Íslendingadrápa Hauks Valdísarsonar’, in Deutsch-nordische Begegnungen. 9. Arbeitstagung der Skandinavisten des deutschen Sprachgebiets 1989 in Svendborg, ed. Kurt Braunmüller and Mogens Brøndsted (Odense: Odense University Press, 1991), pp. 96–103). In Bjarni Einarsson’s article, Jónas Kristjánsson’s analysis of the content of Íslendingadrápa and the sagas is largely ignored, and the aðalhending on a: ǫ in Íslendingadrápa, not otherwise found after c. 1200, is not accorded its appropriate weight (stanza 13 line 6; Íslendingadrápa has no skothendingar for aðalhendingar in even lines; see above pp. 60–61).

2.4 The Late Period up to Snorri and his nephews (c. 1120–c. 1300)

77

2.4.1 Einarr Skúlason Guðrún Nordal and Christopher Abram have drawn attention to the importance of the priest and poet Einarr Skúlason for the changes that took place within poetics from this time onward.147 Although the vicissitudes of transmission may have skewed the picture somewhat, it seems safe to assume that he was a pivotal figure also for the revival of the features of diction considered here. Einarr was probably born around 1090, and he is mentioned as a priest in Morkinskinna and in a list of high-born priests for the year 1143.148 He may have been the brother of Snorri’s maternal grandfather. At least, he belonged to Snorri’s family.149 Einarr’s education as a priest is important in this context, since the technical and exemplary nature of his poetry betrays that he had studied diction systematically in the spirit of Latin grammatica, rather than simply learning the trade from his older contemporaries. His learning is also evident in other ways, such as in his updated religious references, as well as by a stanza of his on cosmology.150 Collectively, the two versions of Skáldatal give his name as court poet to no fewer than twelve kings and earls in Scandinavia.151 In Morkinskinna, he is said to have risen to the prominent position of marshal under king Eysteinn (d. 1123).152 Einarr is most famous for having composed the poem Geisli in honour of Saint Óláfr, a poem which was performed in Kristkirkja in Niðaróss in 1153 and accompanied by many portents and a delightful smell.153 Geisli will figure prominently in the discussion below, but other poems by Einarr are equally important for the present argument, since these share features with the most semantically experimental poetry of the tenth century. Furthermore, Einarr’s poetry attracted the attention of Einarr’s younger relatives in the thirteenth century, Snorri Sturluson and Óláfr Þórðarson, the two most important figures for nativizing and Latinizing grammatica, respectively.154 Both drew

147 Guðrún Nordal, Skaldic Versifying; eadem, Tools of Literacy. The Role of Skaldic Verse in Icelandic Textual Culture of the Twelfth and Thirteenth Centuries (Toronto: University of Toronto Press, 2001), pp. 77–78, 84–6, 88, 337; Abram, ‘The Post-Pagan Mythological Kenning’. 148 Morkinskinna, ed. Ármann Jakobsson and Þórður Ingi Guðjónsson, 2, pp. 87, 124; Guðrún Nordal, Skaldic Versifying, p. 4. 149 Finnur Jónsson, Den oldnorske og oldislandske litteraturs historie, 2, p. 62. 150 See Martin Chase, ‘Framir kynnask vátta mál: The Christian Background of Einarr Skúlason’s Geisli’, in Til heiðurs og hugbótar. Greinar um trúarkveðskap fyrri alda, ed. Svanhildur Óskarsdóttir and Anna Guðmundsdóttir (Reykholt: Snorrastofa, 2003), pp. 11–32; Margaret Clunies Ross and Kari Ellen Gade, ‘Cosmology and Skaldic Poetry’, Journal of English and Germanic Philology, 111:2 (2012), pp. 199–207. 151 Guðrún Nordal, Skaldic Versifying, p. 5. 152 Morkinskinna, ed. Ármann Jakobsson and Þórður Ingi Guðjónsson, p. 222. 153 SkP 7, p. 5; Morkinskinna, ed. Ármann Jakobsson and Þórður Ingi Guðjónsson, p. 222. 154 Einarr’s family connections are not well known, but a fourteenth century scribe of Gunnlaugs saga comments on the poetic streak within the family of the Mýrarmenn – Egill, who lived at Borg á mýrum, and his relatives – and singles out Einarr and Snorri. One may also note that the poet

78

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

heavily on Einarr, who is by far the most commonly quoted poet in Skáldskaparmál (35 times), and who is also quoted three times in the Third Grammatical Treatise (c. 1250).155

2.4.2 Øxarflokkr Many of the quotations in Skáldskaparmál share a number of marked features, which have allowed editors to gather them into a poem called Øxarflokkr (poem about an axe; the name is modern). Øxarflokkr is composed in thanks for an axe or a sword or both; the references are somewhat confusing.156 This poem ties into a tradition of ekphrastic poetry – the so-called shield poems – but unlike decorated shields, there was not much imagery to be described, real or fictive, on an axe or a sword.157 The only feature of representative art to be described in the poem is the head of a dragon; the images in the poem are otherwise contained within its rich poetic language. In the poem, we find complex mythological diction, extended metaphor and wordplay in a constellation reminiscent of Hallfreðr’s Hákonardrápa. In the intermediate period after Hákonardrápa, we find nothing that even remotely matches these poems. There are, however, also subtle differences between the two, and I proceed now to an analysis of Øxarflokkr. Even though Øxarflokkr has the traditional form of a poem in thanks for a precious object, it is also structured as a showpiece of diction, almost as an exercise. This is not to say that the poem is purely mnemonic or pedagogical, unlike four stanzas Einarr composed enumerating heiti for the sea involving belts and jewellery (such as ‘Radøy’s [an island’s] band [sea]’).158 The fact that he composed such mnemonic stanzas, however, underlines his studied interest in diction. This is conspicuous also in Øxarflokkr, where Einarr revels in variations on a limited number of mythological kenning types for gold. Outside of Einarr’s own poetry, a similar display is found in stanzas four to six of the elusive poem Bjarkamál, where an enumeration of kennings for gold is given in order to underline the generosity of the ruler (but also, perhaps, for mnemonic purposes).159 Metre and content have suggested to scholars that these stanzas may be

Skúli Þorsteinsson was a grandson of Egill; see Geisli, ed. Martin Chase (Toronto: University of Toronto Press, 2005), p. 9 note 5. 155 Guðrún Nordal, Tools of Literacy, pp. 77–78, 84–6, 88, 337. He is also quoted once in the fragment of the Fifth Grammatical Treatise; see below p. 188. 156 SkP 3, p. 140. 157 On the shield poems and other ekphrastic poetry, see Margaret Clunies Ross, ‘Stylistic and Generic Identifiers of the Old Norse Skaldic Ekphrasis’, Viking and Medieval Scandinavia, 3 (2007), pp. 159–92); Poole, ‘Scholars and Skalds’. 158 SkP 3, pp. 171–72, 177–79. 159 Snorri, Skáldskaparmál, ed. Faulkes, pp. 60–61, 188–89; Finnur Jonsson, Den oldnorske og oldislandske litteraturs historie, 1, p. 464–65.

2.4 The Late Period up to Snorri and his nephews (c. 1120–c. 1300)

79

twelfth-century additions to the poem.160 Whatever the case may be, the exposition in Øxarflokkr is incomparably more complex than in Bjarkamál. Øxarflokkr’s most conspicuous feature is the large amount of specific mythological references that it contains, above all in kennings for gold and treasure involving the goddess Freyja, as in the clause ‘Freys nipt berr sinnar móður bráa driptir’ (Freyr’s niece [Hnoss > hnoss (treasure)] carries her mother’s (Freyja’s) eyelashes’ snowdrifts [tears > gold]). The short poem counts nine such kennings (the original length of Øxarflokkr is not known).161 Five of these kennings, furthermore, involve ofljóst through the substitution of hnoss (treasure) for Hnoss (daughter of Freyja), as seen in the example above.162 Apart from adding to the complexity of the poem, these instances of ofljóst also constitute personification, and the image of the axe as the young daughter of a goddess serves as the basis for extended metaphor, most notably in stanza 5.163 There, a warrior gives away a daughter, as a father would his own daughter in marriage: ‘Váfaðar þing-þrøngvir gaf mér dóttur’ (Váfuðr’s (Óðinn’s) assembly [battle-] inciter [warrior] gave [his] daughter to me). The daughter in question is not actually the warrior’s daughter, but rather Vanabrúðar dóttir (The Vanirs’ bride’s [Freyja’s] daughter [Hnoss > hnoss (treasure)]). The combination of gefa (give) and dóttir (daughter), nonetheless, evokes marital imagery. In the second half, Gefn’s (Freyja’s) maiden is led (as for the marriage night) to the bed of the poet: ‘ríkr mótvaldr leiddi Gefnar mey á beð skáldi’. This particular maiden, however, is really a weapon, and she is therefore accorded the attribute þróttǫflug (powerful), which adds a further level of referentiality. Here Einarr employs tmesis as well: line 3 reads þing-Váfaðar þrøngvir, which must be read as Váfaðar þing-þrøngvir (Váfuðr’s (Óðinn’s) assembly [battle-] inciter [warrior]).164 In this single stanza, then, Einarr employs all four features under study – specific mythological references, wordplay, extended metaphor and tmesis.165

160 See SkP 3, pp. 496, 500; Meissner, Die Kenningar, p. 225. 161 These are, in addition to the two in the clause above: Mardallar grátr (Mardǫll’s (Freyja’s) weeping [gold]), Óðs beðvinu augna regn (Óðr’s bedfellow’s [Freja’s] eyes’ rain [tears > gold]), Hǫrnar hróðrbarn (Hǫrn’s (Freyja’s) celebrated child [Hnoss > hnoss (treasure)]), Njarðar dóttur barn (Njǫrðr’s daughter’s [Freyja’s] child [Hnoss > hnoss (treasure)], Vanabrúðar dóttir (The Vanirs’ bride’s [Freyja’s] daughter [Hnoss > hnoss (treasure)]), Gefnar mœr (Gefn’s (Freyja’s) maiden [Hnoss > hnoss (treasure)]), Freyju hvarmþeyr (Freyja’s eyebrow-thaw [tears > gold]) (SkP 3, pp. 140–51). 162 Hnoss herself probably originated as a personification of treasure and was named accordingly. No matter her obscure personal qualities, though, these kennings clearly remain ofljóst, decodable only through detailed knowledge of the pantheon (cf. above p. 55). 163 SkP 3, pp. 145–46. This is one of the relatively few stanzas to be quoted in full in Skáldskaparmál, rather than being broken up into half stanzas (Snorri, Skáldskaparmál, ed. Faulkes, p. 44). 164 Finnur Jónsson, ‘Kenningers led-omstilling’, p. 6. Accepted by Kock (Kock, Notationes Norroenae, § 2008 H and SkP 3, p. 146). 165 Finnur Jónsson also claims to have found two instances of tmesis in Geisli, but these are not wholly convincing (‘Kenningers led-omstilling’, p. 20; cf. SkP 7, pp. 16–17, 22–23).

80

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

At one point in Øxarflokkr, Einarr investigates the referential possibilities of extended metaphor. In the description of the axe in stanza 8, gold decoration on silver is described with the words ‘hafleygr þrymr á hvítum digulskafli’ (the seaflame [gold] rests on the white crucible-snowdrift [silver]), creating an antithesis (fire–snow) and an oxymoron (fire resting on snow) based on kenning structures. In the second half, the oxymoron is topicalized: ‘aldri má fyr áls hrynbrautar eldi skála snæ bræða’ (the scales’ snow [silver] can never be melted by the eel’s surging path’s [sea’s] fire [gold]).166 The image of snow is elaborated by means of the fire, which could have melted the snow if it were really snow. The point of the statement is that this particular fire (gold) cannot melt this kind of snow (silver); that would demand actual fire – as would, of course, melting snow. While still a case of extended metaphor, these words illustrate the absurdity of that figure and betray the metaperspective of a grammatical mind.167 The extreme complexity of his diction suggests that one of Einarr’s aims was to verbally outdo his predecessors. The similarities with Hákonardrápa are particularly striking: female personification (Jǫrð/Hnoss) combined with recurrent wordplay (Jǫrð–jǫrð/Hnoss–hnoss), extended metaphor on a sexual and marital theme and interplay between the primary and secondary level of referentiality. Indeed, the persistent use of these devices is nowhere as pronounced as in these two poems in the preserved corpus, suggesting that Einarr is paraphrasing the older poet. If this is the case, it is important for understanding the dynamics of the revival of mythology and complex diction, and the matter thus warrants further exploration. Meissner in Die Kenningar discusses Hnoss as daughter of Freyja and signifying ‘precious object’ only in relation to Øxarflokkr.168 Nonetheless, the kenning type does occur in one more place, in the first stanza in a cluster of six at the beginning of the sixth chapter of Hallfreðar saga. When thanking the king for a sword, much as Einarr does for an axe and/or a sword, Hallfreðr says ‘nú ák Sýrar mey dýra’ (now I own Sýr’s [Freyja’s] precious daughter [Hnoss > hnoss]).169 The verb eiga, here in the form ák (I own), is typically used of being married, and this is how Hallfreðr uses it in Hákonardrápa (see p. 55). Just as in Øxarflokkr, Hallfreðr here combines thanking the ruler for the gift of a weapon with wordplay on Hnoss and erotic/marital imagery. The matter seems straightforward: Hallfreðr is the poet whom Einarr chose to emulate.

166 Snorri, Skáldskaparmál, ed. Faulkes, p. 62; SkP 3, pp. 148–49. 167 Poole, ‘Identity Poetics among the Icelandic Skalds’, p. 155, thinks that the imagery here alludes to the Icelandic landscape with its extremes of fire and snow. There is, however, no mention of the land in the poem, and the focus lies on self-contained, poetic semantics throughout. 168 Meissner, Die Kenningar, discusses the occurrences in Øxarflokkr on p. 66. 169 Vatnsdœla saga, ed. Einar Ól. Sveinsson, p. 156.

2.4 The Late Period up to Snorri and his nephews (c. 1120–c. 1300)

81

The only problem is that in Möðruvallabók, this line reads ‘nú flaustr burar Austra’ (now Austri’s [a dwarf’s] son’s [a dwarf’s] boat [poetry]). This reading is adopted, with emendation, in Finnur Jónsson’s restored text in the B-volumes of Skj, and since Meissner based his catalogue on this text, the kenning Sýrar mœr never made it into Die Kenningar. The text in Möðruvallabók does not make syntactical sense in the stanza, and Finnur Jónsson thus constructed the word núflaust, based on a Norwegian dialectal word nuv, meaning ‘gloomy (of person or horse)’.170 Finnur Jónsson translates his own coinage as ‘gloom-less’, that is, ‘happy’, and this reading was accepted by Kock.171 Flaust(r) is a skaldic word, however, and replacing it with the unattested núflaust is certainly unwarranted. This half stanza has not fared well in Möðruvallabók. Line two has mér where the other manuscripts have konungr, which is clearly the correct reading. Line three has ‘við nǫkkva’ (for the boat), which is tautological to flaustr in the fourth. The explanation for the startlingly different reading in Möðruvallabók probably lies in the extreme rarity of kennings for Hnoss, which are found only here and in Øxarflokkr. This hypothesis is supported by the fact that the kenning Sýrar mœr (Sýr’s [Freyja’s] daughter [Hnoss > hnoss]) has caused trouble also for the scribes of Bergsbók and Flateyjarbók, who write ‘syriar’ (?) and ‘skyrra’ (more clear). A likely explanation for the reading in Möðruvallabók is thus that the kenning was unfamiliar to the scribe or his predecessor and that he therefore altered the line, apparently based on what he thought to be the mention of a boat in line three (nǫkkvi, found in the other manuscripts, can be neuter dative of nǫkkurr ‘someone, something’ or nominative of nǫkkvi, obl. nǫkkva ‘boat’).172 Nú had to be retained because it carries alliteration. The scribe ended up with two boats and an impossible nú, but managed to convey an overall sense that Hallfreðr was rewarded for his poetry. The resulting kenning (‘dwarf’s boat [poetry]’) suggests that the scribe may have been influenced by Skáldskaparmál, since such a kenning is found in one preserved stanza only, transmitted and discussed there (see below pp. 161–64). The same chapter of Skáldskaparmál also quotes stanza three of Vellekla, containing the rhyme pair austr (bilge water) – flausta (of ships). The semantics of austr (bilge water) and Austra (Austri’s) differs, but the convergence of the unique kenning type and the same rhyme pair, albeit with one paronym, suggests that the alteration in Möðruvallabók was made based on this chapter of Skáldskaparmál. Several factors thus combine to indicate that the other manuscripts have the older reading. Based on these observations, we may conclude that Einarr paraphrased Hallfreðr’s stanza, using wordplay on Hnoss in expressing his gratitude for a precious object.

170 Finnur Jónsson, ‘Versene i Hallfredssaga’, Arkiv för nordisk filologi, 18 (1902), p. 322. 171 Kock, Den norsk-isländska skaldediktningen, 1, p. 85. 172 On the form nǫkkvi, see Noreen, Altisländische Grammatik, p. 322.

82

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

In stanza five of Øxarflokkr, Einarr even paraphrases Hallfreðr’s Sýrar mœr in the kenning Gefnar mœr (Gefn’s [Freyja’s] daughter [Hnoss]).173 We shall soon see that he also paraphrased Hallfreðr’s following stanza in Hallfreðar saga in the poem Geisli. In light of these paraphrases, it becomes all the more likely that Hallfreðr also provided Einarr with the model for erotic wordplay with a female deity in Hákonardrápa. The importance of emulation with the old poets for the re-emergence of mythological and advanced diction is nowhere as clear as in the case of Hallfreðr’s poetry and Einarr’s Øxarflokkr.174 Enarr’s paraphrase also shows that he did not invent his mythological imagery, as one might have surmised since the kenning Sýrar mœr is absent from Meissner’s Die Kenningar. Rather, he opted for finding the most obscure references he could in the existing corpus. He went further in this regard than the old poets. Even though the combination of wordplay and personification in Hákonardrápa demands both knowledge and skill in the recipient, the employment of the mythological motif of a god who has sexual intercourse with the earth to portray a conqueror could hardly have been better suited to a warrior culture. When Hnoss is made to serve the same function in Øxarflokkr, however, there is reason to suspect that the imagery itself is largely a by-product of stylistic emulation with the old masters of the skaldic art; Einarr has found a new and contrived way of reviving an old topos.175 The oxymoron-metaphor of fire and snow brings out his studied interests even more clearly, inviting us to what amounts to little more than a game of words. Finally, if Øxarflokkr is compared to earlier ekphrastic poetry, the extent to which the imagery of the poem is contained within diction itself, without reference to a world outside, is noteworthy. Technically, Einarr does indeed outshine his predecessors. This came at a price, however, and it is difficult to find in these stanzas anything other than a poetic exercise for the student of skaldic diction.176

2.4.3 Geisli Einarr apparently chose to emulate Hallfreðr in particular. This is evident not only from Øxarflokkr, but also from Geisli, composed in 1153. Geisli is a hagiographic poem, and presumably because of its topic – the sanctity of the rex perpetuus of

173 SkP 3, p. 145. 174 On skaldic emulation through paraphrase, see Klaus von See, ‘Polemische Zitate in der Skaldendichtung. Hallfrøðr vandræðaskáld und Haldórr ókristni’, Skandinavistik, 7 (1977), pp. 115–19. 175 The simile of a ruler who sexually conquers Óðinn’s mistress (the earth) was a topos (see, for instance, SkP 1, pp. 211–12, 460–61; SkP 2, pp. 114–16; SkP 3, p 238). 176 In Abram’s words, Øxarflokkr ‘smells more of the classroom than the pagan temple’ (‘The PostPagan Mythological Kenning’, p. 50).

2.4 The Late Period up to Snorri and his nephews (c. 1120–c. 1300)

83

Norway, Óláfr Haraldsson – even generic mythological references are rare and inconspicuous, in particular if one considers the poem’s length (71 stanzas).177 Even so, Einarr actively engages with the pagan past in two instances. The first is found in stanza 19. The first line of the second half of Geisli 19 runs: ‘fyrr vas hitt es harra’ (it happened previously that the lord). The same line begins the half stanza that follows Hallfreðr’s Hnoss stanza in Hallfreðar saga.178 Again we see Einarr paraphrasing, or indeed quoting, Hallfreðr. In Hallfreðr’s stanza, harri (lord) refers to Óðinn, to whom Hallfreðr says that he used to sacrifice: Fyrr vas hitt es harra Hliðskjalfar gatk sjalfan – skipt es á gumna giptu – geðskjótan vel blóta.179 It was in the past that I could well perform sacrifices to Hliðskǫlf’s quick-minded lord [Óðinn]. The fortune of men has changed.

Here Óðinn is not only the object of mythological narrative, but of cult, which is very rare in skaldic poetry. In Geisli, by contrast, harri refers to Christ at the moment when he died on the cross. The second half of stanza 19 reads: Fyrr vas hitt es180 harra hauðrtjalda brá dauða happ- (nýtast mér) -mætu (máltól) skini solar.181 It was in the past that the fortunate and excellent light of the sun ceased through the landtents’ [sky’s] lord’s [God’s] death. Speech-tools are useful to me.

177 The poet uses the name Huginn three times. He has the kenning hvítings Hǫrn (the ember’s Hǫrn (Freyja) [woman]) (SkP 7, p. 37) and hjalm-Nirðingar (helmet-Nirðingar (derived from Njǫrðr) [warriors]) (SkP 7, p. 51), and at one point he refers to Sigurðr jórsalafari as Hildar leikmildr (generous with Hildr’s game [war]) (SkP 7, p. 64). 178 This was noted by Bjarni Einarsson, ‘The Last Hour of Hallfreðr vandræðaskáld as Described in Hallfreðarsaga’, in Proceedings of the Eighth Viking Congress, Århus 24–31 August 1977, ed. Hans Bekker-Nielsen, Peter Foote and Olaf Olsen (Odense: Odense University Press, 1981), p. 218. Bjarni Einarsson argues that the term harri is late and therefore concludes that the paraphrase was made by a later impostor, but this is contradicted by skaldic evidence (LP, s.v.). 179 Skj B 1, p. 158; A 1, p. 168; Vatnsdœla saga, ed. Einar Ól. Sveinsson, p. 157. All follow Möðruvallabók without emendations. I have removed the comma in the first line since it suggests that there is a syntactical difference between the stanzas with regard to clause boundaries. There is not. 180 ‘er’ Bergsbók (here normalized to es): ‘at’ Flateyjarbók. 181 SkP 7, p. 22. Unlike SkP, I choose es rather than at in the first line in accordance with Hallfreðr’s stanza (which is not mentioned in SkP), but this does not affect the semantics or the question of whether this is a quotation or not, since minor changes were common in skaldic paraphrase/quotation.

84

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

In the line ‘fyrr vas hit es harra’, Einarr quotes one of the few references to pagan cult, as opposed to mythology, and appropriates it for the only true object of worship, namely the Christian God. The rhetorical emphasis could hardly have been stronger, even if it is left to the listener to identify and interpret the reference. Before we move on to the next instance where Einarr engages with the pagan past in Geisli, I wish to offer some evidence in support of the fact that we are really dealing with Einarr’s paraphrase of Hallfreðr, rather than a later impostor’s paraphrase of Einarr when composing in the name of Hallfreðr.182 Einarr paraphrases or quotes Hallfreðr twice (Hnoss and ‘fyrr vas hitt es harra’), and the two instances are found in two very different poems: the secular, paganizing Øxarflokkr and the devotional Geisli. Øxarflokkr and Geisli have come down to us through two separate lines of transmission: namely, Skáldskaparmál and histories of Norway.183 Furthermore, medieval manuscripts containing poetry suggest that it was collected based on metre and content, not based on the name of the poet. The transmission of Einarr’s poetry in particular and the selection criteria of skaldic poetry in general therefore make it unlikely that a later impostor would have had access to a collection where Einarr’s two stanzas occurred together (the relevant stanza from Geisli is not found in any grammatical manuscripts). Finally, it is difficult to see why a later saga author would single out Einarr as his model when composing in the name of Hallfreðr. All of this speaks against that the paraphrases were performed by an impostor. The tendency towards emulation among skalds, by contrast, provides a plausible explanation to why Einarr would have paraphrased two stanzas by one of his predecessors. Einarr’s paraphrase suggests that at least the first two of a cluster of six stanzas transmitted in the beginning of the sixth chapter of Hallfreðar saga are authentic. Einarr presumably thought that the stanzas to which he alluded were authentic, since emulating a pseudonymous or anonymous poet would not provide much rhetorical edge. Furthermore, his repeated allusions to Hallfreðr, as opposed to a more random distribution of allusions to various poets, suggest that the identity of the poet was important to Einarr. Einarr was born sometime around 1090, and his views on Hallfreðr therefore bring us to a point far earlier than that of the saga authors. Furthermore, the aðalhending in a: ǫ in the fifth stanza in the cluster points to a date before c. 1200.184 As I argue on p. 218 below, the main context for the pseudonymous composition of poetry seems to have been the writing of sagas in the thirteenth and fourteenth century: if poetry can be dated to a period before 1200 or 1150, this probably means that the stanzas are authentic. Since two independent

182 See above p. 83 n. 178. 183 SkP 7, pp. 6–7, 20, 55. The portions of three stanzas of Geisli that are quoted in Snorri’s Edda and the Third Grammatical Treatise are not relevant here. 184 See above pp. 60–61.

2.4 The Late Period up to Snorri and his nephews (c. 1120–c. 1300)

85

criteria – linguistic development and paraphrase – point to composition before this time, the case for the authenticity of Hallfreðar saga’s cluster of stanzas is strong. Einarr’s paraphrases have some bearing on the oral transmission of Hallfreðr’s stanzas. Modern scholars refer to all but the first of this group as Hallfreðr’s ‘conversion stanzas’. Einarr’s paraphrases suggest, however, that the first two stanzas of the cluster were transmitted together in the twelfth century, almost as if they were part of the same poem. Furthermore, the second stanza makes little sense without at least some of the ones that follow in the saga, and we may probably conclude that most or all of the stanzas were transmitted as a cluster, especially given the linguistic criterion in the fifth stanza. The cluster indicates a slightly different focus from the modern one, namely one found time and time again in sagas of Icelanders and kings’ sagas: the skald’s encounter with the king and the challenges involved in that encounter. This is a plausible narrative or conceptual frame for the transmission of this group of stanzas down to the time of saga writing, and that transmission probably accounts for the fact that Einarr refers to two stanzas within a cluster that modern scholarly preferences have dispersed. The next instance of paganizing paraphrase in Geisli involves a poet other than Hallfreðr. As noted above, even generic mythological references are few in Geisli, and only once does Einarr depart from this practice, but he does so markedly. To refer to the sword of Óláfr, he uses the kenning gylðis kindar gómsparri (the wolf’s offspring’s [wolf’s] gum-spar [sword]) (stanza 48).185 This kenning can only be decoded through knowledge of the underlying story, namely that the gods neutralized the wolf Fenrir by placing a sword upright in its mouth. The kenning type is earlier used only once, namely by Eyvindr skáldaspillir in the second half of the tenth century, who employs the kenning Fenris varra sparri (Fenrir’s lips’ spar [sword]).186 Einarr’s use of the word sparri is probably intended to signal his awareness of the old poet’s work, much like Gefnar mœr in Øxarflokkr. It would seem that Einarr, even within the devotional context of Geisli, wanted to display his firm grasp on the most inaccessible part of traditional poetic diction – the mythological. There is probably more at work than mere display here, however. Eyvindr’s kenning refers to a sword reddened in battle and thus to a typical skaldic sword: the only thing setting it apart is the rare mythological kenning used for it. Einarr then appropriates this kenning for a very particular sword, namely the saint’s sword which miraculously moved for three consecutive nights and which was then hung as a relic over an altar by the emperor of Constantinople. According to Snorri, Fenrir was a monstrous wolf which would return to wreak havoc in the last days.

185 The kenning is identical in the two medieval manuscripts (Bergsbók and Flateyjarbók) and thus in all probability belongs to the original composition (SkP 7, p. 46; Skj A 1, p. 468). 186 SkP 1, pp. 223–24.

86

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

For the time being, however, he was neutralized by a strong fetter and by the sword in his mouth. Eyvindr’s kenning supports that the motif of the neutralization of the wolf has a pagan background, and according to Vǫluspá, Óðinn would fall fighting a wolf in the last battle.187 It would therefore appear that the essentials of Snorri’s description are old, and we may assume that Einarr’s perception of Fenrir was roughly similar to that of Snorri. This allows us to see what he is doing in his paraphrase. The kenning gylðis kindar gómsparri refers to the instrument which incapacitated an eschatological beast until the last days, and by using it for Óláfr’s sword, Einarr lets that relic serve as a symbol of faith, or even Christ, neutralizing the destructive force of the Devil until the final battle comes. Just like in the case of harri, he has appropriated a pagan motif and in a sophisticated way turned it to a Christian purpose. The kenning gylðis kindar gómsparri is interesting not only with regard to Einarr’s use of tradition. It also appears to conform to a didactic and interpretive approach of the eleventh and twelfth centuries. There are some traces of a typological equation between Norse gods, monsters and heroes on one side and Christ and Satan on the other. Carved images of Sigurðr slaying the dragon in connection to runic inscriptions with Christian content in the eleventh century and on church portals in the twelfth offer one example. In this case we have no texts that explain the Christian connection, but Michael’s and Christ’s defeat of the Devil seem to be a likely point of reference.188 More relevant to Einarr’s treatment of Fenrir is the motif of Þórr’s fishing for the Miðgarðsormr, an eschatological serpent lying in the sea.189 This motif can be found on Christian decorative stones and rune stones in the eleventh century. Its function in a Christian context can be gleaned from two Old Norse texts, both dating to the twelfth century. In the Old Norse translation of the Gospel of Nicodemus, Christ on the cross is said to be bait for the Miðgarðsormr. In an Icelandic homily on Upprisa Dróttins (The Resurrection of the Lord), preserved in a hand from c. 1200, God’s fishing for Leviathan is described, and above Leviathan’s name the

187 Bugge, Norrœn fornkvæði, pp. 17, 24. 188 See Elisabeth Ashman Rowe, ‘Quid Sigvardus cum Christo? Moral Interpretations of Sigurðr Fáfnisbani in Old Norse Literature’, Viking and Medieval Scandinavia, 2 (2006), pp. 167–200. 189 On this topic, see K. Brøndsted, ‘En kirkelig allegori og en nordisk myte’, Historisk Tidskrift (norsk), 2:3 (1882), pp. 21–43; Otto Gschwantler, ‘Christus, Thor und die Midgardschlange’, in Festschrift für Otto Höfler zum 65. Geburtstag, ed. Helmut Birkhan and Otto Gschwantler, 2 vols (Wien: Notring, 1968), 1, pp. 145–68; James Marchand, ‘Leviathan and the Mousetrap in the Niðrstigningasaga’, Scandinavian Studies, 47 (1975), pp. 328–38; Henrik Janson, ‘Edda and “Oral Christianity”: Apocryphal Leaves of the Early Medieval Storyworld in the North’, in The Performance of Christian and Pagan Storyworlds. Non-Canonical Chapters of Nordic Medieval Literature, eds. Lars Boje Mortensen and Tuomas M. S. Lehtonen with Alexandra Bergholm (Turnhout: Brepols, 2013), pp. 171–97.

2.4 The Late Period up to Snorri and his nephews (c. 1120–c. 1300)

87

same hand has written miðgarðarormr (a variant of miðgarðsormr).190 In both instances, the text goes on to clarify that Christ’s humanity is the bait and his divinity the hook, which gets stuck in the serpent’s mouth. A similar interpretive frame probably explains the function of the motif in a Christian context in the eleventh century.

The word ‘miþgarþar ormr’ glossing ‘leviaþan’ in the Icelandic Book of Homilies (Holm. Perg. 15 4to 35v; Image: Stockholm, Kungliga biblioteket).

The other major eschatological beast was Fenrir, and through Einarr’s sophisticated use of skaldic paraphrase, we see how its neutralization through the insertion of an instrument into its mouth could be framed as a symbol of the victory of faith over the forces of evil, just as Þórr’s fishing for the Miðgarðsormr was. With Sigurðr and the dragon, Þórr and the Miðgarðsormr and finally Fenrir, we get a glimpse of how pagan motifs were used to accommodate the basic tenets of the Christian faith during the first two centuries of Christianity. Einarr’s use of this kenning did not go unnoticed by the like-minded. Snorri, when reporting the story of the binding of Fenrir, adds these seemingly superfluous words about the sword: ‘þat er gómsparri hans’ (that is his gum-spar).191 In all likelihood, Snorri is alluding to the famous poem of his older relative where the word gómsparri is used.192 Snorri’s quotation highlights just how much he and Einarr had in common. To both of these poets, an obscure mythological coinage like gómsparri was a thing to relish. It is also possible that Snorri, by quoting a kenning which refers to the sword of Óláfr, subtly suggests that the correct interpretation of his mythological narrative is another than the pagan one he is reporting.

190 The genitival ending -ar¸rather than -s, is unexpected in an a-stem. The ending is a superscript note, and it is perhaps conceivable that the note’s common use for genitival endings has led to some scribal confusion (Holm. Perg. 15 4to 35v l. 11). 191 Snorri, Prologue and ‘Gylfaginning’, ed. Faulkes, p. 29. 192 Based on the character of the kenning system, this phrase can be identified as a direct quotation from Geisli. Some common kenning types may be found more than once in the corpus, but this is seldom the case with rare kenning types (see above pp. 62–63). Given that gómsparri (gum-spar) is already a kenning for an object which regulates some creature’s mouth or jaws, and given that it is unique even as a type in the preserved corpus, it is unlikely that precisely this kenning had been used elsewhere.

88

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

2.4.4 Einarr and Ofljóst We have now seen Einarr probing the limits of extended metaphor and negotiating the ways in which myth can be used. We have also seen him appropriating pagan myth for Christian purposes. In another instance, he brings wordplay to a new extreme by having it encompass an entire couplet. In the Third Grammatical Treatise – a translation of Donatus’s Barbarismus adapted to skaldic poetry by Snorri’s nephew Óláfr Þórðarson c. 1250 – a quotation by Einarr includes an exceedingly complex example of ofljóst (Óláfr explicitly uses this term here)193: Víst erumk hermð á hesti hefir fljóð ef vill góðan. I am certainly angry at the horse, the woman has [= may have] the good [horse], if she wants.

Óláfr explains: Ok skal svá skilja: ‘Víst erumk hermð á hesti’: Víst legg ek á jó reiðiþokka. Hér er máli skipt. En þat skal svá punkta ok sundr taka: Legg ek á Jóreiði þokka góðan. Hér er seni skipt. ‘Hefir fljóð ef vill’: konu má ná. Hér er máli skipt. Konu Mána. Hér er seni skipt.194 And it should be understood thus: ‘I am certainly angry at the horse (víst erumk hermð á hesti)’: I am certainly angry at the horse (víst legg ek á jó reiðiþokka). Here, the language is changed [that is, the expressions are synonymous]. This should be punctuated and divided thus: I take a fancy for Jóreiðr (legg ek á Jóreiði þokka góðan). Here, the meaning is changed. ‘The woman has, if she wants (hefir fljóð ef vill)’: it is possible to get the woman (konu má ná). Here, language is changed. The wife of Máni (konu Mána).195 Here, the meaning is changed.

Óláfr’s explanation partly defies translation. Thus, for instance, because of the morphology of Old Norse, Óláfr can construe ‘hefir fljóð ef vill’ (the woman has, if she wants) and ‘konu má ná’ (it is possible to get the woman) as synonymous expressions. The complexity of the explanation is predicated on the staggering array of parameters involved in construing the second level of meaning in the couplet (synonymy, homonymy, word division, case, etc.). Again, Einarr has attracted the attention of one of the thirteenth-century grammarians, and again, obscurity and complexity lies at the heart of the attraction. Time and time again, we see Einarr’s fascination with diction, and mythological diction, not least – sometimes to the extent that the content of his poetry is

193 The topic in the treatise is barbarism through shift of accent. 194 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, pp. 66–67, 174–75. 195 Óláfr’s claims that the change from má ná to Mána amounts to changing both vowels from circumflex to acute accent, a statement which has been analysed as an indication of tonemes in Old Icelandic (see Klaus Johan Myrvoll and Tryggve Skomedal, ‘Tonelagsskilnad i islendsk i Tridje grammatiske avhandling’, Maal og Minne (2010), pp. 87–89).

2.4 The Late Period up to Snorri and his nephews (c. 1120–c. 1300)

89

rendered more or less irrelevant. Einarr was a new kind of poet, no longer just the craftsman of poetry, learning the trade from his older contemporaries, but also a theoretician of the art whose approach to poetry was coloured by the methods and prescriptive attitudes of the classroom. I turn now to another poet active before the time of the composition of Snorri’s Edda who shared Einarr’s interests.

2.4.5 Hallar-Steinn Hallar-Steinn seems to belong to the twelfth century, but beyond that, we have little information about him.196 There is, however, internal evidence that he had received some formal schooling. In one stanza, he describes a battle scene with the words bókar sól fló hvasst (the book’s sun flew hard).197 The stanza is quoted in Laufásedda, and in the commentary, we learn that ‘the book’s sun’ is colour, which lights up the page.198 In Old Norse, one word for colour is steinn, which also means ‘stone’. The meaning is thus that ‘the stones flew hard’ in the battle. HallarSteinn employs wordplay here, and he does so with an allusion to manuscript illumination. References to books and book learning are rare in skaldic poetry, but a few occurences are found from this time onwards in monastic poetry or in poetry otherwise connected to formal learning.199 Although this reference is all we have in order to reconstruct his intellectual background, it does mark off Hallar-Steinn from the earlier tradition, and it suggests that he had some experience of learning through books. The possibility that Hallar-Steinn possessed grammatical leanings is also supported by his one preserved longer poem, Rekstefja, albeit obliquely. As

196 See p. 33 n. 55. 197 SkP 3, p. 208. 198 Two Versions of ‘Snorra Edda’, ed. Faulkes, 1, p. 307. 199 LP, s.v. bók. There is an earlier reference in Þórarinn loftunga’s Glælognskviða (1032), in the compound bókamál (book language, presumably Latin) (ibid). The passage is obscure, but it would seem that the poet refers to the language of the clergy and so to the language of a group to which he himself did not belong (SkP, 1: pp. 875–76; see also Margaret Clunies Ross, ‘Reginnaglar’, in News from Other Worlds.‘Tíðendi ór ǫðrum heimum’. Studies in Folklore, Mythology and Culture in Honor of John F. Lindow, ed. Merrill Kaplan and Timothy R. Tangherlini (Berkeley: North Pinehurst Press, 2012), pp. 3–21). Another, equally elusive reference to book script is found in a stanza by Sighvatr, where the last two lines read ‘skrifnask skírinafna | skript, þjóðkonungr, niptar’, which Judith Jesch translates as ‘(his) kinswoman’s document was written for (my) godson’. Both skrifnask and skírinafni are hapax, but skript is not, though it remains unclear what kind of document we are dealing with. Again, the perspective seems to be that of a symbolic value being attributed to book script, but the poet does not associate himself with it or suggest that he saw other functions in it than the symbolic one (Judith Jesch, ‘In Praise of Ástríðr Óláfsdóttir’, Saga-Book, 24 (1994–97), pp. 10–11; Heimskringla, ed. Bjarni Aðalbjarnarson, 3, pp. 18–19; SkP, 1: pp. 734–35). I am not aware of other early references.

90

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

noted in Chapter One, Rekstefja carries earlier metrical possibilities to an extreme that severely circumscribes the syntactical liberty of the poet, suggesting a studied approach to tradition. In Rekstefja, Hallar-Steinn also employs tmesis: stétthrings stofnar (path-sword’s poles) for hring-stéttar stofnar (sword-path’s [shield’s] poles [warriors]).200 More remarkable, however, is his use of two specific mythological kennings for poetry: Þundregn (Þundr- [Óðinn-] rain) and dvergregn (dwarf-rain).201 In the preceding tradition, regn is only used twice as a base-word in kennings for poetry.202 One instance is found in a stanza by Egill: Hǫ́ars þegna regn (Hǫ́arr’s [Óðinn’s] men’s [the gods’/the poets’] rain [poetry]).203 The other is found in Þórðr Kolbeinsson’s Eiríksdrápa 14: Þorins rekka regn (Þorinn’s [a dwarf’s] men’s [the dwarves’] rain [poetry]).204 In skaldic poetry, regn (rain) is otherwise used of things that are metaphorically ‘raining’, such as tears or weapons. This is clearly how Egill understood his own kenning, since he says ‘rigna getr at regni [. . .] Hǫ́ars þegna’ (it is raining [. . .] Hǫ́arr’s [Óðinn’s] men’s [the gods’/the poets’] rain [poetry]), and since he in another kenning refers to how Óðinn vomited forth the mead of poetry when he was in an eagle’s shape (see below pp. 254–55). That mythological incident probably accounts for his choice of regn, that is, the mead of poetry is raining down from Óðinn onto the gods or the poets. Þórðr, by contrast, imitates the structure of Egill’s kenning but seems to disregard its specific mythological underpinning, since he replaces both Óðinn and the recipients of the mead with dwarves, who also had the mead for a time, but on whom it never rained down. In his kenning, regn simply becomes a synonym for ‘alcoholic beverage’, which is typologically deviant and semantically far-fetched.205 The semantic implausibility and rarity of regn as base-word in kennings for poetry make Hallar-Steinn’s kennings highly marked. The structural overlap of Óðinn + regn (Egill and Hallar-Steinn) and dwarves + regn (Þórðr and HallarSteinn) suggests that Hallar-Steinn has sought out these particularly contrived kennings in the preceding tradition. At the same time he achieves a degree of coherence by using the same base-word twice. Indeed, he seems to have analysed the kennings as structurally identical and adapted both in the same way, removing the first definer ( þegna and rekka) and retaining only the second (Hǫ́arr = Þundr and Þorinn = dvergr). Here and in the previous analysis of Einarr’s use of paraphrase, we see that mythological references and literary allusions overlapped to the learned twelfth-century poets. The mythological revival of the

200 Finnur Jónsson, ‘Kenningers led-omstilling’, p. 7; SkP 1, pp. 910–11. 201 SkP 1, pp. 905, 933. 202 See Meissner, Die Kenningar, pp. 427–30, but note that the occurrence in Þórðr’s poem is not listed, due to emendation in Skj (Skj B 1, p. 206; A 1, p. 216; cf. SkP 1, pp. 508–10). 203 Egils saga, ed. Sigurður Nordal, p. 110; Skj B 1, p. 43; A 1, p. 49. 204 SkP 1, pp. 508–10. 205 LP s.v. regn.

2.4 The Late Period up to Snorri and his nephews (c. 1120–c. 1300)

91

twelfth century thus appears not as a vague trend, but as the result of a number of individual choices to display one’s learning and sophistication through extremely demanding allusions to the preceding tradition. In the Third Grammatical treatise, under the heading amphibologia,206 a couplet by Hallar-Steinn with a ‘hidden topic’ (fólgit svá málit) is quoted: Hólmleggjar viðr hilmir hring-Skǫglar mik þǫglan.207 The island-bone’s [stone’s] ruler [dwarf] of the rings’ Skǫgull (Valkyrie name) [woman] makes me silent.

Óláfr explains: ‘Hér kallar hann dverg Lit ok svá fegrð konunnar’ (Here he calls a dwarf Litr, and so also the beauty of the woman). What he means is that we should not understand the kenning hólmleggjar hilmir (the sea-bone’s [stone’s] ruler [dwarf]) as designating any dwarf, but use the name of the individual dwarf Litr, and that we should then substitute it with its homonym litr (colour, hue, looks). The stanza then reads: The woman’s looks make me silent [i.e. strike me dumb]. Could we have understood this stanza without Óláfr’s commentary? Perhaps, but it still remains unlikely that such competence was cultivated outside of a small learned elite.208 It should also be noted that the mythological knowledge needed for decoding this couplet is very specific: namely, that the audience must be able to produce the names of important dwarves. This again draws attention to the strong connection between mythology and diction.

2.4.6 The Thirteenth Century In the thirteenth century, the powerful chieftain Snorri Sturluson and his nephews, Óláfr and Sturla, stand out as cultivators of complex, archaizing poetry, and they also produced texts where they described or used the features under discussion

206 The form of the word is the result of a reanalysis of amphibolia (ambiguity). 207 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 85; SkP 3, pp. 202–03. 208 When Óláfr says about such amphibologia that it is ‘er víða sett í skáldskap’ (widely used in poetry), there is reason to suspect that he overstates the case (Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, pp. 84–85). At least, examples as demanding as this one, where the correct name a concept has to be discovered and then substituted by a homonym, are almost only found in the learned poets. Among earlier examples, only that of Þórsdrápa (gammleið > lopt > Lopt (= Loki)) lies on roughly the same level of complexity. The example from Egill (ekkja > hæll (widow) > hæll (heel)) does not involve a kenning.

92

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

here.209 Skaldic wordplay was well suited for those who had made poetry an object of study and commentary along the lines of the study of the classics. Both Einarr and Hallar-Steinn are likely to have been such poets, and their most complex examples of wordplay are transmitted to us by Óláfr in the Third Grammatical Treatise. Óláfr’s choice of examples thus attests to the grammatical interest in wordplay in the thirteenth century. Snorri himself provides a staggering display of wordplay in his examples of different refhvarf metres – too complex to be treated here.210 Snorri and Óláfr both used specific mythological references in their poetry.211 Óláfr’s brother Sturla composed a rich interlace of specific mythological references and extended metaphor in his Hákonardrápa.212 Somewhat surprisingly, neither Snorri, Óláfr nor Sturla appear to have favoured tmesis: only one instance, in Háttatal, may be considered as secure.213 The twelfth-century revival of archaic diction took place within poetry, and while Snorri, Óláfr and Sturla all continued this trend, Snorri and Óláfr also took it to the next level by composing treatises on diction and mythology: namely, Snorri’s Edda and the Third Grammatical Treatise. These were novel initiatives, but they were nonetheless a natural continuation of the poetic developments of the twelfth century: the twelfth-century poets began to take make use of the tools of grammatical analysis in their compositions, whereas Snorri and Óláfr went on to describe the analysis itself. The cumulative and organic nature of this evolution will become clearer when we analyze the treatise on diction today called Litla Skálda, which

209 Snorri and Óláfr authored treatises where ofljóst and nýgerving are discussed, and Snorri also composed poetry containing both features (Snorri, Háttatal, ed. Faulkes, pp. 6–7, 11–14; Snorri, Skáldskaparmál, ed. Faulkes, p. 109; Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, pp. 66, 84, 89). 210 These metres are defined by diction, namely by using words of opposing meaning such as ‘water’ and ‘fire’ etc. (see Snorri, Háttatal, ed. Faulkes, pp. 11–14). 211 This is common in Háttatal. For examples in Óláfr’s poetry, see SkP 3, pp. 302, 304, 307. 212 The most obvious example is stanza 16: ‘Ok inndrótt | elda kyndi | böðvar-Týs | við baugrenni | svát eggfárs | eisur stukku | of ilflet | Aurnis spjalla’ (And the retinue kindled the battle-god’s [Óðinn’s] fires [swords] against the ring-advancer [man], so that the edge-wrath’s [battle’s] sparks [swords] flew over Aurnir’s [a giant’s] friend’s [giant’s] footsole-floor [shield]) (SkP 2, p. 711). The last kenning requires knowledge of the myth where the giant Hrungnir stands on his shield. Other noteworthy examples of extended metaphor are found in stanzas 7, 8, 17, 22 (SkP 2, pp. 704–05, 712, 715; see also p. 47 above). Specific mythological references are abundant in the poem. 213 Finnur Jónsson lists five instances altogether in Snorri and Sturla (‘Kenningers led-omstilling’, pp. 7–8). One in Háttatal is clearly accepted by Kock, the other perhaps not (Den norsk-isländska skaldediktningen, ed. Kock, pp. 39, 46). Of the examples in Sturla, one is rejected on good grounds in SkP (SkP 2, pp. 728–29). The other two have not yet been edited there and Kock’s view of them remains unclear (Den norsk-isländska skaldediktningen, ed. Kock, pp. 59, 73). To my mind they are doubtful.

2.4 The Late Period up to Snorri and his nephews (c. 1120–c. 1300)

93

appears to represent a transitional stage before the full-blown exposition of poetics of Snorri’s Edda (see pp. 129–47).

2.4.7 Concluding Remarks on Diction The chronological covariation of all features under study suggests that they served related functions for poets, both before c. 995 and after c. 1120. The poetic corpus at large bears this out, as well as individual poems, where such features tend either to be interwoven or to be absent altogether. This is particularly true of specific mythological references, wordplay and extended metaphor. While tmesis does occur in the same poets and poems, it is often independent of the dense web formed by the other three. Perhaps tmesis’ purely formal character and relation to word-order and word-division explains this difference to the other three features, which are connected to semantics. We may also recall that Sighvatr used tmesis, although he avoided specific mythological references, and the same is true of some religious poetry of the late period, such as Harmsól and Leiðarvísan.214 This difference notwithstanding, however, the chronological gap and the poets involved suggest that tmesis, like the other features, offered poets ways to add complexity to their compositions using the same techniques as in the early period. In their emulation of the tenth-century poets, the learned poets of the twelfth century did not single out one feature for producing additional semantic complexity; rather, they displayed their poetic dexterity through combining them all.215 As we have seen, they often carefully chose which poets and kennings to paraphrase in what appears to have been a conscious revival of an elaborated and demanding style. If we compare the analyses of this and the preceding chapter, we may note that Latin learning produced similar results for diction and metre. In neither case are we dealing with a direct transfer. Latin grammatica provided useful tools for metrical analysis, but its specific rules were not directly applicable to skaldic poetry. Indeed, as we shall see in the discussion of 3GT, adapting these rules to the native tradition would demand a concerted effort that, while impressive, really caused more problems than it solved. Rather, what proved most useful to Icelandic poets

214 Finnur Jónsson, ‘Kenningers led-omstilling’, pp. 7, 22. Accepted by Kock and SkP: Den norskisländska skaldediktningen, ed. Kock, pp. 266, 302; SkP 7, pp. 77, 144, An occurrence from Máríuvísur 1 is not accepted in SkP (Finnur Jónsson, ‘Kenningers led-omstilling’, p. 22; SkP 7, p. 685). 215 Relevant poetry that has been left out of this overview includes, for instance: Earl Rǫgnvaldr’s Háttalykill (a list of metres, 1140s), which combines interest in heroic lore, metre, and diction. In other poetry Rǫgnvaldr revives the use of mythological kennings for poetry (SkP 2, pp. 583, 588). Two additional poems that I do not analyse here – Jómsvíkingadrápa and Málsháttakvæði (c. 1200) – indicate a combined interest in rhyme and heroic and mythological lore.

94

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

was the structured and normative perspective of grammatical studies. In the case of metre, this perspective gave an impulse towards systematic analysis and towards observing various patterns and setting up rules and categories based on these observations. Basic features such as alliteration and hendingar remained unaltered. We see here a mixture of methodological innovation and formal conservatism. With regard to diction, too, Latin did not so much provide a model for imitation as suggest a new way of approaching the native tradition. The old poems must have been transmitted through the eleventh century, so interest in them was not new in itself. The revival of early and complex features of diction in composition, however, suggests that the poets’ sense of diachronic awareness had sharpened, and learned poets in particular felt the impulse to emulate the bold imagery of the tenth-century poets. In the context of Latin learning, mythology was harmless, homonymy was the focus of didactic interest, and poetic texts were central for pedagogical and stylistic reasons alike. It is easy to see, then, how Latin schooling could provide students with new ways of approaching their own traditions and enabling them to see that in an earlier period, Old Norse poetry had exhibited many features similar to those so highly prized in Latin poetry. It is presumably for this reason that the poets who had made the acquaintance of Latin took the lead in reviving the referential sophistication of earlier times. Although its study falls outside of the scope of this book, it is worth noting that this situation would repeat itself once again after the medieval period. When the stylistic currents of the Baroque reached Iceland, local poets responded partly by adaptation (for example by composing in new genres), but partly also by reviving traditional diction. They did not just imitate the verbal flowers of Baroque diction outright, but once again they immersed themselves into the intricacies of mythological diction.216 Mythological creativity would increasingly gain ground until it finally culminated in the poetry of Sigurður Breiðfjörð in the first half of the nineteenth century. He portrayed himself as an equally intimate friend of Óðinn as any of the pagan poets of some 800 years earlier.217

2.5 Mythology in Medieval Europe In order to provide a broad cultural contextualization to the developments described in this chapter, I give here a concise overview of the treatment of mythology in medieval Europe. This will clarify some of the cultural rationale behind

216 See Margrét Eggertsdóttir, ‘Eddulist og barokk í íslenskum kveðskap á 17. öld’, in Guðamjöður og arnarleir. Safn ritgerða um eddulist, ed. Sverrir Tómasson (Reykjavík: Háskólaútgáfan, 1996), pp. 91–116. 217 See Arndís Hulda Auðunsdóttir and Aðalheiður Guðmundsdóttir, ‘“Berðu mér ei blandað vín”’. Um rímnakveðskap Sigurðar Breiðfjörð, skáldskaparmjöðinn og áfengið’, Són, 13 (2015), pp. 11–35.

2.5 Mythology in Medieval Europe

95

mythology-as-diction, as well the extent to which Icelandic developments adhered to or differed from those of the rest of Europe. Outside theology, the use of mythology in the Middle Ages may be divided into three main contexts: the historiographical, the grammatical and the philosophical.218 Of these, the grammatical context will be of particular importance. By grammatica, I here intend the more inclusive of the two medieval meanings of the word: that is, not only the study of language, but also of texts.219 Translating it as philology may not be too far off the mark. The historiographical function of mythology is the most difficult to delimitate, since the dividing line between god and human ruler sometimes blurred when speaking of the hoary past. This distinction was, furthermore, largely irrelevant to the Middle Ages. Whatever the gods really were, they were not deities; at best they were human, at worst demons. Historiographical use of mythology was widespread, and it was not limited to classical mythology. Ireland, for instance, had a rich literature on local mythology, though it is often difficult to tell where history ends and mythology begins. This is partly attributable to the fact that there was no exact correspondence to the modern concepts of myth and mythology in the Middle Ages. The closest match was fabula, which designated stories about things that were neither true nor likely to happen.220 The presence or absence of myth in historiography was thus largely a matter of credibility, and what was seen as credible in the context of local historiography may have varied with time, place and discourse. Mythology’s grammatical and philosophical functions partly bled into each other, since philosophical treatment of mythology often took the form of commentary on the classics. Broad mythological knowledge, however, was cultivated mainly within grammatical studies, where it served as a tool for access to the canonical poets. Commentaries, such as Servius’s on the Æneid and the various commentaries

218 Such a claim must necessarily be simplistic, and one could for instance ask how relevant the dividing lines between the different disciplines are, or whether references to the gods in sermons are not more properly described as pastoral rather than as theological, or where encyclopedia fits into this picture. For the present purposes, however, the above division clarifies the relevant functions of mythology. 219 Martin Irvine, The Making of Textual Culture. ‘Grammatica’ and Literary Theory 350–1100 (Cambridge: Cambridge University Press, 1994), p. 5. In Latin, both meanings were subsumed under the word grammatica, but Greek upheld a verbal distinction between the two: γραμματιστική designated the study of letters and language and γραμματική the study of literature (Donat et la tradition de l’enseignement grammatical, ed. Holtz, p. 4). 220 Thus Cicero, De inventione 1,19,27; Ad Herennium 1,8,13; Isidore, Etymologiae 1,44,5. The words mythus and mythologia were used very rarely, and when they were, like the term fabula, they referred not only to stories about the gods, but also to Aesop’s fables. See the discussion in Jan Ziolkowski, ‘Medieval Latin Mythography as Death and Resurrection of Myth’, in Writing Down the Myths, ed. Joseph Falaky Nagy. Cursor Mundi, 17 (Turnhout: Brepols, 2013), pp. 92–95.

96

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

on Ovid’s Metamorphoses, contained a deal of mythological material, both ‘factual’ and interpretive (allegorical etc.). Since this discourse centred on the classics, only classical mythology was studied in a grammatical context throughout medieval Christendom. Focus on mythology increased from c. 1100 onwards in Germany, France and England. A renewed interest in Ovid’s Metamorphoses engendered a rich commentary literature, and in France, Neoplatonism and mythology went hand in hand.221 As discussed in the Introduction, manuscripts of the so-called Third Vatican Mythographer spread across Europe, beginning in the second half of the twelfth century. In Italy, interest in mythology is mainly evident from Dante onwards, most manifestly so in Boccaccio’s Genealogie deorum gentilium libri.222 Only in one place did grammatical study of myth centre on non-classical mythology: Iceland. There, local mythography came to dominate completely – nearly all references to mythology in connection to literature and poetry concern the local tradition, rather than the classical pantheon. Since there is nothing to indicate that Latin was taught differently in Iceland, this is a most unexpected development. In the following, I shall point to some important factors that probably contributed to this outcome. Today, we often assume that mythological narratives somehow remain authentic, as long as members of a society believe in them. When mythology and belief part ways, we enter the realm of literature.223 In the Middle Ages, by contrast, disbelief was the fundamental precondition for the transmission of myths and even defined them. With that proviso, mythology could be studied as anything from a respectable literary tradition to a clouded vision of God, since he often preferred to manifest himself in ways that barred access to simple minds.224 Some critics of mythology existed, but while they may have considered mythological study a waste of time, they did not experience it as a threat. In the twelfth century, for instance, Conrad of Hirsau wrote a treatise on the classical authors

221 Frank T. Coulson, ‘Ovid’s Metamophoses in the School Tradition of France’; Dronke, Fabula. 222 On the interpretive methods used by Boccaccio, see David Lummus, ‘Bocaccio’s Poetic Anthropology: Allegories of History in Genealogie deorum gentilium libri’, Speculum, 87 (2012), pp. 724–65. 223 This is most often articulated in comparisons between the ‘folkish’ Hesiod and Homer and the urbane Vergil and Ovid. See, for instance, David Adams Leeming, The World of Myth. An Anthology (Oxford: Oxford University Press, 1990), p. 7. The connection between belief and myth cannot do justice to all modern analyses of myth, and not least within psychology, belief may often not be an applicable parameter. 224 Comments on the usefulness of Ovid’s Metamorphoses belong to the clearest examples of how the study of myth was, on the whole, unproblematic during the Middle Ages. See Fausto Ghisalberti, ‘Medieval Biographies of Ovid’, Journal of the Warburg and Courtauld Institutes, 9 (1946), pp. 10–59; Frank T. Coulson, ‘Hitherto Unedited Medieval and Renaissance Lives of Ovid’, Mediaeval Studies, 49 (1987), pp. 152–207.

2.5 Mythology in Medieval Europe

97

where he launched into an attack on Ovid.225 He makes it clear that to his mind, the ratio between mythology and wisdom in the Metamorphoses was too uneven for that work to merit serious study, whereas other classical authors had a more sound distribution of content. This is the voice of moral rather than doctrinal concern. Conrad uses of the words secularis disciplina to refer to problematic aspects of Terence, Vergil, Homer and others, indicating that the problem was mainly one of distraction from issues of direct importance for salvation via fabulous tales, descriptions of sex or burlesque humour.226 The other Ovid, the one of Ars amatoria and carnal love, caused much more concern, and it is this work which the future bishop Klœngr was forbidden to read according to Jóns saga helga.227 The Metamorphoses, by contrast, were seen as conveying profound theological and cosmological insights, and in the High Middle Ages they were at times used to vindicate the amorous Ovid from the moral condemnation he might have otherwise incurred.228 An ethical rather than a theological view on pagan poetry – as opposed to pagan cult – had prevailed at least since the time of Augustine. Augustine’s searing critique of pagan poetry in book one of the Confessiones does not centre on belief, and it may be compared to his discussion of grammar generally. Poetry and grammar are both meaningless for salvation, but grammar has the benefit of being slightly more useful in the here and now.229 In De doctrina Christiana he says that the poets describe, but do not teach superstition.230 His attacks on all aspects of paganism in books 1–10 of De civitate Dei are relentless, but even here, mythological poetry appears to enter the discussion more as an easy victim than as a cause for serious concern. In general, Augustine seems less worried about pagan poetry than about many other distractions of this world, and the real danger lies not in fanciful mythology, but in things that one is actually expected to believe, such as Manicheism. The main problem with mythography, then, was its secular nature, and the critique of it was eventually heard in the vernaculars as well. In the German tradition we find it in the ninth-century Otfrid of Weissenburg, the eleventh-century Annolied and the twelfth-century Kaiserchronik, whereas a famous example in Old Icelandic literature appears in the prologue to the saga of Óláfr Tryggvason by Oddr

225 Accessus ad auctores. Bernard d’Utrecht. Conrad d’Hirsau: ‘Dialogus super auctores’, ed. R. B. C. Huygens (Leiden: Brill, 1970), pp. 114–118. 226 Accessus ad auctores, ed. Huygens, p. 116. 227 Biskupa sögur, 1, ed. Sigurgeir Steingrímsson, Ólafur Halldórsson and Peter Foote, 2 vols. Íslenzk fornrit, 15 (Reykjavík: Hið íslenzka fornritafélag, 2003), 2, pp. 211–12 (though note variants). For other references to Ovid in Old Norse literature (all to erotic matters), see Wellendorf, ‘No need for Mead’, p. 139 n. 19. For a European background, see, for instance, Jean Leclercq, Monks and Love in Twelfthcentury France. Psycho-Historical Essays (Oxford: Oxford University Press, 1979), pp. 65–67. 228 See, for instance, Leclercq, Monks and Love, p. 68. 229 Confessions, ed. William Watts, 2 vols. Loeb Classical Library, 26–27 (Cambridge: Harvard University Press, 1912), 1, pp. 38–58. 230 De doctrina Christiana, ed. and trans. R. P. H. Green (Oxford: Clarendon Press, 1995), p. 90.

98

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

munkr.231 Such critique, however, could do little to stem the growing tide of interest in heroic and amusing tales in general and in mythology in particular. Furthermore, since mythological references and stylistic elevation often went hand in hand in the classical authors who taught medieval clerics how to write, it was necessary to find a fruitful mode of coexistence, even for a critic like Conrad of Hirsau. He concludes that if we were to get rid of all pagan literature, we would also have to repudiate the Books of Moses, which quote such sources, as well as Saints Paul and Augustine.232 Mythology, however, was more than a necessary but largely indifferent component of the classical heritage. According to medieval ideas that intellectual challenges were valuable in their own right and important for the continued search for truth, the very obscurity of mythology was an asset. This assumption was generally left implicit, but this is true also of obscurity in the Biblical exegetical tradition. De doctrina Christiana is a rare case where Augustine spells out the logic with regard to the Old Testament prophets: ‘Quae quanto magis translatis verbis videntur operiri, tanto magis cum fuerint aperta dulcescunt’233 (The more these things [divine truths] seem concealed by metaphors, the sweeter they become when revealed). Here and elsewhere, Augustine places great emphasis on the sweetness of obscurity and how it stimulates the believer and keeps the heretic shut out. Medieval scholars thought of mythological poetry in much the same way, as truth hidden under a beautiful veil, and the intellectual efforts its interpretation demanded not only kept the mind alert, but also retained the attention of those who needed such stimuli to keep on pursuing the road towards truth.234 Intellectual playfulness and conceptual dissonance were part of an esthetics that gave precedence to striking sensory impressions, including both the inner images, created by mythological tales and the like, and external perception.235 Such intellectual endeavours and sensory provocations were all

231 Walter Haug, Vernacular Literary Theory in the Middle Ages. The German Tradition, 800–1300, in its European Context, trans. Joanna M. Catling (Cambridge: Cambridge University Press, 1997 [1985]), pp. 31, 60–70; Færeyinga saga. Óláfs saga Tryggvasonar eftir Odd munk Snorrason, ed. Jónas Kristjánsson and Þórður Ingi Guðjónsson. Íslenzk fornrit, 25 (Reykjavík: Hið íslenzka fornritafélag, 2006), p. 126 Alcuin’s famous critique in 797 of the (probably Mercian) monks’ interest in heroic lore, ‘quid Hinieldus cum Christo?’ (What has Ingeld to do with Christ) was written in Latin, but clearly aimed at English, vernacular discourse. Again, the main problem was one of distraction (see Thomas G. Duncan, ‘”Quid Hinieldus cum Christo”. The Secular Expression of the Sacred in Old and Middle English Lyrics’, in Sacred and Secular in Medieval and Early Modern Cultures. New Essays, ed. Lawrence Besserman (Palgrave Macmillan: New York, 2006), pp. 29–30). 232 Accessus ad auctores, ed. Huygens, pp. 116–117. 233 De doctrina Christiana, ed. and trans. R. P. H. Green, p. 216. 234 See Jan M. Ziolkowski, ‘Theories of Obscurity in the Latin Tradition’, Mediaevalia, 19 (1996), pp. 143–53; Dronke, Fabula. 235 Mary Carruthers, The Experience of Beauty in the Middle Ages (Oxford: Oxford University Press, 2013), pp. 61–70 et passim. Carruthers does not explicitly treat mythology, but it forms part of her overall analysis, such as when she quotes what Boccaccio writes about mythological poetry (p. 64).

2.5 Mythology in Medieval Europe

99

equally ‘sweet’. Not only the use of mythology, but also the intricacies of the skaldic tradition generally, bear witness to a similar perception in Iceland. The arrival of Christian learned culture only sharpened these ideas. These intellectual preconditions were largely the same throughout medieval Europe, and they served to clear the way of any obstacles to the study of mythology. In Iceland, poets and scholars availed themselves of this freedom by producing a remarkable body of texts, beginning with poetic compositions in the twelfth century and culminating in the thirteenth in Snorri’s Edda (c. 1220–40) and in compilations of mythological poetry, such as the manuscripts GKS 2635 4to (the so-called Poetic Edda, c. 1270) and AM 748 Ia 4to (c. 1300–1325).236 The manuscript record bears clear witness to which of the several medieval contexts of mythological study (theology, historiography, philosophy, grammatica) these Icelandic mythological texts belonged. Snorri’s Edda and eddic mythological poetry were consistently compiled with various grammatical texts which are either translations of the most important Latin texts (Priscian, Donatus and the Doctrinale) or original compositions based on the Latin grammatical tradition (the First and Second Grammatical Treatises). The manuscripts reveal a learned continuum from orthography to the Old Norse pantheon, clearly indicating the grammatical nature of the mythological corpus (see below pp. 107–09).237 Guðrún Nordal proposes that some of the fascination with native mythology derived from its function as a native counterpart to classical mythology.238 Another reason for the emergence of Icelandic mythography is apparent in the many complex mythological references within skaldic poetry: native mythology had to be preserved to ensure proper understanding of older poetry.239 Both of these explanations have much to commend them, and I would add two more, namely the medieval attraction to the ‘sweetness of obscurity’ mentioned above and the idea of emulation with the poets of the local past.240 The ‘sweetness of obscurity’ explanation is supported by the fact that it was not only mythology that was revived, but obscure language

236 GKS 2635 4to contains eddic poetry only, with some explanatory prose narrative. AM 748 Ia 4to contains only mythological eddic poems, but the manuscript was clearly produced to be used in tandem with (and to be bound with?) A, which contains a fragment of the so-called Fifth Grammatical Treatise, the Third Grammatical Treatise, Skáldskaparmál and the poem Íslendingadrápa (see Guðrún Nordal, Tools of Literacy, pp. 57–64). 237 Mythology also crops up in many places outside of what I here call ‘the mythological corpus’. Jonas Wellendorf, Gods and Humans in Medieval Scandinavia. Retying the Bonds (Cambridge: Cambridge University Press, 2018), studies perspectives on pre-Christian mythology and religion in contexts that may be defined as hagiographic, pastoral and historiographic, in addition to the grammatical context of Snorri’s Edda. 238 Guðrún Nordal, Tools of Literacy, pp. 22–23. 239 Thus, for instance, Snorri, Prologue and ‘Gylfaginning’, ed. Faulkes, p. xvi. 240 Although Guðrún Nordal does not focus on the concept of emulation, it is implicit in many of her writings, in particular Skaldic Versifying and Social Discrimination in Medieval Iceland. The

100

2 Diction: Mythology, Wordplay, Metaphor and Tmesis

generally. With regard to emulation, skaldic poetry had always been fiercely individualistic and agonistic. Once the acquaintance with Latin learning had demonstrated that there was nothing to fear in the abstruse language of the tenth century, the way was open for applying the competitive impulse to games of diction and allusion, as we have seen in the analyses of Einarr and Hallar-Steinn.241 These are some of the factors that may have propelled the revival of mythological and obscure diction. On one hand, internal factors were primary, since if there had been no tradition for obscure and mythological diction, Latin learning would have been powerless to revive it. On the other hand, if poets had not become learned but retained the rudimentary, intolerant views of the missionary kings, mythological creativity would have remained a thing of the past and the texts that have transmitted it to us would never have been written. It is important, therefore, to note some of the basic preconditions that were specific to a learned and more precisely a grammatical setting. The general tolerance of pagan mythology belongs to that setting, but that tolerance could do no more than make a revival possible. Awareness of the classical heritage and the pedagogical premium on obscurity made a revival not only possible but also desirable. Furthermore, the strongly emulative tradition enhanced both the drive to come up with a counterpart to the classical tradition and the ambition to produce obscure language to display the ability of the poet and to test listeners. Finally, none of this would have been possible if the pre-Christian poetry had not been preserved and if the poetic conventions of the ninth and tenth centuries had not made mythology an intrinsic part of the referential system.

2.6 Summary In this chapter, we have seen that the earlier hypothesis of a partial decline in mythological kennings around the year 1000 must be revised considerably. The break can be dated more precisely, to c. 995, with a further decisive break after c. 1015, and mythological kennings do not form one group in poetic practice, but two. Kennings for persons and battle (generic mythological references) were used throughout the eleventh century, whereas all other mythological kennings and

Dorothea Coke Memorial Lecture in Northern Studies delivered at the University College London 15 March 2001 (London: Viking Society for Northern Research, 2003). 241 A related aspect of complex mythological diction is that it makes the poetry accessible only to the ones in the know, and to that extent, it serves as a formal marker of the ‘in’ group, not unlike the trobar clus of troubadour poetry. See, for instance, Laura Kendrick, The Game of Love. Troubadour Wordplay (Berkeley: University of California Press, 1988), pp. 72–73. For obscure language as a marker for the ‘in’ group more generally, see Daniel Heller-Roazen, Dark Tongues. The Art of Rogues and Riddlers (New York: Zone Books, 2013).

2.6 Summary

101

references (specific mythological references) disappeared almost completely. The reasons appear to have been religious and connected to royal patronage, since stray occurrences can still be found, but with the exception of Haraldr harðráði, very rarely in connection to kings. The revival in the twelfth century was therefore a true revival: for more than a hundred years, specific mythological references had lived almost exclusively in poetry that was transmitted in order that poets learn their trade. Furthermore, several other features of diction followed the same trajectory, with an almost complete demise c. 995 and a marked revival in the twelfth century. It would appear that all of these advanced features of diction were closely interconnected, so that when specific mythological references had to go, the others went with it. In the twelfth century, some poets had become learned enough to see that mythology was no threat but could, together with other features of advanced diction, be an asset instead. At this point, the entire complex was revived. This revival would be the heritage of the twelfth-century poets to the thirteenth-century grammarians and mythologizers, to Snorri chief among them. We turn now to grammatical literature and to Snorri himself.

3 Grammatical Literature This chapter examines the Icelandic grammatical texts from the viewpoint of Snorri’s Edda. This perspective makes it possible to place earlier developments in relation to Snorri and thus to see not only how he availed himself of them, but also how he differed from them. Furthermore, nothing was ever the same in Icelandic literature after the composition of Snorri’s Edda, and later developments can therefore fruitfully be analysed in light of it. I treat three of the four parts of the Edda: Gylfaginning, Skáldskaparmál and Háttatal. The prologue is left out except where relevant, since it does not directly engage with poetry.1 The previous chapters outline the general trends to which Snorri was heir. This chapter analyses his use of specific models and sources, both Latin and vernacular. Sources such as Háttalykill and eddic poetry are commonly acknowledged by scholars, whereas others, such as Donatus’s Ars minor (or a similar work) and Horace’s Ars poetica, have received insufficient consideration. I shall also draw attention to a not yet recognized source for Snorri’s Edda: Litla Skálda. Snorri’s foundation in learned treatises and literary scholarhip has been underestimated, which has hidden from view an evolution that actually makes a great deal of sense: the formal study of text gave rise to new poetic developments in the twelfth century, and they led to the first attempts at writing original treatises on orthography and poetics in Icelandic. This chapter demonstrates that Snorri was aware of this tradition and that his Edda was part of it. Far from operating in a vacuum, he was part of a communal effort among the learned elite, and he did not hesitate to avail himself of the work of his predecessors when this was expedient. I begin with a description of the corpus of Old Icelandic grammatical literature, to serve as a background to the rest of the chapter. I then turn to an analysis of Snorri’s Edda which proceeds backwards through that work, since Snorri’s models and methods are clearest at the end and become increasingly more difficult to identify towards the beginning. It is expedient to establish some of his methods before turning to parts of the analysis where indications are less straightforward. After the first section there follows a discussion of the models, sources and rhetorical strategies of Háttatal. Thereafter I turn to Snorri’s use of Litla Skálda as a model for Skáldskaparmál and as a source for Skáldskaparmál and Gylfaginning alike. After having analysed Snorri’s use of learned materials, I proceed to a description of how he approached his poetic sources. His interpretive

1 There is no scholarly consensus on the authorship of the prologue. A debate between Klaus von See and Lars Lönnroth on its authenticity in 1988–90 emitted a rare amount of heat (see Klaus von See, Mythos und Theologie im skandinavischen Hochmittelalter (Heidelberg: Carl Winter, 1988), pp. 18–30; Lars Lönnroth, book review, Skandinavistik, 20 (1990), pp. 43–47; Klaus von See, ‘Zum Prolog der Snorra Edda’, Skandinavistik, 20 (1990), pp. 111–26). https://doi.org/10.1515/9783110643930-004

3.1 Manuscripts and the Scope of Old Icelandic Grammatica

103

strategies turn out to be somewhat different in Gylfaginning and Skáldskaparmál, with an emphasis on religion in Gylfaginning that is absent from Skáldskaparmál. In Skáldskaparmál in particular, he appears to be a more trustworthy transmitter of tradition than he is sometimes credited with having been. We then take leave of Snorri to consider instead how Óláfr Þórðarson positioned himself in relation to Snorri’s legacy, and how he opted for an even fiercer, but less original, mode of appropriation of his learned models.

3.1 Manuscripts and the Scope of Old Icelandic Grammatica Although Latin grammatica in a general sense established the parameters of its Old Icelandic counterpart, vernacular grammatica depended on local traditions and interests, as well. It is therefore worthwhile to describe these traditions and interests in their own right. For instance, even though Latin grammatica largely centred on poetry as the most formalized and prestigious kind of language and text, this tendency became even more pronounced in an Icelandic setting. It is furthermore difficult to tell to what extent Icelandic grammatica constituted a formal study tied to schools, as in the Latin discipline. Instead we should perhaps think of vernacular grammatica as a transfer of analytic tools and prescriptive attitudes from the Latin classroom to vernacular poetry, whether or not this process took place in a school.2 Manuscripts offer the surest foundation for describing the scope of Old Icelandic grammatica. With the possible exception of GKS 2365 4to (the so-called Poetic Edda), there is little doubt as to which manuscripts served a grammatical function, since they all contain texts that clearly relate to grammar and they bear no signs of being thoughtlessly compiled. Additional content of a more ambiguous nature can thus be assumed to belong to the grammatical corpus as well. Below, I first give short introductions to the use of the term grammatica in Old Norse scholarship and to some of the relevant socio-cultural background, after which I proceed to a description of the contents of relevant manuscripts. Finally, I analyse Snorri’s Edda in the same vein, since its internal structure corresponds to that of the grammatical manuscripts, and since nearly all of these contain the Edda either in part or in full. In Old Norse scholarship, only a restricted body of texts is normally referred to as grammatical: namely, the four so-called grammatical treatises (below 1GT, 2GT, etc). These are all contained, together with Snorri’s Edda, in the largest Old Icelandic grammatical manuscript, Codex Wormianus (AM 242 fol, below W), likely produced

2 Little is known about schools in medieval Iceland, and the precise meaning of ‘classroom’ thus remains unclear. I use the term here as shorthand for prescriptive attitudes to language and literature that are likely to have been similar across Europe. A good overview of medieval Icelandic schooling is Sverrir Tómasson, Formálar íslenskra sagnaritara á miðöldum. Rannsókn bókmenntahefðar (Reykjavík: Stofnun Árna Magnússonar, 1988), pp. 15–35.

104

3 Grammatical Literature

at the Benedictine monastery of Þingeyrar c. 1350 (see below p. 277). The treatises are numbered in order of their occurrence in this manuscript, and their original date of composition probably almost corresponds to that order, except that 2GT and 3GT must likely switch place: 1GT c. 1150, 3GT c. 1250, 2GT c. 1270 and 4GT c. 1319–1340.3 1GT and 4GT are attested only in this manuscript, 2GT in one more and 3GT in two more and in one fragment. A fragment of a fifth treatise (5GT) is found only in AM 748 1b 4to. 1GT, 2GT and the first part of 3GT pertain to orthography and phonology, whereas the second half of 3GT and all of 4GT treat figures and tropes. 3GT and 4GT are translations of Latin texts but have been adapted to become more fully-fledged treatises of poetics. The fragment of 5GT treats poetics. 1GT, 3GT and 5GT are treated on pp. 175–92. W is an elaborate compilation, and its collocation of Snorri’s Edda with 1–4GT indicates that medieval Icelandic grammatica included more than 1–4GT alone. Other grammatical manuscripts corroborate and complement this impression. Based on studies of these, Guðrún Nordal has opened Old Norse scholarship to a wider definition of grammatica.4 I turn now to a brief discussion of the socio-cultural background to grammatical studies in Iceland, and then to a description of what kinds of texts formed the basis for these studies according to manuscript witnesses. Even though runic inscriptions from Norway from the late twelfth and the thirteenth century attest to some interest in skaldic poetry there, the remarkable fusion of native poetics and a grammatical mindset seems to have occurred almost exclusively in Orkney and Iceland.5 Part of the explanation for this probably lies with the very close ties between Icelandic and Orcadian secular chieftains – predisposed towards traditional secular poetry and the vernacular – and the learned men whom they patronized. In Iceland, it was in the dialogue between chieftains and their priests that this kind of innovation took place. The Icelandic system of privately owned churches (staðir and bœndkirkjur) resulted in close interaction between the two groups, and

3 See The First Grammatical Treatise, ed. Hreinn Benediktsson, pp. 22–33; The So-Called Second Grammatical Treatise. An Orthographic Pattern of Late Thirteenth-Century Icelandic. Edition, Translation and Commentary, ed. Fabrizio Raschellà (Firenze: Felice Le Monnier, 1982), pp. 126–32; Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, pp. xxxii–xxxvii; The Fourth Grammatical Treatise, ed. and trans. Margaret Clunies Ross and Jonas Wellendorf (London: Viking Society for Northern Research, 2014), pp. xi–xiv. 4 Guðrún Nordal, Tools of Literacy. 5 For Norwegian runic inscriptions in dróttkvætt, see, for instance, SkP 2, pp. 35, 41; James E. Knirk, ‘Runes from Trondheim and a Stanza by Egill Skalla-Grímsson’, in Studien zum Altgermaischen. Festschrift für Heinrich Beck, ed. Heiko Uecker. Ergänzungsbände zum Reallexikon der Germanischen Altertumskunde, 11 (Berlin: de Gruyter, 1994), pp. 411–20; Karin Fjellhammer Seim, ‘Er “et upåaktet gammelnorsk ord: hausi” belagt på Bryggen i Bergen?’, Maal og Minne (1984), pp. 85–89; Liestøl, Krause and Jón Helgason, ‘Drottkvætt-vers fra Bryggen i Bergen’; Norges innskrifter med de yngre runer, 2, ed. Magnus Olsen (Oslo: Jacob Dybwad, 1951), pp. 268–322.

3.1 Manuscripts and the Scope of Old Icelandic Grammatica

105

some of these staðir saw a considerable output of vernacular literature and poetry (notably Oddi in the late twelfth and Reykholt in the early thirteenth century).6 Sometimes chieftains were clerics in their own right, as was the case with Gizurr Þorvaldsson: subdeacon, court poet and first earl of Iceland.7 More often, perhaps, they were younger brothers or half-brothers of a chieftain, such as Óláfr, the poet and author of 3GT. He was an illegitimate younger son who was ordained subdeacon, and similar levels of schooling among chieftains and their families seem to have been common throughout the Middle Ages.8 Gradually after 1275 the church gained control over the staðir and the influence of chieftains over clerics declined.9 Although this break was far from absolute, ecclesiastical and monastic milieux dominate the literary landscape of the fourteenth century. When secular influence over ecclesiastical matters waned, so did some of the intrepid spirit of innovation (see Chapter Five on fourteenth-century developments). How the fusion of grammatical and traditional poetic learning took place on the practical level is difficult to tell, and it probably changed during the period under study and varied between secular seats of learning, cathedral schools and monasteries. The most extensive studies of the relation between Old Norse poetry and vernacular grammatica have been conducted by Guðrún Nordal, who sees grammatical features in poetry and manuscripts as indications of regular schoolroom use.10 It should be noted, however, that schoolroom use would be a rarity in medieval Europe. Something similar may possibly have been the case with the activities of the filid in the Irish monasteries of an earlier period, but the evidence is late and there is a risk of projecting later practices back into an early period.11 Occitan works on poetry and grammar do not seem to have been used in any formal

6 For the staðir generally, see Magnús Stefansson, ‘De islandske stadenes egenart og eldste historie’, in Church Centres. Church Centres in Iceland from the 11th to the 13th Century and their Parallels in other Countries, ed. Helgi Þorláksson (Reykholt: Snorrastofa, 2005), pp. 117–25; for the number of clerics on the staðir, see Benedikt Eyþórsson, ‘Reykholt and Church Centres’, in Church Centres, ed. Helgi Þorláksson, pp. 105–16. 7 Guðrún Nordal, Tools of Literacy, pp. 162–63. 8 Sverrir Tómasson, Formálar íslenskra sagnaritara á miðöldum, p. 35. 9 This process is described in Árna saga biskups (Biskupa sögur, 3, ed. Guðrún Ása Grímsdóttir. Íslenzk fornrit, 17 (Reykjavík: Hið íslenzka fornritafélag, 1998). The collectively owned bœndakirkjur, however, were allowed to remain outside church control. 10 Tools of Literacy, p. 23; Skaldic Versifying and Social Discrimination, pp. 8, 15–16; ‘Samhengið í íslenskum fornbókmenntum’, in Sagnaheimur. Studies in Honour of Hermann Pálsson on his 80th Birthday, 26th May 2001, ed. Ásdís Egilsdóttir and Rudolf Simek (Wien: Fassbaender, 2001), pp. 93–95. 11 For the filid and their connections to monasteries and monastic learning, see Patrick SimsWilliams and Erich Poppe, ‘Medieval Irish Literary Theory and Criticism’, in The Cambridge History of Literary Criticism, 2: The Middle Ages, ed. Alastair Minnis and Ian Johnson (Cambridge: Cambridge University Press, 2005), pp. 293–94; Katherine Simms, ‘Literacy and the Irish Bards’, in Literacy in Medieval Celtic Societies, ed. Huw Pryce (Cambridge: Cambridge University Press, 1998),

106

3 Grammatical Literature

way in schools until c. 1330, and when they did, this was in response to a perceived need to uphold Occitan culture in the face of the Inquisition – a situation very different from that of twelfth- and thirteenth-century Iceland.12 Furthermore, the sparse information given in the sources, almost exclusively found in the Icelandic sagas of bishops, indicates that schoolroom teaching of grammatica was concerned only with Latin.13 Old Icelandic manuscripts in general contain few glosses that suggest schoolroom use, but this may partly be due to the fact that they were written in the vernacular and needed little additional explanation. On balance, the claim that Old Icelandic grammatical texts were used in the schoolroom is somewhat tenuous. Nonetheless, their composition must have answered a perceived need, and their impact on other literature shows that they were studied extensively, whether they were actually present in the schoolroom or they were studied by the interested in their spare time or by monks in the generous time allotted to private reading in Benedictine monasteries.14 The content of the Icelandic grammatical manuscripts suggests a syllabus of texts modelled on Latin grammatica adapted to vernacular needs. Most of the relevant manuscripts date to c. 1300–c. 1400. Their full coverage of vernacular grammatica is likely the result of a prolonged, cumulative process. The structural affinities between the content of these manuscripts and the fluid corpus of texts used within Latin grammatica suggests that the Latin discipline functioned as a model for the addition of new topics to the vernacular corpus throughout the formative period of Old Icelandic grammatica.

p. 239; T. M. Charles-Edwards, ‘The Context and Uses of Literacy in Early Christian Ireland’, in Literacy in Medieval Celtic Societies, ed. Huw Pryce, pp. 70, 76–77. 12 Sarah Kay, ‘Occitan Grammar as a Science of Endings’, New Medieval Literatures, 11 (2009), pp. 60–61. In some English fourteenth-century schoolroom manuscripts, there are examples of translations between vernacular and Latin poetry (Christopher Cannon, ‘The Middle English Writer’s Schoolroom: Fourteenth-Century English Schoolbooks and Their Contents’, New Medieval Literatures, 11 (2009), pp. 19–38). While vernacular poetry may sometimes have served as a pedagogical asset in this context, proficiency in Latin language and literature was the goal. For grammatical schooling in Europe generally, see, for instance, Rita Copeland, ‘Naming, Knowing and the Object of Language in Alexander Neckham’s Grammar Curriculum’, The Journal of Medieval Latin, 20 (2010), pp. 38–39; Manfred Kraus, ‘Grammatical and Rhetorical Exercises in the Medieval Classroom’, New Medieval Literatures, 11 (2009), pp. 63–89; Suzanne Reynolds, Medieval reading. Grammar, Rhetoric and the Classical Texts (Cambridge: Cambridge University Press, 1996); Tony Hunt, Teaching and Learning Latin in 13th-Century England, 3 vols (Cambridge: D. S. Brewer, 1991), 1, pp. 59–98 13 In the saga that gives by far the most information on schooling, Lárentíus saga, only Latin schooling, including poetic composition (versificatio), is mentioned (Biskupa sögur, 3, ed. Guðrún Ása Grímsdóttir, passim). 14 Benedicti Regula, ed. Rudolph Hanslik. Corpus Scriptorum Ecclesiasticorum Latinorum, 75 (Vindobonae: Hoelder/Pichler/Tempsky, 1960), pp. 114–19; Anna Grotans, Reading in Medieval St. Gall (Cambridge: Cambridge University Press, 2006), p. 106.

3.1 Manuscripts and the Scope of Old Icelandic Grammatica

107

3.1.1 Grammatical Manuscripts Based on Guðrún Nordal’s description, I list here the contents of the Old Icelandic grammatical manuscripts.15 R (AM 2367 4to; c. 1300–1325) includes Snorri’s Edda, Bjarni Kolbeinsson’s Jómsvíkingadrápa and Málsháttakvæði,16 probably also composed by Bjarni. Jómsvíkingadrápa treats pagan warriors in archaizing metre and diction (see above pp. 34–35). Málsháttakvæði is composed in end-rhyming hrynhenda and combines aphorisms with references to gods and heroes. One of the first texts that was read in the Latin grammatical curriculum after Psalms was generally a collection of aphorisms, Disticha Catonis, which was translated into Old Norse as Hugsvinnsmál.17 A collection of aphorisms like Málsháttakvæði thus fits well into a grammatical context. Both poems contain heroic and mythological lore, which is a conspicuous component of the Icelandic grammatical manuscripts. The placing of Jómsvíkingadrápa and Málsháttakvæði after Háttatal in R, furthermore, indicates that they may have been included not only for their content, but also for their metrical form. The fact the poems are likely by the same poet is probably irrelevant, since they are anonymous in R. U (DG 11 4to; c. 1300–1325) has Snorri’s Edda, 2GT, a genealogy of the Sturlungs, a list of lawspeakers and Skáldatal – a list of rulers and their poets. W (AM 242 fol; c. 1350) has Snorri’s Edda, 1–4GT, Rígsþula (an eddic poem) and some stanzas on Mary in a fifteenth-century hand on the last page as well as on an earlier page where space had been left blank.18 A (AM 748 Ib 4to; c. 1300–1325) has a fragment of 5GT, 3GT, Litla Skálda (a treatise on kennings), Skáldskaparmál from chapter 45 onwards and Haukr’s Íslendingadrápa: a poem dense in mythological references, enumerating Icelandic heroes of the Saga Age. The sister manuscript of A, AM 748 Ia 4to, was written by the same hand or a similar one, and though the two may not have been originally bound together, they seem intended to complement each other.19 This manuscript contains eddic poetry on the gods. B (AM 757 a 4to; c. 1400) has 3GT, Litla Skálda, Skáldskaparmál from chapter 45 onwards, but also with a number of earlier chapters,20 and devotional poetry. As compilations, A and B are closely related, although B lacks substantial amounts of text from 3GT,

15 Guðrún Nordal, Tools of Literacy, pp. 44–72. 16 See SkP 3, pp. 1213–14. Note, in addition to the arguments presented there, that Málsháttakvæði combines a love theme and myth, which suggests Ovidian inspiration, just as in Bjarni’s other poem Jómsvíkingadrápa (see above p. 34). 17 See, for instance, Kraus, ‘Grammatical and Rhetorical Exercises’, pp. 63–89 (p. 73); Hunt, Teaching and Learning Latin, 1, pp. 66–67. 18 Karl G. Johansson, Studier i Codex Wormianus. Skrifttradition och avskriftsverksamhet vid ett isländskt skriptorium under 1300-talet (Göteborg: Acta universitatis gothoburgensis, 1997), pp. 19–20. 19 Guðrún Nordal, Tools of Literacy, p. 58. 20 See Finnur Jónsson, ‘Edda Snorra Sturlusonar – dens oprindelige form og sammensætning’, Aarbøger for nordisk oldkyndighed og historie, 2nd series, 13 (1898), p. 297.

108

3 Grammatical Literature

such as the discussion of runes and the material following the descriptions of barbarisms.21 A and B are further discussed on pp. 130–33 below. C (AM 748 II 4to; c. 1400) contains only a fragment of Skáldskaparmál and genealogies. Finally, AM 756 4to and 757 b 4to both date to the fifteenth century. AM 756 4to is a transcription of Snorri’s Edda from W, whereas 757 b 4to is a two-leaf fragment of 3GT.22 In summary, Icelandic grammatical manuscripts cover: 1. Orthography and phonology (1–3GT) 2. Figures and tropes (3–4GT) 3. Poetics (3–5GT, Snorri’s Edda) 4. Mythological and heroic lore (Snorri’s Edda, Jómsvíkingadrápa, Málsháttakvæði, Íslendingadrápa, eddic poetry) 5. Aphorisms (Málsháttakvæði) 6. Longer poems of metrical interest (?) (Jómsvíkingadrápa and Málsháttakvæði; munnvǫrp and hrynhenda) 7. Devotional poetry The contents of these manuscripts are wholly consistent with a discipline dealing with language and poetry. Although the stanzas on Mary were added later in W, they presumably did not seem out of place, and the devotional poetry in B is contemporary with the remaining text.23 This may reflect the prominence of hymns in grammatical studies, attested to in many European manuscripts.24 In a similar vein, Snorri included a section on religious language in Skáldskaparmál. The focus on mythological and heroic lore in these manuscripts is prominent and becomes even more so if the most important collection of eddic poetry, GKS 2365 4to (c. 1270), is considered. This manuscript is normally referred to as the main witness to the Poetic Edda, but from this point onward, I shall avoid that title. In the Middle Ages, the texts within the manuscript probably had no collective name, although the individual poems did. The title suggests that we are dealing with a coherent literary work, like Snorri’s Edda or 3GT, and this is also the impression one gets from many modern editions and translations of eddic poetry. This is not the case, but rather the manuscript contains an anthology that was never treated as a work in its own right in the Middle Ages. The title is therefore 21 See Edda Snorra Sturlusonar, ed. Jón Sigurðsson et al., 2, pp. 501–11. 22 See Edda Snorra Sturlusonar, ed. Finnur Jónsson, p. xvii; Kristian Kålund, Katalog over den arnamagnæanske håndskriftsamling, 2 vols (København: Komissionen for det arnamagnæanske legat, 1889–94), 2, p. 180. 23 Guðrún Nordal, Tools of Literacy, p. 64. 24 See Helmut Gneuss, Hymnar und Hymnen in englischen Mittelalter. Studien zur Überlieferung, Glossierung und Übersetzung lateinischen Hymnen in England. Mit einer Textausgabe der lateinischaltenglischen Expositio Hymnorum (Tübingen: Max Niemeyer, 1968), pp. 194–206; Katherine Zieman, Singing the New Song. Literacy and Liturgy in Late Medieval England (Philadelphia: University of Pennsylvania Press, 2008), pp. 1–39.

3.1 Manuscripts and the Scope of Old Icelandic Grammatica

109

misleading if one wishes to study the Middle Ages rather than modern reception. Therefore, I refer to the manuscript as CR (for Codex Regius), to distinguish it from R (Codex Regius of Snorri’s Edda). The manuscript contains only eddic poetry about gods and heroes, and the inclusion of such poetry in the manuscripts above suggests that CR also belongs to the grammatical corpus. In high medieval grammatica, mythology belonged to a relatively advanced level and was studied above all in connection with Vergil’s Æneid, Ovid’s Metamorphoses and commentaries on those works.25 Some independent overviews were also written. In Icelandic grammatica, by contrast, different levels of grammatical material are mixed, perhaps because learning the language itself was not a problem. I turn now to the structure of Snorri’s Edda. In modern editions, it has four parts. It begins with a prologue which explains the rise of paganism, then comes Gylfaginning, which describes the mythology. Its narrative is framed as a lie that the false god Óðinn, in three persons, presents to king Gylfi. Then comes Skáldskaparmál, treating kennings and heiti (poetic synonyms), but also containing some mythological narrative relevant to the poetic examples. In the end comes Háttatal, which is a poem by Snorri himself containing 102 stanzas illustrating 95 metrical forms (hættir), accompanied by commentary. In U, probably the oldest of the manuscripts containing Snorri’s entire Edda, its text is preceded by a rubric in red letters. This is the most unambiguous source to the name, contents and authorship of that work: Bók þessi heitir Edda. Hana hefir saman setta Snorri Sturluson eptir þeim hætti sem hér er skipat. Er fyrst frá ásum ok Ymi, þar næst skáldskaparmál ok heiti margra hluta, síðast háttatal er Snorri hefir ort um Hákon konung ok Skúla hertuga.26 This book is called Edda. Snorri Sturluson has compiled it in the way it is ordered here: it first has [the story] of the æsir and Ymir, thereafter poetic diction and poetic synonyms for many things, and finally the metrical list that Snorri has composed about king Hákon and duke Skúli.

The contents given in the rubric can be listed as: 1. Prologue and Gylfaginning,27 2. Skáldskaparmál on kennings,28 3. Skáldskaparmál on heiti, 4. Háttatal. It is noteworthy that the rubric stresses that Snorri compiled his Edda in the way it is presented in the manuscript. This may perhaps be a reaction against reworkings of Snorri’s Edda, such as found in the contemporary A. Whatever the scribe’s reasons for stressing the order of the Edda, his ordering is corroborated by other manuscripts

25 See, for instance, Nicolette Zeeman, ‘In the Schoolroom with the ‘Vulgate’ Commentary on ‘Metamorphoses’ I’, New Medieval Literatures, 11 (2009), pp. 1–5. 26 Snorri Sturluson, The Uppsala ‘Edda’, ed. Heimir Pálsson and trans. Anthony Faulkes (London: Viking Society for Northern Research, 2012), p. 6. 27 U is the only medieval manuscript to have a separate rubric for Gylfaginning: hér hefr Gylfaginning (here Gylfaginning begins). 28 Chapters 1–53, in U preceded by the end of Gylfaginning.

110

3 Grammatical Literature

(R, W and Codex Trajectinus (T)), and there is thus good reason to believe it to be correct, even though both U and W have inserted additional material between the different parts of the Edda.29 Snorri’s Edda and the grammatical manuscripts are organized according to the same basic principle – a point not noted in previous scholarship. Snorri’s Edda, and Skáldskaparmál within it, has a structure where theory and description are given first, but where the text becomes more and more of an inventory (of poetry, synonyms and metre) as it progresses. It shares this structure with all grammatical manuscripts except the short fragment AM 757 b 4to, whose incompleteness might lead us to set it aside, and CR and AM 748 Ia 4to, where the almost exclusively poetic content would make such a structure impossible. The structure of the manuscripts may well derive from the internal logic of Snorri’s Edda, since it would appear that medieval editors would either insert material between the different parts of Snorri’s Edda or rework parts of it in a way that retained the basic progression from description to inventory. This is most obvious in W, where the second half of Skáldskaparmál, the socalled ókennd heiti listing poetic synonyms, has been placed together with Háttatal after 1–4GT. The redactor seems to have felt that lists should come after the treatises, as in Snorri’s Edda. R adds poetic samples after Háttatal. U inserts a genealogy and a list of lawspeakers after the descriptive, introductory part of Skáldskaparmál – a largely indifferent alteration of the scheme. It has 2GT between Skáldskaparmál and Háttatal, providing orthographic information before the technical description and inventory of metres. A and B have Litla Skálda before Skáldskaparmál and contain a version of Skáldskaparmál with almost no narrative, making it into a collection of poetic examples to follow the descriptive text of Litla Skálda. Both have 3GT in the beginning, in B reworked to serve more or less the same function as 2GT in U, and both manuscripts end with poetic samples. The corpus is thus structurally coherent and allows its users to read up on theory and description in the first part of each manuscript and consult the second part as an inventory.

3.2 Háttatal: Models, Sources and Rhetoric Of the three parts of Snorri’s Edda to be analysed in this chapter, Háttatal has the most obvious model and displays Snorri’s strategies most clearly. It is therefore a good place to start the exploration of Snorri’s sources and methods. Háttatal is a metrical list with commentary and in that respect, it amounts to a treatise on poetics. With regard to the poem’s contents, it is a panegyric about king Hákon and earl Skúli of Norway, and it is the one of Snorri’s extant works that clearly reveals his political ambitions.

29 See Guðrún Nordal, Tools of Literacy, pp. 46, 50, 55–56.

3.2 Háttatal: Models, Sources and Rhetoric

111

If Snorri was the greatest known author of the thirteenth century, he also had an extraordinary lust for power and riches. Largely through wives and concubines he became the richest man in Iceland, and he had befriended the king and the earl. His ambitions were too great for his own good, and when Snorri defied the king’s prohibition to leave Norway, the king sent a letter to Gizurr Þorvaldsson, asking him to either bring Snorri to Norway or to kill him. Shortly afterwards, on 23 September 1241, Snorri was cut down in his basement, where he had fled to find shelter.30 In Háttatal, the two Snorris – the politician and the antiquarian poet – are both clearly visible.31 But one may also find a third Snorri here, a kind of hybrid of the two others. Háttatal is where his claims to authority come through most clearly, revealing an ambition akin to that of Snorri the politician, but on a literary level. Below, I analyse how he went about constructing that authority. First, however, we consider his treatment of his model, Háttalykill. In Orkneyinga saga, we read that earl Rǫgnvaldr of Orkney and the otherwise largely unknown Hallr Þórarinsson in the 1140s composed a poem which the saga calls either Háttalykill hinn forni (the old metrical key) or Háttatal hit forna (the old metrical list), today generally referred to as Háttalykill.32 There may thus have been at least one older poem similar to Snorri’s Háttatal, but there are two fundamental problems connected to it. First, we have a poem in two late paper manuscripts that matches the description of ‘metrical key’, but is it the one mentioned in the saga?33 It bears the right title, but that is likely to be the guess of the scribe, Jón Rugman.34 30 See Sigurður Nordal, Snorri Sturluson (Reykjavík: Þór B. Þorláksson, 1920), pp. 11–21. 31 The tension between the two Snorris has attracted some scholarly interest, and a solution has been suggested by Kevin Wanner, Snorri Sturluson and the Edda. The Conversion of Cultural Capital in Medieval Scandinavia (Toronto: University of Toronto Press, 2008). In the tradition of Pierre Bourdieu, Wanner sees Snorri’s literary activity as a quest to enhance his own cultural capital, and thus in service to his political ambitions. Biographies of Snorri include: Sigurður Nordal, Snorri Sturluson; Fredrik Paasche, Snorre Sturlason (Kristiania: Cappelen, 1922); Óskar Guðmundsson, Snorri. Ævisaga Snorra Sturlusonar 1179–1241 (Rekjavík: Forlagið, 2009). 32 For the use of the two names, see Háttalykill enn forni, ed. Jón Helgason and Holtsmark, p. 5. Regarding Hallr, a single stanza of his is found in Orkneyinga saga (SkP 2, pp. 610–11). 33 The poem is preserved in Stockholm, Kungliga biblioteket, Holm. Papp. 25 8vo and Uppsala, Universitetsbiblioteket, R: 683, Salanska samlingen 28 fol, both from the second half of the seventeenth century and written by Jón Rugman (Háttalykill enn forni, ed. Jón Helgason and Holtsmark, pp. 7–21; Vilhelm Gödel, Katalog öfver Upsala universitets biblioteks fornisländska och fornnorska handskrifter. Skrifter utgifna af Humanistiska Vetenskapssamfundet i Upsala, 2:1 (Uppsala: Almqvist & Wiksell, 1892), pp. 27–29). In the description of the poem in Orkneyinga saga, different versions preserve different traditions about how many stanzas were used to exemplify each metre. The oldest variant reading seems to be that there were three stanzas for each metre, whereas the actual poem has two. Even so, the passage supports, rather than contradicts, the attribution, since it suggests that the author knew that in contrast to Háttatal, Háttalykill had more than one stanza for each metre (cf. Háttalykill enn forni, ed. Jón Helgason and Holtsmark, pp. 5, 118, although negative weight is there accorded to this incorrect description). 34 Háttalykill enn forni, ed. Jón Helgason and Holtsmark, p. 13.

112

3 Grammatical Literature

The saga does not quote the poem, so we have no text for comparison. And second, if it is the right poem, can we know whether Snorri used it in the composition of Háttatal? The saga is silent on this point. As we shall see, the two questions are interrelated. Earlier research has answered both in the affirmative, but since the issue is so important for understanding how Snorri worked, I review the main arguments and add a few observations. Orkneyinga saga speaks of the old metrical key, suggesting that there was one old and one new. Considering the impact of Snorri’s Edda soon after its composition (see pp. 252–96), the *‘new key’ would in all likelihood have been Snorri’s Háttatal. Although the original version of Orkneyinga saga was probably composed around 1200, it has undergone later reworking, and Snorri’s historiographical writings are once explicitly mentioned.35 The reference to Háttalykill hinn forni is therefore likely to have been inserted during the reworking, when Snorri’s Edda had become a point of reference. Rǫgnvaldr’s poem, accordingly, became the ‘old’ metrical key.36 This suggests that the author/interpolator knew of only one poem like this before Snorri. Thus, if the poem in the paper manuscripts can be shown to be older than Háttatal, it is most likely the right one. As it turns out, a survey of letter forms, orthography and language in the paper manuscripts has shown that its lost exemplar probably dated to the twelfth century, indicating that the poem is older than Háttatal.37 If Snorri can be shown to have used the poem, this would further corroborate its identification. This seems indeed to be the case. There is considerable metrical overlap between the two poems, and Háttatal appears to be an expansion of Háttalykill – only one metre is found in Háttalykill which is not, apart from very minor variations, also found in Háttatal, whereas Háttatal has many additional metres.38 This overlap would perhaps not be very noteworthy if the metres in question were traditional, but most of them are not found in earlier poetry.39 Furthermore, Háttatal uses the same names or similar ones to those found in some of the headings in Háttalykill.40 We seem to be dealing with a situation of dependence and expansion. These observations are, I believe, sufficient to confirm that we have the right poem and that Snorri used it. I proceed now to how he used, altered and expanded on his model. On point after point, Snorri presents more elaborate versions of the metres found in Háttalykill, subdividing and regularizing them further. For a full discussion, I refer the reader to the analyses by Anthony Faulkes and Jón Helgason and Anne

35 Háttalykill enn forni, ed. Jón Helgason and Holtsmark, pp. 5–6; Orkneyinga saga, ed. Finnbogi Guðmundsson, pp. xlii–xlii, 101. 36 Cf. Háttalykill enn forni, ed. Jón Helgason and Holtsmark, p. 118. 37 Háttalykill enn forni, ed. Jón Helgason and Holtsmark, pp. 99–118. 38 Háttalykill enn forni, ed. Jón Helgason and Holtsmark, p. 119. 39 Háttalykill enn forni, ed. Jón Helgason and Holtsmark, pp. 119–21. 40 Snorri, Háttatal, ed. Faulkes, p. xviii; SkP 3, 1101–1102.

3.2 Háttatal: Models, Sources and Rhetoric

113

Holtsmark.41 By way of example, I shall describe one instance of elaboration which Snorri has undertaken and which has not been discussed by these scholars. In Háttalykill and Háttatal alike, several metres are defined by interlinear hendingar, that is, the latter hending in one line carries over into the next: Haraldr magnaði hildi, hildr tíddisk gram mildum42 Haraldr increased the battle; the battle was pleasing to the generous ruler.

In Háttalykill, this can be performed in two ways. In the example above, the second hending in the first line rhymes with the first hending in the second. In five instances, however, the second hending in the first line rhymes with a syllable which does not form part of the hendingar in the second line: Morðs lét hilmir Hǫrða hǫrð reyr lituð dreyra.43 The Hǫrðar’s ruler [king of Norway] caused the hard murder-reeds [swords] to be reddened in blood.

Here, hǫrð rhymes with morðs and hǫrða, but not with reyr and dreyra. The pattern thus breaks the regular hending scheme and produces five rhyming syllables. After the tenth century, hendingar could occasionally be retained between lines, as in the first example above, but couplets with five or three hendingar, rather than the ordinary four, were disallowed (couplets containing three hendingar are discussed below on pp. 228–31).44 The fixing of the number of hendingar to four per couplet was one of the factors involved in the evolution of strict, classical dróttkvætt. Rǫgnvaldr and Hallr apparently modelled these metres on early poetry, without respecting the four-hending rule that had prevailed for a century and a half when they composed their poem. In Háttatal, however, no examples of the pattern with five hendings occur, and we thus see that Snorri has disallowed this possibility.45 This is one of several instances where Háttatal represents a further development than 41 Snorri, Háttatal, ed. Faulkes, pp. xiii–xv, 76–77, 85–86; Háttalykill enn forni, ed. Jón Helgason and Holtsmark, pp. 118–21, 140. 42 SkP 3, p. 1073. 43 SkP 3, p. 1073. Finnur Jónsson emended morð to morðs, which has since been confirmed by a second manuscript (Háttalykill enn forni, ed. Jón Helgason and Holtsmark, p. 85). 44 On the chronological distribution of couplets with five rhymes, see Myrvoll, ‘The Authenticity of Gísli’s Verse’, Journal of English and Germanic Philology, forthcoming. On that of couplets with three rhymes, see p. 230 n. 126. 45 Cf. the statement: ‘Víða er þat í fornskálda verka er í einni vísu eru ýmsir hættir eða háttafǫll, ok má eigi yrkja eptir því, þó at þat þykki eigi spilla í fornkvæðum’. (Snorri, Háttatal, ed. Faulkes, p. 26) (It is common in the works of the early poets that the same stanza will display several metres or irregular metre, and we may not compose in that manner, even though it is not seen as unbefitting in old poems.)

114

3 Grammatical Literature

Háttalykill towards regularization in the spirit of normative grammatica.46 The example also highlights another difference between the two poems: in Háttalykill, the lines above belong to one of the two metres which are defined by interlinear hendingar. Háttatal, by contrast, has four such metres.47 The example thus illustrates how Snorri not only regularized, but also further subdivided the metres he found in his exemplar.

3.2.1 Authorship of the Commentary Snorri did more than increase the number of metres and regularizing them: he also added commentary. There has been some debate about the authorship of the commentary. Theodore Möbius and R. C. Boer took a number of discrepancies between the commentary and the poem to indicate that Snorri was not the author of the commentary.48 Finnur Jónsson convincingly explained these discrepancies as due to faulty interpretation of the text by Möbius and Boer, to scribal intervention (during dictation or transmission) and to imprecision in the treatment of a topic for which a fully developed terminology did not yet exist in Snorri’s time.49 He also added positive arguments in favour of Snorri’s authorship (see below). Finnur Jónsson’s arguments are on the whole so good that they50 should have settled the matter. Anthony Faulkes and SkP do not seem to have been aware of Finnur Jónsson’s article, however, and the current standard editions therefore give the impression that the matter is less certain than it really is. For instance, Faulkes has noted ten minor discrepancies between the stanzas and the commentary, but several of these have been explained by Finnur Jónsson. Furthermore, discrepancies can also be found at several points in Skáldskaparmál.51 Finnur Jónsson lists a number of indications in support of Snorri’s authorship of the commentary. One is that the wording in Skáldskaparmál and the Háttatal

46 As far as I am aware, there is only one exception to this: Háttlausa has a more consistent placing of hǫfuðstafr in Háttalykill than in Háttatal (Snorri, Háttatal, ed. Faulkes, p. 81). It is presumably because of the archaic nature of that metre that Snorri for once let some of the inconsistency of the early poets slip through. 47 Háttalykill has dunhenda and iðrmælt, but Háttatal in addition has klifat and liðhendur (Snorri, Háttatal, ed. Faulkes, p. 80). 48 Snorri Sturluson, Háttatal Snorra Sturlusonar, ed. Th. Möbius, 2 vols (Halle: Verlag der Buchhandlung des Waisenhauses, 1879–1881); R. C. Boer, ‘Om kommentaren til Háttatal’, Arkiv för nordisk filologi, 43 (1927), pp. 262–309. 49 Finnur Jónsson, ‘Snorre Sturlusons Háttatal’, Arkiv för nordisk filologi, 45 (1929), pp. 229–69. 50 Snorri, Háttatal, ed. Faulkes, pp. vii–ix; cf. SkP 3, p. 1095.. 51 Snorri, Skáldskaparmál, ed. Faulkes, p. xxi. To Faulkes examples may be added the flat contradiction between the two explanations for the kenning type ‘dwarves’ ship [poetry]’ (Snorri, Skáldskaparmál, ed. Faulkes, pp. 11, 14).

3.2 Háttatal: Models, Sources and Rhetoric

115

commentary reveals an awareness of the overall architecture of Snorri’s Edda from an authorial perspective. In the beginning of Skáldskaparmál, ‘mál ok hættir’ (diction and metre) are said to be the constitutive features of poetry. Hættir are thereafter left out of the discussion for the duration of Skáldskaparmál, until it is picked up again early in the Háttatal commentary. Here one sees an overall plan of commenting on one constitutive feature first, then the other.52 Another indication is found in the commentary preceding stanza 17. The semantics of this stanza are so complex that the stanza is followed by a long explanation of almost every word. Before the stanza the commentary says that it is difficult to find all the right words for this metre, and then: ‘En sýnt er í þessi vísu þat er orðin munu finnask ef vandlega er leitat, ok mun hér þat sýnask at flest frumsmíð stendr til bóta’ (But in this stanza one sees that the [right] words will be found if one inquires thoroughly, and one may see here that most initial compositions may be improved). The poet is here referring to the considerable effort he spent in composing this stanza and to his own experience that the right words will be found in the end. I would add additional positive indications of Snorri’s authorship of the commentary to those of Finnur Jónsson. One is the use of the term nykrat in the commentary. The term, ‘monsterized or nykr-ized’, refers to an inconsistent use of extended metaphor (exactly which kind of monster the nykr is will be discussed below). Snorri’s nephew, Óláfr Þorðarson, calls the same phenomenon ‘nykrat eða finngálknat’ (monsterized or centaurized), expanding on the term in the commentary in line with Horace’s description (see below). The Fourth Grammarian, in turn, explicitly attributes the term finngálknat to Óláfr: ‘[. . .] ok kallar Óláfr þat finngálknat’ (and Óláfr calls this finngálknat).53 There thus seems to be a progressive development of terminology in the sources. Given the date of 3GT, this would place the commentary before c. 1250, which brings us close to Snorri’s time. There are no manuscript witnesses to a version without commentary. Furthermore, in two medieval sources where Háttatal has been used, the version in question clearly had a commentary: the prologue to 1‒4GT refers to Snorri, meaning the commentary to Háttatal 8,54 whereas a pseudonymous stanza attributed to Grettir makes use of both stanza and commentary in Háttatal (see pp. 268–69). Finally, the words in the rubric in U bear repetition:

52 Finnur Jónsson, ‘Snorre Sturlusons Háttatal’, p. 233. 53 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 131. 54 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 154: (‘[. . .] kenningar eigi lengra reknar en Snorri lofar’ (kennings that are not expanded beyond what Snorri allows)). This reference, however, cannot on its own secure the authorship, since a later reworking may have been attributed to the original author, as in the case of Litla Skálda and Skáldskaparmál, both of which are attributed to Snorri in A (see p. 131).

116

3 Grammatical Literature

Bók þessi heitir Edda. Hana hefir saman setta Snorri Sturluson eptir þeim hætti sem hér er skipat. Er fyrst frá ásum ok Ymi, þar næst skáldskaparmál ok heiti margra hluta, síðast háttatal er Snorri hefir ort um Hákon konung ok Skúla hertuga.55 This book is called Edda. Snorri Sturluson has compiled it in the way it is ordered here: it first has [the story] of the æsir and Ymir, thereafter poetic diction and poetic synonyms for many things, and finally the metrical list that Snorri has composed about king Hákon and duke Skúli.

The implications of this rubric for the authorship of the commentary are not straightforward, and the matter calls for some elaboration. The rubrics in U are written with the same hand as the main text, and scribe here claims that ‘Snorri Sturluson has compiled it [his Edda] in the way it is ordered here’, which suggests that to his mind, Snorri composed Háttatal with a commentary.56 True, U has interpolated lists of poets and lawspeakers, as well as 2GT, into Snorri’s Edda, and Skáldskaparmál has been somewhat rearranged.57 There are a number of abbreviated passages, and Daniel Sävborg has argued convincingly that these are not original to the text, but are rather the result of changes during transmission.58 Lasse Mårtensson has shown that with regard to the rendering of names, the scribe sometimes fails on account of misapprehension but does not generally alter them because of reanalysis.59 Several such reanalyses have, however, been undertaken at an earlier point in transmission.60 These changes do not alter the fundamental character of the different sections of the work, however. Since medieval transmission of vernacular texts allowed for much variation, such changes would probably not affect the scribe’s perception that he was presenting Snorri’s Edda in the way Snorri had composed it. A version of Háttatal without commentary is another matter, since the text would then no longer be a treatise but only a poem. The scribe’s treatment of Háttatal itself may also tell us something about how he perceived that text. In the end of the manuscript, he first gives an index to Háttatal, providing names of metres that lack a name in

55 Snorri, The Uppsala ‘Edda’, ed. Heimir Pálsson and trans. Faulkes, p. 6. 56 On the hand in the rubrics, see Snorri Sturluson, Snorre Sturlasons ‘Edda’. Uppsala-handskriften DG 11, ed. Anders Grape, Gottfrid Kallstenius and Olof Thorell (Uppsala: Almqvist & Wiksell, 1962–1977), 2, p. x. 57 It seems likely, however, that the long poems which are found in the R, T, W branch but which are absent in U have been added to the R, T, W branch rather than cut out in U (Edda Snorra Sturlusonar, ed. Finnur Jónsson, pp. xxi–xxiii; but cf. Snorri, Skáldskaparmál, ed. Faulkes, xlii–xliv). 58 ‘Blockbildningen i Codex Upsaliensis. En ny metod att lösa frågan om Snorra Eddas ursprungsversion’, Maal og Minne (2012), pp. 12–53. 59 Mårtensson, Skrivaren och förlagan, p. 265. 60 Haukur Þorgeirsson has in addition shown stemmatically that this is the case also in one instance where Mårtensson attributes the reanalysis to the U-scribe. Rather, it probably occurred earlier in the transmission; see Review of Mårtensson, Skrivaren och förlagan. Amsterdamer Beiträge der älteren Germanistik, 76 (2016), pp. 152–53.

3.2 Háttatal: Models, Sources and Rhetoric

117

Háttatal.61 Then follows Háttatal with the rubric: ‘Háttal, er Snorri Sturluson orti um Hákon konung ok Skúla hertoga’ (the metrical list that Snorri composed about king Hákon and duke Skúli). He has not inserted the additional names of the list into the commentary of Háttatal, which suggests that he respected its integrity. Both the index and Háttatal are defective. The index stops after stanza 36 (skipping stanza 35), whereas Háttatal stops after stanza 56, in the middle of the metres of the early poets. In both instances, it would appear that the scribe had defective exemplars.62 The text of the index shares some significant variants with the Uversion of Háttatal, but it also shares significant variants with other manuscripts against U. It would therefore appear that the index was not constructed by the scribe based on his exemplar of Háttatal, but that he had an exemplar for the index as well.63 By all appearances, the scribe intended to deliver on his promise to provide Snorri’s entire Edda, but his exemplars did not allow him to do so. Háttatal in U seems to be derived from another exemplar than the rest of Snorri’s Edda, probably one from the early to mid-thirteenth century, and thus during or close to Snorri’s lifetime. The indications are: the graph used for /k:/, the medio-passive endings -sk, -mk and the misreading of a Carolingan f for ∫.64 It would therefore seem that the scribe has actively collected at least two exemplars in order to provide the full text of Snorri’s Edda. But how does this relate to the claim that ‘Snorri Sturluson has compiled it in the way it is ordered here’? Was this rubric taken over from an exemplar or does the scribe refer to his own reconstruction of Snorri’s work? Here, the similarity between the rubric of Snorri’s Edda and that of Háttatal is instructive. The first ends ‘[. . .] háttatal er Snorri hefir ort um Hákon konung ok Skúla hertuga’, whereas the second reads ‘Háttal, er Snorri Sturluson orti um Hákon konung ok Skúla hertoga’. They differ only in the tempus of the verb and in the patronym, which is absent in the first rubric, presumably to avoid repetition. Since Háttatal is transcribed from another exemplar than the rest of Snorri’s Edda, the similarity is not likely to be due to that a scribe at a previous stage in the transmission remembered what he had written before. One of these rubrics therefore appears to have affected the other.

61 See Lasse Mårtensson, ‘Översikten över Háttatal i DG 11 4to – dess funktion och ursprung’, Gripla, 20 (2009), pp. 105–45. It was previously thought that the scribe had first intended to write only the first line or lines of each metre, but then changed his mind and wrote down Háttatal in its entirety (Finnur Jónsson, ‘Edda Snorra Sturlusonar’, pp. 330–31). 62 Snorri, The Uppsala ‘Edda’, ed. Heimir Pálsson and trans. Faulkes, pp. lxxxvi–lxxxvii; Mårtensson, ‘Översikten över Háttatal’, p. 140. 63 Mårtensson, ‘Översikten över Háttatal’, pp. 139–41 (see variants ástar and fagrskjǫlduzustum p. 139). 64 Mårtensson, Skrivaren och förlagan, p. 57–58, 86–89, 239–42, 248, 264. This is also the case in W; see Johansson, Studier i Codex Wormianus, pp. 240.

118

3 Grammatical Literature

The U scribe does not generally add rubrics to subsections, as seen by Skáldatal and the genealogies of the Sturlungs and of the lawspeakers, as well as 2GT and the index to Háttatal. All this is material that the scribe has inserted into the text, and he has not added any rubrics. He is therefore unlikely to have added the rubric to Háttatal. Presenting the entire work and its relation to Snorri’s own version is a different matter, however. This is not a question of indexing, but of attributing authority to the text, and the scribe may well have wanted to do so. It would thus appear that he has added the initial rubric of Snorri’s Edda based on the one found in his exemplar of Háttatal.65 In the rubric to Snorri’s Edda, the scribe clarifies his aims with the text he is about to write, although his deficient examplars to Háttatal and to its index did not allow him to reach them completely. If one may judge by R, W and T what Snorri’s Edda looked like, then the scribe’s efforts must have been successful. His recreation of Snorri’s Edda may account for how he stresses its original composition – he has based his compilation on various exemplars, but the reader should rest assured that the resulting text corresponds to the one gathered by Snorri. All the medieval evidence thus supports a version of Háttatal with commentary, and so do Gylfaginning and Skáldskaparmál on a structural level, because they also consist of poetic quotations with commentary. Snorri, it would seem, was not of a mind to let the analysis be up to the whim of the reader. The minor discrepancies between prose and verse are the only reason that Snorri’s authorship of the prose has been called into doubt, and Finnur Jónsson has dealt convincingly with these. When the above observations are combined, I believe the case to be strong enough to proceed to an analysis of the commentary as a work of Snorri.

3.2.2 Building Authority and Creating Order Initially it may be worth considering the general question of what the commentary does for the poetry on which Snorri is commenting. In the Genealogiae deorum gentilium, Boccaccio laments that poetry is the only type of text to be transmitted without commentary.66 While this is clearly a rhetorical exaggeration, it remains true that nothing like a Glossa ordinaria developed for the classical poets; the closest thing would have been the Servius commentary on the Æneid. In practice, Dante and Boccaccio both addressed this problem with regard to vernacular poetry, Dante by commenting on his own poetry in Vita Nuova and the Convivio and on that of others in De vulgari eloquentia, Boccaccio by editing Dante’s poetry with a commentary or

65 The rubric contains no archaic features that would call this conclusion into question. 66 See Martin Eisner, Boccaccio and the Invention of Italian Literature. Dante, Petrach, Cavalcanti, and the Authority of the Vernacular (Cambridge: Cambridge University Press, 2013), p. 58.

3.2 Háttatal: Models, Sources and Rhetoric

119

even by manipulating the presentation of Vita Nuova to make it graphically appear more like a commented text.67 The authority of poetry, and vernacular poetry not least, was a major concern for Dante, Boccaccio and Snorri, and a commentary was the prime marker of such authority. True, the interpretive ambitions of Dante and Boccaccio were of a higher order than Snorri’s relatively factual comments on forms and kennings, but Snorri’s commentary still amounts to something approaching Donatus’s treatment of Vergil, indicating that his own poem was worthy of serious study and imitation. Furthermore, when he singles out which poets and features should be imitated and which should not be in Skáldskaparmál, he establishes a canon and instructs the reader how to use it. When he thereafter presents his own poetry (Háttatal) as a stylistic digest of the preceding tradition, he not only appropriates the authority of that tradition for himself, but he also reshapes earlier forms in accordance with the norms of Latin grammatica. In the following, I highlight two points in this process, namely how he sought to increase the authority of his own commentary, and how he went about establishing the rules. Snorri’s regularization and subdivision of metres in Háttalykill may partly be attributable to a wish to display his poetic ability and to outshine his predecessors. Thus, Snorri’s first example of refhvǫrf (stanza 17) is staggeringly complex, with sixteen pairs of semantically opposing words which can also be construed as kennings and clauses crammed into a stanza that obeys all the rules of dróttkvætt.68 The examples found in Háttalykill (the metre is there called refrún) are nowhere near as elaborate.69 In other instances, such as in the regularization of hendingar in metres with interlinear rhymes discussed above, Snorri’s aim seems not to have been so much to show his own dexterity as to achieve consistency. The clearest instance is when he discusses the metres of the early poets. In Chapter One, I describe how he regularized and subdivided the early, lax type of dróttkvætt into several metres. I repeat the relevant passage for a closer analysis of Snorri’s rhetorical strategy: Nú skal rita þá háttu er fornskáld hafa kveðit ok eru nú settir saman, þótt þeir hafi ort sumt með háttafǫllum, ok eru þessir hættir dróttkvæðir kallaðir í fornum kvæðum, en sumir finnask í lausum vísum, svá sem orti Ragnar konungr loðbrók með þessum hætti: [. . .] Víða er þat í fornskálda verka er í einni vísu eru ýmsir hættir eða háttafǫll, ok má eigi yrkja eptir því, þó at þat þykki eigi spilla í fornkvæðum.70

67 On Dante’s auto-commentary, see Ascoli, Dante and the Making of a Modern Author, pp. 175–78; on Boccaccio, see Eisner, Boccaccio and the Invention of Italian Literature, pp. 60–61. 68 Snorri, Háttatal, ed. Faulkes, pp. 11–12. 69 SkP 3, pp. 1047–49, 1063–65, 1078–80. 70 Snorri, Háttatal, ed. Faulkes, pp. 24–26.

120

3 Grammatical Literature

Now I shall write down the metres in which the early poets composed, and these are collected here, even though they composed some [poetry] with metrical inconsistencies. These metres are called dróttkvætt in old long poems, but some of them are to be found in single stanzas, as king Ragnarr loðbrók composed in this metre: [. . .] It is common in the works of the early poets that the same stanza will display several metres or metrical inconsistencies, and we may not compose in that manner, even though it is not seen as unbefitting in old poems.

Here Snorri ends up in a tight spot between his deference towards the early poets and his wish to regularize, since they composed in a lax dróttkvætt that did not meet his standards. In order to justify his metrical subdivisions, he speaks of the metres of the early poets, whereas in fact it is clear from his examples that he bases his examples on one single metre, dróttkvætt. He is, of course, well aware of this and seeks a solution in the plural dróttkvæðir hættir (dróttkvætt metres).71 He thereby presents the matter as if the early poets offered a precedent for a subdivision into several metres, but sometimes deviated from the metre in which they composed. With the words ‘as King Ragnarr loðbrók composed in this metre’ Snorri fleshes out the illusion of the several metres among the early poets, but tellingly, he has to compose the poetry himself in order to do so. He seems to suspect that some of his readers will recognize the difference between his own ‘early’ metres and actual early poetry, as exemplified in Skáldskaparmál, and he explains this as well with reference to different metres (‘the same stanza will display several metres or metrical inconsistencies’). The two final clauses, ‘and we may not compose in that manner, even though it is not seen as unbefitting in old poems’, betray what this is really all about: we must have rules that can be recognized as such and we should stick to them, even though the old poets did not. This perspective is inherent to medieval grammatical studies generally. It is this prescriptive attitude that Snorri applies to the early poets and Háttalykill alike. Háttalykill already bore the marks of normativity, and a student of grammar would recognize its implicit rules as such. Snorri could thus make them stricter without justifying what he was doing. With regard to the early poets, however, theirs and Snorri’s views on metre and rules diverged so much that Snorri was obliged to somehow explain the difference. His way of doing so was twofold. For one thing, he presented the common medieval view that the old poets could do things that the student should not. His other strategy, that of projecting several metres onto the past and thus ‘modernizing’ the early poets, was more innovative, but fully in line with the medieval propensity to understand the past in terms of the present.

71 He is, however, consistent in regarding irregular hendingar as constituting ‘full change of metre’ ( fullt háttaskipti), and the subdivision is thus not simply an ad hoc solution for dealing with the early poets (Snorri, Háttatal, ed. Faulkes, p. 16).

3.2 Háttatal: Models, Sources and Rhetoric

121

3.2.3 Snorri and Horace In another instance, Snorri is silent about the difference between his recommendations and the styles of the poets. As we saw in Chapter Two, skaldic poets sometimes used extended metaphor inconsistently. Snorri disapproves of this practice and writes: Þá þykkja nýgjǫrvningar vel kveðnar ef þat mál er upp er tekit haldi um alla vísulengð. En ef sverð er ormr kallaðr, en síðan fiskr eða vǫndr eða annan veg breytt, þat kalla menn nykrat, ok þykkir þat spilla.72 Extended metaphors are thought to be well performed if the chosen subject is retained throughout the stanza. But if a sword is called a snake, and then a fish or a wand, or changed in some other way, that is called a monstrosity (nykrat), and it is seen as a flaw.

The word nykrat is derived from the hybrid monster nykr, on which more below. As discussed on pp. 52–53, poets did not generally abide by the rule presented in this quotation, and given his relatively frequent citation of poetry that disproves his claim, Snorri can hardly have been unaware of this fact. Unlike the case of early metrics, however, where a mere glance at, for instance, Bragi’s poetry is enough to reveal Snorri’s misrepresentation of it, testing the validity of his claim with regard to metaphorical consistency would require a concerted engagement with the texts. Such academic probation seems not to have been part of the medieval Icelandic intellectual discourse.73 Perhaps for this reason, Snorri does not clarify the novelty of his rule but assumes his usual posture of transmitter, rather than co-creator, of tradition. There are several indications that Snorri’s insistence on coherent imagery may have had a more specific source than common grammatical sentiment. The use of a monster for a poetic figure is typologically abnormal. Several naming principles can be observed within Old Icelandic poetics, such as naming metres and poetic devices after a poet or after their sound, according to their function or by means of an analogy to house-building and handicrafts.74 Monsters, however, are otherwise unknown. The term itself therefore suggests the use of an external source. I contend that the source in question is the main theoretical authority on poetic composition up to c. 1200 and beyond, namely Horace’s Ars poetica, and that this source accounts for the prohibition against inconsistent use of extended metaphor

72 Snorri, Háttatal, ed. Faulkes, p. 7. 73 Intellectual debates in medieval Iceland appear to have typically taken the form of producing opposing claims rather than of falsifying the empirical foundations of the claims of others. One such debate is examined in Chapter Five. 74 See, for instance, Sverrir Tómasson, ‘Skáldskapur og fræði fyrir stokk innan’, in Freyjas psalter. En psalter i 40 afdelinger til brug for Jonna Louis-Jensen, ed. Bergljót S. Kristjánsdóttir and Peter Springborg, 2nd ed. (København: Det arnamagnæanske Institut, 1997), pp. 190–92.

122

3 Grammatical Literature

as well as the name nykrat.75 The argument is to some extent straightforward, since it is inherently plausible that the most important Latin text on poetics would have influenced Snorri’s efforts to produce a text on the same topic, but it is somewhat complicated by Scandinavian monster semantics, to be discussed below. Here are the opening lines of the Ars: Humano capiti cervicem pictor equinam iungere si velit, et varias inducere plumas undique collatis membris, ut turpiter atrum desinat in piscem mulier formosa superne, spectatum admissi risum teneatis, amici? If a painter chose to join a human head to the neck of a horse, and to spread feathers of many a hue over limbs plucked up now here now there, so that what at the top is a lovely woman ends below in a black and ugly fish, could you, my friends, if favoured with a private view, refrain from laughing?76

Here we learn that we should avoid producing stylistic hybrids in order not to be laughed at. The first of the three monsters is a centaur and the third is a mermaid, while the precise nature of the feathered hybrid remains elusive. Its classification is of no consequence for the present argument, which concerns only the two identifiable monsters. Nykrat should probably be seen as an allusion to or calque of one of these, since, first, an external source to the term nykrat should be sought, given that its monster reference is typologically deviant; second, it should be sought in the Ars, since Icelandic poets did not, in fact, avoid incoherent imagery, and the basic text conveying this message during the Middle Ages was the Ars; and third, the monsters in question overlap with the meaning of nykr, the noun that forms the base of the adjective nykrat.77 The last claim calls for a discussion of exactly what kind of monster the nykr was. Most sources to this creature are early modern or modern, and its main manifestations are either as a horse or as a man. In mainland Scandinavia, Näcken (Swedish) or Nøkken (Norwegian) is sometimes a man who lives in water and

75 During Snorri’s lifetime, Matthew of Vendôme composed his Poetria nova (new poetics) (c. 1190?) to complement the Poetria vetus (old poetics) of Horace. Until then, however, Horace’s Ars had been the standard text on poetics and its influence remained profound later as well (see Karsten Friis-Jensen, ‘The Reception of Horace in the Middle Ages’, in The Cambridge Companion to Horace, ed. Stephen Harrison (Cambridge: Cambridge University Press, 2007), p. 300). 76 Horace, Satires. Epistles. Ars Poetica, ed. and trans. H. R. Fairclough (Cambridge: Harvard University press, 1929), pp. 450–51. 77 Clunies Ross, ‘Skáldskaparmál’, pp. 76–77; eadem, A history of Old Norse Poetry and Poetics, p. 244, notes that Snorri is likely to have been influenced by the concept of decorum as laid out in the opening of Horace’s Ars, but she does not suggest influence on the terminology itself.

3.2 Háttatal: Models, Sources and Rhetoric

123

drowns people, above all young women.78 In early modern and modern Iceland, nykurinn shares the same habitat and activity, albeit with a preference for children and men. In Iceland, however, he is generally a horse or, in some cases, some other animal, and more specifically a horse with its hoofs and ears turned backwards.79 He is not attested as a man in Iceland. Since most sources are so late, it cannot a priori be excluded that Snorri perceived of the nykr as a man. Several indications tell against this, however. First of all, if Snorri aimed for a parallel to Horace’s mermaid, he could have come up with closer ones, such as margýgr (mermaid) or marmennill (merman). Furthermore, two additional references to the nykr in Old Icelandic literature also speak in favour of the horse rather than the man. Snorri’s nephew Óláfr elaborated on his terminology in saying that the linguistic vice cacenphaton (ugly expression) is called nykrat or finngálknat in the vernacular. Finngálkn is Old Norse for ‘centaur’, that is, a creature which is partly horse, and Óláfr says about this vice: ‘ok er þar svá skipt líkneskjum á hinum sama lut, sem nykrinn skiptiz á margar leiðir’ (and there [in nykrat] there is a shift in metaphors for the same object, just as the nykr is transformed in many ways).80 There is some debate as to whether ‘transformed in many ways’ refers to the twisted hoofs and ears of the Icelandic, horse-shaped nykr or if it means that the nykr actually becomes something else, such as a cow or a fish.81 There is evidence for both in later Icelandic tradition, but there are no such traditions relating to the anthropomorphic manifestation of Näcken/Nøkken.82 This would suggest that Snorri is referring to the horse variety of the monster. The only other use of the word in Old Icelandic literature is found in a translation of the Latin Epistola Alexandri, a spurious letter of Alexander the Great to Aristotle. Nykrar (pl.) there translates ipotami (= hippopotami).83 This is really a Greek word, meaning ‘river horse’, the semantics of which suggests that the translator had the Icelandic water horse in mind. It is not certain and perhaps unlikely that he was able to analyze the Greek elements of the word ( ἵππος ‘horse’ and ποταμός ‘river’), but he

78 See Dag Strömbäck, ‘Näcken’, in Kulturhistoriskt lexikon för nordisk medeltid, 22 vols (Malmö: Allhems, 1955–78), 12, cols 432–38; Jochum Stattin, Näcken (Malmö: Carlssons, 1992). 79 For a collection of narratives, see Íslenzkar þjóðsögur og ævintýri, ed. Jón Árnason, 2nd ed., ed. Árni Böðvarsson and Bjarni Vilhjálmsson, 6 vols (Rekjavík: Bókaútgáfan þjóðsaga, 1954–61), 1, pp. 129–32; 3, pp. 207–12. 80 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 80. The fourth grammarian picks up on the second term, saying ‘ok kallar Óláfr þat finngálknat’ (and Óláfr calls this finngálknat) (Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 131). 81 Einar Sigmarsson, ‘Hamskipti eða endaskipti? Um nykur og nykrað, finngálkn og finngálknað’, Gripla, 16 (2005), pp. 287–98. 82 Dag Strömbäck, ‘Näcken’, col. 433. 83 ‘Bréf Alexandri Magni. Den norrøne oversættelse af Epistola Alexandri Magni ad Aristotelem udgivet sammen med forlægget’, ed. Povl Skårup, in Opuscula, 9. Bibliotheca Arnamagnæana, 39 (Hafniæ: Reitzel, 1991), pp. 62–63.

124

3 Grammatical Literature

was apparently aware that it referred to a kind of animal living in water. This information he may have gleaned from Isidore’s Etymologiae, a work which is well attested in medieval Iceland and Norway.84 Isidore writes: ‘Hippopotamus vocatus, quod sit equo similis dorso, iuba et hinnitu [. . .]. Die in aquis commoratur, nocte segetes depascitur, et hunc Nilus gignit.’ (It is called hippopotamus, since it is like a horse in its back, mane, and neighing [. . .]. It stays in the water during the day, at night it grazes in the fields, and the Nile gives birth to it).85 It seems probable that the translator envisioned some such horse-like creature when he turned the ipotami into nykrar.86 Both 3GT and Epistola Alexandri thus indicate that Snorri was referring to the horse variety of the monster and that the allusion is to the centaur in the very first line of the Ars. It would appear that Snorri, true to his programmatic avoidance of loan words, has here come up with a native equivalent to the centaur.87 His nephew Óláfr probably added the finngálkn to provide an actual centaur and thus highlight the Horatian parallel. This is in keeping with the overall difference between Snorri’s nativizing and Óláfr’s Latinizing approaches. Snorri’s discussion of extended metaphor and nykrat, as well as that of the metres of the early poets reviewed above, demonstrates just how far he was from being a passive recorder of tradition. Even as he insisted on how old and traditional his teachings were, he was not above disregarding important features of early and even recent poetry if they did not conform to his ideas of poetic decorum, informed by both the skaldic and the Latin grammatical traditions. For all his ancient lore, Snorri appears to have been in no mind to come across as backwards.

84 There is at least one fragment of the Etymologiae in Iceland (Merete Geert Andersen, Katalog over AM Accessoria 7. De latinske fragmenter. Bibliotheca Arnamagnæana, 46 (Hafniæ: Reitzel, 2008), p. 132). It is listed in two monastic inventories from 1396 and 1397 (Emil Olmer, Boksamlingar på Island 1179–1490 enligt diplom. (Göteborg: Zachrissons boktryckeri, 1902), p. 53). Large parts of it are translated in the Old Norse Pentateuch paraphrase Stjórn (Reidar Astås, An Old Norse Biblical Compilation. Studies in’ Stjórn’ (New York: Peter Lang, 1991), pp. 25–25, 81–82, 180–81). Finally, there are fragments of two copies from medieval Norway (Espen Karlsen, ‘Fragments of Patristic and Other Ecclesiastical Literature in Norway from c. 1100 until the Fifteenth Century’, in Latin Manuscripts in Medieval Norway. Studies in Memory of Lilli Gjerløw, ed. Espen Karlsen (Oslo: Novus press, 2013), pp. 229–30). 85 Etymologiae 12. 6. 21 (Isidori hispalensis episcopi etymologiarum sive originum libri XX, ed. W. M. Lindsay, 2 vols (Oxford: Clarendon, 1911), 2). 86 It should be noted that Old High German nihhus is used for ‘crocodile’ and Old English nicor for, precisely, ‘hippopotamus’, but also for water monsters generally, and the translator may thus be drawing on earlier conventions for designating water monsters (see Stömbäck, ‘Näcken’, col 432). Nonetheless, since the word means ‘river horse’ and is described as such by Isidore, in my view it remains likely that the translator had the specific horse variety in mind. 87 This practice was not unprecedented, since the substitution of native gods for classical ones is common in the translation of hagiography of the twelfth century and later (see, for instance, Lassen, Odin på kristent pergament, pp. 89–109).

3.2 Háttatal: Models, Sources and Rhetoric

125

3.2.4 Snorri and the Grammarians In various ways, then, Snorri established his authority in Háttatal by aligning himself with the grammatical tradition. Let us now consider how this plays out in the opening of the treatise. Háttatal is the part of Snorri’s Edda which is most reminiscent of a grammatical book of instruction, and the opening is not unlike that of Donatus’s Ars minor, which begins: Partes orationis quot sunt? Octo. Quales? Nomen, praenomen, verbum, adverbium [. . .]88 How many are the parts of speech? Eight. Which ones? Noun, pronoun, verb, adverb [. . .]

Háttatal opens: Hvat eru hættir skáldskapar? Þrent. Hverir? Setning, leyfi, fyrirboðning [. . .].89 Which are the modes of poetics? Three. Which ones? Precept, licence, prohibition [. . .].

For the present purposes, it is not necessary to decide whether Snorri modelled his opening on Ars minor specifically – although that is a likely candidate for a first acquaintance with Latin grammar – or if he had a somewhat vaguer feeling for how a work of this kind ought to begin. Suffice it to say that he opens in the style of a Latin treatise on grammar, and that this affiliation with the Latin tradition is likely to be intentional. Some lines further down we read: Hver er grein setningar háttanna? Tvenn. Hver? Málsgrein ok hljóðsgrein. Hvat er málsgrein?

88 Donat et la tradition de l’enseignement grammatical, ed. Holtz, p. 585. 89 Snorri, Háttatal, ed. Faulkes, p. 3. Cf. observation on ‘mál ok hættir’ on p. 115 below.

126

3 Grammatical Literature

Stafasetning greinir mál allt, en hljóð greinir þat at hafa samstǫfur langar eða skammar, harðar eða linar, ok þat er setning hljóðsgreina er vér kǫllum hendingar, svá sem hér er kveðit: Lætr sá er Hákun heitir (hann rekkir lið) bannat [. . .] Hér er stafasetning sú er hætti ræðr ok kveðandi gerir, þat eru tólf stafir í erindi ok eru þrír settir í hvern fjórðung.90 Which are the distinctions in the metres’ arrangement? Two. Which? Distinction in meaning and distinction in sound. What is distinction in meaning? The arrangement of letters distinguishes all meaning, but sound distinguishes whether we have long or short, hard or soft syllables, and what we call hendingar is the arrangement of distinctions of sound, as it has been composed here: Lætr sá er Hákun heitir (hann rekkir lið) bannat ([The king] who is called Hákun forbids [. . .], he emboldens the men) [. . .] Here it is the arrangement of letters (stafasetning) which governs the metre and makes [the language into] poetry. There are twelve letters/alliterating staves (stafir) in each stanza and three are placed in each couplet.

Snorri here continues the opening of the treatise in a grammatical vein, with letters, sounds and syllables (the order in Priscian and 3GT, for instance, is sounds, letters and syllables). This is the voice of scholarly authority, and when Snorri says that the arrangement of letters distinguishes all meaning, it would appear that he sets out with universal phonological pretensions. It soon becomes clear, however, that he is really only interested in hendingar and alliteration, and that he avails himself of the ambiguity inherent in the word stafr (letter / alliterating stave) to shift tacitly from the universal to the specific: there are, of course, more than twelve letters in a stanza, but there are only twelve alliterating staves. The phrase ‘arrangement of distinctions of sound’ has an authoritative, grammatical ring to it and could mean many things, but to Snorri it simply amounts to rhyme. In the course of the treatise, he makes up for his imprecision by discussing poetry on its own terms, but if he had not wanted to give the text a grammatical air, he could have avoided these discrepancies and ambiguities from the start.

90 Snorri, Háttatal, ed. Faulkes, p. 3.

3.2 Háttatal: Models, Sources and Rhetoric

127

With the description of the ‘hard and soft’ syllables, matters get somewhat complicated. This is a literal translation of asper and lenis, which refer to aspirated or unaspirated sounds in grammatical terminology.91 Since the following stanza begins with alliteration in /h/, it might appear that Snorri is referring to aspiration, but the commentary makes it clear that the stanza is intended to exemplify hendingar, not alliteration. A little further down he repeats the first couplet, and since this time he is focusing precisely on alliteration on /h/, he would have had ample opportunity to discuss aspiration. Nothing of the sort, however, can be found there: Ef hǫfuðstafr er samhljóðandi, þá skulu stuðlar vera enn inn sami stafr, svá sem hér: Lætr sá er Hákun heitir (hann rekkir lið) bannat If the main stave/letter is a consonant, then the supporting staves/letters should be the same letter, as here: Lætr sá er Hakun heitir (hann rekkir lið) bannat ([The king] who is called Hákun forbids [. . .], he emboldens the men)

Here, Snorri does not use words that conform to the grammatical concept of aspiration, but simply refers to h as a ‘letter’ (stafr). The difference may seem a fine one, but it is significant within the field of medieval grammar. The letter/phoneme h was always a problem to grammarians, since it was perceived as breath, rather than as a ‘real’ sound. Priscian has the fullest description of h, and in his words ‘aspiration has no power/sound value’ (‘potestas, qua caret aspiratio’).92 Priscian notes that h does not fit within any category of letters, which he establishes trough phonology, and that it does not affect comprehension, except in Greek aspirated consonants, which had by his time become fricatives.93 Although he does not say so, one might add that aspiration was metrically irrelevant in the classical tradition. Accordingly, h was not a letter, but only a ‘sign of aspiration’ (nota aspirationis), and Priscian construes aspiration as an accident of either the vowel or of the syllable.94 This is also the mainstream position in grammatical literature.95

91 Cf, for instance, Grammatici latini, 2, ed. Heinrich Keil (Leipzig: Teubner, 1855), p. 51. 92 Grammatici latini, 2, ed. Heinrich Keil (Leipzig: Teubner, 1855), p. 13. 93 Grammatici latini, 2, ed. Heinrich Keil (Leipzig: Teubner, 1855), pp. 12–13, 18–19. 94 Grammatici latini, 2, ed. Heinrich Keil (Leipzig: Teubner, 1855), pp. 19, 35, 51. 95 Thus, for instance, Isidore (Isidore, Etymologiae 1.4.11 (Isidori hispalensis episcopi etymologiarum sive originum libri XX, ed. Lindsay, 1, p. 30; for further discussion on h by Priscian, see Priscian, Institutiones, 1. 16, 24–25, 47 (Grammatici latini, 2, ed. Heinrich Keil (Leipzig: Teubner, 1855), pp. 12–13,

128

3 Grammatical Literature

In alliterative poetry, however, this non-existent letter was of crucial importance, and Óláfr devotes a short chapter in 3GT to reaching an ingenious compromise between perceptions of h within the native and the Latin tradition. He remarks that even though h is not a letter, all alliterating staves in poetry should be aspirated, or none of them.96 In other words, like Priscian he construes h as an accident of the syllable, but he then establishes a rule of alliteration for this accident that conforms to the rule for consonants, which are indeed letters in their own right, rather than mere accidents. Snorri’s sample couplet, by contrast, shows that he was not troubled by such excessive theorizing: to him, h was simply a letter that alliterated like other letters.97 If Snorri perceived of h as a letter, this would seem to imply that the words ‘samstǫfur harðar eða linar’ (hard and soft syllables) cannot refer to aspiration, since he is with these words referring to some kind of accident of the syllable, not to a letter/sound, which is how he later perceives of h. Still, no other accidents of the syllable than aspiration even remotely resemble what may reasonably be designated by these words, and this suggests that he is simply inconsistent.98 A comparison to Óláfr confirms that this is the most likely explanation. In referring to aspiration, Óláfr uses the words snarpr and linr, the former being a synonym to harðr. The wording thus indicates that the two authors are discussing the same phenomenon, in particular since they both refer to an accident of the syllable – although only Óláfr explicitly uses those words (tilfelli samstǫfu).99 The descriptions are so similar that one must probably conclude that Snorri is

18–19, 35–36, 51)). One may suspect that this analysis is predicated on the fact that aspiration in Greek, to the extent that it was represented in writing at all, was only marked by an inverted apostrophe, rather than by some letter-sized sign. This confusion carried over into the Latin tradition, even though the Latins did, in fact, have a letter to represent the sound. The perception of h as a consonant was not unknown to the grammatical tradition, and is found in Donatus, Ars Maior, 1. 2: ‘H interdum consonans, interdum adspirationis creditur nota’ (h is sometimes considered a consonant and sometimes a mark of aspiration) (Donat et la tradition de l’enseignement grammatical, ed. Holtz (Paris: CNRS éditions, 1981), p. 605). There are no indications, however, that Ars maior, 1–2 were known in Iceland, whereas Priscian’s opinions on h are well represented by Snorri’s nephew Óláfr. 96 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 54. Óláfr observes that aspiration belongs to the tilfelli (accidentia) of the syllable, which can be either soft or hard (linr or snarpr, that is, lenis or asper). 97 This perception was reasonable on more than metrical grounds, since the sound represented by the letter h was surely much more prominent in Germanic languages than in Latin, Romance and Greek, and it may indeed have been pronounced as a fricative /χ/ in Snorri’s day, rather than as /h/. 98 On the authority of Priscian and Óláfr, the accidents are four: accent (grave, acute, circumflex), aspiration (spiritus asper, lenis), time (length of the syllable) and number of letters (Grammatici latini, 2, ed. Keil, p. 51; Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 50). Óláfr discusses all four, and he translates the three varieties of accent into hvǫss, þung and umbeygilig hljóðsgrein (sharp, heavy, and flexible accent) (Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, pp. 39, 55). 99 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 50–56.

3.3 Litla Skálda, Gylfaginning and Skáldskaparmál

129

indeed referring to aspiration. This leaves us with two options as to what is going on here: 1. Snorri refers to aspiration, but he does not really know what it is, and thus fails to take it into consideration in his discussion of alliteration, or to connect it to the letter h. 2. Snorri knows what aspiration is, but he is unable to accommodate it within his description of poetics, and so he simply disregards the earlier description once it has fulfilled its rhetorical purpose. Both explanations are viable, and they both indicate that the link from the opening words of the treatise to what follows is very weak. In opening with a definition of letters, sounds and syllables, Snorri appropriates the authority of Latin grammatical discourse. He has, however, barely begun defining these concepts before he shifts the meaning of ‘letter’ to ‘alliterating stave’ and leaves all aspects of ‘sound’ and ‘syllable’ which are not directly relevant to rhyme behind. Furthermore, the grammatical definition of aspiration as an accident of the syllable is quickly forgotten in favour of the commonsensical description of h as an alliterating stave/letter. In sum, Snorri’s attempt at a phonological theory is merely rhetorical, the result of a wish to bestow on Háttatal the authority of a Latin grammatical handbook. He does not adapt the following analyses to a Latin framework, and this makes the beginning of the treatise partly contradict its body. More than anything else, the opening of Háttatal is a bid for a hearing. The discrepancies between the beginning and the bulk of the treatise also underline the extent to which local poetics retained its distinctive character.100 As we shall see in the discussion of 3GT below, the thorough application of Latin categories to this native art would demand a much more strenuous and learned effort, only to produce a treatise of more limited practical usefulness, if not for the study of Latin. As stated on p. 93–94, developments within Old Icelandic metrics and diction from the twelfth century onwards was guided by methodological innovation and formal conservatism. This is true also of Háttatal and Háttalykill in particular. New metres were extracted from old ones, but poetry’s basic form did not change. True, the systematic way of concocting metres now conformed to the methods of the schools, and to that extent, the mental landscape had changed, but the building blocks remained the same.

3.3 Litla Skálda, Gylfaginning and Skáldskaparmál101 The only major model for Snorri’s Edda to have been identified by scholars is Háttalykill. With regard to the treatment of diction in Skáldskaparmál, a statement

100 The relative independence of local poetics is stressed also by Stephen N. Tranter, ‘Clavis Metrica’. ‘Háttatal’, ‘Háttalykill’ and the Irish Metrical Tracts. Beiträge zur nordischen Philologie, 25 (Frankfurt: Helbing & Lichtenhahn, 1997), pp. 92–99; Kristján Árnason, ‘Um Háttatal’; idem, ‘Vernacular and Classical Strands in Icelandic Poetics and Grammar in the Middle Ages’, NorthWestern European Language Evolution, 69: 2 (2016), pp. 191–235. 101 I am thankful for fruitful discussions with Inger Helene Solvin during my work on this section.

130

3 Grammatical Literature

by Margaret Clunies Ross is representative of current scholarly opinion: ‘There is no evidence, however, that the largely systematic, theorized exposition of the diction of skaldic poetry that is found in part in Háttatal and more expansively in Skáldskaparmál had ever been attempted before in Iceland or indeed in any other part of Scandinavia or Western Europe.’102 Below I hope to show that, if one allows for earlier attempts to have been less systematic and theorized, such evidence can indeed be produced. Litla Skálda is a short treatise on kennings that has received limited attention. It consists of only a few printed pages, divisible into three sections on the basis of structure, function and content. The first and most substantial part lists kenning types in a somewhat haphazard fashion, with no poetic examples. The second part deals with celestial bodies and the world, providing some mythological description and narrative and quoting two and a half eddic stanzas. The third lists words and names for things connected to the mythological wolf Fenrir, quoting one eddic stanza. As a treatise on kennings, Litla Skálda is more closely related to Skáldskaparmál than to Gylfaginning. The second section, however, bears some resemblance to Gylfaginning, since it combines poetry from one of the main sources of Gylfaginning – Grímnismál – with prose description and narrative. Both Gylfaginning and Skáldskaparmál are, however, vastly more elaborate than Litla Skálda in scope and structure. Skáldskaparmál quotes some 400 poetic examples to illustrate different kenning types and to provide precedence in tradition. The corresponding section of Litla Skálda has no poetry at all. In the following, I address three questions regarding the connection between Litla Skálda and Snorri’s Edda. First, is there any textual connection? Second, if there is such a connection, which work is the giver and which the taker? Third, if the line of influence can be determined, what might this tell us about the evolution of Icelandic literature on poetics in general, and about Snorri’s working methods within this tradition in particular? Before turning to these questions, I begin with a description of the treatise in its manuscript context.

3.3.1 A and B Litla Skálda is found in two medieval manuscripts, A (c. 1300–1325) and B (c. 1400). The two manuscripts are closely related, containing, in sequence, 3GT, Litla Skálda and a shortened redaction of Skáldskaparmál.103 Guðrún Nordal has 102 ‘Snorri Sturluson and the Construction of Norse Mythography’, in Writing Down the Myths, ed. Joseph Falaky Nagy. Cursor Mundi, 17 (Turnhout: Brepols, 2013), p. 202. 103 In B, much material in 3GT, such as the discussion of runes and the sections following the descriptions of barbarisms, has been cut out.

3.3 Litla Skálda, Gylfaginning and Skáldskaparmál

131

argued that Litla Skálda and Skáldskaparmál together form a treatise on kennings in these two manuscripts.104 The rubrication in the manuscripts supports Guðrún Nordal’s view, because it indicates the beginning of a new treatise with Litla Skálda, but there is no rubric between Litla Skálda and Skáldskaparmál. Before Litla Skálda A has: ‘Hér er lykt þeim hlut bókar er Óláfr Þórðarson hefir samansett ok upp hefr skáldskaparmál ok kenningar eptir því sem fyrri fundit var í kvæðum hǫfuðskálda ok Snorri hefir síðan samanfœra látit’105 (Here ends the part of the book that Óláfr Þórðarson has compiled and [the section on] poetic diction and kennings begins, according to what has been found in the poetry of the main poets, and the gathering of which was later commissioned by Snorri). B simply has: ‘Hér byrjaz kenningar skáldskapar’106 (Here begin the kennings of poetry). Guðrún Nordal writes that ‘it was necessary to edit Skáldskaparmál in the light of Litla Skálda to avoid repetition’.107 While some repetition may be avoided – such as gold kennings based on the myth of king Fróði – much is left standing, and Guðrún Nordal’s claim is therefore doubtful. Both A and B contain Skáldskaparmál from chapter 45 onwards, with some material missing. The redaction is closely related to U. Additional chapters seem to have been present at an earlier stage, since A refers to kennings for kings ‘fyrr í bókinni’ (earlier in the book).108 The reference is to chapter 53, one of the chapters missing in A. This and another similar reference probably derive from A’s exemplar, since it appears to make little sense to have other material on kennings before 5GT and Litla Skálda. The archetype does not, however, appear to have contained chapters 1–44. In chapter 64, A has removed the narrative about Hálfdan, which is present in B. As a result, the words ‘enn áttu þau Hálfdan’ (Hálfdan and his wife had a further [. . .]) lack an antecedent.109 While A lacks all of chapters 1–44, B has reintroduced a number of missing chapters from the W-branch, but not the narrative in them. The archetype’s omission of chapters 1–44 is curious, since it does not correspond to the division between kennings and heiti. The latter begin in chapter 54, and this is the division mentioned in the rubric in U. The omission could conceivably be due to a faulty exemplar, but chapter 44 marks another dividing line in Skáldskaparmál: it is the last chapter containing substantial amounts of narrative prose, and narrative prose is prominent up to this point in Skaldskaparmál. This internal dividing line

104 Guðrún Nordal, Tools of Literacy, pp. 224–25, 229. Because of the unified nature of the new treatise, Guðrún Nordal is critical to the common practice of referring to these versions of Skáldskaparmál as fragments. For a description of these manuscripts and the interplay between texts in them, see Guðrún Nordal, Tools of Literacy, pp. 57–66, 213–31. 105 Edda Snorra Sturlusonar, ed. Jón Sigurðsson et al., 2, pp. 427–28. 106 Edda Snorra Sturlusonar, ed. Jón Sigurðsson et al., 2, p. 511. 107 Guðrún Nordal, Tools of Literacy, p. 226. 108 See Finnur Jónsson, ‘Edda Snorra Sturlusonar’, p. 294. 109 Finnur Jónsson, ‘Edda Snorra Sturlusonar’, p. 296.

132

3 Grammatical Literature

suggests that the exclusion of chapters 1–44 is intentional. A has then carried this trend further by removing also the narrative about Halfdan, whereas B has added some of the kenning material that went missing in the process. B seems intended to be comprehensive: not only are kennings designating, for instance, poetry and the gods reintroduced, but a quotation from an earlier part of Snorri’s Edda on what kennings are has been added (see below). Their differences notwithstanding, both manuscripts contain a reworking of Skáldskaparmál that by removing its narrative transforms it into a catalogue of practical poetic examples, demonstrating the use poets made of the kennings theoretically described in Litla Skálda. The lists of heiti in prose and metrical form in the end of Skáldskaparmál have been retained and serve a similar function. The new treatise, combining Litla Skálda and Skáldskaparmál, thus shares a common structure with other medieval Icelandic grammatical manuscripts, as well as with Snorri’s Edda internally: theory comes first, followed by practical examples, including poems (see above pp. 109–10). Some features in the texts of the two manuscripts are suggestive of the perceptions of this new text and the writing process behind it (including its relationship to Snorri and his Edda). First, in A, the entire treatise on kennings is attributed to Snorri. This can hardly be correct. The summary of the contents of Snorri’s Edda in U, as well as the contents of U, R, W and T, suggest that Snorri’s text has been reworked in A, and that Litla Skálda has been added. The new text, however, was still associated with Snorri’s authoritative name. In B something different has happened. In the middle of the text of Skáldskaparmál, under the heading Capitulum, we read: Svá segir í bók þeiri sem Edda heitir at sá maðr sem Ægir hét spurði Braga skáld meðal annarra hluta [. . .] En eigi skulu kristnir men trúa á heiðin goð ok eigi á sannindi þvílíkra frásagna ǫðruvís en svá sem skrifat finnz í fyrsta kapitula greindrar bókar.110 In the book called Edda, it is related how the man called Ægir asked the poet Bragi among other things [what poetry is, whereafter follow discussions of the names of the gods and of poetic tradition]. But Christians should not believe in the pagan gods or in the veracity of such tales in any other way than is written in the first chapter of the aforementioned book.

This passage is recognizable as the beginning of Skáldskaparmál. What is most noteworthy, however, is that the scribe/redactor here refers to the Edda as another book. It would seem that he thought of the Edda as the narrative portions of what U calls the Edda, but he considered what he himself was writing down as a treatise on diction. U itself may offer a clue to why this should be the case, since the passage above is there part of Gylfaginning. Presumably, this was also the case in B’s exemplar, or at least in a manuscript known to the scribe, so that he is really referring to

110 Edda Snorra Sturlusonar, ed. Jón Sigurðsson et al., 2, pp. 532–33.

3.3 Litla Skálda, Gylfaginning and Skáldskaparmál

133

the book from which he extracted the text. This is likely not least since U and B appear to belong to the same branch of the stemma.111 The B-scribe’s opinion of what the Edda contained was probably not a common one in his day (though it corresponds to an early modern distinction between Edda and Skálda), since three poets a few decades earlier refer to Snorri’s Edda as a treatise which contained instructions on how to construct kennings.112 Deviant or not, the distinction that the B-scribe/redactor (or his predecessor) made between the treatise on the gods (prologue and Gylfaginning) and the treatise on kennings may have prompted the interpolation. By inserting the one passage from the treatise on the gods that was relevant to diction into the treatise on kennings, while cutting out the passages of mythical narrative in the latter, the functional distinction between two treatises became complete.

3.3.2 Character of Litla Skálda and Skáldskaparmál Finnur Jónsson considered Litla Skálda to be a compendium which is at least partly based on Skáldskaparmál.113 He does not state his reasons for drawing this conclusion, and it poses some problems, especially since several categories of referents in Litla Skálda are absent from Skáldskaparmál (thus, for instance, kennings for ‘ice’, ‘dog’, ‘horn’ and ‘stone’). Since Finnur Jónsson had a tendency to see ‘old’ and ‘good’ as related concepts, I find it likely that he saw the poor quality of Litla Skálda as a sign of monastic degeneration and that this accounts for his view on the relation between the two treatises. More recently, Sverrir Tómasson has disputed that Litla Skálda is derived from Skáldskaparmál, but nonetheless considers it certain that Litla Skálda cannot be much older than A. This implies a late thirteenth-century date.114 Like Finnur Jónsson, he does not state his reasons. Karl Müllenhoff, to whom Finnur Jónsson and Sverrir Tómasson both refer, considers Litla Skálda to have been composed independently of Snorri’s Edda in the 1230s or 1240s, but again the basis for this remarkably exact date remains elusive.115 The question, therefore, needs to be reconsidered. The additional categories of referents in Litla Skálda (‘ice’, ‘dog’, ‘horn’, etc.) show that it cannot be based on Skáldskaparmál alone, and I see no

111 Haukur Þorgeirsson, ‘A Stemmatic Analysis of the Prose Edda’, Saga-Book, 41 (2017), 49–70. 112 Lilja, Guðmundarkvæði, and Guðmundardrápa. The relevant passages are discussed on pp. 290–96. 113 Edda Snorra Sturlusonar, ed. Finnur Jónsson, p. lix. 114 Sverrir Tómasson, ‘Nýsköpun eða endurtekning? Íslensk skáldmennt og Snorra Edda fram til 1609ʹ, in Guðamjöður og arnarleir. Safn ritgerða um eddulist, ed. Sverrir Tómasson (Reykjavík: Háskólaútgáfan, 1996), pp. 20–22. 115 Karl Müllenhoff, Deutsche Altertumskunde, 5 vols (Berlin: Weidmann, 1887–1900), 5, pp. 214–22.

134

3 Grammatical Literature

compelling reasons to think that it is based on Skáldskaparmál at all or for dating it to the period c. 1230–1300. Anthony Faulkes has suggested and Judith Jesch has explicitly argued that Litla Skálda may be older than Skáldskaparmál.116 Jesch bases her argument on the structures of kennings in Litla Skálda, which resemble kennings in Háttalykill and in other poetry by earl Rǫgnvaldr of Orkney more than the rules given in Skáldskaparmál. Although she focuses on the connection of Litla Skálda to Orkney and is cautious with regard to chronological implications, her observations may contribute to a relative dating of Litla Skálda, if they are corroborated by further evidence.117 I believe that Jesch’s argument carries particular weight in cases where Litla Skálda and Háttalykill are in agreement against Skáldskaparmál in the use of words and kenning types, and where their agreement, furthermore, amounts to breaking the rules given in Skáldskaparmál. Thus, Háttalykill contains the kenning alnar dript (the fore-arm’s snowdrift [gold]) which follows the rule in Litla Skálda: ‘enda má gull kenna til snæs ok íss ok kenna þá til handar’ (gold can also be referred to as snow and ice and then be defined by the hand/arm). Skáldskaparmál, by contrast, distinguishes between kennings using fire, which designate gold because it is red, and kennings using snow or ice, which designate silver because it is white.118 According to this description, snow cannot signify gold, as it does in Háttatal and Litla Skálda. Snorri seems to have been inspired to make this distinction between gold and silver by the third and fourth stanzas, both by Einarr Skúlason, that he quotes after presenting the rule.119 Thus, for instance, Einarr mentions an axe adorned with bæði sjóðs snær ok geima eldr (both the purse’s snow [silver] and the sea’s fire [gold]).120 In light of Einarr’s tendency to elaborate on earlier poetic conventions, it is not unlikely that he was the first to make the distinction.

116 Faulkes, ‘The Sources of Skáldskaparmál’, pp. 61–62; Snorri, Skáldskaparmál, ed. Faulkes, p. xiv; Judith Jesch, ‘The Orcadian Links of Snorra Edda’, in Snorres ‘Edda’ i europeisk og islandsk kultur, ed. Jon Gunnar Jørgensen (Reykholt: Snorrastofa, 2009), pp. 153–59. 117 Her caution is partly attributable to the possibility that the character of the kennings in Litla Skálda rests more on the choice of poetry for kenning extraction than on the time of composition of the treatise. 118 Jesch, ‘The Orcadian Links of Snorra Edda’, pp. 155–56. 119 According to the translation in SkP 3, p. 485, a half stanza by Þorleikr fagri contains one kenning for silver and one for gold. There is fire in the stanza, but no snow or ice, and no opposition between the two kennings is evident. It is possible that both kennings simply mean ‘precious metal’, and the stanza cannot have inspired Snorri to make the distinction between fire [gold] and snow [silver]. 120 SkP 3, pp. 147–48; Snorri, Skáldskaparmál, ed. Faulkes, p. 61. Anthony Faulkes, Poetical Inspiration in Old Norse and Old English Poetry (London: Viking Society for Northern Research, 1997), p. 11, expresses some doubt as to whether Einarr’s stanza does or does not distinguish

3.3 Litla Skálda, Gylfaginning and Skáldskaparmál

135

Another instance where Litla Skálda does not abide by the rules of Skáldskaparmál is in the use of the neuter word gagl (gosling) in kennings for ravens. This is also found in Háttalykill 20a: hjaldrgǫgl (‘battle-goslings [ravens]). According to Skaldskaparmál, this is not a viable kenning for ‘raven’, since the birds used in such kennings must be masculine. In the description of similar kennings in Litla Skálda, no such restriction is given.121 I add one example to those of Jesch, although it does not involve Háttalykill: Litla Skálda says that when sea-kings serve as definers in kennings for ships, only horse names can be used (such as Geitir’s (a sea-king’s) horse [ship]).122 Skáldskaparmál, by contrast, says that a ship can be called ‘hest eða dýr eða skíð sækonunga eða sævar eða skipreiða eða veðrs’ (horse or animal or plank of sea-kings or of the sea or of shipgear or of weather).123 The second stanza quoted after this in Skáldskaparmál has Sveiða hreinar (Sveiði’s reindeer [ships]), and this kenning also serves as rubric to the stanza.124 The two descriptions cannot be reconciled, and it is difficult to see how Litla Skálda’s rule could be derived from a passage which lays particular emphasis on an example that directly contradicts it. The fact that the first two kenning types are found in Háttatal with the same meaning complicates matters. Stanza 43 has Grótta glaðdript (Grótti’s joyful snowdrift [gold]) and 62 has undgagl (wound-gosling [raven]).125 The influence of Háttalykill on Háttatal may account for this, but it also means that Snorri broke his own rules almost as much as Háttalykill and Litla Skálda did. There is broad consensus that Háttatal was the first part of Snorri’s Edda to be composed, and one explanation of Snorri’s disobedience to his own rules may thus be that he had not yet set out to theorize the kenning system. As I argue below (pp. 145–47), this consensus is partly based on an insufficient understanding of Litla Skálda’s role in the literary tradition,

between gold and silver here, but I believe that it would make little sense to say ‘both X and Y’ if there was no distinction between them. The next stanza mentions áls hrynbrautar eldr (eel’s roaring way’s [sea’s] fire [gold]) and skála snær (the scales’ snow [silver]). After this stanza, A and B read ‘hér er gull kallat eldr áls hrynbrautar, en silfr snær skálanna’ (here, gold is called the eel’s roaring way’s [sea’s] fire [gold], and silver the scales’ snow [silver]’, but in RTWCU, the words ‘eldr áls hrynbrautar, en silfr’ have fallen out (Edda Snorra Sturlusonar, ed. Finnur Jónsson, p. 144). The resulting reading, ‘hér er gull kallat snær skálanna’ directly contradicts the stanza as well as Snorri’s rule, and such a faulty reading is not likely to have originated with Snorri, just after he has posited the distinction based on, precisely, Einarr’s stanzas. Rather, A and B have the older reading here, although this is difficult to explain on stemmatic grounds. 121 Jesch, ‘The Orcadian Links of Snorra Edda’, p. 155. 122 Edda Snorra Sturlusonar, ed. Finnur Jónsson, p. 255. 123 Snorri, Skáldskaparmál, ed. Faulkes, p. 74. 124 The manuscripts agree on this reading, except that U has replaced both prose and poetry with the non-grammatical hér er kallaðr sunddýr (here it (masc. for neutr.) is called sea-animal), which appears to be derived from the stanza which is no longer there (Edda Snorra Sturlusonar, ed. Finnur Jónsson, p. 156; U 33v l. 28). 125 Jesch, ‘The Orcadian Links of Snorra Edda’, pp. 155–56.

136

3 Grammatical Literature

but it still remains likely that the poem in Háttatal was the first part of Snorri’s Edda to be fixed in its present state. Another, perhaps less likely, possibility is that in Háttatal Snorri focused on metre, whereas in Skáldskaparmál he focused on diction, and that he held himself up to different standards in each case. Whatever the explanation, the correspondences between Litla Skálda, Háttalykill and Háttatal suggest the possibility that Litla Skálda could be Snorri’s first attempt at composing a treatise on kennings, but one which he later supplanted by Skáldskaparmál, and that he provided additional rules as he went. As we shall see, however, the tenors of the two works, and of Litla Skálda and Snorri’s Edda generally, are fundamentally different, rendering that explanation unlikely. If Skáldskaparmál has made use of Litla Skálda, these discrepancies make sense, whereas influence running in the opposite direction makes them difficult to explain.126 The two first discrepancies between Litla Skálda and Skáldskaparmál are due to the presence of further distinctions and categorizations in Skáldskaparmál (colour, male bird rather than any bird). With regard to Litla Skálda’s claim that only horses can be used with sea-kings, this is not an unreasonable deduction by someone who is well acquainted with skaldic poetry. Indeed, if it were phrased as a strong tendency rather than as a rule, it would have been correct.127 The massive collection of poetic examples that the compilation of Skáldskaparmál entailed, however, would have shown that there were examples of other animals in connection to sea-kings. Since the prose mentions this and a stanza provides an example, Skáldskaparmál would hardly allow for Litla Skálda’s rule to be extracted from it. Instead, the rule given in Litla Skálda appears to be based on independent analysis of skaldic poetry, albeit of a more limited nature than that entailed in the composition of Skáldskaparmál. For this and other reasons given below, if any traces of influence between the two works can be demonstrated, Litla Skálda is likely to be the source. Although Litla Skálda and Skáldskaparmál are both treatises on kennings, they differ in several respects. Litla Skálda is a short, straightforward prose treatise with only three and a half stanzas quoted towards the end. Skáldskaparmál is a long treatise, quoting some four hundred half stanzas or stanzas, apart from the þulur (metrical lists if heiti). Litla Skálda is a pragmatic description of kennings for the most common referents in skaldic poetry, but their rationale is often not explained. Skáldskaparmál is an extensive and explanatory description with many examples of a more restricted number of referents. Litla Skálda does not comment on tradition, but seems rather to aim at providing a basic understanding of the kenning system

126 I here exclude the possibility of influence of Háttatal on Litla Skálda, for which I have found no indications. 127 See Meissner, Die Kenningar, pp. 209–13, 218–20. It may be noted that Háttatal abides by the rule in Litla Skálda, but since that is the general tendency in skaldic poetry, that observation is of limited value.

3.3 Litla Skálda, Gylfaginning and Skáldskaparmál

137

for further composition. In Skáldskaparmál, by contrast, the importance of tradition is strongly emphasized (see below pp. 144–45). Litla Skálda provides a pragmatic description of the kenning system and excludes kennings for gods, presumably because Christian poets do not need them for composition. The absence of any narrative explanation of mythological kennings for poetry in Litla Skálda might be interpreted as a lack of interest in mythology, but this is uncertain, since Litla Skálda gives no explanations at all in the first section where these kennings occur. Skáldskaparmál lacks a number of referents that are present in Litla Skálda and is to that extent less comprehensive. Kennings for gods, however, come first and are treated at length. The section on kennings for gods is a clear example of the importance of other than pragmatic considerations in Skáldskaparmál. The focus on the gods ties in nicely with the premium placed on tradition, and it stands in stark contrast to the pragmatic approach of Litla Skálda. Litla Skálda can appear somewhat incoherent. The ordering of kennings is largely based on semantic categories, but another principle is also at work, namely that the author picks up on a word contained in a preceding kenning, such as: [. . .] lagvápn má kalla fiska heitum ok orma, ok kenna við herklæði ok hlífar, sár eða blóð. Blóð er kallat sjóar heitum ok vatna, ok kent við hræ eða ben, sár eða undir. Sjó má kalla [. . .]128 [. . .] piercing weapons may be called by the names of fishes and snakes, and be defined by armour and shields, wounds or blood. Blood is called by the names of the sea and bodies of water, and defined by carrion or injuries, wounds or gashes. The sea may be called [. . .]

Because of this associative procedure, kennings for ‘sea’ round off the section on battle-related kennings, and the sea in turn leads to Miðgarðsormr, the great serpent which abides there. Similarly, kennings for ‘arm’ end up with kennings for gold, rather than with other body parts, because gold is worn on the arm. Litla Skálda seems to be composed ‘as you go’ to some extent, and other indications support this idea. The treatise starts by enumerating different kenning types, but after a few entries, the author seems to realize that he has not stated the basic principle for kenning formation: ‘Ekki skal kenna þat er sitt nafn hefir sjálfs, en kenna alt þat, er annars nafni er nefnt en sínu’129 (One should not define that which has its own name, but define everything that is called by another name than its own). After this, the list of kenning types resumes. Towards the end, two stanzas from Grímnismál (40–41) are quoted to explain why the earth, sea and clouds can be referred to as parts of Ymir’s body. The author has now shifted to poetic quotation as a mode of explanation. Then, we get a

128 Edda Snorra Sturlusonar, ed. Finnur Jónsson, p. 256. I thank Inger Helene Solvin for pointing out this tendency. 129 Edda Snorra Sturlusonar, ed. Finnur Jónsson, p. 255.

138

3 Grammatical Literature

short narrative about Fróði and his slave women and an etiology about how the sea became salt. This section seems to constitute an explaination of the gold kennings relating to Fróði given earlier in the treatise, before the author started to provide explanations of obscure references (which he has now begun to do, through the quotation from Grímnismál).130 The treatise has not been reworked to give this explanation where the reader needs it. These features all suggest that the author of Litla Skálda had little in the way of established norms for how to organize a treatise on poetic diction to draw on. After the narrative about Fróði follows a section on the wolf Fenrir, again quoting a stanza. This part of the treatise is a kind of vocabulary of the places and implements connected to the binding of Fenrir. In A, this section has its own heading – Frá Fenrisúlfi (about the wolf Fenrir) – and Finnur Jónsson did not consider it to be part of the treatise.131 This conclusion strikes me as somewhat arbitrary. At this point, the text has just shifted from pure poetics to a narrative about the magic millstone Grótti/Grótta, and how king Mýsingr took it and ground salt with it until his ships sank in the Pentland Firth, and the sea became salt. The author has thus established a connection between mythology and place, and the section on Fenrir opens in the same vein: ‘Ámsvartnir heitir vatn, en Lyngvi hólmi í vatninu [. . .]’ (A lake is called Ámsvartnir, and an island in the lake [is called] Lyngvi [. . .]). Because of this thematic connection, I see no reason why the section on Fenrir should not originally have been part of Litla Skálda and have received a heading in A at some later point, in particular since the work is otherwise of a relatively incoherent nature. In any event, even if the section on Fenrir might possibly constitute a later addition, this must have been done quite early on, since Snorri has made use of this section in much the same way as he did with other parts of the treatise (see below). A gradually increased focus on the connection between mythology and place thus unites the second part, containing cosmology and narrative, and the third part, on Fenrir. There is also another link between the two. In the second part, kennings proper are largely abandoned in favour of names and synonyms (heiti). The third part treats heiti exclusively and thus forms a natural continuation of the second part. The shift to heiti in the second part is interesting in yet another respect. In Skáldskaparmál, heiti are treated from chapter 54 onwards, and the treatise is thus divided into two parts: the first treating kennings, the second heiti. Litla Skálda and Skáldskaparmál thus share the same overall structure, but there is one important difference: in the second part of Litla Skálda, the author jumps back and forth between kennings and heiti (and narrative). Only in the third part is there

130 Edda Snorra Sturlusonar, ed. Finnur Jónsson, p. 256. 131 Edda Snorra Sturlusonar, ed. Finnur Jónsson, pp. lix, 259; Guðrún Nordal, Tools of Literacy, pp. 228–29.

3.3 Litla Skálda, Gylfaginning and Skáldskaparmál

139

finally a consistent focus on heiti. The shift to heiti in the second part seems to be triggered by some kennings where many synonyms are given: Hjarta er kallat steinn eða korn, hnot eða epli, en kent til hugar. Hjarta heitir ok akarn ok eiskǫld. Gollorr heitir þat, er næst er hjarta manns. Hnetr heita fylvingar. Heimr er kallaðr hús eða ker veðra [. . .]132 The heart is called ‘stone’ or ‘grain’, ‘nut’ or ‘apple’, and is defined by the mind/courage. The heart is also called acorn or eiskǫld.133 That which is closest to a person’s heart [pericardium] is called gollorr.134 Nuts are called fylvingar.135 The world is called the house or vessel of bad weather [. . .]

Here, the description moves from kennings for ‘heart’ to heiti and then back to kennings, now for ‘world’, and this ambivalence carries on until the narrative about Grótti. The shift to heiti, then, is not primarily predicated on an overall plan for the structure of the work, but on which kennings are treated. In Skáldskaparmál, by contrast, the shift is very clearly part of the plan of the composition.136 It is easy to

132 Edda Snorra Sturlusonar, ed. Finnur Jónsson, p. 258. 133 The word occurs only in the A-version of Litla Skálda, but it is found in both A and B in their version of heiti for bodyparts in Skáldskaparmál – a version which is partly based on Litla Skálda (Edda Snorra Sturlusonar, ed. Jón Sigurðsson et al., 2, pp. 467, 550). Elsewhere, it is found in Fáfnismál and in a stanza by Illugi Bryndœlaskáld. The latter stanza is only preserved at the end of Skáldskaparmál in A (LP, s.v. eiskald; SkP 2, pp. 283–84; Edda Snorra Sturlusonar, ed. Jón Sigurðsson et al., 2, p. 493). It might therefore appear that the word as well as the stanza have been added to Litla Skálda and Skáldskaparmál in A, but not in B. This is unlikely to be the case, however. First of all, we do not know whether the stanza was originally in B, since the manuscript breaks off before this point (see Edda Snorra Sturlusonar, ed. Jón Sigurðsson et al., 2, pp. 488, 572). More importantly, both A and B add eiskǫld, gollorr to the heiti for ‘heart’ in Skáldskaparmál, which corresponds exactly to the order of the words in A, whereas no other manuscripts have this addition. It seems clear, then, that the addition of the word and the stanza has been triggered by Litla Skálda, which is present in both manuscripts, and that eiskǫld has simply fallen out in the B text of Litla Skálda. (The stanza would then be intended to serve as the poetic source for the heiti, like other stanzas in Skáldskaparmál.) Such minor differences between the A and B version are found throughout Litla Skálda (see Edda Snorra Sturlusonar, ed. Finnur Jónsson, pp. 255–59). 134 This word is also found in the A and B versions of Skáldskaparmál, dependent on Litla Skálda (see previous note). It is otherwise only found in þulur and in Merlínuspá, where it is twice used in kennings: gollorheimr (pericardium-world [breast]) and gollorhǫll (pericardium-hall [breast]) (LP, s.v.). 135 This word is only used by Gísli Súrsson (Skj B 1, p. 97; A 1, p. 102). In that stanza, both hnetr and fylvingar are used for ‘tears’, based on an extended metaphor where the eyes are referred to as hazelnuts. As Inger Helene Solvin has shown, the use of both words here probably indicates that the stanza by Gísli belonged to Litla Skálda’s sources, in particular since the words occur among kennings for bodyparts. Only in the stanza by Gísli do they belong there (Inger Helene Solvin, ‘Litla Skálda’ – Islands første poetiske avhandling? Et forsøk på å etablere en relative kronologi mellom ‘Skáldskaparmál’ og ‘Litla Skálda’ (MA thesis, University of Oslo, 2015), p. 75). 136 Faulkes notes that a few kennings are found also in the section on heiti, but I do not think that his conclusion that Snorri first did not intend to separate the two categories is warranted (Snorri, Skáldskaparmál, ed. Faulkes, pp. xi, 87, 90, 209). As he himself notes, there are many inconsistencies

140

3 Grammatical Literature

see how Litla Skálda could have inspired such a structure, especially since the focus on heiti is consistent in the third part. The opposite line of influence, by contrast, is unlikely: in that case, one would expect the author to show some awareness of the shift, and not to repeatedly fall back into describing kennings. Litla Skálda bears all the hallmarks of an early attempt at listing kenning types – plus whatever difficult features of diction the author came across in the process. To translate this observation into a relative chronology, however, it would be necessary to find traces of influence between Litla Skálda and Skáldskaparmál. As argued above, because of the additional referents in Litla Skálda, it cannot simply be an abbreviation of Skáldskaparmál. If the author used Skáldskaparmál, but expanded on its number of referents, it is difficult to see why he should have so completey disregarded its ordering principles. The repeated breaking of the rules of Skáldskaparmál points in the same direction. Thus, if there is any influence at all, it is highly likely that it runs from Litla Skálda to Skáldskaparmál. In fact, Litla Skálda seems to have exerted such influence not only on Skáldskaparmál, but also on Gylfaginning.

3.3.3 Intertextual Connections In several instances, I have found similar formulations in Litla Skálda and Snorri’s Edda involving a number of unique words or names, or collocations of the same kennings. Of these words, names and kennings, only axlar fótr and Gleipnir are found in the meaning they have in Litla Skálda and Skáldskaparmál in any known poetry that is likely to be older than Skáldskaparmál.137 Furthermore, they all belong in passages without a basis in poetry – except two stanzas found in Litla Skálda itself. It is thus unlikely that the similarities are due to independent use of the same poetic sources. Rather, we are probably dealing with direct influence. Here follows the first poetic or quasi-poetic (the last line is too short) quotation in Litla Skálda. I have italicizied unique words as well as the words sár, stǫng and bera, which reoccur in Snorri’s Edda: Sœgr heitir sár, en Simul stǫng, Bil ok Hjúki

in Skáldskaparmál, but the overall structure of the treatise is very clear in this regard (Snorri, Skáldskaparmál, ed. Faulkes, p. xxi). 137 Axlar fótr is found in a stanza in Grettis saga (Skj B 2, p. 476; A 2, p. 445). The age of this stanza is somewhat difficult to assess (see p. 257 n. 239 and p. 269 n. 281). In any event, the stanza cannot account for the collocation of axlar fótr and bognauð. Gleipnir is attested in the same meaning in Stríðkeravísur, a poem which appears to belong to the revival of advanced diction in the twelfth century (or possibly later) (LP, s.v.; see above, p. 76 n. 145).

3.3 Litla Skálda, Gylfaginning and Skáldskaparmál

141

bera hann, þat er kallat at þau sé í tungli.138 A trough is called Sœgr / and [its] pole Simul / Bil and Hjúki / carry it – it is said that they are in the moon.

Gylfaginning has a prose description: ‘Bil ok Hjúki [. . .] báru á ǫxlum sér sá er heitir Sœgr, en stǫngin Simul’ (Bil and Hjúki [. . .] carried on their shoulders the trough that is called Sœgr, and the pole [is called] Simul).139 There is also overlap in a number of unique kennings. Litla Skálda has: ‘Eyru manns er kallat hlustar auga eða sjónir’ (a person’s ears are called the eye or sight of hearing).140 Skáldskaparmál has: ‘Eyru heita [. . .] sjón eða augu heyrnarinnar’ (the ears are called [. . .] the sight or eyes of hearing).141 Litla Skálda has: ‘Hǫnd er [. . .] nauð boga eða tǫng. Kalla má hana fót eða lim axlar’142 (The hand is the bow’s harm or thongs. It may be called the shoulder’s foot or limb). Skáldskaparmál has: ‘Hǫnd má kalla [. . .] fót axlar, bognauð’143 (The hand may be called [. . .] the shoulder’s foot, bowharm). Háttatal 48, furthermore, has boga nauðir (the bows’ harms [hands]).144 Litla Skálda’s section on Fenrir has dense verbal echoes in Gylfaginning, mostly in names that are not attested elsewhere, or which otherwise carry another meaning: Ámsvartnir (the lake containing the island Lyngvi), Lyngvi (the island where Fenrir was fettered),145 Gleipnir (the fetter), Drómi, Læðingr (fetters that could not hold Fenrir), Gnjǫll/Gjǫll (a hole in the stone where Gleipnir is fastened [LS]/the stone itself [SkskM]),146 Þviti (the stone where Gleipnir is fastened [LS]/an additional stone for further security [SkskM], Gelgja (the peg put in the loop at the end of the rope [LS]/the loop itself [SkskM]).147

138 Edda Snorra Sturlusonar, ed. Finnur Jónsson, p. 259. Only Sophus Bugge has identified the half stanza in this passage, see Norrœn fornkvæði, ed. Bugge, pp. xxxiii, 334. 139 Edda Snorra Sturlusonar, ed. Finnur Jónsson, p. 18. 140 Edda Snorra Sturlusonar, ed. Finnur Jónsson, p. 256. 141 Edda Snorra Sturlusonar, ed. Finnur Jónsson, p. 190. 142 Edda Snorra Sturlusonar, ed. Finnur Jónsson, p. 256. 143 Edda Snorra Sturlusonar, ed. Finnur Jónsson, p. 192. U has baugnauð, which is clearly due to corruption (saute du même au même). 144 LP, s.v. nauðr; Snorri, Háttatal, ed. Faulkes, p. 22. The kenning is not found elsewhere, but the parallel dalnauð (bow-harm [hand]) is used in Bjarkamál 5 (LP, s.v.). 145 In Litla Skálda and Skáldskaparmál, Lyngvi designates an island, but in the þulur, it is a heiti for a sea-king (LP, s.v.). 146 B has ‘ginul’ (Edda Snorra Sturlusonar, ed. Jón Sigurðsson et al., 2, p. 515). Gjǫll can elsewhere mean ‘battle’ or be the name of a mythological river (LP, s.v.). 147 Gelgja may have been a poetic word for ‘band’, but is attested only once (Egill, lausavísa 12, see LP, s.v. Note, however, the discussion on etymology and semantics in Fritzner, Ordbog, 1, s.v.). All these names/words are found in Edda Snorra Sturlusonar, ed. Jón Sigurðsson et al., 2, pp. 431, 515; Snorri, Prologue and ‘Gylfaginning’, ed. Faulkes, pp. 27–29.

142

3 Grammatical Literature

In this section, Litla Skálda also contains a stanza on the components of Gleipnir – the fetter that was finally able to hold Fenrir: Ór kattar dyn ok ór konu skeggi, ór fisks anda ok ór fugla mjólk, ór bergs rótum ok bjarnar sinum, ór því var hann Gleipnir gerr.148 From the noise of the cat and from the beard of the woman, from the breath of the fish and from the milk of birds, from the roots of the mountain and from the sinews of the bear – from this Gleipnir was made.

Gylfaginning has: Hann [Gleipnir] var gǫrr af sex hlutum: af dyn kattarins ok af skeggi konunnar ok af rótum bjargsins ok af sinum bjarnarins ok af anda fisksins ok af fogls hráka.149 It [Gleipnir] was made from six things: from the noise of the cat and from the beard of the woman and from the roots of the mountain and from the sinews of the bear and from the breath of the fish and from the spittle of the bird.

By inverting the order of the words, Snorri has removed the metre from the stanza, much as he has done with the half stanza previously discussed. Why he did so in these cases is not entirely clear, but he was certainly selective regarding which poetry he included, and the lines bera hann and bjarnar sinum do not conform to Seiver’s metrical types. He may thus have had aesthetical reasons for rendering them in prose. There are otherwise two marked differences between the text in Litla Skálda and that in Gylfaginning. First, the second and third couplets have switched places. Second, the stanza has ‘the milk of birds’ and Gylfaginning ‘the spittle of the bird’. This difference is probably conditioned by metrics: fugla is the only plural genitive in the stanza, and if it were changed to the singular to achieve symmetry, fugls mjólk would create metrical dissonance (even if the line would not be unmetrical), whereas fugls hráka works well and is metrically identical to fisks anda and bergs rótum. The differences in order and wording between the stanza in Litla Skálda and the prose in Gylfaginning were presumably not introduced by Snorri, but by someone who chose to retain the stanza in its poetic form. Since Snorri takes words both from the surrounding prose and from the stanza, I think it likely that he had

148 Edda Snorra Sturlusonar, ed. Jón Sigurðsson et al., 2, pp. 432, 515. 149 Snorri, Prologue and ‘Gylfaginning’, ed. Faulkes, p. 28.

3.3 Litla Skálda, Gylfaginning and Skáldskaparmál

143

access to both through a written text, albeit one that differed slightly from what is preserved in Litla Skálda. Some of the hierarchical shifts in meaning from one implement for tying the wolf to the next may also be due to variants in the source he used. It could be argued that the similarities as well as the minor discrepancies are based on oral tradition, even though they may not be based on poetry, except for the two stanzas in Litla Skálda which are echoed in Gylfaginning. The fact that the verbal echoes are found in all three sections of Litla Skálda makes this unlikely, however. In order for these affinities to derive from oral tradition, their distribution would make it necessary to postulate something like an oral Litla Skálda. This seems implausible and unwarranted. The fact remains, however, that Snorri’s exemplar contained considerable variants in both the poetry and the prose of one of the three sections, namely that on Fenrir. This probably suggests some degree of oral interference in the transmission of Litla Skálda, since it would appear that someone has revised the stanza and the meaning of the implements in accordance with his own perception of traditional lore relating to this topic. This does probably not imply separate written transmission of the section on Fenrir, since Snorri used all three sections, but it may indicate that this section circulated orally as well. If so, it offers a fascinating glimpse into the oral transmission of lore that was not exclusively in poetic form. It should be remembered, however, that written transmission of Old Norse texts sometimes resulted in considerable variation, the dynamics of which have not yet been fully investigated. This is shown, for instance, by the þáttr about Þorgils discussed on pp. 259–62, as well as by the interpolations in the Greater Saga of Saint Óláfr discussed on pp. 72–75. The suggestion of an oral background to the variants in the section of Fenrir must therefore remain tentative. The semantic differences notwithstanding, the Fenrir passage makes Snorri’s working method relatively clear. In Litla Skálda, this section enumerates the places and implements connected to the binding of Fenrir – in practice it serves as a kind of vocabulary. Snorri has turned it into narrative, in conformity with the rest of Gylfaginning. With regard to the content of Gylfaginning at large, the eddic poems were no doubt vastly more important than Litla Skálda. Litla Skálda may, however, have suggested the possibility of composing a poetic treatise containing descriptive and narrative prose accompanied by eddic poetry, since the second and third section of Litla Skálda attempt this in a rudimentary form. Possible prose sources to Gylfaginning have proved elusive, but it seems likely that Snorri drew on some version of Trójumanna saga and possibly Elucidarius.150 These texts have no bearing

150 The information about the Trojans in the end of Gylfaginning may be derived from Trójumanna saga, and it has been suggested that Snorri also drew on a now lost *Sigurðar saga (Snorri, Prologue and ‘Gylfaginning’, ed. Faulkes, p. xxiii). On Elucidarius, see The Arnamagnæan Manuscript 674A, 4to. ‘Elucidarius’, ed. facs. Jón Helgason, p. xxxi; Margaret Clunies Ross, ‘Skáldskaparmál’. Snorri

144

3 Grammatical Literature

on Snorri’s poetic project, however. Since Litla Skálda is the only text on poetics comparable to either Gylfaginning or Skáldskaparmál that we know of, and since many indications that Snorri used it have here been presented, it is highly likely that the form of Gylfaginning has developed in a dialogue with it. The relationship between Litla Skálda and Skáldskaparmál is more straightforward: they are both treatises on kennings. It is therefore easier to get to grips with how Snorri used his model with regard to structure, function and emphasis. On point after point, Snorri’s treatment of his model is analogous to how he used Háttalykill in the composition of Háttatal. In both cases, he expanded considerably on his model and added rules and subdivisions, as well as a coherent structure. In the process, he blended poetic examples and commentary.

3.3.4 Litla Skálda, Snorri and the Traditionalizing Position In his treatment of Litla Skálda, however, Snorri not only transformed and expanded on his model, but he also shifted the work’s emphasis. The key word here is tradition. To the extent that skaldic poetry was an archaic and archaizing genre, the content of Litla Skálda is traditional. Nevertheless, Litla Skálda does not promote or contradict tradition or any particular version thereof either in its explicit injunctions nor in its mode of composition. In a famous and often quoted passage in Skáldskaparmál, by contrast, we read: En þetta er nú at segja ungum skáldum þeim er girnask at nema mál skáldskapar ok heyja sér orðfjǫlða með fornum heitum eða girnask þeir at kunna skilja þat er hulit er kveðit: þá skili hann þessa bók til fróðleiks ok skemtunar. En ekki er at gleyma eða ósanna svá þessar sǫgur at taka ór skáldskapinum fornar kenningar þær er hǫfuðskáld hafa sér líka látit. En eigi skulu kristnir menn trúa á heiðin goð ok eigi á sannyndi þessar sagnar á annan veg en svá sem hér finnsk í upphafi bókar er sagt er frá atburðum þeim er mannfólkit viltisk frá réttri trú [. . .]151 And this one shall tell the young poets who wish to learn poetic diction and acquire a vocabulary rich in old expressions, or who wish to understand that which is opaquely expressed: Then may he understand this book for learning and for pleasure. And one should not forget or disprove these tales by taking away old kennings that the canonical poets found pleasing. But Christians should not believe in the pagan gods or in the truth of this narrative in any other way than what can be found here in the beginning of the book, where the events are described when mankind strayed from correct belief [. . .]

Sturluson’s ‘Ars Poetica’ and Medieval Theories of Language (Odense: Odense University Press, 1987), pp. 55–58; Sveinbjörn Rafnsson, Um ‘Snorra Eddu’ og ‘Munkagaman’. Drög til menningarsögu íslenskra miðalda. Ritsafn Sagnfræðistofnunar, 43 (Reykjavík: Háskólaútgáfan, 2016), pp. 42, 76. Sveinbjörn Rafnsson claims to identify a rich body of additional sources, but his empirical foundation is not always convincing. 151 Snorri, Skáldskaparmál, ed. Faulkes, p. 5

3.3 Litla Skálda, Gylfaginning and Skáldskaparmál

145

If young poets were to stick to tradition, they also needed to have access to it. Accordingly, Snorri provided hundreds of poetic quotations of poets worthy of emulation. This difference in the emphasis on tradition may also explain another notable divergence between Litla Skálda and Skáldskaparmál. In the manuscripts that are most likely to reflect the original structure of the Edda, the passage quoted above precedes a long section of kennings for Óðinn, poetry (since it is connected to Óðinn, god of poetry) and the other gods, whereas Litla Skálda simply opens with kennings for poetry.152 The names of the gods were embedded in many traditional kenning types, and these are given in Litla Skálda, but as subject matter for skaldic poetry, the author seems to have perceived of the gods as obsolete. Such a commonsensical approach is foreign to Snorri’s Edda, which is aimed not only at assisting composition, but also at directing it and at promoting a particular view of the poetic tradition. This is also borne out by the prominence of mythological narrative in Skáldskaparmál, which far exceeds what is required for the interpretation or composition of poetry. While Litla Skálda gives a sober description of a traditional genre, Skáldskaparmál is actively traditionalising. This difference has some bearing on the likelihood that Litla Skálda is an early work by Snorri himself. Insistence on tradition, defined by mythology and poetics, is a recurrent theme throughout Snorri’s Edda, and this marked shift in emphasis between Litla Skálda and Snorri’s Edda makes it unlikely that the former was composed by Snorri. Snorri’s older relative Einarr Skúlason, who is one of the best-known proponents of neo-traditional poetry, is quoted far more often than any other poet in Skáldskaparmál, and it is likely that he served as a paragon for Snorri from early on, through family traditions.153 In general, the long-term development of the traditionalizing position rules out Snorri as a likely candidate for authorship and makes it tempting to follow Jesch’s lead towards dating the earliest attempts at a vernacular poetics around the middle of the twelfth century, or at least to sometime c. 1200 or before. We saw in Chapter One that metre became the object of learned study in the twelfth century, and Chapter Two described the revival of mythology and advanced diction which had been under way for at least two or three generations by Snorri’s time. Finally, with Litla Skálda, we get a treatise on diction which may also be understood as a piece of mythography in embryo. What do these developments tell us about Snorri’s motivations in composing his Edda? Elias Wessén’s answer to this question in his introduction to the edition of A and AM 748 Ia 4to, published in 1940, has become very influential. Wessén admits that Latin artes poeticae could have contributed to the general idea of composing Snorri’s Edda, but Snorri’s real reason for doing so was his despair at the rapid disappearance

152 R, W, T. U has a somewhat different redaction, with parts of the prose of Skáldskaparmál placed towards the end of Gylfaginning, and lists of poets and lawspeakers are placed between Gylfaginning and Skáldskaparmál, resulting in a less direct relation between this passage and kennings for gods and poetry (The Uppsala Edda, ed. Heimir Pálsson and trans. Faulkes, pp. lxxv, 90). 153 See Guðrún Nordal, Tools of Literacy, pp. 77–78, 84–6, 88, 337.

146

3 Grammatical Literature

of the skaldic art.154 Wessén here adheres to the views of his older contemporary, Sigurður Nordal. According to him, Snorri’s aim was nothing less than that of arresting ‘the victorious advance of the Roman Church in Iceland’.155 Later scholars have done much to contextualize Snorri’s Edda within a European framework, but they have tended to be more circumspect than Sigurður Nordal or Wessén with regard to the reasons for its composition.156 While European parallels are valuable for understanding how Snorri’s Edda could find its place within a culture which was, after all, European, the very fact that these preconditions were present across Latin Christendom means that they cannot explain why a work like this would have been composed in Iceland but not elsewhere. One is therefore easily left with the impression of Snorri as a creative genius who composed his Edda more or less ex nihilo. According to the analysis presented in this book, the development of a focus on advanced metre and diction in the twelfth century situates Snorri’s Edda within a larger poetic trend. When these changing poetic interests are viewed in tandem with Snorri’s use of Litla Skálda, a more organic and likely development presents itself: from poetic developments to the first attempts at treating poetics in the form of a treatise and then, finally, to the redaction of the extraordinary work that is Snorri’s Edda. This explanation – that Snorri was part of a trend that was gradually gaining force – is almost diametrically opposite to that of Wessén, where the demise of poetic activities was what spurred Snorri’s rescue mission. The present analysis also means that Wessén’s reconstruction of the order in which Snorri composed his Edda must be subject to serious doubt. Wessén claims that Snorri composed his Edda backwards, beginning with metrics under the influence of Háttalykill and then explaining and justifying each section as he went.157 This reconstruction has become a truism within Old Norse studies. Everything except the metrical poem Háttatal thus materialized as monumental afterthoughts without precedence. In light of earlier poetic developments, however, as well as the use of Litla Skálda in both Skáldskaparmál and Gylfaginning, there is no reason to

154 Codex Regius of the Younger ‘Edda’, ed. Wessén, pp. 12–13, 30. 155 Codex Wormianus (The Younger Edda). MS. No. 242 fol. in The Arnemagnean Collection in the University Library of Copenhagen, ed. Sigurður Nordal. Corpus codicum Islandorum medii aevi, 2 (Copenhagen: Levin and Munksgaard, 1931), p. 12. 156 Cautious opinions about why Snorri’s Edda was composed and why it assumed the character it did may be found in, for instance, Guðrún Nordal, Sverrir Tómasson and Vésteinn Ólason, Íslensk bókmenntasaga, 1 (Reykjavík: Mál og Menning, 1992), pp. 535–37; Clunies Ross, A history of Old Norse Poetry and Poetics, pp. 183–84. Important works regarding European contextualization are, for instance, Ursula and Peter Dronke, ‘The Prologue of the Prose Edda: Explorations of a Latin Background’, in Sjötíu ritgerðir helgaðar Jakobi Benediktssyni, ed. Einar G. Pétursson and Jónas Kristjánsson (Reykjavík: Stofnun Árna Magnússonar, 1977), pp. 153–76; Clunies Ross, ‘Skáldskaparmál’; Faulkes, ‘The Sources of Skáldskaparmál’; Háttalykill enn forni, ed. Jón Helgason and Holtsmark. 157 Codex Regius of the Younger ‘Edda’. MS No. 2367 4ᵗᵒ in the Old Royal Collection in the Royal Library of Copenhagen, ed. Elias Wessén (Copenhagen: Munksgaard, 1940), p. 32.

3.4 Snorri’s Reception of Poetry: Skáldskaparmál

147

suppose that Snorri could not have been working on these two texts at an early stage. The process of composition may have been considerably less linear than in Wessén’s psychologizing reconstruction, but the details remains elusive.158 What can be said with some confidence is that Gylfaginning, Skaldskaparmál and Háttatal were all composed as parts of a complex but coherent whole. As noted on pp. 114–15, the complementary relation between Skáldskaparmál and Háttatal is indicated by the prose of both sections. The complementary relation between Gylfaginning and Skáldskaparmál is clear from their precise lack of narrative overlap: Skáldskaparmál reports the narrative necessary to contextualize the kennings under discussion, whereas Gylfaginning gives a mythographical overview which leaves out the motifs treated in Skáldskaparmál.159 It has often been noted that Snorri’s Edda displays inconsistencies, and this is true, but it is important also to note these indications of a coherent authorial vision in order to appreciate the ambitious scope of Snorri’s undertaking. To date, Guðrún Nordal is probably the one scholar who has done most to show how Snorri’s Edda relates to a learned discourse on skaldic poetry that developed from the twelfth century onwards.160 Litla Skálda was a key component in this process, and through the analysis of its relation to Snorri’s Edda, as well as the delineation of developments within metrics and diction in preceding chapters, I hope to have added to Nordal’s analysis. The picture that emerges is one where Snorri was doing much the same thing as others had been doing for more than half a century before him, though it is probably fair to say that he outdid them all in creativity, scope and self-esteem.

3.4 Snorri’s Reception of Poetry: Skáldskaparmál Snorri’s handling of the mythological tradition has been the object of many studies.161 In contrast to most of that research, the discussion here has focused on how

158 Similarly Heimir Pálsson, ‘Reflections on the Creation of Snorri Sturluson’s Prose Edda’, Scripta Islandica, 68 (2017), pp. 189–232. 159 Heimir Pálsson argues that the lack of narrative overlap between Gylfaginning and Skáldskaparmál indicates that they were conceived as separate entities (‘Reflections on the Creation of Snorri Sturluson’s Prose Edda’, pp. 202–03, 217–18). He therefore concludes that Snorri may have been working on both at various times. I do not contest the second statement, but I would argue that the lack of narrative overlap indicates the opposite of Heimir Pálsson’s first claim, namely that Gylfaginning and Skáldskaparmál were composed to complement each other. 160 Nordal, Tools of Literacy. 161 See, for instance, Edith Marold, ‘Der Dialog in Snorris Gylfaginning’, in Snorri Sturluson. Beiträge zu Werk und Rezeption, ed. Hans Fix (Berlin: De Gruyter, 1998), pp. 131–80; Margaret Clunies Ross, ‘Snorri’s Edda as Medieval Religionsgeschichte’, in Germanische Religionsgeschichte. Quellen und Quellenprobleme, ed. Heinrich Beck, Detlev Ellmers and Kurt Schier (Berlin: De Gruyter, 1992), pp. 633–55; Klaus von See, Mythos und Theologie; Gerd Wolfgang Weber,

148

3 Grammatical Literature

Snorri treated identifiable sources and how he reshaped and reframed them to suit his interests. This method allows for some degree of testing of the hypotheses. I turn now to the most self-evident of these sources: namely, poetry. The most substantial portion of Snorri’s source material lies in plain sight in the form of hundreds of poetic quotations. Here, I wish to consider the cases when his interpretations of these sources appear to go beyond what the poet probably wished to convey. I begin with Skáldskaparmál, since this work has been the focus of the most thorough attempt so far to describe Snorri’s creative use of his poetic sources, and I thereafter proceed to Gylfaginning. To date, the main study of this topic has been conducted by Roberta Frank.162 Frank investigates how Snorri constructed the narrative about the mead of poetry in Skáldskaparmál. Although Frank’s arguments have not been universally accepted, they deserve careful consideration, given their implications for the working methods of Iceland’s most important mythographer.163 Here I shall focus on Snorri’s sources, and in Chapter Four I shall turn to Frank’s arguments about the later use of Skáldskaparmál. The main component of Frank’s argument regards Snorri’s description of the being Kvasir, whose blood became the mead of poetry, and of the three vats in which the mead of poetry was brewed: Són, Boðn and Óðrerir. Frank argues that Kvasir and the three vats are the products of a transformation of kennings into narrative. The names of Kvasir and the vats are all attested in stanzas quoted in Skáldskaparmál to exemplify kennings for poetry, according to a narrative which is given somewhat earlier in the text.164 Three of the four examples are taken from the opening stanzas of Einarr skálaglamm’s Vellekla, where the poet makes a

‘Intellegere historiam. Typological perspectives of Nordic prehistory (in Snorri, Saxo, Widukind and others)’, in Tradition og historieskrivning: Kilderne til Nordens ældste historie, ed. Kirsten Hastrup and Preben Meulengracht Sørensen (Århus: Aarhus universitetsforlag, 1987), pp. 95–141: idem, ‘Siðaskipti. Das religionsgeschichtliche Modell Snorri Sturlusons in Edda und Heimskringla’, in Sagnaskemmtun. Studies in Honour of Hermann Pálsson, ed. Rudolph Simek, Jónas Kristjánsson and Hans Bekker Nielsen (Wien: Böhlau, 1986), pp. 309–29; Heinz Klingenberg, ‘Gylfaginning. Tres vidit unum adoravit’, in Germanic Dialects. Linguistic and Philological Investigations, ed. Bela Brogyani and Thomas Krömmelbein (Amsterdam: John Benjamin, 1986), pp. 627–93; Anne Holtsmark, Studier i Snorres mytologi (Oslo: Universitetsforlaget, 1964). 162 Roberta Frank, ‘Snorri and the Mead of Poetry’, in Speculum Norroenum: Norse Studies in Memory of Gabriel Turville-Petre, ed. Ursula Dronke et al. (Odense, 1981), 155–70. 163 Most of the criticism raised up until now is of such a nature that makes it difficult to verify. For instance, Thomas D. Hill thinks that traditional myth must probably account for the the motif of the mead of poetry, since many elements in the story are not used in kennings and it is unlikely that Snorri would have invented so much. This assumption, however, is not easily tested (‘Beer, Vomit, Blood, and Poetry. Egils saga, Chapters 44–45ʹ, in New Norse Studies. Essays on the Literature and Culture of Medieval Scandinavia, ed. Jeffrey Turco. Islandica 58 (Ithaca, NY: Cornell University Library, 2015), pp. 248 and n. 5). 164 Snorri, Skáldskaparmál, ed. Faulkes, 12–13. The story is given on pp. 3–4.

3.4 Snorri’s Reception of Poetry: Skáldskaparmál

149

point of varying kennings for poetry. The fourth, the one involving Són, is found in a stanza by Eilífr Goðrúnarson, the most obscure of Old Norse poets. The stanza in question is no exception. Earlier editions assign this stanza to an otherwise lost poem about Hákon jarl, whereas SkP presents it as part of Þórsdrápa.165 Snorri’s introduction suggests that one should find ‘Són’s liquid [poetry]’ in this stanza, but no liquid is forthcoming. Frank construes són with sefrein, giving sónar sefrein (reconciliation’s mind-land), where ‘mind-land’ means ‘breast’, and ‘reconciliation’ would presumably be intended to underline the theme of friendship in the stanza (‘you must govern friendly gifts’, the poet says). SkP has Sónar orða sáð (Són’s words’ [poetry] seed [poetry]). Neither of these interpretations accords with Snorri’s description, although the one in SkP is closer to it. I have not found evidence in support of either reading which is sufficient to test Frank’s hypothesis with regard to this stanza.166 One difficulty in choosing between the readings is that, even if SkP is right in assigning the stanza to Þórsdrápa, there are no similar expressions in the rest of the poem which can indicate which reading is the right one. Because of the elusive nature of Eilífr’s poetry and the lack of material for internal comparison, I base the following argument on the stanzas from Vellekla, where such comparisons can be made. The stanzas from Vellekla contain the following relevant kennings, according to Snorri’s reading: Kvasis dreyri (Kvasir’s blood [poetry]), Óðreris alda (Óðrerir’s wave [poetry]) and Boðnar bára (Boðn’s wave [poetry]). Frank, however, sees no proper names here and reads kvasis dreyri (ferment’s blood [mead > poetry]) and Óðreris alda (the mind-stirrer’s / poetry-stirrer’s [mead of poetry’s] wave [poetry]).167 The final kenning she sees as part of a longer one: berg-Saxa boðnar bára (the mountainSaxons’ [giants’] vat’s wave [mead > poetry]). Boðn seems to be an obsolete word meaning ‘vat’ and cognate with Old English byden and Modern Icelandic byðna of the same meaning.168 If Frank is right this word retains its original meaning in the stanza, whereas Snorri’s description presupposes that it has become a proper name. I turn now to an evaluation of whether Snorri’s, Frank’s or other editors’ interpretation of these kennings is more likely to correspond to that of the poet. I begin with the boðn stanza:

165 SkP 3, pp. 124–26. 166 Kormákr, who like Eilífr composed to the earls of Hlaðir, uses the word haptsœni (gods’ reconciliation [poetry]) in Sigurðardrápa. The use of the root sœn- (umlaut of són-) in a kenning for poetry in a social setting close to that of Eilífr suggests that one should follow Snorri and connect sónar in Eilífr’s stanza to poetry. In Sigurðardrápa, the word haptsœni stands in the genitive (haptsœnis), and it is therefore on formal grounds possible that it derives from a word haptsœnir (reconciler of the gods). Some scholars have thus interpreted it to refer to Óðinn (see SkP 3, p. 282). This interpretation, however, entails attributing a quality of reconcilitation to Óðinn which is contrary to what we otherwise know about him. It therefore seems relatively certain that the reference is to poetry, not to Óðinn. 167 Frank, ‘Snorri and the Mead of Poetry’, p. 162. 168 Frank, ‘Snorri and the Mead of Poetry’, 162 note 17.

150

3 Grammatical Literature

3.4.1 Boðn Nú es þats boðnar bára berg-Saxa tér vaxa, gørvi í hǫll ok hlýði hljóð fleyjǫfurs þjóðir.

In Frank’s reading, the word order and content of this stanza are relatively straightforward: Nú er þats berg-Saxa boðnar bára tér vaxa, fleyjǫfurs þjóðir gørvi hljóð í hǫll ok hlýði. Now it is that the mountain-Saxons’ [giants’] vat’s wave [mead = poetry] begins to swell. May the ship-king’s men be silent in the hall and pay attention.

In the reading of Skj and SkP, the word order is unusually contorted, even for skaldic poetry: Nú er þats Boðnar bára tér vaxa, jǫfurs þjóðir gørvi hljóð í hǫll ok hlýði [SkP] /heyri [Skj] bergSaxa fley. Now it is that Boðn’s wave [poetry] begins to swell, may the king’s men be silent in the hall and pay attention to/hear the mountain-Saxons’ [giants’] ship [poetry].

Berg-Saxa in the second verse here goes with fley in the middle of the fourth. The editors in SkP do not posit such a solution for any of the other opening stanzas of Vellekla, and this may raise doubts as to whether this is the right reading. Vellekla also offers internal support for the connection of fley to jǫfurs, as in Frank’s reading: elsewhere in the poem, its recipient Hákon is referred to as flausta ǫrþeysir (the ships’ swift-mover [captain]), hranna hrafna vǫrðr (the waves’ ravens’ guardian [captain]) and haffaxa hald-Viðurr (the sea-horses’ controlling Viðurr (name for Óðinn) [captain]).169 These are kennings, whereas fleyjǫfurr would be an epithet where jǫfurr retains its proper meaning. Stanza four contains an analogue to this in the expression Ullar asksǫgn (Ullr’s ash- [ship] crew [ship-crew]) or simply asksǫgn (ship-crew).170 Asksǫgn is is structurally identical to fleyjǫfurr, retaining the meaning of its final element, but focusing on the crew rather than the captain. The original order of the five opening half stanzas is not certain, except that the one with the opening formula was first, but the themes and mode of composition suggests that they belonged together. Fleyjǫfurr and asksǫgn would thus have been close to each other, possibly even in the same

169 Kock, Notationes norroenae, § 392. 170 The first reading is that of Skj B 1, p. 117, the second is that of SkP 1, pp. 287–88. The first may be preferable since it is more specific.

3.4 Snorri’s Reception of Poetry: Skáldskaparmál

151

stanza. Their proximity, as well as their structural and semantic affinities, suggests that fleyjǫfurr is the right reading. There is another piece of evidence against taking fley as an independent word. The verb hlýða takes the dative, and a neuter ja-stem like fley always retained the dative -i until the twelfth century – thereafter, -i in rare instances fell.171 The expected form would therefore be *fleyi, for which there is no room in the metre. Finnur Jónsson emended hlýði to heyri (hear), which takes the accusative – fley – and gives hending with gørvi.172 This emendation is undertaken against the unequivocal evidence of the manuscripts. More importantly, with the reading hlýði, we get lack of hending in the odd line which instead ties into the next (hlýði: hljóð in this instance). This is a trademark of Einarr’s, and one that later poets did not reproduce.173 A scribal change from heyri to hlýði, presupposed by Finnur Jónsson, would have created what poets in Snorri’s day considered to be a flaw, and cannot thus be explained on text-critical grounds. Absolute use of hlýða, however, as in Frank’s reading, is common, and no object is thus necessary.174 SkP and Anthony Faulkes both retain the manuscript reading hlýði and treat fley as its dative object. SkP does not mention the morphological problem, but Faulkes suggests that the dative ending may have been elided before j-.175 This, however, is clearly an ad hoc solution, since while a few common words can have initial elision (ák for á ek etc.), final elision occurs almost exclusively in the fourth position and not before semi-vowels.176 All scholars who have addressed this problem find a dative without -i to be problematic, and this suggests that we should look for a solution where fley is not in the dative. There is yet another argument against the reading berg-Saxa fley and in favour of berg-Saxa boðnar/Boðnar bára. In the first half of the eleventh century, Hofgarða-Refr in his bid for a hearing in a poem about a certain Þorsteinn appears to allude to the

171 Neuter a-, ja- and wa-stems ending in a vowel always show dative -i before the twelfth century (Finnur Jónsson, Det norsk-islandske skjaldesprog omtr. 800–1300 (København: Møllers bogtrykkeri, 1901), pp. 26–27, 31, 35). With regard to dative -i, these three groups may be treated as belonging to the same declension. Outside of the present stanza, fley never occurs in the dative in the Old Norse corpus, and the proposition can thus not be verified by this particular word. The closest match is hey. In Old Norse manuscripts, the forms with or without -i of that word vary, but the ones without are rare (Adolf Noreen, Altisländische und altnorwegische Grammatik (Laut- und Flexionslehre), 4th ed. (Halle: Max Niemeyer, 1923), p. 257). One would assume that the variation between the two forms cannot have begun before other similar contractions took place in the twelfth century. There is no metrical evidence for this with regard to hey, but there is for the a-stem fé etc. See Finnur Jónsson, Det norsk-islandske skjaldesprog, pp. 26–27. 172 Thus Skj B 1, p. 117; A 1, p. 123. Finnur Jónsson remarks that hlýði ‘must be wrong for heyri’ (LP, s.v. hlýða). 173 Kuhn, ‘Vor tausend Jahren’. The one exception is Snorri, Háttatal stanza 58 (Snorri Sturluson, Edda. Háttatal, ed. Faulkes, pp. 26, 78). 174 See LP, p. 266. 175 SkP 1, pp. 284–85; Snorri, Skáldskaparmál, ed. Faulkes, p. 163. 176 Kuhn, Das Dróttkvætt, pp. 70–72.

152

3 Grammatical Literature

kenning berg-Saxa boðnar/Boðnar bára in Vellekla. Refr says ‘berg-Mœra glymr bára’ (the mountain-Mœrirs’ [giants’] wave [poetry] resounds).177 The collocation of bergand bára in tandem with a kenning for giants of the type berg-(ethnonym), only found in the kennings berg-Danir (twice), berg-Saxar and berg-Mœrir, together indicate that there is intertextual dependence here.178 The fact that both occurrences are found in the bid for a hearing further corroborates this. There is no ship in Hofgarða-Refr’s kenning, and this presumably means that he construed the kenning in Vellekla as bergSaxa boðnar/Boðnar bára. This also means that he must have construed fleyjǫfurr as a compound, since fley syntactically has nowhere else to go and precedes jǫfurr. Finally, Snorri does not appear to have construed the stanza in Vellekla with the kenning berg-Saxa fley. According to his introduction to this group of stanzas, the ship belonged to the dwarves alone. SkP solves this problem by explaining this as a case of ofljóst involving lið ‘ship’ and líð ‘beer’, and the drink did indeed belong to the giants for a time. While Snorri discusses such an instance of ofljóst further down, at this point he still focuses exclusively on mythical explanations.179 In Skáldskaparmál, the stanza exemplifies the use of Boðn, and in Snorri’s description, Boðn, unlike the ship, belonged to the giants after they received the mead from the dwarves. Snorri’s reading would thus in all likelihood have been ‘the giants’ Boðn/ boðn’, which would have left fley to be attached to jǫfurr, just as in Frank’s interpretation. There are thus many reasons to accept Frank’s reading, but two factors complicate the matter somewhat. First, fley- is not attested in compounds for people, and -jǫfurr is a rare element in compounds.180 The same could, however, be said of asksǫgn, where ask- is not attested in compounds for people and -sǫgn is very rare in the meaning ‘crew’.181 This point would thus appear to be irrelevant to the argument. Second, the words fley and jǫfurr are written separately in the manuscripts. Again, a comparison with asksǫgn is informative. Here, we are clearly dealing with a compound, and in R, asksǫgn is written as one word, whereas in U, W and B it is written as two (‘ask song’ in U). We can thus see a tendency at work to separate the elements of a compound, which is not surprising given that the elements can function as independent words and that the compounds are unique formations, which means that the linguistic competence of a scribe would make him recognize the elements, but not the compound. I therefore believe that this argument has little bearing on the issue,

177 SkP 3, p. 250. 178 Meissner, Die Kenningar, pp. 256–57. The collocation of berg- and bára is subject only to alliteration and not hending constraint, and should thus be seen as a free choice on the part of the poet. 179 SkP 1, p. 285. 180 LP, p. 141, s.v. jǫfurr. 181 LP, s.vv. askr, asksǫgn, sǫgn.

3.4 Snorri’s Reception of Poetry: Skáldskaparmál

153

whereas the ones regarding word order, morphology and the analoguous asksǫgn, when taken together, speak strongly in favour of rejecting the Skj and SkP readings and construing fleyjǫfurr as one word.

The word asksǫgn in R (21v; Image: Reykjavík, Stofnun Árna Magnússonar) and W (23v; Image: Copenhagen, Den Arnamagnæanske Samling. Photo: Suzanne Reitz).

Another question is whether Frank’s word order necessitates her interpretation. At least at first glance, this would seem to be the case. Frank’s reading of the kenning in question is berg-Saxa boðnar bára (the mountain-Saxons’ [giants’] vat’s wave [poetry]), and the reading berg-Saxa Boðnar bára (the mountain-Saxons’ [giants’] Boðn’s wave [poetry]) would render berg-Saxa redundant, as noted in SkP.182 The validity of this argument is not clear, however. Scholars have been long opposed to redundant elements in kennings, and this means that they may be rarer in modern analyses of skaldic poetry than they were in medieval ones. Regrettably, while Meissner discusses the dropping of elements in kennings, he does not treat redundant ones, and to the best of my knowledge this topic has not been investigated. Neither Skj nor SkP use the concept of redundancy in a rigorous way: the editions sometimes exclude an element to avoid redundancy, at other times not, and the problem is not explicitly addressed. In several instances, modern scholars have possibly seen redundancy where the poets did not, which adds to the problems in using this criterion to guide the interpretation of kennings. The seemingly most obvious type of redundancy may be the repetition of entities from the same class, such as in the kenning Þorins rekka regn (Þorinn’s [a dwarf’s] men’s [the dwarves’] rain [poetry]), discussed on p. 90. Here, however, it is probably relevant that the mead belongs to more than one dwarf, and no element is therefore really redundant. In several other instances, structurally redundant elements appear to make the description inherent to the kenning more vivid. In fact, such a practice appears to be remarkably common in Vellekla itself. Stanza 1 contains the kenning fjarðleggjar fyrða dreggjar brim (the fjord-bone’s [stone’s] men’s

182 SkP 1, p. 285.

154

3 Grammatical Literature

[dwarves’] dreg’s surf [poetry]), where the meaning ‘poetry’ could be produced without the element dreggjar.183 Stanza 3 contains the kenning Óðreris hafs alda (Óðrerir’s sea’s [poetry’s] wave [poetry]), where alda is structurally redundant. In stanza 10, the easiest solution would be to take Hǫ́ars drífu hauldar/hǫlðar as a kenning (Hǫ́arr’s snow-storm’s [battles] men [men, warriors]).184 Here, hauldar/ hǫlðar already means ‘men’ and Hǫ́ars drífu is thus structurally redundant. In order to avoid this, scholars have taken Hǫ́ars drífu as an independent, locative dative or genitive (that is, ‘in the battle’),185 but such datives and genitives are very rare and, as far as I am aware, they refer to physical spaces, not to activities such as battle.186 Furthermore, locative datives and genitives occur in words which have unambiguous case endings, whereas drífu could be either accusative, dative or genitive, and they are found with verbs which suggest location or movement, while the verbal in this half stanza is lét vaxa (caused to grow).187 Hǫ́ars drífu can therefore hardly be a locative dative or genitive.188 Other scholars have suggested reading Hǫ́ars drífu as genitive and construing the words as part of other kennings. The resulting structures lack clear analogues in the corpus and are thus hardly tenable.189 For these reasons, it would appear that Hǫ́ars drífu hauldar/hǫlðar is indeed a kenning or rather a kenning-like structure whose definers are descriptive, just as in the reading berg-Saxa Boðnar bára (the mountain-Saxons’ [giants’] Boðn’s wave [poetry]). Vellekla stanza 22 has Heðins veggjar eggþings Ullar (Heðinn’s wall’s [shield’s] edge-meeting’s [battle’s] Ullar [pl. of Ullr; warriors]).190 Since Heðins veggjar Ullar or eggþings Ullar could designate warriors independently, editors have construed eggþings as a locative genitive meaning ‘in the battle’ in order to avoid redundancy.

183 SkP 1, pp. 283–84. 184 SkP 1, p. 295. 185 SkP 1, 296; Russell Poole, ‘Some Examples of the Adverbial Genitive in Skaldic Poetry’, Mediaeval Scandinavia, 14 (2004), p. 122. 186 SkP 1, p. 296. I have found no full overview of independent locative datives in poetry, but all examples of such datives and genitives given by Nygaard refer to space, not to an activity or to space by means of a metonym such as ‘battle’ (Nygaard, Norrøn syntax, pp. 126, 152). 187 See examples in Nygaard, Norrøn syntax, pp. 126, 152 188 The same editors opt for a similar, and equally problematic, solution in SkP 1, pp. 326–27. 189 Konráð Gíslason reads haffaxa Hǫ́ars drífu hald-Viðurr (sea-horse’s [ship’s] Hǫ́arr’s snow-storm’s [battle’s] holding-Viðurr [sea-warrior]), but ‘sea-horse’s [ship’s] Hǫ́arr’ would mean ‘man’, so that Hǫ́arr would no longer be available in the meaning ‘Óðinn’ to construe the kenning for ‘battle’. Furthermore, a kenning like ‘ship’s battle’ is unattested (Konráð Gíslason, Efterladte skrifter, 1: Forelæsninger over oldnordiske Skjaldekvad (København: Gyldendal, 1895), p. 112). Kock takes Hǫ́ars drífu as apposition to Laufa veðr (Laufi’s [sword’s] storm [battle]). SkP states that ‘apposition involving kennings is rare at best’ (SkP 1, p. 297; Kock, Notationes norroenae, § 2241). This is a doubtful statement and one that is contradicted by the reading of two kennings for ‘ruler’ by the same editors in SkP 1, p. 314, but it is probably true that it is rare or non-existent in kennings for battle, as opposed to kennings for rulers. 190 SkP 1, pp. 309–10.

3.4 Snorri’s Reception of Poetry: Skáldskaparmál

155

This interpretation is complicated by the fact that all poetic examples of locative genitive involving verbs given by Marius Nygaard contain verbs of motion, not rest. It would therefore appear that there was no locative genitive, but only a genitive of direction.191 The kenning eggþings may be construed with the predicate ‘saman fóru’ (came together), which might support the interpretation, but it is still somewhat doubtful whether genitives of direction can occur in kennings. Furthermore, like Hǫ́ars drífu, eggþings designates an activity, not a space.192 For these reasons, it is unlikely that we are dealing with a genitive of direction. There is a less problematic solution, however: in Rekstefja 20 we find an analogous instance with egg and a definer: mækis eggja hregg (the sword’s edges’ storm [battle]), even though eggja hregg would have been enough.193 There is therefore no need to posit the rare phenomenon of genitive of direction. Rather, we have here a kenning with a descriptive or disambiguating element. Finally, stanza 32 has Heðins reikar fúrs folkleikr (Heðinn’s hair-parting’s fire’s [sword’s] army-game [battle]), where folkleikr already means ‘battle’.194 Outside of Vellekla, a stanza by Rǫgnvaldr appears to have the kenning Svǫlnis slíðrvǫndr (Svǫlnir’s (Óðinn’s) sheath-staff [sword]) or Svǫlnis Vára slíðrvǫndr (Svǫlnir’s (Óðinn’s) Vǫ́rs’ (Valkyries’) sheath-staff [sword]), where ‘sheath-staff’ is already a kenning for ‘sword’.195 A stanza preserved in Njáls saga, Kristni saga and the Greater Saga of Óláfr Tryggvason has the kenning hattar hauðrs steði (the hat’s land’s [head’s] anvil [head]), where the redundant element ‘anvil’ is dependent on the surrounding imagery, where the warrior lets his morðhamarr (murder-hammer [axe]) gjalla (ring) against this anvil.196 These examples may suffice to suggest that ‘overdetermined’ kennings might not have been avoided to the same degree by skalds as they are by modern editors. More importantly, it would appear that Einarr used a number of overdetermined

191 Nygaard, Norrøn syntax, p. 152. In prose, some adverbial uses are idiomatic and in some cases no distinction between direction and location is upheld (see examples in Nygaard, Norrøn syntax, p. 153). 192 Such genitives can be found on rare occasions in a limited number of lexemes designating spaces, including hafs (to the sea) in Vellekla 24 (SkP 1, p. 312). 193 SkP 1, p. 920. 194 SkP 1, pp. 323–25. 195 Meissner, Die Kenningar, p. 157; SkP 2, pp. 590–91. 196 Skj B 1, p. 166; A 1, p. 176; Brennu-Njáls saga, ed. Einar Ól. Sveinsson, pp. 260–61; Biskupa sögur, 1, ed. Sigurgeir Steingrímsson et al., 2, pp. 22, 134–35. In Biskupa sögur, the redundant element has led the editors to interpret the head’s anvil as ‘neck’, but the skaldic convention is to cut people in the head, and the expression is clearly based on the surrounding imagery. Some manuscripts have moldhamarr (earth-hammer), which does not seem to make sense and is in any event irrelevant to the present argument. Two have gjalda for gjalla, but this is clearly secondary since it ruins the aðalhending (it does create skothending with mold-, but this line should have aðalhending and mold- is found together with gjalda in one manuscript only; see Skj A 1, p. 176).

156

3 Grammatical Literature

kennings in Vellekla, and Frank’s reading may therefore be adopted even if Boðn is retained as a proper name. Her hypothesis that boðn should here be understood as a common noun, unlike in Snorri’s narrative, therefore has little internal support in Vellekla.

3.4.2 Óðrerir In the case of Óðrerir ‘mind-stirrer’ or ‘poetry-stirrer’, Franks takes this as a descriptive name for the mead of poetry itself. This is a probable ultimate origin of the name, but it is used twice in Hávamál, and one of these instances indicates that the semantic transfer to the vat containing the mead took place at a stage before Snorri. The first time the word is used in Hávamál, it refers either to the drink or to the vat (stanza 107). The text reads: ‘því at Óðrerir | er nú upp kominn | á alda vés jarðar’197 (because Óðrerir has now come up on †the men’s sanctuary’s earth’s†). Scholars do not agree on how to resolve the crux, although they do agree that it is in need of emendation, and suffice it to say they it has been interpreted as Miðgarðr, Ásgarðr or Valhǫll.198 It is clear from this stanza and others around it that what has happened is that Óðinn has brought the mead of poetry from the giant Suttungr to the gods. If the underlying story was similar to the one found in Skáldskaparmál, Óðinn did not bring the container along with the mead, but it remains possible that this poet thought that he did so. It is also possible that the name of the container is here used metonymically for its contents. It therefore remains unclear exactly how the use of Óðrerir in stanza 107 relates to how Snorri used it or to the stories in Skáldskaparmál. In stanza 140 we read: ‘en ek drykk of gat | ens dýra mjaðar | ausinn Óðreri’ (and I got a drink | of the precious mead | poured from / with [that is, by means of] Óðrerir) (stanza 140).199 Here, Óðrerir is clearly perceived as something other than the mead, and more precisely as something from or with which the mead can be

197 Norrœn fornkvæði. Islandsk samling af folkelige oldtidsdigte om Nordens guder og heroer, almindelig kaldet Sæmundar Edda hins fróða, ed. Sophus Bugge (Christiania: Mallings, 1867), pp. 55–56; B. Sijmons and Hugo Gering, Kommentar zu den Liedern der Edda, 2 vols (Halle (Saale): Buchhandlung des Waisenhauses, 1927–31), 1, p. 129. 198 See Sijmons and Gering, Kommentar zu den Liedern der Edda, 1, p. 129–30; Eddukvæði, ed. Jónas Kristjánsson and Vésteinn Ólason, 2 vols. Íslenzk Fornrit (Reykjavík: Hið íslenzka fornritafélag, 2014), p. 343. 199 Norrœn fornkvæði, ed. Bugge, p. 61. On syntax and translation, see Sijmons and Gering, Kommentar zu den Liedern der Edda, 1, p. 151. Heimir Pálsson, ‘Alltaf sama sagan?’, Tímarit Máls og Menningar 71: 4 (2010), pp. 65–79 (p. 71), sees Óðrerir as a kenning for a poet, i.e. ‘[the mead] poured out for the poet’, and this reading could in principle be reconciled with the wording of Vellekla and Hávamál alike. Even though Óðinn is the god of poetry and that art belongs to him, however, he is not elsewhere referred to as a poet within the poetic corpus (see, for instance,

3.4 Snorri’s Reception of Poetry: Skáldskaparmál

157

poured: that is, as some sort of container, as in Snorri’s description. Frank mentions this occurrence, but she does not discuss its implications for her interpretation.200 It does, however, pose serious obstacles to her claim that Snorri invented the interpretation of Óðrerir as a vat, unless one supposes that the passage in Hávamál was composed under the influence of Snorri’s Edda. Snorri himself, however, drew on Hávamál in a form that had a similar beginning and end to the one transmitted in CR (see below pp. 165–66). Furthermore, the presence of alliteration in vr- in stanza 32 suggests that at least the first part of the poem should be dated to before the eleventh century.201 Finally, Gustaf Lindblad sets the ante quem for the exemplars of CR at c. 1240, and he seems to lean towards a date c. 1225–1235, that is, when Snorri was still active and presumably working on his Edda.202 For Frank’s hypothesis to hold, one must therefore assume that a collector – contemporary to and quite likely a personal acquaintance of Snorri – would have used Snorri’s work in order to compose spurious poetry for insertion into the poems he was collecting. I have found no indications that this would be the case, and Snorri’s use of a version of Hávamál that has the same beginning and end as the preserved one makes this unlikely, even if not impossible. Finally, spurious composition would appear to defeat the purpose of collecting poetic sources for grammatical study. By all appearances, then, Hávamál is really an independent witness to the perception of Óðrerir as a vat, showing that Snorri did not invent this idea. Regarding Frank’s interpretation of the two kennings reviewed so far, it remains unclear whether Snorri invented the interpretation of Boðn as a proper name rather than a common noun, whereas Hávamál shows that he did not do so in the case of Óðrerir.

Meissner, Die Kenningar, pp. 252–53, 363–64; also the list of Óðinn’s names in Grímnismál, see Norrœn fornkvæði, ed. Bugge, pp. 85–88). A further obstacle to this interpretation is found in Snorri’s treatment of Óðrerir in Vellekla. Snorri would always, as far as we can tell, try to extract as many concrete and animate referents out of kennings as possible. A shift from ‘poet/Óðinn’ to the vat would run counter to this ‘animate principle’ as well as to Snorri’s near-obsession with Óðinn generally (on the ‘animate principle’, see Frank, ‘Snorri and the Mead of Poetry’, p. 158; Margaret Clunies Ross, ‘Skáldskaparmál’: Snorri Sturluson’s Ars Poetica and Medieval Theories of Language (Odense: Odense University Press, 1987), pp. 16, 30–31, 104–05, 115). Old Norse poetic practice and Snorri’s working methods thus both contradict Heimir Pálsson’s interpretation. Ursula Dronke has picked up on an early suggestion to read ausinn with ek, i.e. ‘I who have been showered with Óðrerir’, but there are no indications that Old Norse culture found other uses in alcoholic beverages than that of drinking them (The Poetic Edda, 3 vols, ed. and trans. Ursula Dronke (Oxford: Oxford University Press, 1969–2011), 3, pp. 31, 57–58, 62; cf. Sijmons and Gering, Kommentar zu den Liedern der Edda, 1, p. 151). 200 ‘Snorri and the Mead of Poetry’, p. 161. 201 Haukur Þorgeirsson, ‘The Dating of Eddic Poetry’, p. 37. 202 Gustaf Lindblad, Studier i Codex Regius av Äldre Eddan (Lund, Gleerup, 1965), pp. 273, 276. Lindblad sees the exemplar of Hávamál as independent from the exemplar(s) of the rest of CR, but he does not posit an individual ante quem for it (Lindblad, Studier i Codex Regius, pp. 263–64).

158

3 Grammatical Literature

3.4.3 Kvasir The last relevant kenning in Vellekla reads, according to Snorri: Kvasis dreyri (Kvasir’s blood [the mythological mead of poetry = poetry]), whereas Frank reads it as kvasis dreyri (ferment’s blood [mead > poetry]). Here, there is a problem in the last semantic transfer, from mead to poetry. Mead figures prominently in kennings for poetry, but it always comes with a definer: ‘Óðinn’s mead’, ‘the mead of the gods’, and so on. In Frank’s reading, this kenning means only ‘mead’, and she therefore assumes either that the definer is provided by the next kenning, giving fjarðleggjar fyrða kvasis dreyra (the fjord-bone’s [stone’s] men’s [dwarves] ferment’s blood [mead > poetry]), or that the context makes it clear that ‘mead’ here means ‘poetry’.203 Frank suggests that the situation in the Óðrerir stanza is similar, where the kenning ǫldrhafs vágr (ale’sea’s wave [ale]) would signify poetry, without further definition. In this instance, however, SkP provides a plausible reading where both kennings for poetry in the stanza have definers.204 Frank also mentions so-called ‘half-kennings’ – kennings which lack one of two definers – but these refer to concrete entities like people, ships and weapons, not to abstract ones like poetry.205 The lack of clear analogues is problematic, since no matter how complex skaldic poetry was, composition and decoding must have been based on convention in order for the poetry to be comprehensible. Frank’s reading thus runs counter to the referential system of skaldic poetry – at least as it has been reconstructed by modern scholars – where ‘mead’ simply means ‘mead’, rather than ‘poetry’. This referential system may perhaps have been somewhat more open-ended than we think: scholars adopt variants and emendations based on their preconceived notions of the system, thus further strengthening it for the next repetition of the same procedure. Specifically, it seems that a number of scribes have seen a pure mead-kenning as a kenning for poetry in a stanza by Hofgarða-Refr. Editors read bǫðgœðis ógnstǫðvar ægir (battle-enhancer’s [Óðinn’s] fear-home’s [breast’s] sea [mead > poetry]), but the manuscripts where the words make any sense at all have bǫðgœði (with some orthographic variation) in apposition with Þorsteini, thus leaving the mead-kenning to signify poetry on its own206 That we should really expect a kenning for poetry here is supported by repetition of this motif in the poem, the coherence of which is corroborated by the repetition of the name Þorsteinn, even though the stanzas have been collected from various parts of Skáldskaparmál. There is thus some evidence, at least on a scribal level,

203 ‘Snorri and the Mead of Poetry’, p. 160. 204 SkP 1, pp. 285–87. Finnur Jónsson’s rendering is discussed by Faulkes, ‘Poetical Inspiration’, pp. 20–21. 205 ‘Snorri and the Mead of Poetry’, p. 160 n. 12; Meissner, Die Kenningar, pp. 74–80. 206 SkP 3, pp. 251–52; Snorri, Skáldskaparmál, ed. Faulkes, pp. 66 (216), 143; Edda Snorra Sturlusonar, ed. Jón Sigurðsson et al., 1, pp. 416–17 (though the interpretation there is problematic).

3.4 Snorri’s Reception of Poetry: Skáldskaparmál

159

that Frank’s reading would be possible, albeit exceptional. Nonetheless, this strikes me as insufficient support for her hypothesis. When Kvasis dreyri is viewed in tandem with the mention of Óðrerir in Hávamál, I believe that the evidence is sufficient to corroborate the fact that Snorri reported older lore about the mythical genesis of poetry. Since the evidence for Són and Boðn is equivocal, the possibility remains that Snorri knew only of a mythological being and of one, rather than three, vats, and that he interpreted the Són and Boðn stanzas creatively in order to produce two more vats. I do not, however, see any compelling reasons to think that this would be the case, and even if it were, this would be a considerably less invasive act than that of inventing the entire complex of the mythological being and the vats. If Frank’s hypothesis falls, so does the strongest case put forward so far for Snorri’s manipulation of his sources in Skáldskaparmál. I have not, for my part, been able to detect any clear indications that Snorri invented major mythological motifs in Skáldskaparmál, at least in cases where he supported his narrative through poetic quotation. The poetic evidence suggests that Snorri has made a relatively truthful report of the kenning system as he perceived it.

3.4.4 Kenning Formation and Narrative Elements In a few cases, however, the presentation of elements belonging to mythical motifs seems to be based on kenning formation, although it is not always clear whether these elements originated with Snorri or earlier. These instances are different from the ones discussed so far in that Snorri does not quote poetry to corroborate his description and may thus have felt himself more entitled to take liberties. In the opening of Skáldskaparmál, we read that Óðinn invited Ægir to a feast where the hall was lit up with shining swords, which served as the only source of light during the evening.207 In the background, we hear the kenning type ‘Óðinn’s fire [sword]’, which is by far the most frequent in sword-kennings defined by Óðinn.208 The gleaming of swords has given rise to several kenning types (‘battle’s fire’ etc.), which suggests that the type is not based on a given narrative, but on metaphor. It would thus appear that Snorri has produced a narrative element from a kenning type. By analogy, one may infer that Snorri has used the same technique later in Skáldskaparmál. Ægir now invites the æsir to come to him, and this time the hall is lit up with shining gold (lýsigull) ‘svá sem í Valhǫllu váru sverðin fyrir eld’ (just as the swords were used as fire in Valhǫll).209 Since

207 Snorri, Skáldskaparmál, ed. Faulkes, p. 1. 208 For this type, see Meissner, Die kenningar, p. 157. 209 Snorri, Skáldskaparmál, ed. Faulkes, pp. 40–41.

160

3 Grammatical Literature

Ægir is also a name for the sea and since gold served as fire at this feast, Snorri explains, we get the kenning type ‘sea’s fire [gold]’. This time, the connection between the kenning and the mythical motif is explicit, and again, the description is in all likelihood based on the kenning type, rather than the way around. Exactly why gold is called ‘water’s fire’ is unclear, but may perhaps have to do with the story of the Rhine gold.210 Other features in Snorri’s description of the mead of poetry than Kvasir and the three vats may indicate his working methods. One such instance is discussed by Frank, namely the detail that a portion of the mead came out through the rear end of Óðinn in an eagle’s shape. Frank suggests that Snorri invented this motif and that this passage has given rise to the kenning arnar leir(r) ‘eagle’s clay [bad poetry]’, which is found three times in the skaldic corpus.211 Her argument requires one stanza dated to c. 1120 to be spurious and that another, dated to 1220, be recognized as depending on Skáldskaparmál. The second stanza is found in Íslendinga saga and relates to the circumstances that gave rise to a feud with the Southerners. It cannot easily be dismissed as a composition by Sturla Þórðarson, author of the saga, and this in turn makes it unlikely that it is composed under the influence of Skáldskaparmál at such an early date. Furthermore, the actual wording arnar leir is found only in U in the phrase ‘ok heitir arnar leir’ (and that is called eagle’s clay), which disrupts the narrative and looks like an adaptation of the text to skaldic instruction. This suggests that it may have been added to, rather than taken from, Skáldskaparmál, although it cannot on principle be ruled out that the phrase is a remnant from an early draft.212 Finally, a kenning by Auðunn illskælda (late ninth–early tenth century), Hǫ́ars lǫggvar kleppdǫgg (Hǫ́ar’s [Óðinn’s] cask-rim’s lump-dew [dregs; bad poetry]), suggests that parodic kennings referring to the myth of poetry may have a long pedigree.213 Such kennings have an analogue in parodic kennings for people, such as ostmýgir (cheese-subduer), meaning something like ‘bad warrior’ and based on kennings

210 Cf. Meissner, Die Kenningar, p. 225, though he remarks that the Rhine is not prominent in such kennings. Given the skalds’ propensity to use a variation of heiti, however, that is perhaps not unexpected. 211 Frank, ‘Snorri and the Mead of Poetry’, pp. 168–69. 212 Edda Snorra Sturlusonar, ed. Finnur Jónsson, p. 85. Lasse Mårtensson has found traces of two strata of exemplars for U, one from the early thirteenth century and one from the middle or slightly later (Skrivaren och förlagan. Norm och normbrott i Codex Upsaliensis av ‘Snorra Edda’ (Oslo: Novus forlag, 2013), pp. 263–66). Some of the differences between U and other manuscripts may thus go back to an early date. 213 SkP 3, pp. 18–19; SkP 1, p. 120. Kock, Notationes, § 1006, sees a positive kenning here, but he does not consider that the stanza breaks the normal poetic convention by asking the poet, rather than the audience, to be quiet. The negative reading of SkP thus has contextual support. On the conventional bid for a hearing, see Cecil Wood, ‘The Skald’s Bid for a Hearing’, Journal of English and Germanic Philology, 59 (1960), pp. 240–54.

3.4 Snorri’s Reception of Poetry: Skáldskaparmál

161

like Vinða mýgir (the Wends’ subduer [warrior]).214 The use of parodic kennings appears to be based on conventions within the kenning system rather than on narrative (there are, for instance, no stories about warriors fighting cheese). The kind of kennings that were parodied in arnar leir may perhaps be gleaned from Egill’s kenning arnar kjapta ǫrð (‘the eagle’s jaws’ [the beak’s] seeds [vomit = poetry]’). Several factors thus suggest that the first thing to happen was that positive, eagle-based kennings for poetry were parodied in new kennings, and that the narrative element of the eagle’s droppings evolved out of these. This last step may have been taken by Snorri, but it is also possible that it happened earlier, for instance among the learned poets of the twelfth century. It therefore remains an open question whether Snorri has here invented a narrative element based on kenning formation. Someone has, however, and whoever this person was, his action bears witness to the methodology behind the evolution of the mytho-poetic landscape. In contrast to Frank’s interpretation, we would not be dealing here with a pure invention, but with a transformation of kenning structures into narrative elements. These elements would not amount to complex motifs in their own right, since two kennings in Egill’s poetry indicate that the motif of Óðinn in the form of an eagle and vomiting the mead of poetry was known in the tenth century (see below pp. 254–55). Snorri would therefore have added at most a detail to a pre-existent motif. Finally, another possible transformation of kennings into narrative involving the mead of poetry concerns the kenning type ‘dwarves’ ship [poetry]’. Snorri in his narrative reports that the dwarves were abandoned on an islet soon to be swept by the tide and that they paid their way out of there with the mead, which thus functioned as their ‘ship’.215 Somewhat later, however, he says that we are here actually dealing with ofljóst: Enn er kallaðr skáldskaprinn far eða lið dverganna; líð heitir ǫl ok lið heitir skip. Svá er tekit til dœma at skáldskapr er nú kallaðr fyrir því skip dverga.216 Poetry is also called the vessel or ship of the dwarves; líð is beer and lið is a ship. This illustrates that poetry is now for that reason called ‘dwarves’ ship’.

Here, Snorri explains that there are phonetic reasons for the kenning formation (líð (beer) > lið (ship)) and suggests that such kennings really do not refer to ships at all, and he then quotes an anonymous stanza, the only one to have come down to us with a ‘dwarves’ ship’-kenning. The stanza reads:

214 Meissner, Die Kenningar, pp. 365–67. Meissner informs us that the kenning ostmýgir is ‘sehr lustig’ (p. 366). 215 Snorri, Skáldskaparmál, ed. Faulkes, pp. 3–4. 216 Snorri, Skáldskaparmál, ed. Faulkes, p. 14, cf. p. 109.

162

3 Grammatical Literature

Bæði ák til brúðar bergjarls ok skip dverga sollinn vind at senda seinfyrnð gǫtu eina.217 I am in possession of both, to send in the same direction: the mountain-earl’s [giant’s] wife’s [giant woman’s] swollen wind [thought] and the slowly forgotten dwarves’ ships [poetry]. . .

The typical context when skalds express that they send their thoughts in some direction is in erotic poetry, so that is probably what we have here. We have seen comparable examples of ofljóst where, for instance, the dwarf name Litr does not refer to a dwarf, but to the looks (litr) of a beautiful woman (p. 91). That example differs from the pair lið/líð in that litr/Litr are true homonyms, whereas lið/líð are strictly speaking paronyms, since they differ in vowel length. Different vowel length (as opposed to quality), however, appears to have been seen as permissible in ofljóst, to judge by the pairs presented in the discussion of ofljóst in Skáldskaparmál (far/fár, hlið/hlíð, lið/ líð), as well as by the stanza by Einarr Skúlason analyzed on p. 88 above. It may be that this possibility developed when the skalds had been exposed to writing, since a literate person would be more likely than an illiterate to see long and short /i/ as the same, based on their realization in script. If so, the stanza should probably be dated to the twelfth century. Of Snorri’s two explanations of the kenning type ‘dwarves’ ship’, it would appear that the mythical one was primary, at first glance. If ‘ship’ is simply ofljóst for ‘beer’, there is no obvious reason why the use of this ofljóst should be restricted to dwarves, unless there was some underlying narrative involving ships and dwarves. The matter is, however, complicated by Litla Skálda, which opens: ‘Skáldskapr er kallaðr skip dverga ok jǫtna ok Óðins [. . .]’218 (poetry is called the ship of dwarves and of giants and of Óðinn [. . .]). Here, unlike Snorri’s description, the ship belongs to all of these mythical beings. The first part of Litla Skálda gives only rules of kenning formation without explaining them, and it therefore remains unclear to what extent the author had particular explanations in mind. If he did, the reason that the ship can belong to several categories of beings is presumably that the ship is really not a ship (lið) at all, but a drink (líð), and one which did indeed at one point in time or other belong to them all. I see no way of testing whether this was his understanding, however. Ship-based kennings for poetry are extremely rare according to Skj, Meissner and SkP, and I would argue that the example quoted by Snorri is actually the only occurrence in the corpus. One more ‘dwarves’ ship’ kenning is found, according to Finnur Jónsson and Meissner, in Hallfreðr’s Austra burar nǫkkvi (Austri’s (a dwarf’s) son’s [a dwarf’s] ship).219 As shown on p. 81, however, this line is based on the

217 SkP 3, p. 512; Snorri, Skáldskaparmál, ed. Faulkes, p. 14. 218 Edda Snorra Sturlusonar, ed. Finnur Jónsson, p. 255. 219 Cf. Meissner, Die Kenningar, p. 428.

3.4 Snorri’s Reception of Poetry: Skáldskaparmál

163

corrupt reading of Möðruvallabók, and if that scribe understood the kenning as Finnur Jónsson and Meissner do, this is probably due to influence from Snorri’s Edda. This kenning is thus irrelevant for the situation before the composition of Snorri’s Edda. The dwarves’ ship found in a stanza by Egill is based on an emendation to which there is strong contrary evidence (see below p. 229 n. 122). Bjarni Aðalbjarnarson has hypothesised another instance in Bandadrápa 3, but this is convincingly rebutted in SkP.220 If one includes giants and Óðinn as definers, Meissner also lists berg-Saxa fley (the mountain-Saxons’ [giants’] ship [poetry]) and hertýs víngnóð (the battle-god’s [Óðinn’s] wine-ship [vat]), but as noted above, many factors indicate that fley should not be construed with berg-Saxa, and although Gnóð is a heiti for a ship, víngnóð is a kenning for a vat, not a ship.221 Only dwarves are thus plausibly attested in shipbased kennings for poetry, and only in one kenning. Litla Skálda’s description, which

220 Heimskringla, ed. Bjarni Aðalbjarnarson, 1, pp. 250–51; SkP 1, p. 460. 221 This kenning is peculiar in that víngnóð remains a kenning for ‘vat’, although one would normally expect the meaning of the element vín- to be changed by previous definers, so that ‘Óðinn’s wine’ would mean ‘poetry’. ‘Poetry’s ship ’, however, makes little sense, and we must conclude that the base-word is an independent kenning. The situation is similar in stanza eleven of Vellekla, where we find Lopts vinar vínheimr (Loptr’s [Loki ’s] friend’s [Óðinn’s] wineworld [vat]) (SkP 1, p. 297). Here, ‘wine-world’ still means ‘vat’. It would appear that ‘Óðinn’s wine’ in these instances means not ‘poetry’, but specifically ‘mead of poetry’, whose ship or world is the specific vat in which it was contained. ‘Wine’ thus retains its meaning, although it is used metonymically for ‘mead’. A similar structure may account for Eyvindr skáldaspillir’s kenning galga farms hverlǫgr (the gallow’s cargoe’s [Óðinn ’s] pot-liquid [mead > poetry]) (SkP 1, p. 197). ‘Pot-liquid’ is a conventional kenning for mead or beer (Meissner, Die Kenningar, pp. 41, 432). ‘Óðinn’s pot’ here seems to denote one of the pots from which Óðinn drank the mead or into which he vomited it. Like ‘wine’, ‘pot’ here retains its meaning, but it is a pot belonging to Óðinn specifically, and the expression ‘pot-liquid’ works according to ordinary kenning dynamics. This also appears to be how Snorri interprets the kenning: ‘Kvasis blóð var lǫgr í Óðreri áðr mjǫðrinn væri gjǫrr […] er hann [mjǫðrinn] kallaðr fyrir því hverlǫgr Óðins’ (Kvasir’s blood was liquid in Óðrerir before the mead was made […] the mead is therefore called Óðinn’s pot-liquid) (Snorri, Skáldskaparmál, ed. Faulkes, p. 14). This interpretation is supported by the stanza, where the kenning refers to the poem in which it occurs, but where Óðinn is also said to have carried the liquid in question when flying from the giants – that is, we are dealing with liquid in a concrete sense (SkP 1, p. 197). Another example of this kind of structure is found in stanza eleven of Haraldskvæði. The kenning is Sváfnis salnæfrar (Sváfnir’s [Óðinn ’s] hall-birch-barks [roof-shingles = shields]), and Snorri quotes it as a witness to the perception that Óðinn’s hall, Valhǫll, had a roof of shields (SkP 1, p. 106; Snorri, Prologue and ‘Gylfaginning’, ed. Faulkes, p. 7). Here the definer ‘Sváfnis’ indicates the specific hall in question, and the meaning ‘hall’ is retained for the kenning ‘hall-birchbarks [roof-shingles]’. These shingles become shields only because they are found on the roof of this particular hall. Kennings with an independent kenning as base-word are relatively rare (see Meissner, Die Kenningar, pp. 41–42; of the kennings discussed in this note, Meissner has only noted galga farms hverlǫgr. The instance noted in SkP 3, p. 448, is wrong – in that case, ǫl-Gefn ‘beer-Gefn [woman]’ is a fully independent kenning, and hundr is the base-word of the other kenning). Almost without exception, the kenningas-base-word consists of a compound, a structure which signals its semantic integrity.

164

3 Grammatical Literature

does not restrict the ship to dwarves, and the tradition of ofljóst, which is based on phonetics rather than narrative, together suggest that Snorri has added the element of the ship to his narrative based on the reference in the stanza which he quotes. In these four examples from Skáldskaparmál – the illumination at two feasts, the eagle’s droppings and the ship of the dwarves – elements within larger mythical motifs appear to derive from kenning formation. The mythologization of kenning types may have appeared before Snorri, and the instances rehearsed above may therefore not all have originated with him. It seems likely, however, that he made the most of these possibilities. His use of these poetic resources had some narrative consequences, in that he provided a richer depiction of feasts, dwarves and eagle’s droppings than would otherwise have been the case, but its effect was not as drastic as that of inventing the man Kvasir and the three mead vats outright, as in Frank’s interpretation. Even if one assumes that Snorri invented all narrative elements that the evidence suggests that he could have, including Són and Boðn, this still only amounts to adding details to pre-existent motifs. In this analysis, Snorri thus comes out as a more trustworthy witness to the mythical background of poetic diction than he is sometimes credited with having been. He added details and colour, but not much more.

3.5 Snorri’s Reception of Poetry: Gylfaginning In Gylfaginning, not treated by Frank, Snorri is using rather than describing poetry, as he does in Skáldskaparmál. He seems to feel correspondingly more at liberty to interpret it according to the logic of his own narrative. Gylfaginning gives an overview of the mythological world and of mythological history from beginning to end. Three eddic poems – Vǫluspá, Vafþrúðnismál and Grímnismál – are woven into the narrative and serve as its main sources. As we saw in the previous section, Litla Skálda probably provided him with a rudimentary model for mythological prose narrative with some eddic poetry quoted for corroboration. As an eddic, mythographic prosimetrum, however, Gylfaginning is vastly more elaborate than Litla Skálda, and it is by far the most informative witness to how eddic poetry was received and interpreted in the thirteenth century. This aspect of Gylfaginning has received only scattered attention. Given that eddic poetry is such a large sub-field within Old Norse studies and recent scholarly focus on reception, the lack is somewhat remarkable. I wish here to focus on Gylfaginning as a source to eddic reception, but in other regards, in this book I treat eddic poetry mainly as a point of comparison. There are a number of reasons for this: first, eddic poetry is notoriously difficult to date; second, it does not display certain features, like hendingar, that are important for my analyses, or it displays such features, like kennings, only rarely; third, the collections of eddic poetry lack authorial voice or most

3.5 Snorri’s Reception of Poetry: Gylfaginning

165

other indications as to what their function is, at least as compared to other grammatical texts. I argued on pp. 107–09 that the eddic compilations belong to the grammatical corpus, and they appear to have been collected in the decades before and around 1270 at the latest, and probably some decades earlier.222 Since it is possible to follow the development of grammatical interest from the twelfth through the thirteenth century, and since eddic poetry appears to have been written down in order to accompany grammatical texts, one may conclude that eddic compilations would in all likelihood never have come into existence if it were not for the new and systematic interest in certain poetic features. Allusions to the eddic mythological world can be found already in the earliest datable grammatical text, 1GT (c. 1150), where we read: ‘Hǫ́ dó, þá er Hǫlgatroll dó, en heyrði til hǫddu, þá er Þórr bar hverinn’223 (A tall [woman] died when the Hǫlgatrǫll [Þorgerðr hǫlgabrúðr] died, and one could hear the [sound of the] handle, when Þórr carried the kettle). Here, we recognize the description of Þórr carrying the kettle which we know from stanza 34 of Hymiskviða, including the detail of the handles which, according to Hymiskviða, ‘skullu á hælum’ (banged around his heels).224 The mythological reference appears to point outside of the learned discourse, into a field of lore as yet largely oral.225 This impression is corroborated by the lack of any mention of poetry among the types of texts that the First Grammarian says had been committed to writing in his day, as well as by his reference to how people would scan a particular half stanza before he wrote it down.226 A comparison between Gylfaginning and CR suggests that eddic poetry remained within the oral sphere for some time and that large-scale writing down of eddic poetry commenced under the influence of Snorri’s Edda. This claim requires some elaboration. Gylfaginning is the first datable collection of eddic poetry. As such, its treatment of its poetic sources may have some bearing on our understanding of the impetus behind later collections. For this purpose, it is important to note that although three eddic poems – Vǫluspá, Vafþrúðnismál and Grímnismál – are the main visible sources in Gylfaginning, another poem fulfils an equally important function, although the poem itself remains largely invisible. 222 The two main manuscripts are CR and AM 748 Ia 4to, dating to c. 1270 and c. 1300–1325, respectively. Gustaf Lindblad’s investigation of CR indicates that it had more than one exemplar and that these were no younger than c. 1240, suggesting that collections of mythological and heroic poetry, though perhaps not as large as CR, may have existed in the first half of the century (Lindblad, Studier i Codex Regius, pp. 273–76). 223 The First Grammatical Treatise, ed. Hreinn Benediktsson, p. 244. 224 Norrœn fornkvæði, ed. Sophus Bugge, p. 111. 225 Cf. The First Grammatical Treatise. The Earliest Germanic Phonology. An Edition, Translation and Commentary, ed. Einar Haugen, 2nd ed. (London: Longman, 1972), p. 77. 226 The First Grammatical Treatise, ed. Hreinn Benediktsson, p. 226.

166

3 Grammatical Literature

The use of Hávamál as a framing device for the discourse of the High One in Gylfaginning (Hávamál means ‘the speech of the High One’) has attracted little scholarly interest. Gylfaginning quotes the first stanza when the discourse begins and paraphrases the last when it ends. The paraphrase is found in the words ‘ok njóttu nú sem þú namt’ (and now make use of what you have learned).227 Although these words are lacking in U,228 they are present in R, W and T. The exemplar of T can, through a number of morphological and orthographical parameters, be dated perhaps as early as the mid-thirteenth century.229 Even in U, Gylfaginning amounts to something of a paraphrase of Hávamál, since both are didactic discourses presented by Hárr/Hávi (although he has become a Trinitarian pastiche in Gylfaginning). This means that while CR begins with the sequence of poems Vǫluspá, Hávamál, Vafþrúðnismál and Grímnismál, Gylfaginning quotes Vǫluspá, Vafþrúðnismál and Grímnismál as its main sources and uses Hávamál as a poetic backdrop for the entire narrative. The poetic overlap in the first sections of the two works is thus complete and this can hardly be coincidental. Furthermore, Snorri’s use of Hávamál presupposes that the reader be well acquainted with it and it therefore seems reasonable to suppose that the first part of CR works as a textbook to be used in tandem with Gylfaginning, although much additional poetry was also gathered into the collection. This chimes well with Gustaf Lindblad’s dating of the earliest exemplar of CR, before or around 1240.230 The somewhat later collection of eddic poetry AM 748 Ia 4to (c. 1300–1325) has a different and, it would appear, less coherent arrangement of poems. This is, however, the sister of the A manuscript of Skáldskaparmál, and the connection to Snorri’s Edda is thus abundantly clear.231 It thus seems likely that we, in an oblique way, have Snorri to thank also for the collection of the eddic poems.232 The development of eddic poetry before the thirteenth century is much more unclear than that of skaldic poetry.233 Eddic poetry is typically anonymous and lacks rhymes and a number of additional features that are helpful for dating skaldic poetry. The methodology for dating eddic poetry needs to be further developed in order to allow for firmer conclusions, and in particular, the datable skaldic corpus should be taken into account to test the hypotheses. Thus, for instance, Haukur Þorgeirsson has through such comparison recently demonstrated that alliteration in vr- as opposed to r- was never used after the end 227 Snorri, Prologue and ‘Gylfaginning’, ed. Faulkes, pp. xxiv–xxv, 8, 54. 228 For a discussion of this, see, for instance, John Lindow, ‘Two Skaldic Stanzas in Gylfaginning: Notes on Sources and Text History’, Arkiv för nordisk filologi, 92 (1977), pp. 106–24. 229 See Codex Trajectinus. The Utrecht Manuscript of the ‘Prose Edda’, ed. Anthony Faulkes. Early Icelandic Manuscripts in Facsimile, 15 (Copenhagen: Rosenkilde and Bagger, 1985), pp. 19–20. 230 See p. 157. 231 See Guðrún Nordal, Tools of Literacy, pp. 58–59. 232 Faulkes suggest this in general terms; see Snorri, Prologue and ‘Gylfaginning’, ed. Faulkes, p. xvi. 233 For an overview, see Fidjestøl, The Dating of Eddic Poetry.

3.5 Snorri’s Reception of Poetry: Gylfaginning

167

of the tenth century, when initial vr- had become r- in Icelandic, and that medieval scribes failed to understand the phonetic background of the phenomenon.234 Earlier claims that alliteration in vr- might be due to archaization are thus strongly contradicted by the sources, and such alliteration now appears to be a useful dating criterion. This is relevant here, since both Hávamál and Vafþrúðnismál display one instance each of unambiguous alliteration in vr-, and since the independent criterion of a high incidence of the expletive particle of/um also supports that these poems predate the eleventh century.235 In the same vein, Haukur Þorgeirsson has revisited the old criteria of breaks to the V2 principle in unbound clauses and use of the particle of/um, to test whether these yield the same results when they are used in tandem as when they are applied individually. It turns out that they do, which means that these criteria are all the more likely to be useful, but also that Vǫluspá is likely to belong to the early eddic poems.236 Hávamál, Vafþrúðnismál and Grímnismál are composed in ljóðaháttr and therefore not included in Haukur Þorgeirsson’s analysis. Much work remains to be done, however, in order to arrive at a more secure dating of the eddic poems. In this context I shall not attempt further investigations along the promising, comparative lines suggested by Haukur Þorgeirsson’s studies. Rather, in keeping with my present focus on Snorri’s reception, a few words on his approach to these poems may suffice. Snorri clearly presents the poetry as old, most notably perhaps when introducing a stanza with the words ‘svá er hér sagt í orðum Ásanna sjálfra’ (as it is said here in the words of the gods themselves).237 Likewise, Snorri’s nephew Óláfr argued forcefully that the poetry of the euhemerized gods had actually been preserved in the Old Norse tradition (see below pp. 178–80). On this as on many other points, he probably shared the outlook of his uncle, with whom he had spent much time and with whose work he was evidently well acquainted.238 The assertions of the two grammarians may seem extravagant, but they are not more far-fetched than belief in the prophecies of the patriarchs or of the Sibyl, which served to authenticate Christian truth. I do not, therefore, think that these claims should be rejected out of hand. Gylfaginning presents pagan lies and misconceptions, but not everything in it was seen as false. The gods, for instance, had actually existed, but they were not gods. Their poetry, it may be inferred, also existed, and their falsehood was not one of dating, but of content. From a literary point of view, Gylfaginning is a complex text, not least with regard to its various levels of pagan lies. Even so, Snorri appears to have used his

234 Haukur Þorgeirsson, ‘The Dating of Eddic Poetry’. 235 Haukur Þorgeirsson, ‘The Dating of Eddic Poetry’, pp. 37, 54. 236 Haukur Þorgeirsson, ‘Late Placement of the Finite Verb in Old Norse Fornyrðislag Meter’, Journal of Germanic Linguistics, 24 (2012), pp. 264–65. 237 Snorri, Prologue and ‘Gylfaginning’, ed. Faulkes, p. 34. 238 Guðrún Nordal, Tools of Literacy, p. 181.

168

3 Grammatical Literature

ordinary method of establishing historical truth – in this case the historical truth of the pagan lies – through quotation of poetry. It would therefore seem likely that he really thought that the eddic poems were old and that they could for that reason serve as the backbone of his narrative. The matter is, however, somewhat complicated by the fact that Snorri quotes Hyndluljóð, which he refers to as Vǫluspá in skamma (the short Vǫluspá) in much the same way as he quotes Vǫluspá (‘svá segir í X’ (as it says in X); ‘svá er sagt í X’ (as it is said in X)).239 Several factors indicate that Hyndluljóð was composed in the twelfth century. It exhibits no alliteration in vr-, which would have placed the poem in the period before c. 1000. The combined ratio of breaks to the V2 principle in unbound clauses and use of the particle of/um in Hyndluljóð places it well below Vǫluspá, but Hyndluljóð contains more of these features than Gunnlaugr Leifsson’s Merlínuspá (c. 1200).240 Given that the period c. 1015–1120 is unlikely for composition of a heavily mythologizing poem like Hyndluljóð, composition around the middle of the twelfth century seems likely. This is also contextually plausible, given the mythological revival at the time. The words ‘æ trúði Óttarr | á ásynjur’ (Óttarr always believed in the ásynjur) in stanza 10 may suggest a Christian perspective, since belief is not an issue in sources that can plausibly be considered pagan. Unlike Vǫluspá, there is very little unique mythological material in Hyndluljóð, and it may all or nearly all have been gleaned from older poems (Vǫluspá and Grímnismál in particular, but the poet also appears to have used Eiríksmál and Eyvindr skáldaspillir’s Hákonarmál, both from the tenth century) and historical lore.241 Hyndluljóð exhibits an encyclopedic tendency, gathering the major royal families with a divine background into the lineage of the otherwise unknown Óttarr

239 The designation ‘short’ has caused some concern among scholars, since the preserved Hyndluljóð is 50 stanzas long. Vǫluspá is either 62 (R) or 59 (H) stanzas long. This is longer than Hyndluljóð, but perhaps not long enough for Hyndluljóð to merit the epithet ‘short’. Scholars have therefore suggested that Snorri was referring only to the mythological stanzas 29–44, and that these have been inserted into a genealogical poem at a later point (Kommentar zu den Liedern der Edda, 3: Götterlieder, ed. Klaus von See et al. (Heidelberg: Carl Winter, 2000), pp. 687–88, 773–74). This suggestion poses some problems, however, the main one being that this short poem would have no clear beginning and no frame presenting the seeress, which is the reason for the name Vǫluspá (the prophesy of the vǫlva) in the first place. Furthermore, all other poems presenting mythological and historical lore, like Hyndluljóð does, have a frame comparable to that found in Hyndluljóð (Havamál lacks a frame, but it begins as a collection of proverbs, not of historical lore). It is therefore likely that the poem was always long, and the reason for Snorri’s designation must be sought elsewhere: the part of the poem that matches Vǫluspá is indeed stanzas is stanzas 29–44, and this section of 16 stanzas may have given rise to a designation for the poem as a whole, or the designation may have been used for this section only (cf. Kommentar zu den Liedern der Edda, 3, ed. Klaus von See et al., pp. 687–88, 773–74, but note that the editors there emphasise the stylistic coherence of Hyndluljóð more than the implications of its frame). 240 Haukur Þorgeirsson, ‘Late Placement of the Finite Verb in Old Norse Fornyrðislag Meter’, p. 264. 241 Kommentar zu den Liedern der Edda, 3, ed. Klaus von See et al., p. 687.

3.5 Snorri’s Reception of Poetry: Gylfaginning

169

heimski. It does not have the linearity of earlier genealogical poems, but rather, it is a kind of mythological and legendary ‘best of’. In this regard, it bears comparison to Háttalykill (c. 1140) and Íslendingadrápa (second half of the twelfth century), which gather much historical material into one poetic compendium. Málsháttakvæði (c. 1200) also collects much seemingly random mythological lore. The unknown and presumably non-existent addressee, Óttarr ‘the stupid’, supports that we are dealing with learned speculation. In general, historical poetry had its greatest flowering in the twelfth century, before the sagas came to fulfil that role (see below p. 195). The twelfth century thus appears as a plausible date of composition for Hyndluljóð. Is it likely that Snorri saw nothing of this, and that he thought of Hyndluljóð as equally old as Vǫluspá? This question is probably impossible to answer, but at least the quotation of Hyndluljóð suggests that Snorri’s use of a poetic text in Gylfaginning cannot be taken as a strong indication of its high age. Whatever Snorri’s real opinion was, however, and no matter the actual date of composition of the eddic poems, it is clear that their contents are generally not derived from Latin tradition in any straightforward way.242 They are based on traditional material, and they do not lend themselves more easily to Christian interpretation than, say, classical poets such as Ovid or Vergil.243 They thus demanded a treatment as pre-Christian sources and that may suffice for an analysis of Snorri’s working methods. In the following, four examples of creative readings are analysed to give an impression of how Snorri’s interpretations and selections informed the framing and narrative of Gylfaginning. In all cases, he has availed himself of the poetry in such a way as to align Old Norse mythology with Christian motifs. His preferred method seems to have been to draw on poetry, rather than inventing motifs outright. On one hand, Snorri appears to have been anxious not to present any overtly Christian material in the pre-Christian setting of Gylfaginning. On the other hand,

242 Christian parallels have been adduced in the case of Vǫluspá in particular, but at least to my mind, a clearer methodology for comparison has yet to be formulated and a likely context for composition has yet to be proposed (Gro Steinsland, ‘Vǫluspá and the Sibylline Oracles with a Focus on the ‘Myth of the Future’’, in The Nordic Apocalypse. Approaches to Vǫluspá and Nordic Days of Judgement, ed. Terry Gunnell and Annette Lassen (Turnhout: Brepols, 2013), pp. 147–60; Karl G. Johansson, ‘Vǫluspá, the Tiburtine Sibyl, and the Apocalypse in the North’, in The Nordic Apocalypse, pp. 161–84; Pétur Pétursson, ‘Manifest and Latent Biblical Themes in Vǫluspá’, in The Nordic Apocalypse, pp. 185–201; see further references in these articles). Matthias Egeler, ‘Eikþyrnir and the Rivers of Paradise. Cosmological Perspectives on Dating Grímnismál 26–28ʹ, Arkiv för nordisk filologi, 128 (2013), pp. 17–39, argues that the hart with rivers flowing from its horns in Grímnismál has been influenced by Christian images and that at least this part of the poem should be dated to the twelfth or early thirteenth century. 243 Thus also Preben Meulengracht Sørensen, ‘Om eddadigtenes alder’, in At fortælle Historien – Telling History. Studier i den gamle nordiske litteratur – Studies in Norse Literature, ed. Preben and Sofie Meulengracht Sørensen (Trieste: Edizioni Parnaso, 2001 [1991]), pp. 143–50.

170

3 Grammatical Literature

he also went to great lengths to present pre-Christian religion as a parallel to Christian history. Perhaps the most striking example, as argued by Anne Holtsmark and others, is his insertion of the Flood story into the narrative, based on a stanza that mentions nothing of the sort (Vafþrúðnismál 35).244 The passage merits quoting in full: Þá svarar Hár: ‘Synir Bors drápu Ymi jǫtun. En er hann fell, þá hljóp svá mikit blóð ór sárum hans at með því drektu þeir allri ætt hrímþursa, nema einn komsk undan með sínu hýski. Þann kalla jǫtnar Bergelmi. Hann fór upp á lúðr sinn ok kona hans ok helzk þar, ok eru af þeim komnar hrímþursa ættir, svá sem hér segir: Ørófi vetra áðr væri jǫrð skǫpuð þá var Bergelmir borinn; þat ek fyrst of man er sá hinn fróði jǫtunn á var lúðr of lagiðr.’245 Then Hár answers: ‘The sons of Borr killed the giant Ymir. When he fell, so much blood gushed forth from his wounds that with it they drenched the entire race of frost-giants, except one, who got away with his household. That one the giants call Bergelmir. He and his wife climbed up onto his lúðr and hung on to it, and from them come the races of frost-giants, as described here: Innumerable winters before the creation of the earth then Bergelmir was born; the first thing I remember was when that wise giant was laid on a lúðr.’

I have left the word lúðr untranslated. It can mean, among other things, ‘cradle’ or ‘coffin’. In the stanza above, then, the first thing the speaker (Óðinn) remembers is that the proto-giant Bergelmir was laid in (lit. ‘on’) his cradle or coffin.246 Lúðr, however, can also mean ‘box’, and this meaning it shares with the word ǫrk. The latter word typically designates the Ark of the Covenant or the Ark of Noah.247 244 See Anne Holtsmark, ‘Det norrøne ord lúðr’, Maal og Minne (1946), pp. 53–54, and references there. See also Heimir Pálsson, ‘Að sætta heimildir. Lítið eitt um starfsaðferðir Snorra Sturlusonar’, in Heiðin minni. Greinar um fornar bókmenntir, ed. Haraldur Bessason and Baldur Hafstað (Reykjavík: Heimskringla, 1999), p. 165. 245 Snorri, Prologue and ‘Gylfaginning‘, ed. Faulkes, p. 11. 246 Holtsmark argues that the word should here be understood as ‘coffin’ and that the meaning ‘cradle’ should be deleted from dictionaries (‘Det norrøne ord lúðr’, p. 65). For the present argument, ‘cradle’ and ‘coffin’ are equivalent. 247 Ǫrk is a loan from Latin arca. Finnish has borrowed the word from proto-Scandinavian in the form arkku, which suggests a loan before c. 700 (see de Vries, Altnordisches Etymologisches Wörterbuch, p. 688).

3.5 Snorri’s Reception of Poetry: Gylfaginning

171

This is clearly the meaning in Snorri’s prose introduction to the stanza, where Noah’s Flood is given a pre-Christian equivalent. To arrive at this point, Snorri pushes the etymological method to the limits even of that remarkably flexible branch of learning.248 Why, then, did Snorri not simply use the word ǫrk in his description of the Flood? For one thing, Snorri had to retain the word from the stanza if he was to quote a supposedly pre-Christian testimony to Noah’s Flood. For another, although he could have clarified that lúðr is here to be understood as ǫrk, Snorri’s conceit in Gylfaginning is one of presenting no material with an obviously Christian background. The pagan cosmos and history of Gylfaginning is a world onto itself. Snorri’s interpretation does considerable violence to the stanza. The Flood is not mentioned there and the giant is laid on the lúðr, rather than climbing onto it. The stanza clearly had nothing to do with a flood before Snorri directed its interpretation that way, and this is borne out by Vafþrúðnismál, from where Snorri quoted the stanza. Here, as often in medieval textual culture, context was not allowed to stand in the way of an attractive interpretation. The next example is a quotation of Grímnismál stanzas 46–50 and 54, listing the many names of Óðinn, 49 in all. Again, though more subtly in this instance, the interpretation is slanted to align it with a Biblical motif. Gylfi is baffled by the many names, and Hár replies: Mikil skynsemi er at rifja vandliga þat upp. En þó er þér skjótast at segja at flest heiti hafa verit gefin af þeim atburð at svá margar sem eru greinir tungnanna í verǫldunni, þá þykkjask allar þjóðir þurfa at breyta nafni hans til sinnar tungu til ákalls ok bœna fyrir sjálfum sér, en sumir atburðir til þessa heita hafa gerzk í ferðum hans ok er þat fœrt í frásagnir, ok muntu eigi mega fróðr maðr heita ef þú skalt eigi kunna segja frá þeim stórtíðindum.249 It is very wise to rehearse it diligently. But what I can tell you in a hurry is that most names have been given because of the circumstance that as many as the distinctions of languages are in the world, then all peoples find it necessary to change his name into their own language to invoke him or to pray for themselves, but some events that have given rise to these names have taken place in his travels and are reported in stories, and you will not be called a wise man unless you can recount these great events.

248 Medieval etymology was based on a liberal use of homonyms and paronyms, but this strategy was normally not combined with the use of synonyms, as in Snorri’s lúðr (coffin) > lúðr (box) > ǫrk (box) > ǫrk (the Ark). As Anne Holtsmark suggests, this interpretive strategy may be inspired by the poetic figure ofljóst (‘Det norrøne ord lúðr’, p. 53; for a description and discussion of ofljóst, see above pp. 44–45). 249 Snorri, Prologue and ‘Gylfaginning’, ed. Faulkes, p. 22.

172

3 Grammatical Literature

The second explanation is familiar from the sagas and eddic poetry, where Óðinn repeatedly appears under a pseudonym to hide his true identity.250 Later, Snorri explains Freyja’s many names in the same way.251 The first explanation, however, is somewhat curious, since the names that have just been rehearsed are all patently Norse and for the most part semantically transparent. Rather than having different names among different peoples, it is clear that many of Óðinn’s names describe his different qualities, such as Herteitr (battle-happy).252 Just before the stanza, Snorri has explained two of Óðinn’s names – Alfǫðr and Valfǫðr – in this vein. Here, however, he abandons this mode of explanation in favour of one that is not borne out by his example. The reason for this counterintuitive reading again probably lies in Snorri’s endeavor to create parallels between pagan and Christian motifs, in this instance with the common motif of the connection between the division of tongues at the Tower of Babel and the origin of polytheism connected to this event.253 This alignment with a Biblical motif does not focus on a given event – there is no pagan Tower of Babel in Gylfaginning – but rather on a mode of historical understanding: the many languages of the world have led to many names for the one God/Óðinn. The æsir, however, did not understand – or at least did not wish for Gylfi to understand – that the many names really explain polytheism as a whole. Rather, the explanation only addresses the large number of names. This may serve as an illustration of how, in the words of the prologue to Snorri’s Edda, pagans understood the world ‘jarðlegri skilningu’ (with earthly understanding).254 Only from a retrospective, Christian viewpoint could the causal relationship between the division of tongues and polytheism be discerned, and such a perspective is absent from the perverted historical understanding that is presented by the æsir in Gylfaginning. The third example is based on the same quotation from Grímnismál and concerns the framing of the narrative. This time, the use of the stanzas is left implicit. Gylfaginning features many pagan misconceptions of Christian doctrine, and one of them is the mock Trinity whose three persons act as informants to king Gylfi.

250 Thus, for instance, in Grímnismál itself, where Óðinn disguises himself under the name Grímnir, but the motif is a common one (see Annette Lassen, Odin på kristent pergament. En teksthistorisk studie (København: Museum Tusculanum, 2011), passim). 251 Snorri, Prologue and ‘Gylfaginning’, ed. Faulkes, p. 29. 252 This is not unlike the many synonyms given in the eddic poem Alvíssmál, a poem unique in its focus on language and in its affinity to þulur. There, different names for the same thing are said to belong to different groups of supernatural beings. Alvíssmál may to some extent have served as a source of inspiration to the passage above (it is quoted twice in Skáldskaparmál; see Kommentar zu den Liedern der Edda, 3, ed. Klaus von See et al., pp. 268, 289). The connection of many names for a god to the division of tongues, however, is indebted to the Babel motif. 253 Thus, for instance, Petrus Comestor, Historia scholastica. Liber Genesis 40 (Petri Comestoris Scholastica historia. Liber Genesis, ed. A. Sylwan (Turnhout: Brepols, 2005), p. 76. 254 Snorri, Prologue and ‘Gylfaginning’, ed. Faulkes, p. 4.

3.5 Snorri’s Reception of Poetry: Gylfaginning

173

Again, it seems that Snorri did not wish to produce a Trinity out of nothing, and he ingeniously extracted three names from the stanzas: Hár, Jafnhár and Þriði (the High One, the Just-As-High and the Third). While ‘the High One’ is an apt epithet for a god, the origin of the other two is enigmatic, but when these three names are spread among the other 46, the Trinity does not easily spring to mind. When Snorri exstracts these three only they become doctrinal and heterodox at the same time: God is the High One, as one might expect. The three persons of the Trinity, furthermore, are coequal, coeternal and have one will, which is borne out by the name Just-As-High. Snorri further underlines their unity by having them play the part of one respondent in the dialogue. Furthermore, the persons of the Trinity are no more and no less than three, and the Third thus indicates that we are indeed dealing with a Trinity. Thus far orthodoxy. In other regards, however, the setting is heterodox. Even though the three play the part of one, they do so in three separate voices, which is a heretical understanding of the Trinity. More importantly, they are liars, which is contrary to the nature of God. Again, the understanding of the Trinity is staunchly temporal, in accordance with the conceit of Snorri’s Edda. Anthony Faulkes has noted a final instance of creative reading. It regards stanzas 38–39 of Vǫluspá.255 In Gylfaginning, these stanzas occur just after the last battle and describe places of punishment for perjurers and murderers after the Last Judgement, as it were. In both redactions of Vǫluspá, however, they are part of the description of the awful state of affairs before the last battle, and no punishment follows (but a possible reward is hinted at in stanza R 61/H57).256 Again, Snorri has not altered the text itself, apart from quoting the stanzas in the order he saw fit, but he has adjusted the setting, and thus the interpretation, to that of salvation history. These creative readings – the Flood motif, the Babel motif, the Trinity and the final punishment – may serve as examples of Snorri’s use of eddic poetry. They are all parallel to some Christian motif or event, which presumably serves to show God’s hand in history, but they are nonetheless presented as merely historia, without spiritual meaning, nor indeed with any obvious purpose at all, with the possible exception of the final judgement. The Flood drenched the giants as an accidental consequence of the killing of Ymir, not in just retribution for sins. The division of tongues gave rise to many names for one god, but there is no sign of an understanding of the connection of this event to the rise of polytheism, as one may find in the prologue and especially in the later interpolations into it (see pp. 281–86). On the contrary, Hár stresses that these names all belong to one god and that it is necessary to know the circumstances that gave rise to all of them. The Trinity is nothing but the close collaboration of three human individuals with magical powers. Only with regard to

255 Snorri, Prologue and ‘Gylfaginning’, ed. Faulkes, p. xxvii. 256 It is also quoted in Gylfaginning (Snorri, Prologue and ‘Gylfaginning’, ed. Faulkes, p. 20).

174

3 Grammatical Literature

the last punishment do the pagans appear to have grasped some fundamentals about the workings of divine justice. Some conclusions about Snorri’s working methods can now be drawn. In Gylfaginning and Skáldskaparmál alike, Snorri corroborates many of his claims by quoting older poetry, and he seems not to have fabricated any of that poetry himself. Had he done so, he would not have had to go to such lengths to interpret it in a desirable way. Eugen Mogk and Anne Holtsmark claim that in one instance, Snorri seems to have made a change to Vǫluspá 3. Whereas both Codex Regius and Hauksbók have ‘Ár var alda | þar er Ymir byggði’ (In the beginning of time | when Ymir settled down), Gylfaginning has ‘Ár var alda | þat er ekki var’ (In the beginning of time | when nothing existed).257 The reason for this change is, supposedly, that Ymir at this point in the narrative has yet to be created. The stanzas of Vǫluspá which Snorri quotes, however, contain many variant readings as compared to the versions in CR and Hauksbók, whereas those of Vafþrúðnismál and Grímnismál are largely consonant with CR, and it is not clear why he would have revised this poem only.258 It may thus be that Snorri is just quoting a variant reading, undertaken at some earlier point in the tradition to bring the text closer to the Christian creatio ex nihilo. It cannot, furthermore, be conclusively ruled out that such perceptions have a long prehistory in the pagan era, whether they originated with Christianity or not.259 If Mogk’s and Holtsmark’s hypothesis is correct, this is the only instance to have so far been identified where Snorri actively redacted a poem to suit his purposes. On the whole, his method was one of creative reading, not of altering the content of a poem, and certainly not of composing it outright. With regard to his use of older poetic sources for corroboration of his text, Snorri’s working method appears to have been similar in Gylfaginning and in Skáldskaparmál. When interpreting these sources, however, he worked differently in the two texts. The religious nature of Snorri’s reading of older poetry in Gylfaginning differs from what appears to be his purely narrative use of kenning formation in Skáldskaparmál. This is probably not due to the different types of poetry involved – eddic as opposed to skaldic – but is rather a product of context: in Gylfaginning, Snorri produces a world history, whereas in Skáldskaparmál he discusses poetry. The setting thus conditions the choice of poetry and the mode of interpretation alike.

257 Eugen Mogk, ‘Zur Bewertung der Snorra-Edda als religionsgeschichtliche und mythologische Quelle des nordgermanischen Heidentums’, Berichte über die Verhandlungen der Sächsischen Akademie der Wissenschaften zu Leipzig. Philologisch-historische Klasse, 84 (1932), p. 13; Holtsmark, Studier i Snorres mytologi, pp. 28–29. 258 See Heimir Pálsson, ‘Að sætta heimildir’, p. 162. 259 Although most scholars see the CR and Hauksbók reading as primary, Sigurður Nordal does not (‘Völuspá’ gefin út med skýringum, ed. Sigurður Nordal (Reykjavík: Prentsmiðjan Gutenberg, 1923), p. 38–39). Even in the CR/Hauksbók version, the stanza suggests an ex nihilo perception, except for the presence of Ymir.

3.6 The First, Third and Fifth Grammatical Treatises

175

While Snorri in Gylfaginning establishes a number of parallels between Old Norse mythology and the Bible, their meaning appears to have been lost on the protagonists.260 For the intended reader, however, such parallels presumably serve to show that the meaning was there all along, even though the pagans did not see it. In Skáldskaparmál, Snorri’s creativity has little to do with religion. Here, his main concern seems to be to provide a concrete narrative background to the obscure references of the early poets. The different aims of the two sections, then, conditioned which motifs Snorri tried to extract from his sources. Now, however, it is time to leave Snorri and other nativizing grammarians to turn instead to those of a more Latinate disposition.

3.6 The First, Third and Fifth Grammatical Treatises Above, I treated Snorri and the nativizing grammatical corpus that is mainly predicated on the local poetic tradition, even if it is loosely connected to currents within Latin literature. I turn now to the Latinizing corpus, 1–4GT, which is based on foreign models, as well as 5GT, which is a hybrid of the nativizing and Latinizing approaches. Since 2GT builds on 1GT and 4GT is written as a continuation of 3GT, I focus on 1GT, 3GT and 5GT as the main witnesses to their respective analytical and rhetorical approaches. Although 1–4GT are all heavily indebted to Latin grammatica, they also bear the marks of considerable intellectual independence, and I shall argue below that this is to a large extent due to the local poetic discourse. In 3GT and 4GT, the poetic component is clearly visible, since these treatises deal with and quote poetry extensively, and a study of 3GT follows below. 1GT and 2GT, however, do not explicitly treat poetry, and 1GT is by far the earliest of the grammatical treatises, as well as being universally admired as something of a linguistic masterpiece. Before moving on to 3GT, I wish to attempt at least a partial answer to a simple question: How can 1GT be so good? 2GT, highly original though it may be, elaborates on 1GT, and I therefore let 1GT serve as primary witness to the precocious Icelandic fusion of Latin and vernacular grammatica.

3.6.1 The First Grammatical Treatise The First Grammarian was well acquainted with the grammatical tradition, but unlike his celebrated Latin contemporaries such as Abelard or Peter Helias, his greatness

260 Ursula and Peter Dronke have outlined the intellectual background of similar perceptions in the Prologue in ‘The Prologue of the Prose Edda: Explorations of a Latin Background’.

176

3 Grammatical Literature

does not lie in an elaboration of the finer points of Priscian’s Institutiones. Rather, he is at his best when he departs from his Latin models. The treatise was written for the stated purpose of introducing a workable orthography for Old Icelandic, and although the Institutiones provided a helpful framework for thinking about these matters, Priscian’s description could not accommodate Old Icelandic’s rich vowel system.261 Latin literature had other, more useful models for this purpose. The First Grammarian establishes vowel phonology through minimal pairs, that is, words where only the phonemes under discussion change, such as far (vessel) and fár (danger, harm). He presents nine vowels which can each be realized in four ways, and these should all, he contends, be represented in script: he recommends ordinary letters for short vowels, accented ones for long, a dot above the letter for nasalisation and a dot plus accent for long nasalized vowels. He gives sample sentences containing the minimal pairs, often in the form of imperfect rhymes: Sú kona gǫfgar Goð er sjǫlf er góð (the woman who herself is good honours God).262 This method seems to be based on Latin so-called differentiae, the most famous of which were composed by a possibly somewhat younger contemporary of the First Grammarian, Serlo of Wilton.263 Serlo presents minimal pairs in the form of Leonine hexameters: docto crede dǔci si vis ab eo bene dūci (have faith in the wise leader if you wish to be well led by him).264 The First Grammarian had not only studied Latin, however. He also discusses skaldic poetry with the precision of an avid student, and phonetic variation reminiscent of minimal pairs permeates that entire tradition in the form of hendingar. Through skaldic poetry, the ears of the First Grammarian would have been finely tuned to vowel harmony, and his stringent analysis, based on minimal pairs, was thus most likely achieved through a blend of the tools supplied by Old Norse poetics and Latin grammatica.265 On a surface level, the poetic presence in 1GT is not strong, since the author quotes skaldic poetry only twice (and the Latin Distichs of Cato once). One of these quotations, however, is telling. When advocating the spelling for the

261 The fullest overview of the background and linguistic foundation of 1GT is found in the edition The First Grammatical Treatise, ed. Hreinn Benediktsson. 262 The First Grammatical Treatise, ed. Hreinn Benediktsson, pp. 220–222. 263 Anne Holtsmark, En islandsk scholasticus fra det 12. århundre (Oslo: Jacob Dybwad, 1936), pp. 89–90; Vivien Law, The History of Linguistics in Europe. From Plato to 1600 (Cambridge: Cambridge University Press, 2003), pp. 180, 201. 264 See Holtsmark, En islandsk scholasticus, p. 89. 265 Cf. Roberta Frank, Old Norse Court poetry, p. 37. Note, however, that Frank does not presuppose any Latin background. Unlike the scholars mentioned, I do not assume that the use of minimal pairs in 1GT has to be derived from either the Latin tradition or from skaldic poetry. Rather, while recourse to this strategy in a grammatical context is likely to have been inspired by Latin precedents, the analysis was no doubt aided by the skaldic training that the First Grammarian evidently possessed.

3.6 The First, Third and Fifth Grammatical Treatises

177

diphthong /ja:/, he addresses his fictive opponent, who is indifferent to phonographematic correspondences, by appealing to the authority of the poets: ‘skáld eru hǫfundar allrar rýnni eða málsgreinar sem smiðir [smíðar] eða lǫgmenn laga’ (poets are the authorities on all matters concerning literature or linguistic distinctions, just as craftsmen are [in their craft] or lawyers in laws). 266 He then quotes a stanza containing the word form éarn (iron), which must in this context be scanned as disyllabic, rather than with the ordinary pronunciation with a diphthong (jarn or járn). Many wise men, he assures his opponent, have pronounced this word with a syllabic /e:/, and they have also told him that they have heard others (presumably older men) do so as well. If the opponent will not be convinced even by this, the First Grammarian says that he can do no more than have recourse to Cato’s words: ‘contra verbosos noli contendere verbis | sermo datur cunctis, animi sapientia paucis’ (Do not quarrel with babblers | speech is given to all, but wisdom to few). In other words: the authority of poets is like that of lawyers, and they should get the final word in all matters relating to language. If you will not listen even to them, it is no use talking to you. After the Irish Auraicept na n-éces, 1GT is the first treatise in Europe on vernacular grammatica (Ælfric wrote his grammar in English, but about Latin). Even though it is heavily indebted to Latin grammatica, then, the undertaking is quite original, and one must ask whence the self-confidence came which prompted its composition. I believe that the author provides at least part of the explanation himself: if poets are the authorities on all matters concerning literature or linguistic distinctions, then his project may not have seemed as novel to him as it appears to us. They had already produced a normative body of texts on which to establish linguistic correctness, and by drawing on that authority in the form of a treatise, derived from the Latin tradition, the author combined the best of two worlds. The extraordinary quality of 1GT, which has baffled some modern linguists, is not primarily due to its author’s Latin learning.267 Rather, it is predicated on the relatively pragmatic tradition of skaldic poetry, which remained close to linguistic realities, and which the author allowed to interfere freely with the normative system of grammatica. Thus, although skaldic poetry is quoted only twice in the treatise, it is of fundamental importance for understanding its background and its character. By contrast, the importance of poetry is readily evident in 3GT.

266 The First Grammatical Treatise, ed. Hreinn Benediktsson, pp. 224–26. The word smíðar is left out in the manuscript, but the emendation may be regarded as relatively secure due to the parallel construction lǫgmenn laga. 267 Hreinn Benediktsson writes: ‘FGT [i.e. 1GT] may appropriately be said to bear greater resemblance, in its methodology, to linguistics in the twentieth century than in any other period.’ (The First Grammatical Treatise, ed. Hreinn Benediktsson, p. 38).

178

3 Grammatical Literature

3.6.2 The Third Grammatical Treatise 3GT (c. 1250) was composed by Óláfr Þórðarson (d. 1259), nephew of Snorri Sturluson and member of the powerful Sturlung family.268 Óláfr was ordained subdeacon. His treatise attests to an interest in both the classics and skaldic poetry, and he was also the founder of some kind of school, presumably a grammar school.269 The Sturlung milieu in the first half of the thirteenth century seems to have consisted of learned chieftains and their ecclesiastics collaborating in common intellectual endeavours.270 As a member of the family and as a subdeacon, Óláfr combined these qualities in one person. The first part of 3GT is based on Priscian’s Institutiones 1–2, probably in tandem with Ælfric’s Excerptiones de arte grammatica anglice, and it treats the topic of letters and sounds.271 The second and larger portion of the treatise is a translation of Donatus’s Barbarismus (the third book of the Ars Maior), dealing with figures and tropes. The prose roughly follows Barbarismus, but the examples given by Donatus, mainly Latin hexameters from the Æneid, are all replaced by skaldic poetry. These two poetic traditions differ radically in metre and style, but in the introduction to the translation of Barbarismus, their supposed common background is used to argue that they are the same art: Í þessi bók má gerla skilja, at ǫll er ein listin skáldskapr sá, er rómverskir spekingar námu í Athenisborg á Grikklandi ok sneru síðan í latínumál, ok sá ljóðaháttr eða skáldskapr, er Óðinn ok aðrir Asíamenn fluttu norðr higat í norðrhálfu heimsins, ok kenndu mǫnnum á sína tungu

268 The attribution is based on: 1. The colophon in A. 2. Two references in 4GT naming Óláfr as the author of 3GT. 3. A reference by the author himself to ‘his lord Valdimarr’, the king of Denmark with whom Óláfr is known to have stayed (see Edda Snorra Sturlusonar, ed. Jón Sigurðsson et al., 2, pp. 62–63 note 1). 269 Sverrir Tómasson, Formálar íslenskra sagnaritara á miðöldum, p. 35 and n. 130. 270 Snorri, for instance, was a friend of the prior Styrmir and appears to have used his saga of Saint Óláfr as the chief source for his own saga about him (Turville-Petre, Origins of Icelandic Literature, pp. 219, 222). Snorri was the fosterson of Jón Loptsson, who was himself both chieftain and deacon (Turville-Petre, Origins, p. 220). 271 Kari Ellen Gade has demonstrated that the source is Priscian’s Institutiones rather than Excerptiones de Prisciano, which has sometimes been suggested, and that there is considerable terminological overlap between 3GT and Ælfric’s Excerptiones. She has also shown that at least parts of Ælfric’s Excerptiones were known in Iceland c. 1400, which makes it increasingly likely that they were used in the composition of 3GT (‘Ælfric in Iceland’, in Learning and Understanding in the Old Norse World. Essays in Honour of Margaret Clunies Ross, ed. Judy Quinn, Kate Heslop, and Tarrin Wills (Turnhout: Brepols, 2007), pp. 321–39). Ælfric’s grammar held a very strong position in twelfth-century England (it was even glossed in French), and it must presumably have reached Iceland in a period when it was still studied in England, and thus in the early thirteenth century at the very latest. See Vivien Law, ‘Anglo-Saxon England: Aelfric’s “Excerptiones de arte grammatica anglice”’, Histoire Épistémologie Langage, 9 (1987), pp. 63–64.

3.6 The First, Third and Fifth Grammatical Treatises

179

þesskonar list, svá sem þeir hǫfðu skipat ok numit í sjálfu Asíalandi, þar sem mest var frægð ok ríkdómr ok fróðleikr veraldarinnar.272 From this book it can be clearly understood that the poetry that the Romans learned in Athens in Greece and then translated into Latin, and the verse form or poetry that Óðinn and the other Asians transferred up here to the northern part of the world, is all the same art. And they taught people this kind of art in their own tongue, in the manner that they had arranged and learnt it in Asia itself, where there was more glory, riches and wisdom than anywhere in the world.

Stories about Trojan descent were told all over Europe in the Middle Ages, the ultimate source of inspiration being the Æneid.273 In this vein, Snorri describes the gods as Trojan heroes, and the passage above echoes the prologue to Snorri’s Edda.274 Óláfr’s description differs from others, including Snorri’s Edda, however, in claiming that Latin and skaldic poetry are ‘the same art’.275 Indeed, Óláfr is implying more than mere equality. He does not activate the Trojan background for the Romans; their poetry (epic hexameter, as found in Barbarismus) was transmitted through Athens and based on translations. The ancestors of the northern peoples, by contrast, were the original composers of this poetry and never had to translate it (note the careful opposition of ‘Athens [. . .] translated [. . .] their own tongue [. . .] Asia itself’).276 With these claims, Óláfr proceeds to the second part of the treatise, where he replaces all Latin poetry with skaldic stanzas. In so doing, according to

272 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 60. My normalization. 273 The fullest overview of the Old Norse version of the Troy story is Andreas Heusler, Die gelehrte Urgeschichte im altisländischen Schrifttum. Abhandlungen der Königlichen Preussischen Akademie der Wissenschaften philosophisch-historische Classe, 3 (Berlin: Verlag der Königlichen Preussischen Akademie der Wissenschaften, 1908). In an addition to the prologue of Snorri’s Edda in Codex Wormianus there is an explicit comment on this widespread practice: ‘[. . .] ok af þeira frægð gáfu eptirkomandi menn sér þeira virðingarnǫfn ok einkannlega svá sem Rómverjar [. . .]’ ([. . .] and because of their [the Trojans’] fame, later people appropriated their dignified name, particularly so the Romans [. . .]) (Edda Snorra Sturlusonar. Codex Wormianus AM 242, fol, ed. Finnur Jónsson (København and Kristiania: Gyldendal, 1924), p. 6. My normalization). 274 The statement about the ‘glory, riches and wisdom’ of Asia is a condensed paraphrase of the end of chapter 3 of the prologue (Prologue and ‘Gylfaginning’, ed. Faulkes, p. 4 (21–26). 275 As Rita Copeland notes, ‘it is only in medieval Scandinavia that the intellectual traditions of grammar and figurative language find a decisive vernacular reception’, with particular reference to 3GT (‘Grammar, Rhetoric, and Figurative Language: Learned Innovations and Vernacular Receptions’, in Intellectual Culture in Medieval Scandinavia c. 1100–1350, ed. Stefka G. Eriksen (Turnhout: Brepols, 2016), p. 237). The atypical use of the Troy story for equating the arts is part of this larger picture of ‘a decisive vernacular reception’. 276 The same idea is presented in the prologue to Snorri’s Edda: the Asians imposed their tongue on Northern Europe, and this can be seen by place names in Great Britain that are clearly not from the same linguistic root as Old Norse (the reference seems to be to Celtic names) (Snorri, Prologue and ‘Gylfaginning’, ed. Faulkes, pp. 6, 57; for a further discussion of Icelandic perceptions of linguistic relationships, see Gunnar Harðarson, ‘“Alls vér erum einnar tungu”. Um skyldleika ensku og íslensku í Fyrstu málfræðiritgerðinni’, Íslenskt mál, 21 (1999), pp. 11–30).

180

3 Grammatical Literature

his own version of the Troy story, he is only restoring Barbarismus to the form it ought originally to have had. The claim that Latin and Old Norse poetry are the same art is a bold one, but Óláfr manages to deliver on it remarkably well. True, he must admit that ellipsis of final -m is irrelevant to Old Norse, and some of the vices become meaningless if applied to skaldic poetry.277 He manages, however, to accommodate the rich kenning system by expanding some categories of tropes considerably, such as metaphora, metonymia and epitheton, making 3GT much longer than its source text (for some qualification of this statement, see below).278 The best example of Óláfr’s working method may be his ingenious adaptation of the passage on paronomasia (near homonymy) and parhomoeon (alliteration) in order to make room for hendingar and alliteration. In Donatus’s example of paronomasia, the key words are amentium/amantium, differing only in one letter/phoneme.279 Óláfr’s example is ‘heldr vill hilmir | herja en erja’ (the ruler prefers | warfare to ploughing), where herja and erja also differ only in one letter.280 Hendingar, however, are contained in one syllable only, and the end-rhymes that do occur are optional and rare. Óláfr has masked the difference between the two traditions by choosing an atypical example.281 Óláfr treats parhomoeon (in 3GT called paranomeon) in a similar way. Donatus’s example is from Ennius: ‘O Tite tute Tati tibi tanta, tyranne, tulisti’ (Oh tyrant Titus Tatius, you drew such great [outrages] upon yourself).282 This example differs from Old Norse alliteration in that medial as well as initial consonants play a role in it. The reader, however, does not have to worry about the applicability of paranomeon to skaldic poetry, since Óláfr has made the choice for him, opting for initial alliteration and thus the single most fundamental feature of Old Norse poetry. Óláfr’s example is ‘sterkum stilli | styrjar væni’ (to the powerful ruler | eager for battle). He comments: Þessi figúra er mjǫk hǫfð í málsnilldarlist er rethorica heitir ok er hon upphaf til kveðandi þeirrar er saman heldr norrœnum skáldskap, svá sem naglar halda skipi saman, er smiðr gerir, ok ferr sundrlaust ella borð frá borði – svá heldr ok þessi figúra saman kveðandi í skáldskap með stǫfum þeim er stuðlar heita ok hǫfuðstafir. Hin fyrri figúra gerir fegrð með hljóðsgreinum

277 Such meaningless vices are, for instance, hiatus and retained -m (Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 71). For elision of -m, see Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 90. 278 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, pp. 102–08 (unfortunately, there is a lacuna in the section on metonymia). 279 Donat et la tradition de l’enseignement grammatical, ed. Holtz, p. 665. 280 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 96. 281 See Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, pp. 195, 208. 282 Donat et la tradition de l’enseignement grammatical, ed. Holtz, p. 665.

3.6 The First, Third and Fifth Grammatical Treatises

181

í skáldskap, svá sem felling skipsborða. En þó eru fastir viðir saman, þeir sem negldir eru, at eigi sé vel feldir, sem kveðandi helz í hendingarlausum háttum.283 This figure [parhomoeon] is common in the art of eloquence which is called rhetoric, and it is the origin of the scansion which holds Norse poetry together, just as nails hold a ship which a carpenter makes together, and the planks will otherwise fall apart – just so does this figure hold scansion together in poetry with the letters that are called props and main letters [i.e. auxiliary and main alliterating staves]. The former figure [paronomasia] creates beauty by phonetic distinctions in poetry, just as the seams of planks. But if the pieces of wood have been nailed, they will stick together even though the seams have not been well executed, just as scansion is retained in metres without hendingar.

Here, Óláfr allows us to see to what extent, as he has said earlier, classical and Norse poetics are one and the same art. It is almost as if Donatus was really thinking of skaldic poetry all along – only, of course, that Óláfr had to redefine both parhomoeon and paronomasia in order to produce that result. The first part of 3GT is based on Priscian’s treatment of letters and syllables, and Óláfr does two things that are relevant to our analysis there. First, he relates his discussion to poetic discourse. For instance, he discusses hendingar and endrhyme and adapts the discussion of aspiration to the demands of Old Norse alliteration (see p. 128).284 In the end of the treatment of the accidents of the syllable he writes: ‘En með því at þesskonar greinir heyra lítt norrœnu[m] skáldskap at flestra manna ætlan, þá tala ek ekki fleira at sinni’285 (but since these distinctions in the opinion of most have little to do with Norse poetry, I shall speak no more [about them] at this time). Here, he explicitly mentions his poetic priorities, even though this part of the treatise centres on orthography and phonology.

283 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, pp. 96–97. Óláfr’s claim that alliteration is common in rhetoric is curious, since the Latin tradition generally warns against profuse use of alliteration and since his exemplar in all likelihood contained such a warning at this point (Victor Frans, ‘Sub regulis Donati’. ‘Ars Brugensis’ och Den Tredje grammatiska avhandlingen (MA thesis, University of Oslo, 2019), pp. 60, 66). His description matches the Irish use of the word retoiric, however: rhythmical, alliterative text (on retoiric or roscada, see, for instance, Liam Breatnach, ‘Zur Frage der roscada im Irischen’, in Metrik und Medienwechsel. Metrics and Media, ed. Hildegard L. C. Tristram (Tübingen: Gunter Narr, 1991), pp. 197–205). The semantic overlap is so precise and the contrast with the Latin tradition is so stark that it would appear that Óláfr here draws on the Irish use. A partial parallel to such awareness of Irish literary culture may be fond in the First Grammarian, who displays some knowledge of Irish pronunciation of Latin (The First Grammatical Treatise, ed. Hreinn Benediktsson, p. 234). This information seems to have been mediated by Latin, since the author refers to the Irish as skotar rather than írar, that is, by their Latin rather than by their Norse designation (Holtsmark, En islandsk scholasticus, p. 59). I have found no additional traces of Irish learned culture in thirteenth-century Norse texts, however, and I therefore cannot rule out that the semantic overlap between Irish retoiric and Óláfr’s use of rethorica is coincidental and based on a wish to find a place for the most fundamental feature of Old Norse poetry among the disciplines of the curriculum. 284 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, pp. 51–52, 54. 285 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 56.

182

3 Grammatical Literature

Second, Óláfr adapts Priscian’s discussion to treat runes rather than Latin letters. His method in both parts of the treatise is thus one of replacement: hexameters with dróttkvætt in the second part, letters with runes in the first. He takes every opportunity to argue for the respectable pedigree of the runes, as well as their rational correlation to phonology. His underlying assumptions are mostly implicit, but they come to the fore on closer scrutiny: when he seeks to justify different features of the runes, he refers to Greek and Hebrew, but when he wishes to clarify rather than defend a given feature, he compares the runes to Latin script. Thus, he writes: ‘Stafanǫfn eru .xvi. í danskri tungu, í þá líking sem girkir hǫfðu forðum daga’ (the names of the letters are sixteen in the Danish tongue, just as among the Greeks in the early days). This information goes back to Priscian.286 Actually, Óláfr discusses more than sixteen runes, but he appears to be referring to those which belonged to the fuþark proper and which had their own traditional names. The fact that he mentions the number of names, rather than the number of runes, is interesting, since Priscian only mentions the number of letters (literae), but not their names. Óláfr is probably thinking of names with a meaning, such as f fé (cattle, wealth) and so on. Primarily through Jerome’s Letter 30 and texts derived from it, this was known to be the case for Hebrew as well: alef was translated as doctrina (doctrine), beth as domus (house), and so on.287 A digest of Jerome’s letter is found in AM 732 b 4to 6v (c. 1300–1325), showing that a text about the actual and spiritual meanings of the Hebrew letters was present in Iceland at a relatively early date. This manuscript contains the names of the letters and some discussion of their spiritual meaning, but not the letter forms themselves. These are found, together with the Greek letters and the names and phonetic values of both, in AM 685d 4to 30v–31r (fifteenth century), without translations of the names. Much earlier, the First Grammarian named and attempted to reproduce the letters deleth and sade, and he knew that in Greek, long and short vowels are written with different letters. This combines to suggest that he had access to a comparable alphabet list around 1150.288 The Icelandic transmission of lore relating to the Hebrew and Greek letters, as well as the general spread of such lore across Europe, make it likely that Óláfr was aware that their forms were different from those of Latin letters and that their names were words, not simply representations of the sounds of the letters, such as Latin a, be, ce, etc.289 Óláfr’s awareness of the Hebrew letter names is attested by the fact that he

286 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 40; Grammatici latini, 2, ed. Keil, p. 11. 287 PL, 22, cols 441–45. 288 The First Grammatical Treatise, ed. Hreinn Benediktsson, pp. 218, 238. See Richard Cole, The Jew Who Wasn’t There: Studies on Jews and their Absence in Old Norse Literature (PhD Diss., Harvard University, 2015), pp. 75–80. 289 The fullest overview of such lists that also contain runes is René Derolez, Runica Manuscripta. The English Tradition (Brugge: Rijkuniversiteit te Gent, 1954).

3.6 The First, Third and Fifth Grammatical Treatises

183

names alef and yod (see below). He probably knew that the Hebrew letter names were words that carried meaning, just like the names of runes, and he may have thought that this was true of the Greek letter names, as well.290 The fact that Óláfr departs from Priscian in mentioning the names of the runes, rather than runes as signs only, suggests that he saw some affinity between the Hebrew-Greek and runic traditions of having letter names with a meaning. The runic tradition was thus aligned with the most pristine of the Sacred Languages, and it was different from the derivative Latin tradition. It is also worth noticing that in mentioning the number sixteen, he makes no comparison to the more numerous Latin letters – partly, perhaps, because their higher number might induce the reader to ponder which script is the more precise of the two. Rather, the implied connection to pre-Roman antiquity remains the only justification for the number. Shortly afterwards, Óláfr justifies the imprecision of the rune íss (its values are /i/ , /e/, and sometimes /æ/) with reference to Hebrew. He seems to have had some notion of Hebrew matres lectionis – representing long vowels through the use of the letters alef, he, waw, yod – since he writes: ‘[. . .] ok er íss stundum settr fyrir æ, þá er hann er stunginn, svá sem alæph eða ioth setiaz fyrir .ij. raddarstǫfum í ebresku máli’ (and íss is sometimes used for /æ/, when it is punctuated, just as alef or yod are used for two vowels in the Hebrew language).291 He does not say why he provides this information, but the reason may be inferred to be that Latin script displays no such imprecision, at least in theory. He then proceeds to claim that when the masters of runes needed a sign for the phoneme /y/, they took it from Hebrew. This claim seems to be based on an equation of the Icelandic rune ↕ (as opposed to mainland Z) to the shape of yod in some of the the alphabet lists – in particular the De inventione linguarum tradition (see image).292 In other cases, when there are no troubling discrepancies between Latin

290 In some English and Norman manuscripts from the period c. 900–1200, the Greek letter names, just like the Hebrew ones, are accorded their own meaning, rather than just a numerical value, which was more common (Bernard Bishoff, ‘Das griechische Element in der abendländischen Bildung des Mittelalters’, Byzantinische Zeitschrift, 44 (1951), pp. 33–34). If Óláfr had some awareness of that tradition, the Greek letter names would have appeared very similar to those of runes, but this possibility is difficult to verify. 291 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 42. It is remarkable that Óláfr was aware of matres lectionis, since they belonged to the practical sphere of Hebrew literacy and not to the body of lore which is found in alphabet lists or in texts derived from Jerome. He must probably have gotten this information from an informant who had studied at some monastic centre in England (such as Ramsey Abbey or Lincoln) or Paris (St Victor), where Hebrew was written alongside Latin texts. See Judith Olszowy-Schlanger, ‘The Knowledge and Practice of Hebrew Grammar among Christian Scholars in Pre-Expulsion England: The Evidence of “Bilingual” Hebrew-Latin Manuscripts’, in Hebrew Scholarship and the Medieval World, ed. Nicholas de Lange (Cambridge: Cambridge University Press, 2001), pp. 107–28; Cole, The Jew Who Wasn’t There, pp. 71–74. 292 I thank Victor Frans for this observation. See Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, pp. 42–43 and Derolez, Runica Manuscripta, p. 350. The manuscripts strongly suggest that the rune ýr occurred three times in the archetype. W is first missing ýr in the enumeration of the

184

3 Grammatical Literature

letters and runes, Óláfr simply drops all references to Greek and Hebrew. He seems content with mere equality and only resorts to the other sacred languages when the shortcomings of runic script leave him in a tight spot.

A 2r ll. 18–23 (Image: Reykjavík, Stofnun Árna Magnússonar) and W 48v ll. 9–13 (Image: Copenhagen, Den Arnamagnæanske Samling. Photo: Suzanne Reitz). Here one sees the use of ↕ in two instances in A, where W is lacking the rune in the first instance and has a h in the second instance (see p. 183 n. 292).

five vowels and then, when it should have occurred for the second time, has the Latin letter h (see image). This shape is clearly corrupt, but may reflect a Z in the exemplar (Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 5). Finally, it has the shape Z in the pangram. A has the Icelandic shape of the rune, ↕, rather than the older Z, in all three instances. Since ↕ would appear to be the lectio facilior in an Icelandic setting, it might at face value appear likely that Óláfr’s text had Z both in the text and in the pangram. Because of the similarity of ↕ to yod in the alphabet lists and Óláfr’s special interest in aleph and yod, however, it seems likely that he had ↕ in the text but Z in the pangram, presumably due to its royal Danish provenance, and that a later scribe in the A tradition opted for one shape in all instances, whereas the Icelandic shape was subject to corruption in the W tradition. Tarrin Wills reconstructs ↕ in all three instances, however (Tarrin Wills, The Foundation of Grammar. An Edition of the First Section of Óláfr Þórðarson’s Grammatical Treatise (PhD diss., University of Sydney, 2001), pp. 84, 88). A further complication is that is the common rune for /y/ in Danish inscriptions. On a possible background of the Icelandic shape, see the text under the image below p. 223. On the rune ýr, see Islands runeindskrifter, ed. Ander Bæksted. Bibliotheca Arnamagnæana, 2 (Hafniæ: Munksgaard, 1942), pp. 46, 52; Larsson, Patrik, Yrrunan: Använding och ljudvärde i nordiska runinskrifter, Runrön, 17 (Institutionen för nordiska språk vid Uppsala universitet: Uppsala, 2002), pp. 138–39, 185–86. Richard Cole has suggested that Óláfr was thinking of ‫( ש‬šin) rather than yod, but šin has a consonantal value in the alphabet lists and in Hebrew (see Cole, The Jew Who Wasn’t There, p. 81).

3.6 The First, Third and Fifth Grammatical Treatises

185

A 2v ll. 12–14 (Image: Reykjavík, Stofnun Árna Magnússonar) and W 49r ll. 1–3 (Image: Copenhagen, Den Arnamagnæanske Samling. Photo: Suzanne Reitz). Valdimarr’s runic pangram, with the rune ↕ in A and Z in W (see p. 183 n. 292). The text reads: ‘þessa stafi ok þeirra merkingar compileraði minn herra Vildimarr [W dana-] konungr með skjótu orðtœki á þessa lund: “Sprengd manns hǫk flýði tovi [W tvvi] bǫll”’ (my lord king Valdimarr [W of Denmark] compiled these letters and their meaning in this way: ‘Sprengd manns hǫk flýði tovi [W tvvi] bǫll’ [meaning disputed])

AM 685 d 4to 30v–31r: The Greek and Hebrew alphabets, with letters, names and values (Image: Reykjavík, Stofnun Árna Magnússonar).

186

3 Grammatical Literature

Paris BNF MS Latin 5239, 235r (tenth century, Limoges): Part of the Hebrew alphabet in the beginning of the treatise known as De inventione linguarum, showing a form of the Hebrew letter yod (here ‘Ioth’) which is likely to have informed Óláfr’s argument regarding the rune ↕.

Óláfr also takes pains to describe the order of the runes as rational. A twelfth-century commentary on Priscian gives the physical rationale of the order of Latin vowels, namely that /a/ is spoken in the chest and the others progressively higher up, ending with the labial /u/.293 Óláfr may not have used this particular commentary, but he refers to this ordering principle among the ‘authorities on Latin’ (látínumenn) and claims that the reasoning behind the order of runic vowels is the opposite one, beginning with /u/. When he reaches the last of the vowels, which in the runic alphabet is not /a/ but /y/, the argument is dropped, possibly because he cannot bring himself to claim that /y/ is spoken deep down in the chest. Instead, as noted above, he informs the reader that this vowel is taken from Hebrew. After some further discussion, Óláfr states: ‘[. . .] ok hafa því hvárir tveggju meistarar vel ok náttúrliga skipat stǫfunum í sínu máli’ ([. . .] and thus the masters of both tongues have ordered the letters well and in accordance with nature in their respective languages).294 To reach this conclusion, however, he had to withhold about about /y/ the information that he gives about the other letters. Immediately after this discussion of order, Óláfr gives the reason for the different names of runic vowels: they all begin with the sound of the vowel and end with the sounds /r/ and /s/.295 Here, he does not say two things. First of all, he does not make any explicit comparison to Latin, where the name and the sound of vowels are the same and where their names illustrate their definition: vowels can be spoken without the support of other sounds. In the grammatical tradition, that information would have been expected at this point. In such a comparison the runic vowel names could hardly have come across as anything but awkward, since they offer no indication of what

293 Notices et extraits de divers manuscrits Latins pour servir à l’histoire des doctrines grammaticales au moyen âge, ed. Charles Thurot (Paris: Imprimerie impériale, 1868), pp. 135–36. 294 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 43. 295 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 43.

3.6 The First, Third and Fifth Grammatical Treatises

187

constitutes a vowel. In consequence, the ordinary definition of vowels is simply dropped. Furthermore, Óláfr does not give the basic characteristics of names for runic vowels, namely that they are short words beginning with the vowel in question (for instance íss (ice), the name of the rune for /i/). Rather, he strives to find some pattern, and the /-r/–/-s/ one seems to be the best he can come up with under the circumstances. Once more, he shows that the runic alphabet is rational by leaving out relevant details. The section on runes is Óláfr’s most independent adaptation of Priscian, and the topic is one that enables him to argue that Old Norse script is as rational as, and even older than, its Latin counterpart. Óláfr’s treatment of runes and poetry is a valuable complement to Snorri’s Edda as a witness to the construction of Iceland’s poetic past. With his arguments about equivalence and sameness, Óláfr gives a theoretical rationale for why the traditional types of literature may be used to replace Latin texts. In practice, as we shall see in Chapter Four, many saga authors were at this time following his lead in creating a literature about Icelandic hero-poets which could form a homegrown counterpart to the classical poets. Another noteworthy aspect of Óláfr’s translation of Barbarismus is shared with the Fourth Grammarian’s translation of Alexander’s Doctrinale. Both Old Icelandic texts are much longer and more complex than their Latin sources. This claim requires some elaboration, since Óláfr seems to have used a rather lengthy commentary on the Barbarismus and since the Doctrinale was often heavily glossed.296 The second part of 3GT and 4GT are thus not necessarily longer than their actual exemplars. Both treatises, however, speak with an authorial voice, rather than that of a commentator giving further distinctions, definitions and background. Óláfr may thus be said to appropriate the voice of Donatus, whereas the Fourth Grammarian writes a continuation of 3GT and retains its style, so that his voice may perhaps be said to be that of Óláfr impersonating Donatus, even if his exemplar was the Doctinale. The result is the same in both treatises: their discourse is much longer and more complex than that of the authority which they are appropriating (4GT is based on less than a hundred lines of the Doctrinale). The opposite is in general the norm in

296 On some close matches to 3GT among the commentaries on Barbarismus, see Valeria Micillo, ‘Classical Tradition and Norse Tradition in the “Third Grammatical Treatise”’, Arkiv för nordisk filologi, 108 (1993), pp. 68–79; eadem, ‘Die grammatische Tradition des insularen Mittelalters in Island. Spuren insularer Einflüsse im Dritten Grammatischen Traktat’, in Übersetzung, Adaptation und Akkulturation im insularen Mittelalter, ed. Erich Poppe and Hildegard L. C. Tristram. Studien und Texte zur Keltologie (Münster: Nodus Publikationen, 1999), pp. 215–29. Micillo found affinities, but no single text which could have served as Óláfr’s exemplar. I have identified his likely exemplar; see Mikael Males, ‘PseudoRemigius and the Old Icelandic Barbarismus: A Pilot Study’, in Reading Slowly. A Festschrift for Jens E. Braarvig, ed. Lutz Edzard, Jens Wilhelm Borgland and Ute Hüsken (Wiesbaden: Harrassowitz Verlag, 2018), pp. 321–331. This hypothesis has since been corroborated by Frans, ‘Sub regulis Donati’, pp. 13–28.

188

3 Grammatical Literature

translations into Old Icelandic. Latin pseudo-historical works such as Sallust, Lucan, Dares, Geoffrey of Monmouth and Walter of Châtillon were abbreviated, the time-line was straightened out, and authorial and ethical commentary was removed.297 In other cases, the translation was mainly literal, as in Elucidarius. The considerable elaboration with regard to both length and complexity attested in 3GT and 4GT is unparalleled within the corpus, but for 3GT, this is valid only for the second part, which treats poetry. It would seem that it is the topic itself, namely poetry, which provided the impetus to these unrepresentative outbursts of creative energy.298 The step from Háttalykill’s, Litla Skálda’s and Snorri’s nativizing approach, and indeed from the relatively independent adaptation of Latin learning of 1GT, to that of translation and explicit appropriation in 3GT was a big one. Translation may seem like a relatively straightforward way of importing foreign learning, but that was not how the poetic discourse had developed during the period from Einarr Skúlason to Snorri. 3GT represents a radical break with the methods of several generations of authoritative poets and scholars. One short and rarely mentioned text – 5GT – suggests that the process leading up to the translation of Priscian and Donatus was one of trial and error. 3.6.3 The Fifth Grammatical Treatise 5GT only survives as a fragment of eight lines and two words of a ninth that precedes 3GT in A. After 3GT follows a colophon saying: Hér er lykt þeim hlut bókar er Óláfr Þórðarson hefir samansett ok upp hefr skáldskaparmál ok kenningar eftir því sem fyri fundit var í kvæðum hǫfuðskálda ok Snorri hefir síðan samanfœra látit.299 Here ends the part of the book that Óláfr Þórðarson has compiled and [the section on] poetic diction and kennings begins, according to what has been found in the poetry of the main poets, and which Snorri has since then gathered.

This may be taken to mean that the scribe attributes both 5GT and 3GT to Óláfr. It is conceivable, however, that the ‘part of the book’ he is referring to is only 3GT, so I shall return to the question of authorship below. After being edited in 1884, 5GT had not been the subject of a detailed analysis until I published an article on the topic in 2017.300 As it turns out, it is possible to say

297 Stephanie Würth, Der ‘Antikenroman’ in der isländischen Literatur des Mittelalters. Eine Untersuchung zur Übersetzung und Rezeption lateinischer Literatur im Norden (Basel: Helbing und Lichtenhahn, 1998), pp. 119–23. 298 There may, of course, also be additional factors contributing to the greater length and complexity of the treatises, such as pedagogical deliberations or the Fourth Grammarian’s penchant for allegorical interpretation, but even the latter seems to be triggered by the poetic topic. 299 AM 748 Ib 4to, 8v ll. 6–8; Jón Sigurðsson et al. (ed.) 1848–87, vol. 2: 427–28. 300 Mikael Males, ‘Character, Provenance, and Use of the Icelandic Fifth Grammatical Treatise’, Arkiv för nordisk filologi, 132 (2017), pp. 121–38.

3.6 The First, Third and Fifth Grammatical Treatises

189

much more about the character and provenance of the treatise than its fragmentary state might suggest. I refer the reader to the article for a detailed analysis, but 5GT is so important for understanding the development of Icelandic poetics that I shall briefly rehearse the most relevant points. Here follows the text of the fragment in full: [. . .] garð er þat kallat ef lengi er talat um hit sama efni, sem Guðbrandr kvað í Svǫlu: Upp dregr mǫkk hinn mikla ok allar vísur þær er eptir fara eru af einu efni allt til stefja fram. Snyrða þykkir heldr bragarbót en spell ok skulu þá standa margir samhljóðendr eptir raddarstaf hinir sǫmu, ok síðast t eða r, sem Einarr kvað: Elfr varð unda gjalfri eitrkǫld roðin sveita vitnis fell með vatni varmt ǫldr í men Karmtar

[. . .] It is called (-)garð if you stick to the same subject for a long time, as Guðbrandr í Svǫlu composed:

The great fog gathers and all stanzas that follow treat one and the same subject all the way to the refrain. Snyrða is considered a poetic virtue rather than a vice, and then many of the same consonants should stand after the vowel, and finally t or r, as Einarr composed

The poison-cold Göta älv was reddened by the wounds’ sweat’s [blood’s] sea [blood]. The warm wolf’s beer [blood] fell, along with the water, into Kǫrmt’s necklace [sea].

Skarbrot er þat ef aukit er samstǫfu skammri í fyrsta vísuorði, svá at ófegra þykki, sem Óláfr Leggsson kvað:

It is skarbrot if one short syllable is added in the first verse, so that it is less pleasant, as Óláfr Leggsson composed:

‘Skular’ (=Skúla’r) bezt und báli byrræfrs skǫpuð ævi.

The best life under the wind-roof’s [heaven’s] bonfire [the sun] has been created for Skúli.

Unlike 3GT, this text uses no Latin terminology. Like 3GT, however, it gives examples of virtues and vices, although these Latinate concepts are veiled by the terms bragarbót (poetic virtue) and ófegra (less pleasant). The previous tradition had only given examples of what to do, not of what to avoid. Snorri even replaced the poetry of the early poets with examples where their vices were no longer present (see above pp. 35–37, 119–20). Examples of vices, by contrast, belong to the Latin grammatical tradition. The phenomena described, furthermore, appear to be inspired by Latin grammatical concepts. 301 Bragarbót here clearly retains its natural meaning of ‘poetic virtue’ rather than being used as the name of a metre, as in Háttatal (Faulkes (ed.) 2007: 17; Finnur Jónsson 1931, s.v bragarbót). 302 Björn M. Ólsen (ed.) 1884: 297, introduces the variant heitu for sveita, giving ‘the wounds’ hot sea [blood]’. 303 AM 748 Ib 4to, 1r ll. 1–9; Björn M. Ólsen (ed.) 1884: 159. As Björn M. Ólsen notes, the rhyme -ræfrs: ævi shows that /f/ must have merged with /v/ in this position at the time of composition (Björn M. Ólsen (ed.) 1884: 298).

190

3 Grammatical Literature

5GT in A (1r ll. 1–9; Image: Reykjavík, Stofnun Árna Magnússonar). The treatise ends with the two last words of the ninth line, which are preceded by the first rubric of 3GT and the first words of that treatise: ‘Allt er hljóð þat er [um] kvikvendis eyru [má skilja]’ (It is all sound, that which may be perceived through the ears of living creatures).

The term (-)garð is the first word in the fragment and may have been a compound, but its neuter form is problematic, since garðr (yard, fence) is masculine and would have remained so also in a compound. The semantics of the (non-existent) lemma thus appear to be unhelpful, but its meaning is specified in 5GT: it means sticking to the same subject for a long time. The regulation of subject matter rather than form or semantics is generally absent from the nativizing corpus of texts. The Latin extraction is borne out by 3GT, where the same poetic example, this time with the second line of the couplet, illustrates the same phenomenon under the heading macrologia.304 The figure described in the second example, snyrða, entails a longer cluster than usual of consonants in the hending, even when ending in an -r or a -t which could, in the author’s view, have been left out of the rhyme.305 The figure appears to be constructed ad hoc, and the phenomenon is not described in any other text. In the description of the last figure, the term skarbrot (gap in a seam between planks) seems to be a calque on hiatus (gap between vowels), even though it would appear that the scribe in the poetic example has produced its opposite, namely elision. The prose is unequivocal, however, and there can be no doubt that its meaning is that all vowels in ‘Skúla er’ should be pronounced. It is possible that the scribe wanted to produce a graphic anomaly to match the vice described in the prose.

304 The couplet in 3GT contains a semantically indifferent variant (setr for dregr). If there is intertextual dependency between 5GT and 3GT, this variant suggests that the source may have been mediated by memory. 305 In skaldic practice and according to the principle of the sonority minimum, only nominative -r (as in Elfr) is actually optional as part of the hending, whereas -t is not, at least not in this cluster (-rmt). See p. 22 n. 6.

3.6 The First, Third and Fifth Grammatical Treatises

191

The Latinate characteristics of 5GT set it apart from nativizing texts on poetics. Unlike 3GT and 4GT, however, it imitates the nativizers by using no Latin terminology. Skarbrot alone appears to be a calque (on hiatus), and not a very transparent one. In further contrast to 3GT and 4GT, 5GT does not seem to be systematically organized. The first figure may be a vice, since the words ‘Snyrða þykkir heldr bragarbót en spell’ (snyrða is considered a poetic virtue rather than a vice) appear to express a contrast to the previous figure. Snyrða is clearly described as a virtue, and skarbrot as a vice. There is no indication of a shift from a section on virtues to one on vices, and it would seem that these are simply mixed, unlike in Latin treatises. The use of Latinate concepts is thus not paired with the use of Latinate organizing principles. 5GT is a hybrid, sharing characteristics with both the nativizing and the Latinizing texts. It may therefore tell us much about the process that resulted in the shift from a nativizing to a Latinizing approach, provided that it can be historically contextualized. I believe that this is possible. The use of the same poetic example to illustrate the same concept, albeit under different names ((-)garð and macrologia), in 5GT and 3GT suggests that the texts are related, either by having the same author or by one author using the text of the other. The calque on hiatus indicates that the author of 5GT had access to a Latin text that treats this figure, since 3GT has omitted this term, even though the phenomenon is described there. The two likely candidates for a Latin source to hiatus in medieval Iceland are Barbarismus with commentary and book one of Isidore’s Etymologiae. If 5GT used 3GT and took the concept of (-)garð with its poetic example from the description of macrologia, it is difficult to see why the author would in addition search out a Latin text describing hiatus, which is already described, but not named, in 3GT. To go through this trouble only to then ignore the organizing principles of 3GT and the Latin texts would seem strange indeed. Rather, the influence is likely to have gone the other way, either by Óláfr writing 5GT before 3GT or by his use of 5GT before or during the composition of 3GT. The last poet mentioned in 5GT, Óláfr Leggson svartaskáld, is known to have been in Bergen in 1231, where he killed Snorri’s son Jón murtr in a drunken brawl.306 This circumstance appears not to have harmed his relations with the Sturlung family as much as one might expect, since he was later quoted by Ólafr Þórðarson in 3GT.307 The quotation in 5GT must probably date from his Norwegian sojourn or afterwards, since in it he praises Hákon’s earl and would-be rival Skúli (d. 1240). This gives us a post quem for 5GT of c. 1231. The author is thus contemporaneous with Óláfr Þórðarson hvítaskáld – our Óláfr – and clearly had similar interests. Óláfr himself is a likely candidate for authorship, and if not him, then someone moving in the same circles, discussing the same matters, reading the same books and feeling the same

306 SkP 3, p. 311. 307 SkP 3, p. 317.

192

3 Grammatical Literature

urge to position himself as a more Latinate kind of vernacular grammarian than Óláfr’s imposing uncle Snorri. Óláfr’s use or authorship of 5GT implies that the assertive and elaborate appropriation of Latin learning in 3GT did not arrive out of the blue. It had been preceded by less successful attempts, and it is noteworthy that it should have taken such a prolonged intellectual effort to arrive at the seemingly most obvious way of importing Latin learning, namely that of translation. This can, I believe, be explained by the staunchly traditionalizing character of skaldic poetry and the discourse that evolved around it. Snorri displayed his Latin learning with vague references to Latin grammarians and poets, but his analyses remained pragmatic. With Snorri’s Edda, the nativizing approach had reached its apex and Óláfr appears to have wanted to do something different. 5GT is evidence of an early stage in that process, adopting Latin concepts but retaining the nativizing feature of using vernacular terminology. 3GT is the final product, and the strong rhetorical emphasis on the greatness of local poetics suggests that Óláfr was acutely aware of the experimental nature of his undertaking. With 3GT, this approach was established as a possible one, and when 4GT was written as a continuation of 3GT in the following century, the rhetorical edge towards Latin is gone.

3.7 Summary In this chapter, we have seen how Snorri availed himself of his models, expanding, structuring and subdividing them if they were of local origin (Háttalykill and Litla Skálda), but tacitly importing and distorting them if they were Latin (Horace and grammatical treatises). In comparison to Litla Skálda, his mythological emphasis is marked. We have also observed how he used his poetic sources, aligning them with Biblical narrative in Gylfaginning and extracting mythological motifs from them in Skáldskaparmál, though in doing so, he probably relied more on tradition than has sometimes been supposed. We have noted that CR may well have Snorri to thank for its existence, although the poems themselves are clearly older. In general, Snorri’s reception of the poetic tradition was deeply influenced by grammatical learning – to a large extent probably not directly so, but rather as it had been internalized within the learned poetic tradition itself for at least some two or three generations. He went further than any of his predecessors in his structured and comprehensive description of poetics and mythology, however. These observations are strongly at odds with an older interpretation of why Snorri composed his Edda, namely that he did so out of fear that the skaldic tradition was dying out. The data reviewed in this and the two previous chapters suggest the opposite explanation, namely that learned poetics had been gaining ground for almost a century and that Snorri continued this trend. After this study of Snorri the nativizing grammarian, we considered two Latinizing grammarians, namely the First and the Third, as well as the hybrid Fifth Grammarian. The First Grammarian was active around 1150 and we noted that although he quotes

3.7 Summary

193

skaldic poetry only twice, that tradition nonetheless had a profound impact on the character of his treatise. The Fifth Grammarian, who was either Snorri’s nephew Ólafr Þórðarson or someone close to him, took the first step towards a Latinizing poetics, using concepts derived from Latin but retaining an exclusively Icelandic terminology. In 3GT, Óláfr took full leave of his uncle’s native ways by translating texts by Priscian and Donatus. Even though his undertaking was derivative in so far as he translated Latin texts, in some ways it was even more assertive than Snorri’s: Ólafr claimed that Norse runes and poetry were at least as respectable as their Latin counterparts, and he ingeniously ‘proved’ this to be the case by replacing Latin letters and hexameters with runes and skaldic stanzas. In the end, Snorri’s programme was the more pragmatic and successful one, importing structures and systematic analyses from Latin but basing the discussion on the actual characteristics of skaldic poetry. This may to some extent account for the fact that Snorri’s Edda and its reworkings are found in a wealth of early modern paper manuscripts, and that its stylistics are largely retained in late medieval and early modern Icelandic rímur. Now, however, it is time to leave the grammarians behind and to move to the other branch of Old Icelandic literature which has made a profound impact on posterity, namely the sagas. Most of these were composed in a prosimetrical form and much of the authors’ creativity revolved around that form. This will be the topic of the next chapter.

4 Prosimetrical Narrative This chapter shifts our focus from poetry and grammatical literature to sagas. Modern readers of the sagas would be well advised not to engage too deeply with the hopelessly complex poetry in the sagas if they wish for the experience to bear any resemblance to other forms of leisure. This was not how medieval Icelanders read the sagas, however. The saga authors did their best to produce a rich prosimetrical interlace and the stanzas are often integral to the action in the sagas or to their historical credibility. These authors relished the verse and expected their readers to do likewise. They gathered as much poetry by their protagonists as they could, and when they ran out, they occasionally composed additional poetry themselves. Saga literature, therefore, has more to do with poetry than one might initially think. Before reaching the heights of classical saga narrative, the authors needed to create the form itself. Many factors were involved in this process, and equal justice cannot be done to all of them here. This chapter investigates the importance of poetry for the emergence of the sagas in the form in which we have them, with particular emphasis on the development of their prosimetrical form and the methods used for pseudonymous poetic composition. Pseudonymous composition was not a standard procedure amoung saga authors. On the contrary, it appears rather to have been the exception than the rule. But identifiable cases remain many, and since pseudonymous stanzas are the work of the saga authors they are apt to shed light on their literary interests and their methods. In this chapter, I first treat the age and development of the prosimetrical form. I then turn to a description of the corpus and the methods to be used in the remainder of the chapter, with particular focus on the circumstances that led Icelandic authors to compose pseudonymous poetry and those circumstances that would not. After this I analyse examples of pseudonymous composition which can be identified on the basis of metrical archaization, and in this context I also discuss the authorship of Egils saga. This is followed by a treatment of pseudonymous archaization of diction. Finally, I focus on what may be the most gifted pseudonymous poet of all, pseudo-Grettir, whose creativity is remarkable enough to merit its own analysis.

4.1 Age and Development of the Prosimetrical Form The textual record suggests that around 1200 and in the decades that followed, historical, non-fantastic narrative that had earlier appeared as either prose alone or poetry alone came to be presented in a prosimetrical form. For fantastic, pre-settlement narrative, which would later be represented in fornaldarsögur drawing on eddic poetry, we have no textual record in the twelfth century, and I https://doi.org/10.1515/9783110643930-005

4.1 Age and Development of the Prosimetrical Form

195

therefore postpone the discussion of this textual tradition until after a description of twelfth- and early thirteenth-century texts. The historiographical material from the period up to c. 1220 is not rich, but if both prose and poetry is taken into account and if the related genre of indigenous hagiography is drawn into the comparison – through Óláfs saga helga, it partly overlaps anyway – a relatively clear picture emerges.

4.1.1 Prose, Poetry and Prosimetra up to the Mid-Thirteenth Century The earliest datable historiographical text, Íslendingabók (c. 1130), quotes only one couplet, and the Norwegian Ágrip (c. 1190) quotes six couplets, half stanzas or stanzas, but since the manuscript is incomplete, the original number may have been slightly higher.1 An alternative form of historiography in the twelfth century was purely poetic, as found in the poems Rekstefja, Óláfs drápa Tryggvasonar, Jómsvíkingadrápa¸ Krákumál and possibly Búadrápa.2 Translated hagiography appears to have been composed in prose or poetry only. Translated hagiography in prose was a rich genre, clearly attested by manuscripts from the late twelfth and early thirteenth centuries.3 Plácitúsdrápa stands out for being a hagiographic translation into poetic form. It is a translation of Vita sancti Eustacii in the form of a traditional skaldic praise-poem, preserved as a consecutive text in the oldest manuscript containing skaldic poetry (c. 1200).4 Several twelfth-century hagiographical or otherwise religious poems are not translations in a strict sense, but treat foreign material. Having been collected in the manuscripts AM 649 4to (after Jóns saga postula: Níkulás Bergsson’s Jónsdrápa postula, Gamli

1 See Ágrip af Nóregskonunga sǫgum. Fagrskinna – Nóregs konunga tal, ed. Bjarni Einarsson. Íslenzk fornrit, 29 (Reykjavík: Hið íslenzka fornritafélag, 1985), pp. v, 51 n. 1, 54 n. 10. 2 On the historiographical poems, see Bjarne Fidjestøl, ‘Sogekvæde’, in Deutsch-nordische Begegnungen. 9. Arbeitstagung der Skandinavisten des deutschen Sprachgebiets 1989 in Svendborg, ed. Kurt Braunmüller and Mogens Brøndsted (Odense: Odense University Press, 1991), pp. 64–65; see also Guðrún Nordal, ‘Samhengið í íslenskum fornbókmenntum’. 3 The collection of hagiographic fragments AM 655 4to has several fragments dating to c. 1200 and one (AM655 4to IX) dating to c. 1150–1200 and containing Placidus saga, Blasius saga and Matheus saga postola (ONP Registre, p. 458–60); AM 645 4to (c. 1220 and, from Martinus saga, c. 1225–1250; ONP Registre, p. 458) has Jarteinabók Þorláks biskups, Clemens saga, Pétrs saga postola, Jakobs saga postola, Bartholomeus saga postola, Matheus saga postola, Andreas saga postola, Páls saga postola, Niðrstigninga saga and Martinus saga biskups; AM 677 4to (c. 1200–1225; ONP Registre, p. 461) has Gregory’s Dialogues in translation, also a hagiographical work. 4 The vita exists also as a prose text, but it has as yet not been possible to determine whether the prose or poetic translation came first. The C-redaction of the prose saga can be shown to have been influenced by the drápa, but the other redactions may be older (Plácidus saga. With an Edition of ‘Plácitus drápa’ by Jonna Louis-Jensen, ed. John Tucker. Editiones Arnamgnæanæ, B 31 (Copenhagen: Reitzel, 1998), pp. ciii–cxxv).

196

4 Prosimetrical Narrative

kanóki’s Jónsdrápa, Kolbeinn Tumason’s Jónsvísur)5 and B (after Skáldskaparmál: Gamli kanóki’s Harmsól, Leiðarvísan),6 they do not seem to derive from a prose context. Similarly, Einarr’s hagiographical poem Geisli seems to have been introduced to a prose context in the thirteenth and fourteenth century.7 Indigenous prose hagiography appears to have begun somewhat later than historiography but follows roughly the same trajectory with regard to poetic quotation. The indigenous bishops’ sagas Jóns saga helga (shortly after 1200)8 and Þorláks saga (c. 1200)9 contain no poetry; Hungrvaka (c. 1200–1215)10 contains one stanza; Páls saga (c. 1200–1215)11 contains four; and Kristni saga (c. 1240–50)12 contains eleven.13 In historiography and hagiography alike, then, we seem to be dealing with a division into prose and poetry, with the first faint beginnings of a mixed form in Ágrip within historiography and somewhat later in Páls saga and Kristni saga within hagiography. In the following discussion, I treat historiography only, since the material becomes abundant in the thirteenth century and historiography develops a much richer and more varied use of poetic quotations than does hagiography. Around 1200, some sagas begin to feature more numerous poetic quotations. Sverris saga displays 18 poetic quotations, though exactly how many of these were found in Karl Jónsson’s original version of the saga from the 1180s – the so-called Grýla – is probably impossible to say.14 In any event, it is highly unlikely that they were more numerous than those of the preserved saga. Óláfs saga Tryggvasonar by Oddr munkr also has 18, but it should be noted that both of these texts are considerably longer than the ones discussed above and that the form is still a far cry from the later rich prosimetra.15

5 SkP 7, pp. 66, 133, 223. 6 SkP 7, p. 70, 140. 7 SkP 7, p. 6. 8 Biskupa sögur, 1, ed. Sigurgeir Steingrímsson et al., 1, p. ccxiv. 9 Biskupa sögur, 2, ed. Ásdís Egilsdóttir. Íslenzk fornrit, 16 (Reykjavík: Hið íslenzka fornritafélag, 2002), p. xxxii. 10 Biskupa sögur, 1, ed. Sigurgeir Steingrímsson et al., 1, pp. xxiv–xxix. 11 Biskupa sögur, 1, ed. Sigurgeir Steingrímsson et al., 1, p. cxxix–cxxxii. 12 Biskupa sögur, 1, ed. Sigurgeir Steingrímsson et al., 1, p. cliv. 13 While later hagiography is not very relevant for the question at hand, it may be noted that Árna saga (1300–1310; Biskupa sögur, 3, ed. Guðrún Ása Grímsdóttir, pp. xxii–xxvii) and Lárentíus saga (mid-fourteenth century; Biskupa sögur, 3, ed. Guðrún Ása Grímsdóttir, pp. lxiv–lxxv) contain no poetry. This may perhaps be due to influence from translated hagiography. The matter of the original amount of poetic quotation in the fourteenth-century Guðmundar sögur is complex and will be described in the coming edition in Íslenzk fornrit. 14 See Sverris saga, ed. Þorleifur Hauksson. Íslenzk fornrit, 30 (Reykjavík: Hið íslenzka fornritafélag, 2007), pp. lv–lx. 15 For the debate on the date and original scope of Sverris saga, see Theodore Andersson, ‘Kings’ Sagas (Konungasögur)’, in Old Norse-Icelandic Literature. A Critical Guide, ed. Carol J. Clover and John Lindow. Islandica, 45 (Ithaca: Cornell University Press, 1985), pp. 214–15, and Sverris saga, ed.

4.1 Age and Development of the Prosimetrical Form

197

Six fragments of the so-called Oldest Saga of Óláfr Haraldsson have been dated to c. 1225, and the redaction itself (the *Oldest Saga) to c. 1200. The fragments are extensive enough to verify that the style is the full-blown prosimetrum of the somewhat later kings’ sagas (they contain the remains of seven stanzas and the first line of a drápa).16 This is further confirmed by the somewhat later Legendary Saga of Óláfr (c. 1225, manuscript c. 1225–1250), and the often exact correspondence in wording shows that the Legendary Saga is an abbreviated witness to the same redaction as the fragments of the *Oldest Saga.17 The relationship between the *Oldest and the Legendary Saga is relevant to the question of whether the emergence of skaldic prosimetra should be seen as primarily driven by Icelandic or Norwegian interests. Given the cultural interaction of Icelandic and Norwegian intellectuals, this distinction cannot be absolute, but Old Norse literature would have been of little significance if it were not for the literary interests of some prominent Icelandic chieftains. These chieftains were not isolated, however, and Icelandic poets and authors working for Norwegian patrons make for a common pattern in Old Norse literature. In order to understand the extent and the character of this dynamic’s influence on the emergence of the prosimetrical form, it is of some importance to discuss whether the *Oldest saga, which may be the earliest example of a fully developed skaldic prosimetrum, should be understood as an Icelandic or a Norwegian product. While the linguistic realization of the Legendary Saga is clearly Norwegian, its prosimetrical form goes back to the *Oldest Saga.18 The provenance of the *Oldest Saga is

Þorleifur Hauksson, pp. xxii–xxiv, lv–lxiv. The case of Óláfs saga is particularly complicated, since Oddr originally composed it in Latin, and only one stanza seems to have belonged in the Latin original. The two principal versions, however, contain 17 ½ stanzas with surrounding prose that are common to both of these otherwise quite different redactions, and which may therefore be assumed to have belonged to the original translation. That translation seems to have taken place no later than the beginning of the thirteenth century (see Sveinbjörn Rafnsson, Ólafs sögur Tryggvasonar, pp. 35–52). Sveinbjörn Rafnsson suggests, based on earlier scholarship, that the stanzas and surrounding prose are interpolated from an *Old Saga of Óláfr Tryggvason in Icelandic, but I contend that it is equally possible, and perhaps more likely, that the surrounding prose may have been added to make the stanzas fit the narrative, and that no older saga need necessarily be posited. 16 Otte bruddstykker av den ældste saga om Olav den Hellige, ed. Gustav Storm (Christiania: Det norske historiske kildeskriftsfond, 1893), pp. 2, 4, 5, 6, 7, 10. Dating according to ONP Registre, p. 351. Note that fragment seven and eight of Storm’s edition have since been shown to belong to another text (see Andersson, ‘Kings’ Sagas’, pp. 212–13 and references there). The form of other historiographical texts that have since been lost cannot be reconstructed with a sufficient degree of certainty (see Andersson, ‘Kings’ Sagas’, pp. 214–15 and references there). 17 Theodore Andersson, The Growth of the Medieval Icelandic Sagas (1180–1280) (Ithaca: Cornell University Press, 2006), pp. 46–47. 18 On the linguistic form, see Marius Hægstad, ‘Um maalet i Ólafs saga hins helga’, in Olafs saga hins helga efter pargamenthaandskrift i Uppsala Universitetsbibliotek, Delagardieske samling nr. 8ᴵᴵ, ed. Oscar Albert Johnsen (Kristiania: Jacob Dybwad, 1922), pp. xxviii–lvii.

198

4 Prosimetrical Narrative

generally held to be Icelandic, because of its marked focus on Icelandic poets.19 In light of the overall literary development, this is a fair guess, albeit one which over time has perhaps come to be treated as somewhat more of a fact than it deserves. I believe that at the very least we may presuppose a considerable Icelandic involvement in the shaping of the *Oldest Saga’s innovative form; much of its very novelty amounts to giving much more attention to Icelandic poets and their poems than before. Norwegian involvement need not be excluded, however; one may for instance recall that the Icelandic abbot Karl Jónsson was commissioned to write the royal history of Sverrir when Karl was in Norway in the 1180s.20 As part of the development of the prosimetrical form, however, the *Oldest saga is probably best described as a witness to Icelandic aestetic and historical interests. Another possible precursor to the full prosimetrical form is Orkneyinga saga, dating to the beginning of the thirteenth century and quoting some 82 stanzas, although all extant manuscripts represent a reworked version and the original number of stanzas may thus have been lower.21 The great kings’ saga compilations of the 1220s – Morkinskinna, Fagrskinna and Heimskringla – quote hundreds of stanzas. While Morkinskinna and Heimskringla are clearly Icelandic, Fagrskinna was composed in Norway, as shown by repeated references to Iceland as ‘there’ rather than ‘here’.22 It is probably impossible to tell whether the author was Icelandic or Norwegian, but whatever the case may be, this is the only saga clearly composed in Norway that exhibits a

19 This argument was put forth by Konrad Maurer, Ueber die Ausdrücke. Altnorwegische, altschwedische und isländische Sprache (München: Straub, 1867), pp. 76–77. Gustav Storm added the argument that a passage from Fóstbrœðra saga had been interpolated into the text, but few scholars would accept this line of influence today (Otte brudstykker af den ældste saga om Olaf den hellige (Christiania: Det norske kildeskriftsfond, 1893), p. 23). Anne Holtsmark notes that in a couple of instances, there is confusion between r and n in the manuscript and draws the conclusion that the exemplar of the Legendary Saga must have contained insular rs, which occur sporadically in Norwegian manuscripts up to c. 1200 but are unknown in Iceland (Legendarisk Olafssaga, ed. facs. Anne Holtsmark. Corpus Codicum Norvegiorum medii aevi. Quarto serie, 2 (Oslo: S.T.U.A.G.N.H, 1956), p. 9; cf. Hreinn Benediktsson, Early Icelandic Script as Illustrated in Vernacular Texts from the Twelfth and Thirteenth Centuries. Íslenzk handrit. Series in Folio, 2 (Reykjavík: The Manuscript Institute of Iceland, 1965), p. 21 n. 2). Confusion between n and r is, however, relatively common also in later, purely Icelandic transmission, and the validity of Holtsmark’s argument is therefore doubtful (see, for instance, SkP 3, pp. 124–25: mæran/mœrar). Although many scholars refer to the Icelandic provenance of the *Oldest Saga, I am not aware of others who present additional evidence in the question. 20 Sverris saga, ed. Þorleifur Hauksson, pp. xxii–xxiv, 3. 21 Orkneyinga saga, ed. Finnbogi Guðmundsson. Íslenzk fornrit, 34 (Reykjavík: Hið íslenzka fornritafélag, 1965), pp. vii–ix, though the editor suggests an earlier date for the original redaction (pp. xc–cviii). On the reworkings and the version represented by the manuscripts, se ibid, pp. xvi–lxxxi, cix. 22 Gustav Indrebø, Fagrskinna (Kristiania: Grøndahl, 1917), p. 263.

4.1 Age and Development of the Prosimetrical Form

199

rich prosimetrical form.23 This form became typical not only of kings’ sagas, but also of many sagas of Icelanders and fornaldarsögur. The latter are legendary prosimetra containing eddic poetry, treating the period before the settlement of Iceland. We are not able to follow their written record further back than 1200, but I shall argue on comparative grounds below that the emergence of skaldic prosimetra may be partly indebted to the oral antecedents of this group of texts. Finally, the explosion of poetic quotation in the sagas had its counterpart in the sphere of poetics: as has been shown above the Litla Skálda used by Snorri quotes only three and a half stanzas or stanza-like units, whereas Skáldskaparmál quotes some 400. While these works cannot easily be compared to the sagas, they possess the same pattern and they occur within the same timeframe. Presumably, they also reflect the same interests at least to some extent. In the period c. 1220–1260 or so, the prosimetrical form seems to have dominated in Icelandic literature. The evidence of the sagas of Icelanders complicates this claim somewhat, since, as Guðrún Nordal notes, some 22 of these (or variants thereof) contain less than five stanzas and thus hardly merit to be counted as prosimetra.24 The earliest securely datable saga, however – Egils saga – is prosimetrical, and so are most other sagas that are likely to belong to the earliest representatives of the genre. With the possible exception of Laxdœla saga, none of the prose sagas belong to what are generally considered to be the earliest sagas, although the dating of sagas is a vexed issue.25 Furthermore, saga style does not respect the boundaries between genres that modern scholars have staked out, and kings’ sagas and hybrid forms like Orkneyinga saga feature a style similar to that of the sagas of Icelanders, not least in descriptions of the interaction between Icelanders and kings. If one views this larger corpus together and takes such dates as we have into account, it seems likely that the sagas as we know them developed in a largely prosimetrical setting. The question then becomes why some of them, such as sagas from Eastern Iceland and Laxdœla saga, are in prose.

23 While Indrebø is reluctant to decide on the matter, Eyvind Fjeld Halvorsen states it as a fact that Fagrskinna was composed by an Icelander in Norway (‘Fagrskinna’, in Kulturhistoriskt lexikon för nordisk medeltid, 22 vols (Malmö: Allhems förlag, 1956–78), 4, cols 139–40). 24 Guðrún Nordal, ‘The Art of Poetry and the Sagas of Icelanders’, in Learning and Understanding in the Old Norse World. Essays in Honour of Margaret Clunies Ross, ed. Judy Quinn, Kate Heslop, and Tarrin Wills (Turnhout: Brepols, 2007), p. 221; see also eadem, ‘Skaldic Citations and Settlement Stories as Parametres for Saga Dating’, in Dating the Sagas. Reviews and Revisions, ed. Else Mundal (Copenhagen: Museum Tusculanum Press, 2013), pp. 195–212). 25 See Einar Ól. Sveinsson, Dating the Icelandic Sagas. An Essay in Method (London: Viking Society for Northern Research, 1958); Theodore M. Andersson, The Growth of the Medieval Icelandic Sagas (1180–1280) (Ithaca: Cornell University Press, 2006); Dating the Sagas. Reviews and Revisions, ed. Else Mundal (Copenhagen: Museum Tusculanum Press, 2013).

200

4 Prosimetrical Narrative

In many, perhaps most, of the 22 prose sagas, the form may not reflect an active choice so much as the circumstance that their authors did not have much poetry to draw on. A special case is Laxdœla saga, since its author quotes only very simple poetry, and only five times.26 He twice refers to more prestigious and complex poems that are relevant to the narrative, but chooses not to quote them.27 This author apparently avoided poetry that would stop the narrative flow and become a topic of interpretation in its own right. Laxdœla saga contains a number of indications that enable us to date it to c. 1230–1260: It mentions people and conditions down to the 1220s in the past tense, giving a post quem of c. 1230.28 It discusses trial by ordeal as a contemporary reality, which is unlikely to reflect perceptions very long after 1247, when Cardinal William of Sabina implemented the prohibition of the Fourth Lateran Council in the archbishopric of Niðaróss.29 1260 seems a reasonable ante quem. Finally, the saga is found in one fragment from the second half of the thirteenth century and two from around 1300, and four other sagas refer to Laxdœla saga (Eyrbyggja saga, Gunnlaugs saga, Grettis saga and the Greater Saga of Óláfr Tryggvason).30 This all suggests that it is a relatively early saga. Laxdœla saga is the most courtly of the sagas of Icelanders, and it dates to the period when courtly literature was being translated into prose at the court of Hákon Hákonarson (from 1226 onward).31 Because of the close contacts between members of the Icelandic elite and Hákon’s court, it seems reasonable to assume that Laxdœla saga’s prose bias reflects the prose form of translated courtly literature.32 In general, the combined evidence of the genres that feature skaldic prosimetra supports the contention that the prosimetrum was the default choice for sagas from the early to at least the middle of the thirteenth century, allowing for some delay within hagiography and for some variation based on the availability of relevant poetry and the impact of translated prose.

26 Laxdœla saga, ed. Einar Ól. Sveinsson. Íslenzk fornrit, 5 (Reykjavík: Hið íslenzka fornritafélag, 1934), pp. 76, 96, 194, 198. 27 The author names two drápur which he does not quote, Ulfr Uggason’s Húsdrápa and the commemorative drápa that Arnórr jarlaskáld composed about Gellir Þorkelsson (Laxdœla saga, ed. Einar Ól. Sveinsson, pp. 80, 229). 28 Laxdœla saga, ed. Einar Ól. Sveinsson, pp. xxv–xxvi. 29 Laxdœla saga, ed. Einar Ól. Sveinsson, p. xxvi. 30 Laxdœla saga, ed. Einar Ól. Sveinsson, p. xxvii; cf. ONP Registre, p. 323. 31 Laxdœla saga, ed. Einar Ól. Sveinsson, p. xxix. 32 Cf. Guðrún Nordal, ‘The Art of Poetry and the Sagas of Icelanders’, pp. 226–27; eadem, ‘Skaldic Poetics and the Making of the Sagas of Icelanders’, in New Norse Studies. Essays on the Literature and Culture of Medieval Scandinavia, ed. Jeffrey Turco. Islandica, 58 (Ithaca, NY: Cornell University Library, 2015), pp. 138–39.

4.1 Age and Development of the Prosimetrical Form

201

4.1.2 Reykjahólar 1119 The background to the prosimetrical ideal remains hazy, as is the precise relationship between the seemingly gradual increase before c. 1220 and the sudden burst in the following decade. Studies of the development of the prosimetrical form have chiefly been concerned with fornaldarsögur. Although stories about some of the heroes of these sagas can be shown to have been told in the twelfth century (there are, for instance, references to some of them in Háttalykill in the 1140s), the form or degree of fixity of such accounts is unknown. In scholarly literature on the subject, considerable emphasis is often placed on a report in Þorgils saga ok Hafliða about the entertainment given at a wedding at Reykjahólar in 1119.33 There, we hear, two sagas were told, and what appears to be their prosimetrical form is mentioned. Since the passage has been so important to the modern discussion of the topic, I give it in full in order to then analyse its implications for the development of Icelandic prosimetra. The saga account probably dates to the first half of the thirteenth century or somewhat later.34 It is thus at about a hundred years’ remove from the events, and possibly more. It reads: Frá því er nǫkkut sagt, er þótti lítit til koma,35 hverir þar skemtu eða hverju skemt var. Þat er í frásǫgn haft, er nú mæla margir í móti ok látaz ekki vitat hafa, því at margir ganga duldir hins sanna ok hyggja þat satt, er skrǫkvat er, en þat logit, sem satt er: Hrólfr frá36 Skálmarnesi sagði sǫgu frá Hrǫngviði víkingi ok frá Óláfi liðsmannakonungi ok haugbroti Þráins37 ok Hrómundi Gripssyni, ok margar vísur með. En þessari sǫgu var skemt Sverri konungi, ok kallaði hann slíkar lygisǫgur skemtiligstar38; ok þó kunna menn at telja ættir sínar til Hrómundar Gripssonar.39 Þessa sǫgu hafði Hrólfr sjálfr saman setta.40 Ingimundr prestr sagði sǫgu Orms Barreyjarskálds ok vísur margar ok flokk góðan við enda sǫgunnar, er Ingimundr hafði ortan, ok hafa þá41 margir fróðir menn þessa sǫgu fyrir satt.42

33 Thus, for instance, SkP 8, pp. lviii–lix, and Joseph Harris, ‘The Prosimetrum of Icelandic Saga and Some Relatives’, in Prosimetrum. Crosscultural Perspectives on Narrative in Prose and Verse, ed. Joseph Harris and Karl Reichl (Cambridge: D. S. Brewer, 1997), pp. 134–35; for a more recent discussion of the passage and further references, see Ralph O’Connor, ‘History or Fiction? Truth-Claims and Defensive Narrators in Icelandic Romance-Sagas’, Mediaeval Scandinavia, 15 (2005), pp. 133–39. 34 See Peter Foote, ‘Notes on the Prepositions of and um(b) in Old Icelandic and Old Norwegian’, Studia Islandica, 14 (1955), p. 68; idem, ‘Sagnaskemtan: Reykjahólar 1119ʹ, in Peter Foote, Aurvandilstá, ed. Michael Barnes, Hans Bekker-Nielsen, and Gerd Wolfgang Weber (Odense: Odense University Press, 1984), p. 79. 35 Reading according to Br, V. Kålund here adopts the conjecture er þó er lítil tilkoma (even though it is of minor importance). 36 frá: af H, Br, V, Ip. 37 Þráins: Þráins berserks H, Br. 38 skemtiligstar: skemtiligar Ip. 39 Gripssonar: Greipssonar Br, V, missing Ip. 40 Þessa–setta missing Ip. 41 þá: þó Br, því V. 42 Sturlunga saga, ed. Kålund, p. 22.

202

4 Prosimetrical Narrative

Even though it was thought of minor importance, it has been related who entertained there and with what they entertained. What is told in the report is now disputed by many, who claim that they have heard [lit. ‘known’] of nothing of the sort, since many are ignorant of the truth and think that to be true which has been made up and that to be mendacious which is true: Hrólfr from Skálmarnes recited43 a saga about Hrǫngviðr the Viking and about Óláfr king-of-warriors and the breaking of the cairn of Þráinn and Hrómundr Gripsson, and many stanzas with it. King Sverrir was entertained with this saga and he said that such lying tales were the most amusing ones; and yet people can reckon their ancestry to Hrómundr Gripsson. Hrólfr had composed this saga himself. The priest Ingimundr recited the saga of Ormr Barreyjarskáld and many stanzas and a good poem which Ingimundr had composed at the end of the saga, and many wise men hold this saga to be true.

The author strongly insists on the truthfulness of his account, and the passage has often been taken as an indication of the existence of prosimetrical sagas in 1119. There are problems with this assumption, however, hinging on which parts of the account the author is claiming to be true and which parts serve to corroborate that truth. But first a few words about the sagas in question. Both of the sagas mentioned are now lost, but a rímur-poem about Hrómundr Gripsson has been preserved.44 It is clear from the poem that the saga would have been a fornaldarsaga and thus, in all likelihood, any poetry connected to it would have been eddic. There are no traces whatsoever of the saga about Ormr Barreyjarskáld (poet from the island of Barra in the Hebrides), but some dróttkvætt poetry by him has been preserved in Skáldskaparmál.45 Ormr’s dates are not known, but he is likely to have lived in the tenth or eleventh century. The poetry recited by Ingimundr in connection to the saga would in all likelihood have been skaldic. Ormr’s Hebridean nickname suggests a skaldic, sagas-ofIcelanders date of action (tenth–early eleventh century) rather than an eddic, fornaldarsögur date of action (before the settlement of Iceland), and all his preserved poetry is skaldic. Because of my focus on skaldic poetry and since the description of the form of *Orms saga is somewhat fuller than that of the saga about Hrómundr, I will devote this discussion primarily to *Orms saga, but the argument has some implications for the trustworthiness of the description of the saga about Hrómundr as well. First of all, it is important to consider what form the author is actually envisaging. The expression ‘ok vísur margar ok flokk góðan við enda sǫgunnar’ (and many stanzas

43 Even though the verb segja literally means ‘to say’, Hermann Pálsson notes that segja sǫgu generally means to read aloud from a written text. This agrees well with the note that Hrolfr had ‘composed’ (saman setta) the saga himself, a calque on Latin componere which also typically refers to written composition (Sagnaskemmtun íslendinga (Reykjavík: Mál og Menning, 1962), p. 52). 44 See Ursula Brown, ‘The Saga of Hrómund Gripsson and Þorgilssaga’, Saga-Book of the Viking Society, 13 (1947–48), pp. 51–77. 45 See Guðrún Nordal, Tools of Literacy, p. 283.

4.1 Age and Development of the Prosimetrical Form

203

and a good poem at the end of the saga) is ambiguous, since it is not clear whether ‘at the end of the saga’ refers only to the poem or to the stanzas as well as the poem. If the phrase describes both, the saga is not represented as prosimetrical, but rather as a kind of opus geminatum with all poetry at the end, something to which we have no real analogues in the corpus.46 If it refers only to the flokkr, however, we are dealing with a prosimetrical saga with a long poem at the end, and we do have some examples of this.47 The semantics of vísur and flokkr may be of help here. The singular vísa refers to a stanza, without further specification. The plural vísur, however, is generally synonymous with flokkr (poem without refrain). Illustrative examples include: ‘hann orti [. . .] flokk þann, er Nesjavísur eru kallaðar’48 (he composed the flokkr which is called Nesjavísur), ‘Einarr orti um Grégóríúm Dagsson flokk þann, er kallaðar eru Elfarvísur’49 (Einarr composed the flokkr which is called Elfarvísur about Grégóríús Dagsson). Vísur can also refer to a number of stanzas composed in response to a given occasion, but again, the semantics remain very close to those of flokkr. This can, however, hardly be what vísur means here, because if it were, there would be no obvious distinction between vísur and flokkr, and there would have been no point in mentioning both. The difference may be one of authorship: the vísur were attributed to the protagonists of the saga whereas the flokkr was Ingimundr’s own composition (‘ortan’ (composed) correlates only with ‘flokk’). If so, one would expect that the vísur would form part of a prosimetrum, since at the end of sagas, we find long poems, not single stanzas by the protagonists. The alternative is to assume that the author is indeed referring to a lost type of saga of a peculiar opus geminatum type. To posit such a form e silentio when the passage may easily be read in conformity with known forms seems unwarranted, however. Furthermore, there are good reasons to doubt that the memory of obsolete saga forms would have been preserved for very long. Discussions about the diachronic development of form seem generally to have been restricted to poetic form rather than the form of prose or prosimetra. In the famous source-critical evaluations in the prologues to Heimskringla and the Separate Saga of Saint Óláfr, for instance, discussions of historical veracity regard content, not form, except in one place in the prologue to the Separate Saga when poetic form becomes a relevant factor for the preservation of knowledge.50 At that point, the author treats poetry and poetry alone,

46 Hákonar saga góða in Heimskringla and some manuscript versions of Egils saga, Ǫrvar-Odds saga and Arngrímr’s Guðmundar saga give a substantial amount of poetry at the end, but these sagas are also prosimetrical in themselves. 47 See previous note. 48 ‘Saga Óláfs konungs hins helga’, ed. Johnsen and Jón Helgason, p. 91. 49 Heimskringla, ed. Bjarni Aðalbjarnarson, 3, p. 359. 50 Heimskringla, ed. Bjarni Aðalbjarnarson, 2, p. 422.

204

4 Prosimetrical Narrative

not some combination of poetry and other material. Although many saga authors exhibit an elaborate prose style, references to and between sagas mention content, not form. The U rubric to Snorri’s Edda is the only reference of which I am aware, apart from the passage treated here, that focuses on the form of a prose or prosimetrical work. But it describes the most authoritative poetic treatise of its day and is therefore not easily comparable to the discourse concerning sagas (see pp. 109–10, 115–18). The overall differences in the Icelandic treatment of sagas and poetry thus suggest that these authors did not so much conceptualize the sagas in terms of form as in terms of content. It is therefore unlikely that they would have attempted to trace the evolution of saga form. In addition, even with regard to poetic form, they sometimes misunderstood or misconstrued obsolete forms to suit a contemporary understanding, as in the case of the hierarchical hending principle of the early poets (so Bjarni Kolbeinsson, Snorri and Pseudo-Ragnarr; see pp. 34–37, 119–20, 228–32, 246–49). All in all, it is unlikely that the author of the passage in Þorgils saga is describing an obsolete saga form. Rather, he has in mind a prosimetrical saga of the kind found in the preserved corpus. This is supported by the reference to the saga about Hrómundr, where the wording ‘many stanzas with it’ appears to suggest a prosimetrical form. It would thus seem that the author is indeed referring to prosimetra as we know them, but since diachronic awareness of form appears to have been restricted to poetic form, and since the passage was probably written a hundred years or more after the events it describes, it is unlikely to be trustworthy on matters of form. The author focuses on the relation between fact and fiction on two levels: regarding the events at the wedding and regarding the content of sagas. He stresses who entertained at the wedding and what sagas were told. He also focuses on authorship: Hrólfr is the author of a saga, whereas Ingimundr had composed a flokkr. The author talks about ‘lying tales’, which nonetheless appear to contain some truth, whereas *Orms saga is apparently ‘true’. All of these claims can be understood as dealing with the veracity of historical events or genealogical information relating to the wedding and the sagas: these men entertained at the wedding and were the authors of these sagas, the content of these sagas relate to historical events and genealogy in this way or that way. Finally, the author repeatedly states that the sagas in question contained many stanzas, and this fact seems to have been important to him – but why? In the text, Sverrir represents those who do not believe in some sagas, whereas the author and ‘wise men’ belong to the other camp, noting that there is either some or much truth in the sagas. Scholars agree that this passage was composed sometime during the first half of the thirteenth century or slightly later, and during this period poetry was often used to authenticate historical narrative. This circumstance, in tandem with the consistent focus on truth in the passage, suggests that the repeated mention of stanzas is intended to scatter the reader’s doubts regarding the veracity of the sagas. If the author meant to focus on saga form, this would leave him alone among medieval Icelandic authors, but if his focus was on historical veracity, then he fully shared the interests of his contemporaries. The tenor of

4.1 Age and Development of the Prosimetrical Form

205

the passage, the focus on historicity in other authors and the use of poetry in the period all suggest that the mentions of stanzas are meant to encourage belief in the historical value of the sagas. In sum then, the author would probably have been unable to give a reliable report of the form of sagas told a hundred years earlier even if he had wanted to, and it is in any event unlikely that this is what he was trying to do. Rather, he wished to focus on the truth value of stories about weddings, poets and legendary heroes, and his mentions of poetry support his claim that there is some truth to be found in these stories. For these reasons, I think that the passage should at best be accorded an ancillary function in modern reconstructions of the early development of saga form.51 The real evidence must be found elsewhere.

4.1.3 A Local Background Nonetheless? Scholars have suggested that eddic prosimetra – fornaldarsögur – grew out of prose commentary to eddic poetry.52 This hypothesis is perhaps not unlikely, but no early examples of these precursors to fornaldarsögur survive. Carol Clover has argued on typological grounds that the prosimetrical form of the sagas is likely to have an oral background, since strictly prosaic narrative is rare in oral societies.53 This argument, while compelling, poses a problem: Clover’s observations cannot easily accommodate the gradual increase of poetic quotation in skaldic prosimetra unless one posits a slow breakthrough of oral narrative forms into a written tradition based on Latin prose. Her analysis also does not explain why, in fornaldarsögur and sagas of Icelanders alike, there is a relatively rich record of poetry that can plausibly be judged as spurious, to be analysed in the remainder of this chapter. This indicates that prosimetrical narrative often did not simply make its way into writing, but rather had to be shaped in accordance with literary taste. For these reasons, it is unlikely that the preceding tradition was generally prosimetrical, though such a form may have existed and served as a source of inspiration. Few examples of spurious poetry can be found in the kings’ sagas, but that does not mean that they need have a prosimetrical background in their own right. Klaus von

51 Similarly, but with a focus on fornaldarsögur, Klaus von See, ‘Das Problem der mündlichen Erzählprosa im Altnordischen. Der Prolog der Þiðriks saga und der Bericht von der Hochzeit im Reykjahólar’, Skandinavistik, 11 (1981), pp. 91–95. 52 Thus Brown, ‘The Saga of Hrómund Gripsson’; Anne Holtsmark, ‘Heroic Poetry and Legendary Sagas’, Bibliography of Old Norse-Icelandic Studies (1965), pp. 9–21; Lars Lönnroth, ‘Hjalmar’s Death-Song and the Delivery of Eddic Poetry’, Speculum 46 (1971), pp. 1–20; Paul Acker, Revising Oral Theory. Formulaic Composition in Old English and Old Icelandic Verse (New York: Garland Publishing, 1998), pp. 85–110. 53 Carol Clover, ‘The Long Prose Form’, Arkiv för nordisk filologi, 101 (1986), pp. 15–19, 27–28.

206

4 Prosimetrical Narrative

See among others has argued that long poems were often broken up by the author to serve specific functions in the narrative.54 In many cases, this is explicitly borne out by the text, through formulations like ‘as poet X says in poem Y: (stanza)’. This procedure probably accounts for the bulk of the poetry in the kings’ sagas. Furthermore, one may wonder how absent prose narrative has really been across the world’s oral societies. Other traditions to which Clover refers are precisely that: established traditions for modes of storytelling. While prose as an artful narrative mode usually develops with writing, this does not imply that people could not tell each other stories in language unbound by the restrictions of art, much like they could relate events in general. The oral ‘saga tradition’ may not have been a tradition with strong formal characteristics, but rather a manifestation of an interest in old stories. If so, ‘prose’ would in this instance signify not so much a literary form as a relative lack thereof. The foregoing overview suggests that the case for widespread use of prosimetrum in preliterate times is tenuous, and that a gradual development of this form in the decades around 1200 has more support in the sources than any possible oral antecedents. Still, there are some tantalizing hints that there may be more to the prosimetrical story than I have presented so far. These are oblique and require us to make a detour into Latin and Irish literature. In Latin literature, the prosimetrical form saw its greatest surge in the historiography of the twelfth century.55 The emergence of a similar ideal only slightly later in corresponding genres in Old Icelandic literature is striking, but no particular lines of influence have been ascertained. Furthermore, in Latin historical prosimetra, poetry mainly functioned as embellishment intended to provide moral lessons or to quote the classics for auctoritas (this is true also of the Norwegian Theodoricus’s Historia de antiquitate regum Norwegiensium, c. 1180). In other cases, the poetry was completely integrated into the narrative and presented explicitly or implicitly as the work of the

54 Klaus von See, ‘Skaldenstrophe und Sagaprosa. Ein Beitrag zum Problem der mündlichen Überlieferung in der Altnordischen Literatur’, Mediaeval Scandinavia, 10 (1977), pp. 78–82. See further idem, ‘Der Skald Torf-Einar’, Beiträge zur Geschichte der deutschen Sprache und Literatur, 82 (1960), pp. 31–43; idem, ‘Mündliche Prosa und Skaldendichtung. Mit einem Exkurs über Skaldensagas undTrobadorbiographien’, Mediaeval Scandinavia, 11 (1981), pp. 82–91; idem, ‘Das Problem der mündlichen Erzählprosa im Altnordischen’. With regard to the prose element of the prosimetra, von See concludes that in preliterate times it must have been very unstable and rudimentary, and finally: ‘Als literaturgeschichtliches Phänomen ist er [der Prosakommentar] deshalb nicht greifbar’ (p. 82). 55 See Laurentius von Durham, Consolatio de morte amici, ed. Udo Kindermann (PhD diss., Friedrich-Alexander-Universität, 1969), pp. 70–74; Friis-Jensen, Saxo Grammaticus, pp. 29–38; Jan Ziolkowski, ‘The Prosimetrum in the Classical Tradition’, in Prosimetrum. Crosscultural Perspectives on Narrative in Prose and Verse, ed. Joseph Harris and Karl Reichl (Cambridge: D. S. Brewer, 1997), pp. 56–57.

4.1 Age and Development of the Prosimetrical Form

207

author. This is the case with the two most influential prosimetrical texts in the Middle Ages, Boethius’s Consolatio Philosophiae and Martianus Capella’s De nuptiis Mercurii et Philologiae. Poetry as embellishment or as an authorial change of form is extremely rare in Old Icelandic prosimetra. The kings’ sagas place considerable emphasis on authenticating stanzas, and in the Old Icelandic corpus generally, the stanzas are attributed to the protagonists rather than to the author. The closest parallel to this within Latin historiography are quotations of epitaphs and popular poetry for documentary evidence, but such quotations are never even remotely as prominent as in Old Icelandic literature.56 Except for Saxo’s Gesta Danorum, which will be treated below, I am aware of only one Latin work which is vaguely reminiscent of Icelandic prosimetra, namely Suetonius’s De vita Caesarum (Life of the Caesars). This text was widely copied in the twelfth and thirteenth centuries, albeit mostly in truncated form.57 Although I am not aware of any clear indications of the study of Suetonius in the Norse area, the fact that his work consists of royal biographies, which is where we see skaldic prosimetra emerge, is conspicuous and may perhaps indicate direct influence.58 Even if that could conceivably be the case, however, the differences in poetic quotation remain conspicuous. In Suetonius, most poetic quotations are either a line or two that are said to have been popular at the time, a few famous words by the emperor or a quotation of general import by a famous poet, which is not historically relevant to the description. The popular verses might be taken as counterparts to authenticating quotations in Icelandic prosimetra, but since they seem mainly to be included in De vita Caesarum to add colour, this is a bit of a stretch. The quotations by the Ceasars are to some extent similar to situational quotations, since they are attributed to the protagonists, but they do not form part of the dialogue or affect the plot as they do in the sagas.

56 The nearest counterpart within Latin historiography to the Old Norse practice may be the six instances where Henry of Huntingdon translates what seems to have been an Old English poem about battles, introducing the quotations with unde dicitur (whence it is said) and similar formulations. These are, however, marginal occurrences in a chronicle where most of the poetry is Henry’s own (Henry Archdeacon of Huntingdon, Historia Anglorum. The History of the English People, ed. and trans. by Diana Greenway (Oxford: Clarendon Press, 1996), p. cvii). 57 Texts and Transmission. A Survey of the Latin Classics, ed. L. D. Reynolds (Oxford: Clarendon, 1983), pp. 399–405. 58 Per Meldahl, Norges keisere. Kongesagaen i klassisk-retorisk perspektiv. Om den Svetonius-imiterende strukturen i Snorres ‘Heimskringla’ og andre kongesagaer (Bergen: University of Stavanger, 2007), studies the similarities between Suetonius and the kings’ sagas and argues that they are due to conscious imitation. He does not, however, discuss what evidence there is of the study of Suetonius in the Norse area. Overall, the disregard for the transmission of Suetonius’s work makes many of Meldahl’s conclusions tenuous, since Suetonius was mostly circulated in truncated form, whereas Meldahl’s analysis is based on the integral text.

208

4 Prosimetrical Narrative

For these reasons, even though the possibility of influence from Suetonius should not be ruled out, his text is too different to explain why some functions of poetic quotation came to dominate the way they did. Overall, the contrasts between Latin and Old Icelandic prosimetra are such that while Latin prosimetrical historiography may have been a source of inspiration in a very general sense, other factors are likely to have been involved in the Icelandic development. These differences, however, partly break down when we get to Saxo’s Gesta Danorum from the beginning of the thirteenth century. Some of the poetry there is translated from the vernacular and consists of speeches by the protagonists. The poetry in the Gesta is formally indebted to Latin literary tradition, but functionally, it is distinctly Norse and un-Latin: authenticating or situational and attributed to the protagonists. I have found no other texts displaying a similar distribution in the Latin tradition. It is most of all reminiscent of the use and function of poetry in the fornaldarsögur, and there is some overlap in content.59 The Gesta thus seem to indicate that there was some form of vernacular, prosimetrical tradition to draw on around the year 1200. A comparison with other historically and culturally related traditions sheds further light on the prehistory of Icelandic prosimetra. The poems entered under a number of years during the tenth century in the Anglo-Saxon Chronicle are interesting in that their function appears to overlap with the royal panegyrics of skaldic poetry, but they are entered withouth further comment or framing and are therefore difficult to compare with poetic quotations in the sagas.60 Nonetheless, the lack of framing in itself represents a different mode of quoting poetry than that found in the Icelandic prosimetra. More promisingly, Irish prosimetra display a distribution of poetic functions that is similar to what we find in Icelandic sagas: like in these and in Saxo, the poetry in Irish prosimetra is authenticating or situational and attributed to the protagonists.61 Although the dating of Old and Middle Irish texts is a complex matter, there is every reason to believe that the Irish prosimetrical form goes back to the early Middle Ages.62 For instance, scholarly consensus assigns the prosimetrical

59 See Friis-Jensen, Saxo Grammaticus, pp. 29–38, 54–55, 58–62; Karsten Friis-Jenssen, ‘Saxo Grammaticus og fornaldarsagaerne’, in Fornaldarsagaerne. Myter og virkelighed, ed. Agneta Ney, Ármann Jakobsson, and Annette Lassen (København: Museum Tusculanum, 2009), pp. 67–77. 60 See, for instance, Matthew Townend, ‘Pre-Cnut Praise-Poetry in Viking Age England’, The Review of English Studies. New Series, 51 (2000), pp. 349–70. 61 For the functions of verse in Irish prosimetra, see Proinsias Mac Cana, Prosimetrum. Crosscultural Perspectives on Narrative in Prose and Verse, ed. Joseph Harris and Karl Reichl (Cambridge: D. S. Brewer, 1997), pp. 110–14. 62 Mac Cana, ‘Prosimetrum’, pp. 99–110. The dating of Old and Middle Irish texts is a demanding field, and the methodology is still being worked out. Linguistically, the dating of a text to the Old or the Middle Irish period (c. 600–900 and c. 900–1200) is generally not very problematic, although some texts may straddle the border. A more precise dating within these periods is difficult to

4.1 Age and Development of the Prosimetrical Form

209

Echtrae Chonnlai (The Expedition of Connlae), Immram Brain (Bran’s Journey) and parts of Táin Bó Cúailnge (The Cattle Raid of Cooley) to the Old Irish period, before c. 900 (and in all likelihood well before that date).63 A number of prosimetrical texts are traceable to the now lost manuscript Cín Dromma Snechtai, which most scholars date to the eighth century.64 Proinsias Mac Cana has argued that the (semi-)poetic passages in the texts of Cín Dromma Snechtai were memorized in their earlier oral transmission, but that the rudimentary and stylistically variable character of the surrounding prose suggests that it was reconstituted with each telling.65 Although Mac Cana’s views on orality are now somewhat dated, this argument strikes me as sound, and it may provide a plausible reconstruction of the prehistory of Irish prosimetra. In Irish and Old Icelandic literature alike, prosimetrum is a feature of local historical narrative, whereas translations were mostly made into prose only, unless the preceding poetic tradition suggested that a certain kind of content should be

achieve without early manuscript or contextual evidence, however. The literature is vast and includes standard overviews as well as editions and articles. A few of these include: Rudolf Thurneysen, A Grammar of Old Irish, trans. and rev. D. A. Binchy and Osborn Bergin (Dublin: Dublin Institute for Advanced Studies, 1946); Liam Breatnach, ‘An Mheán-Ghaeilge’ in Stair na Gaeilge in ómós do Phádraig Ó Fiannachta, ed. Kim McCone et al. (Maigh Nuad: Roinn na SeanGhaeilge, Coláiste Phádraig, 1994), pp. 221–333; ‘Echtrae Chonnlai’ and the Beginnings of Vernacular Writing in Ireland: A Critical Edition with Introduction, Notes, Bibliography and Vocabulary, ed. Kim McCone. Maynooth Medieval Irish Texts, 1 (Maynooth: Department of Old and Middle Irish, National University of Ireland Maynooth, 2000). David Stifter and Gregory Toner are presently working on dating methodology. See also the following three footnotes. It should be noted that the use of poetry within a prose setting may have undergone profound changes in Irish literature, albeit at so early a date that it does not affect the present argument. Stephen N. Tranter, ‘Clavis Metrica’. ‘Háttatal’, ‘Háttalykill’ and the Irish Metrical Tracts. Beiträge zur nordischen Philologie, 25 (Frankfurt: Helbing & Lichtenhahn, 1997), places considerable emphasis on the impact of Latin models on Irish metrical developments, but less so for Old Norse. In particular, he considers the stanzaic form of syllabic metres to be of Latin extraction in Irish (though not in Old Norse), and there is scholarly consensus on this point (see, for instance, David Stifter, ‘Metrical Systems of Celtic Traditions’, North-Western European Language Evolution, 69: 1 (2016), p. 86; Rudolf Thurneysen, ‘Zur irischen Accent- und Verslehre’, Revue celtique, 6 (1883–85), pp. 336–47). Tranter does not explicitly discuss prosimetrum, but if the stanza was of Latin origin, that would presumably leave only the roscada (accentual, alliterating poetry or poetry-like passages) to serve as the poetic component of oral Irish prosimetra (on roscada, see Liam Breatnach, ‘Zur Frage der roscada im Irischen’, in Metrik und Medienwechsel. Metrics and Media, ed. Hildegard L. C. Tristram (Tübingen: Gunter Narr, 1991), pp. 197–205). 63 For a detailed discussion of the dates of Echtrae Chonnlai and Immram Brain, see ‘Echtrae Chonnlai’, ed. Kim McCone. 64 See, for instance, John Carey, ‘On the Interrelationships of Some Cín Dromma Snechtai Texts’, Ériu, 47 (1995), p. 91. 65 Proinsias Mac Cana, ‘Mongán Mac Fiachna and Immram Brain’, Ériu, 23 (1972), pp. 107–08.

210

4 Prosimetrical Narrative

transmitted in poetic form.66 The long-standing cultural exchange of the North Sea area, as well as the strong focus on poetry and poets within Irish and Norse culture alike, suggest that the similar treatment of poetry in prosimetra by both traditions may have deep roots in local culture.67 Since the sources allow us to see that skaldic prosimetra evolved in the decades around 1200, however, this leaves only the eddic prosimetrical tradition as a likely candidate for having such a long prehistory. It is thus plausible that skaldic prosimetra developed under the impact of eddic ones. When taken together, the record of known works, the passage in Þorgils saga ok Hafliða, the breaking up of longer poems and the composition of pseudonymous poetry all suggest that skaldic prosimetra developed in the decades around 1200. Before then, skaldic poetry was probably transmitted in contexts that may not have been independent of prose, but where the two were not integrated. It is true, of course, that skaldic poetry has always been difficult to understand. It must have been the topic of discussion, and some narrative was presumably often required to make sense of it. In some cases, a group of stanzas related to a topic or an event may have been transmitted together, but without necessarily having a fixed account to accompany it. I shall now briefly discuss three examples where the character of the written texts suggests such an oral background. In the encounter of the pagan Steinunn, mother of Hofgarða-Refr, and Þangbrandr, sent by Óláfr Tryggvason to evangelize the Icelanders, the sources allow us to see the development from quoting stanzas about a famous incident to producing an integrated prosimetrical account of the same incident. In Kristni saga, we read that Þangbrandr’s ship broke against the shore and that Steinunn composed two stanzas about this event. The stanzas are then quoted without being connected to the narrative.68 In the later Njáls saga these stanzas are integrated into a hostile dialoge between Steinunn and Þangbrandr.69 We see a development here from reporting stanzas that were known to have been composed in a given context to stanzas functioning as dialogue in a saga account. In another instance, two versions of a þáttr about the poetic exchange between Haraldr harðráði, Þjóðólfr and a fisherman/three men display very little lexical overlap in the prose, and yet there are indications that there is a written connection between 66 Regarding Irish translations, see Mac Cana, ‘Prosimetrum’, pp. 113–14. In Old Norse, instances of verse translation based on the preceding tradition are Hugsvinnsmál, based on Hávamál and gnomic poetry in general, and Merlínuspá, based on Vǫluspá (McKinnell, ‘The Making of Hávamál’; Lavender, ‘Merlin and the Vǫlva’, Viking and Medieval Scandinavia, 2 (2006), pp. 111–39; Johansson, ‘Översättning och originalspråkstext’; Sveinbjörn Rafnsson, ‘Merlínuspá og Völuspá í sögulegu samhengi’). Hagiographical translation in poetic form (Plácitúsdrápa) is probably indebted to the panegyrical tradition. 67 For an assessment of the possible preliterate background of Irish prosimetra, see Mac Cana, ‘Prosimetrum’, pp. 122–26. 68 Biskupa sögur, 1, ed. Sigurgeir Steingrímsson et al., 2, pp. 138–39. 69 Brennu-Njáls saga, ed. Einar Ól. Sveinsson, pp. 265–67.

4.1 Age and Development of the Prosimetrical Form

211

the two versions. These indications are discussed below on pp. 259–62, and the explanation of the startling differences, in spite of a shared written transmission, seems to be that the þáttr originated as a minimal frame holding a number of stanzas together and that it was later adapted to different contexts. At this second stage, the prose around the stanzas grew into the preserved versions of the þáttr. As discussed on pp. 80–85 above, the first two of a cluster of six stanzas in the beginning of the sixth chapter of Hallfreðar saga were paraphrased by Einarr Skúlason sometime during the first half or around the middle of the twelfth century. In the first of the six stanzas, Hallfreðr thanks Óláfr Tryggvason for a sword, whereas in the following five he expresses his difficulties in changing his religious allegiance from Óðinn to Christ. In modern scholarship, these five stanzas are treated as Hallfreðr’s ‘conversion stanzas’, leaving the first one out. Einarr, however, paraphrases the first and the second, suggesting that these were transmitted together in his time and that their conceptual frame was that of Hallfreðr’s encounter with the king, including their religious negotiations but also the typical skaldic topic of the king bestowing a sword upon the skald. In the saga, the six stanzas are connected by minimal passages of prose, mostly consisting in the king saying that the previous stanza was not good enough and that Hallfreðr has to compose another. This is little more than a frame for the stanzas, and this is likely to have been the case also when Einarr treated these stanzas as a group long before the saga was written. In three of these texts – Kristni saga, Njáls saga and Hallfreðar saga – the scene is over when the last stanza is over, whereas the þáttr about Haraldr, Þjóðólfr and the fisherman/the three men ends in a few concluding sentences. This suggests that the scenes are there because of the stanzas in them, and in the case of Kristni saga and Njáls saga, as well as the þáttr, it would appear that we can follow the process towards integration into the narrative. We seem not to be dealing with pre-existent prosimetrical accounts, but with conceptual frames for groups of stanzas, later reworked into saga accounts.70 Even today, students of skaldic poetry discuss the stanzas and their historical background, but they do not opt for a classical saga style when doing so. This may be

70 Jan Ragnar Hagland has suggested that eight stanzas connected to the conflict between Glúmr Geirason and Eyvindr skáldaspillir and to Eyvindr’s shift of allegiance from Hákon góði to Haraldr gráfeldr have been transmitted to Fagrskinna within a similar framework (‘Eit skriftleg førelegg for Øyvind Skaldespillars dikt i Fagrskinna?’, Maal og Minne (1975), pp. 12–19). In Fagrskinna, a stanza by Eyvindr is followed by the words ‘áðr en ek týnda Hákoni’ (before I lost Hákon), which cannot be connected to the following prose and appear to have been retained because they were mistakenly thought to belong to the stanza (Ágrip af Nóregskonunga sǫgum. Fagrskinna – Nóregs konunga tal, ed. Bjarni Einarsson, p. 98). Hagland considers the source to have been a collection of poetry with some connecting prose. Since the first stanza is by Glúmr, we appear to be dealing with a situational frame, rather than a collection of poetry by Eyvindr alone. The extent of the prose in the source is unclear, but it seems likely that it was minimal and therefore had to be reworked.

212

4 Prosimetrical Narrative

a crude point, but it is an important one for reminding us that the discourse surrounding skaldic poetry in an oral setting may have been very different from the one we are presented with in the sagas. The evidence reviewed above suggests that skaldic stanzas, including situational ones like those of Steinunn and Hallfreðr, were transmitted in contexts that bore little resemblance to sagas. For this reason, I avoid here the term Begleitprosa, often used for designating the exposition that may have accompanied the poetry. It conveys a perception of this ‘prose’ as something like the prose surrounding the stanzas in the sagas, and the scholarly discourse about Begleitprosa bears this out.71 More open-ended and to my mind more attractive is Árni Magnússon’s statement that ‘nockur munnleg expositio’ (some oral exposition) must have accompanied skaldic poetry.72 Nonetheless, the functions of poetic quotations which diverge from the Latin tradition and which are found in the sagas, in Saxo and in Irish literature suggest that some sort of prosimetrical tradition was present in the North Sea area and predated the arrival of Latin literature in Iceland. Around 1200, Saxo appears to have drawn on eddic prosimetra – fornaldarsögur – and it thus seems likely that some oral form of these existed before the time of saga writing and affected the emerging skaldic prosimetrum. The functions of poetic quotations in these new skaldic prosimetra were predicated on the preceding tradition, but for various reasons they also developed different features. The action was closer in time and often in space, and this may have contributed to a perception that unlike eddic quotations in fornaldarsögur, skaldic authenticating quotations should indeed be authentic. As will be seen below, none of the examples of spurious stanzas that I have been able to locate occur in skaldic authenticating quotations, except for two examples in Grettis saga (see p. 271). Furthermore, skaldic poetry invited riddling on a purely linguistic level (as opposed to riddles with a focus on content, as in the eddic Hervarar saga). Yet the similarities between eddic and skaldic prosimetra are much more conspicuous than their differences, especially compared to the similarities of either to the Latin prosimetrical tradition. Now that the relatively short prehistory of skaldic prosimetra has been outlined, it is time to consider viable methods of approaching the texts themselves.

4.2 Corpus and Method This section describes the corpus of skaldic prosimetra as well as the dating methods I shall employ in following sections. Different functions of poetry in saga prose 71 The term was introduced by Siegfried Beyschlag, ‘Möglichkeiten mündlicher Überlieferung in der Königssaga’, Arkiv för nordisk filologi, 68 (1953), pp. 109–39. 72 Árni Magnússons levned og skrifter, ed. Finnur Jónsson, 2 vols (Kommissionen for det arnamagnæanske legat: København, 1930), 2, pp. 139–40.

4.2 Corpus and Method

213

evolved over the course of the development of the prosimetrical form. Some uses seem to have been abandoned early, at least in the kings’ sagas. Thus, the portions of Morkinskinna that are likely to be original to the work display many quotations that are of a merely descriptive character, some of which are composed in simple eddic metres.73 The compilations Fagrskinna and Heimskringla, composed only slightly later, avoid quotations of this type. Sverris saga, meanwhile, quotes two stanzas to prove a moral point: this is a common strategy in Latin prosimetra but is almost unheard of in Old Norse.74 Increasingly, two types of quotation came to dominate: namely, what scholars now call authenticating and situational quotation, as discussed by Alois Wolf, Bjarni Einarsson and Diany Whaley (see above pp. 22–23). The distinction between the two types may be used for source-critical sifting of the stanzas. When poetry was used to corroborate the truthfulness of historical narrative, this implies that the poem’s authenticity was important to medieval Icelandic authors. Such quotations are typically introduced with the words svá segir N. N (as N. N. says) or the like. When poetry was used for dramatic effect, by contrast, authors seem not to have necessarily asked questions about authenticity. These quotations are typically introduced with words like þá kvað N. N vísu (then N. N recited a stanza). This division has met with some resistance on account of its crudeness, and it has also been deemed inept as a source-critical tool.75 Two clarifications are in order therefore. First, it is a crude model indeed, and therein lies its usefulness. It does not take rhetorical nuances into account, but it does, on the whole, allow for a distinction into authenticating quotations and ‘others’. True, in some cases the distinction is not articulated in the texts and cannot then be drawn into the analysis, but I do not consider that a valid reason to jettison the majority of cases where it can. Second, situational quotation in itself does not imply that the poetry is spurious, but rather that it may be, since saga authors or redactors would not consider it a breach in narrative responsibility to compose the poetry themselves in such cases. Furthermore, the different uses of authenticating quotations in different settings underline the need for a critical application of this distinction.

73 Morkinskinna. The Earliest Icelandic Chronicle of the Norwegian Kings (1030–1157), trans. Theodore Andersson and Kari Ellen Gade. Islandica, 51 (Ithaca: Cornell University Press, 2000), p. 25. 74 The moral character of both quotations may be due to the particular circumstance that they belong in speeches attributed to Sverrir (Sverris saga, ed. Þorleifur Hauksson, pp. 72, 257). Stylistically, Sverris saga is somewhat more similar to Latin historiographical works than later kings’ sagas (Sverre Bagge, ‘The Old Norse Kings’ Sagas and European Latin Historiography’, pp. 4, 11). 75 See, for instance, Roberta Frank, ‘Skaldic Poetry’, in Old Norse-Icelandic Literature. A Critical Guide, ed. Carol J. Clover and John Lindow. Islandica, 45 (Ithaca: Cornell University Press, 1985), p. 173; Clunies Ross, A history of Old Norse Poetry and Poetics, pp. 71, 78–80.

214

4 Prosimetrical Narrative

Snorri’s Edda, though not treated by Bjarni Einarsson, is a case in point. Quotations in Gylfaginning and Skáldskaparmál may on the whole be called authenticating. It is probably safe to say that Snorri believed the poetry he quoted in Skáldskaparmál to be authentic, with the exception of Bragi’s verbal exchange with a troll-woman.76 Gylfaginning is another matter. Even if Snorri may conceivably have believed that the poetry was, as he says, composed by the gods (that is, by the early immigrants to Scandinavia), no scholar would argue that today.77 The distinction must be used judiciously therefore. The same holds true for fornaldarsögur, where authenticating quotations are considerably more common than in sagas of Icelanders, but where many stanzas appear to be spurious.78 In Gylfaginning and fornaldarsögur, one may perhaps speak of a chronologically conditioned use of authenticating quotations: not the chronology of saga writing, but of the date of the action within the text. Demands of historical truthfulness seem not to set in before the time of the settlement of Iceland.79 This chronological distinction in the reliability of authenticating quotations makes a certain amount of sense, since the authors presumably insist that the fantastic fornaldarsögur are true because the reader is unlikely to think so, whereas in kings’ sagas, they do so because the genre aims at a description which is based on actual events. The chronological dividing line also generally coincides with the quotation of eddic (pre-settlement) versus skaldic (post-settlement) poetry. The higher incidence of pseudonymous poetry in pre-settlement authenticating quotations and in situational quotations need not imply that we are dealing with falsification in a modern sense. It is likely that the authors in these cases aimed not at truth, but at plausibility. If the matter were expressed in the terms used by Cicero and Isidore, one might say that if authenticating quotations from post-settlement times belonged to historia (narrative about true things), pre-settlement poetry and situational quotations belonged to argumentum (narrative about likely things), rather than to fabula (lying narrative about the fantastic).80 The influential

76 Snorri, Skáldskaparmál, ed. Faulkes, pp. 83–84, 206. The stanzas are composed in a lax kind of tøglag, indeed so lax and irregular that one is tempted to think that they may be quite old. Faulkes is hesitant to attribute poetic skill to trolls, but SkP suggests that the stanzas may be authentic (SkP 3, p. 64). 77 Snorri, Prologue and ‘Gylfaginning’, ed. Faulkes, p. 34. On the matter of Snorri’s opinions about the age of the poetry he quoted in Gylfaginning, see above pp. 167–69. 78 Bjarni Einarsson, ‘On the Rôle of Verse in Saga-Literature’, p. 124. Bjarni suggests that the high ratio of authenticating quotations may be due to that some fornaldarsögur are partly based on earlier poems, but this would need considerable elaboration, since most of the poetry in sagas of Icelanders is evidently old and could therefore have been expected to be treated in a similar way. 79 For a general discussion of fictionality and historicity in Old Norse literature, see Ralph O’Connor, ‘History or Fiction?’; Sverrir Tómasson, Formálar íslenskra sagnaritara á miðöldum, 245–60. 80 Thus Cicero, De inventione 1,19,27; Ad Herennium 1,8,13; Isidore, Etymologiae 1,44,5.

4.2 Corpus and Method

215

distinction of Isidore may not have been consciously applied by saga authors, but the archaizing tendencies in poetic quotations, to be outlined below, bespeak a roughly similar framework: old poets were expected to compose in an oldfashioned way. Several scholars believe that Icelandic authors adhered to the argumentum position in situational and authenticating quotations alike, and that source-critical sifting of skaldic stanzas is an all but impossible enterprise.81 An often repeated point is that poets could archaize at will and that dating on formal grounds is thus largely futile.82 Remarkably, none of these scholars have attempted to show that such archaization was actually undertaken or have noticed the various features that never appear in later poetry (several hiatus forms, aðalhending in a: ǫ, compensatory hendingar, etc.). While relatively crude metrical and advanced stylistic archaization may be found, instances of morphological or phonological archaization have either been rejected or have yet to be found, and this is the kind of archaization most often assumed e silentio (see p. 249). It is beyond the scope of this book to undertake a full investigation of all features involved, but the reader should be aware that a necessary precondition for what follows is that skaldic poetry can indeed, under favourable circumstances, be roughly dated.83 Furthermore, I have not, with the methods I employ, found any spurious stanzas in authenticating quotations pertaining to the post-settlement period, except in Grettis saga. I believe that this speaks to the source-critical usefulness of the distinction into situational and authenticating quotations.

81 Preben Meulengracht Sørensen, ‘The Prosimetrum Form, 1: Verses as the Voices of the Past’, in Skaldsagas: Text, Vocation, and Desire in the Icelandic Sagas of Poets, ed. Russell Poole. Ergänzungsbände zum Reallexikon der Germanischen Altertumskunde, 27 (Berlin and New York: Walter de Gruyter, 2001), pp. 172–90; Shami Ghosh, Kings‘ Sagas and Norwegian History. Problems and Perspectives (Leiden: Brill, 2011), pp. 25–69 and references there. 82 Thus, for instance, Gabriel Turville-Petre, ‘Gísli Súrsson and his Poetry: Traditions and Influences’, in Nine Norse Studies (London: Viking Society for Northern Research, 1972 [1944]), pp. 148–150; Peter Foote, ‘An Essay on the Saga of Gisli’, in The Saga of Gisli, trans. George Johnston (Toronto: University of Toronto Press, 1963), pp. 93–134; Peter Foote, ‘Beginnings and Endings: Some Notes on the Study of Skaldic Poetry’, in Peter Foote, Aurvandilstá, ed. Michael Barnes, Hans Bekker-Nielsen, and Gerd Wolfgang Weber (Odense: Odense University Press, 1984 [1976]), pp. 236–48; Bjarni Einarsson, ‘Íslendingadrápa’, Tímarit Háskóla Íslands, 4 (1989), pp. 127–31; Ghosh, Kings‘ Sagas and Norwegian History, pp. 58–59. See also Haukur Þorgeirsson, ‘The Dating of Eddic Poetry’, p. 34. 83 To date, the only exhaustive investigation of dating criteria for skaldic poetry is Klaus Johan Myrvoll, Kronologi i skaldekvæde. See also Klaus Johan Myrvoll, ‘The Authenticity of Gísli’s Verse’; Kari Ellen Gade, ‘The Dating and Attribution of Verses in the Skald Sagas’, in Skaldsagas. Text, Vocation, and Desire in the Icelandic Sagas of Poets, ed. Russell Poole (Berlin: De Gruyter, 2001), pp. 50–7; Males, ‘Egill och Kormákr’; Finnur Jónsson, ‘Sagaernes lausavísur’, Aarbøger for nordisk oldkyndighed og historie 3: 2 (1912), pp. 1–57.

216

4 Prosimetrical Narrative

In modern scholarship, the different uses of skaldic quotations are distributed by genre, where authenticating quotations dominate in the kings’ sagas and situational ones in the sagas of Icelanders.84 This description is more appropriate for the sagas of Icelanders than for the kings’ sagas, where situational quotations are relatively common.85 To name a further problem with this rule, a distinction based on genre may partly mask the rationale for different kinds of quotations, which vary according to the character of the participants and the nature of the actions represented. When authors present Icelandic poets engaging in dialogue, whether in Iceland or Norway, situational quotations are likely to occur. When they do not present such dialogue, situational quotations are likewise absent.86 Morkinskinna is a case in point. It has more situational quotations than most kings’ sagas (some 30% by Bjarni Einarsson’s count), and this is precisely because of its many examples of interactions between Icelandic poets and the king.87 The same holds true of the part of Heimskringla that deals with Saint Óláfr. Together with the Separate Saga of Saint Óláfr, this is the only kings’ saga which displays a larger proportion of situational quotations than Morkinskinna (slightly over a third).88 In other parts of the kings’ saga corpus, there is less interaction between the king and his Icelandic poets and proportionally fewer situational quotations. Many sagas of Icelanders deal with disputes where some Icelandic poet is involved, and situational quotations are thus all but bound to occur. This does not, however, explain the near absence of authenticating quotations in them (Fóstbrœðra saga, with its near affinity to the kings’ sagas, and Eyrbyggja saga are partial exceptions).89 Presumably, the historical veracity of such local lore was not deemed to be of crucial importance, and a good story could stand on its own merits. This criterion of gravity may also have been at work internally in the kings’ sagas; getting a king’s feats and battles right was important enough that it warranted quotation of contemporary witnesses, whereas amusing but idle talk at dinner or on fishing trips was a locus of entertainment where the historiographer could give some rein to creativity. In the source-critical evaluation of skaldic poetry in the prologues to Heimskringla and to the Separate Saga of Saint Óláfr, skaldic poetry is said to verify the kings’ deeds (verk) and battles (orrustur), but there is no corresponding assessment of what might have been said during a battle, let alone at dinner.90 If the distribution of different types of quotation is not a matter of genre alone, neither can the date of composition of a saga easily account for their ratio. Óláfs saga

84 85 86 87 88 89 90

See, for instance, SkP 1, p. xlv; Frank, ‘Skaldic Poetry’, p. 173. Bjarni Einarsson, ‘On the Rôle of Verse in Saga-Literature’, pp. 120–21. Cf. Meulengracht Sørensen, ‘The Prosimetrum Form, 1ʹ, p. 188. Bjarni Einarsson, ‘On the Rôle of Verse in Saga-Literature’, p. 121. Bjarni Einarsson, ‘On the Rôle of Verse in Saga-Literature’, p. 121. Bjarni Einarsson, ‘On the Rôle of Verse in Saga-Literature’, p. 122. Heimskringla, ed. Bjarni Aðalbjarnarson, 1, p. 5; 2, p. 422.

4.2 Corpus and Method

217

Tryggvasonar by Oddr munkr presents almost exclusively authenticating quotations, whereas the near contemporary *Oldest Saga of Óláfr Haraldsson (c. 1200) – as witnessed by the Legendary Saga – contains both authenticating and situational quotations in clusters, with the situational ones found in passages where the king interacts with his Icelandic poets.91 As we have seen, this is a common feature in sagas about Saint Óláfr. Orkneyinga saga (c. 1200) has both kinds of quotations, but authenticating ones dominate in the chronicle-like first part of the saga, whereas situational ones belong to the adventures of earl Rǫgnvaldr, who comes across as something of an amalgam between Icelandic poet and Norwegian king. Neither genre nor date of composition, then, but the character of the action and the elevation of the subject are decisive for what kind of quotation will dominate in a saga or a portion of a saga. With the emergence of the prosimetrical norm in the early thirteenth century, the demand for poetry suited for inclusion in the sagas appears sometimes to have outstripped supply. The strategies that Icelandic authors used to solve this problem with an eye to metrics will be the focus of attention in the following section, followed by the strategies of diction in the next section. Unlike many other components in the development of Old Icelandic secular literature, the shaping of the prosimetrical form by these methods has left some building blocks still visible, bearing mason’s marks, as it were, that allow us to reconstruct the progression of the work. As we have seen, eddic prosimetra are important for reconstructing the development of the prosimetrical form. Eddic poetry does not, however, contain hendingar and some of the other metrical features that are useful for dating. Furthermore, eddic poetry does not use kennings and other relevant features of diction anywhere near as much as skaldic poetry does. For these reasons, I do not treat the main genre of eddic prosimetrum – fornaldarsögur – except in so far as it contains skaldic poetry (notably Ragnars saga loðbrókar). Rather, my focus falls on sagas of Icelanders and kings’ sagas. Scholars are considerably more confident with regard to the dating of skaldic poetry than that of eddic poetry, but one must not underestimate the extent to which this is due to the attribution of skaldic poetry to named poets. Linguistic criteria are often hard to come by, except in rare cases where the poetry was composed as late as the turn of the fourteenth century or later as is the case with Grettis saga. Metrical criteria on the minute level of the Sieversian tradition are most apt for dating extended poetic corpora: for individual stanzas, such criteria are often not sufficient. Phonological features, such as hendingar, are more helpful in so far as they can allow for the dating of single stanzas. They have the potential drawback that later poets were aware of them and at least in principle, they could employ them to make their poetry look older, although, as noted above, there are in fact no indications that they did so. Intertextual dependency can sometimes be helpful, but it is rarely

91 Cf. Andersson, The Growth of the Medieval Icelandic Sagas, p. 47. None of these sagas are treated by Bjarni Einarsson.

218

4 Prosimetrical Narrative

evident which poem is the giver and which the taker. I have argued above that authenticating quotations probably indicate the perceived authenticity of poems composed by poets who were active after the settlement, but the perceived authenticity of a stanza that is purportedly some 300 years older than the context in which it appears is not necessarily trustworthy. In short, if skaldic poetry leaves us on a firmer footing than eddic poetry, the potential sources of error remain legion. I shall use several parameters concurrently to identify pseudonomous compositions. Identification requires either a very strong indicator or several, mutually independent indicators pointing to pseudonymous composition. My method is thus eclectic, yet guided by source-critical deliberations. The distinction between situational and authenticating quotations will be upheld, and all stanzas singled out as pseudonymous are situational quotations (with two exceptions in Grettis saga; see p. 271). In other regards, the method will vary depending on the source. In the section on metrics, I use metrical features, clusters of stanzas, source context and independent attestation in tandem to sift the stanzas. In the section on diction, intertextuality comes to the fore, since the impact of Skáldskaparmál makes this a viable option. For the poetry in Grettis saga, I accept the unusually strong linguistic evidence, and such considerations enter the analysis at some other points as well. The discussion up to this point has an important corollary. We have seen that skaldic prosimetra probably did not develop until c. 1200. Here I have suggested that pseudonymous stanzas are almost exclusively to be found in situational quotations. This means that the content of pseudonymous stanzas depends on the narrative and that they were probably composed for insertion into it. The novelty of the skaldic prosimetrical form makes it likely that the pseudonymous stanzas have been composed by the saga author or some later redactor rather than at some earlier point during the transmission of traditional lore. This conclusion runs counter to the assumptions of many scholars active from the middle of the twentieth century onwards, who have suggested that pseudonymous stanzas were often composed during the twelfth-century transmission of what would eventually become the sagas. Although exceptions are certainly possible, the character of the Old Icelandic corpus at large suggests that the stanzas in the sagas are either roughly contemporary to their purported date or composed for the sagas themselves. If the twelfth-century hypothesis is to be upheld, it must be accompanied by evidence in favour of twelfth-century skaldic prosimetra, since if it is not, the pseudonymous stanzas fulfill no obvious function. As discussed in the previous section, such evidence appears to be lacking, whereas there is much evidence to the contrary.92 To the extent that pseudonymous poetry can be identified in the corpus of a given poet, then, we are probably not seeing the result of prolonged accretion but of authorial intervention.

92 The twelfth-century hypothesis is discussed in my article ‘1100-talets pseudonyma skaldediktning: En kritisk granskning’, Maal og Minne (2017), pp. 1–24.

4.3 Metrical Archaization in Egils saga

219

4.3 Metrical Archaization in Egils saga In this section, I discuss indications of pseudonymous composition of poetry in the sagas, based on metrical features. The aim is not to reveal the scam, as it were, but to come closer to the methods that the saga authors used to arrive at a form they found congenial. Indeed, examples of pseudonymous composition that can be identified with a high degree of plausibility amount to only a fraction of the poetic corpus, and it would seem that scholars have sometimes overestimated their number, at least in some sagas (for example, Egils saga; see below). Even so, the character and distribution of spurious stanzas suggest details of collection and supplementation and offer some of to the clearest indications of how the sagas took shape. These indications are of particular importance since, when classical saga style developed in the first half of the thirteenth century, the prosimetrical form seems to have dominated. Form and content, furthermore, interact in a dynamic way in the sagas, as when Egill uses poetry to confess his love and Gísli a murder; both inviting the reader or listener to ponder the meaning of the poetry and its implications for the plot. The study of pseudonymous composition is a promising way to get to grips with the evolution of the prosimetrical form, which will bring us close to the emergence of the sagas as we know them. Of the early sagas, Egils saga (before c. 1250) is most firmly datable and also provides the clearest evidence of the process leading up to its prosimetrical form. The four manuscripts fragments that date to up to c. 1300 contain all stanzas which are found in the corresponding sections of the main witness, Möðruvallabók (c. 1350) –12 ½ stanzas in all. It would therefore appear that all or nearly all stanzas found in Möðruvallabók belong to the original composition of the saga, and a likely ante quem for any putative interpolation of stanzas is thus c. 1250 at the latest.93 Debates about the authenticity of Egill’s stanzas have been long and hard, and scholars have ranged from declaring all or nearly all stanzas to be authentic to attributing almost all of Egill’s poetry to the author of the saga.94 Many arguments, in

93 AM 162 A θ fol. (c. 1250) has stanzas 23.5–8 (1–4 only in WolfAug 9 10 4to), 24, 25 (25 has been in Möðruvallabók, see Egils saga Skalla-Grímssonar, ed. Bjarni Einarsson, p. 98), AM 162 A ζ fol. (c. 1250–1300) has stanzas 4–7, AM 162 A γ fol. (c. 1275–1300) has stanza 2, AM 162 A δ fol stanza 1–10. Holm.Perg. 7 4to (c. 1300–1325) contains only the beginning of the saga and is thus irrelevant. Dating according to ONP Registre, p. 234. For control of occurrences and length of the fragments I have used Skj A 1 and Egils saga Skallagrímssonar tilligemed Egils större kvad, ed. Finnur Jónsson (København: Samfund til udgivelse av gammel nordisk litteratur, 1886–88). Margaret Clunies Ross also comments on the relative stability regarding which stanzas are included in the saga (‘Verse and Prose in Egils saga Skallagrímssonar’, in Creating the Medieval Saga. Versions, Variability and Editorial Interpretations of Old Norse Saga Literature, ed. Judy Quinn and Emily Lethbridge (Copenhagen: University Press of Southern Denmark, 2010), pp. 202–205). 94 The modern debate goes back to Finnur Jónsson’s doctoral dissertation, Kritiske studier over en del af de ældste norske og islandske skjaldekvad (København: Gyldendal, 1884). Important later

220

4 Prosimetrical Narrative

particular of those who favour the author-poet, have been of a literary kind, the plausibility of which is not easily evaluated. I wish here to avoid both extremes and to focus on one feature that is perhaps the most easily measurable of all, namely the use of hendingar. Irregular hendingar are common in the poetry of Egils saga and are generally considered an archaic feature.95 For this and other reasons that will become evident below, Egils saga may be the most promising text for showing us how a saga author went about producing a full prosimetrical form worthy of his protagonist. When classifying the stanzas, I have counted those that contain at least two lines with irregular hendingar as archaic or archaizing; this is done in order not to include stanzas composed in the classical pattern, but with one deviant line, as sometimes occurs. By this method, eleven stanzas in the saga come out as archaic or archaizing. These fall into two groups: nine stanzas which have irregular hendingar, largely natural word order and a few simple kennings, and two stanzas which make up for irregular hendingar with interlinear hendingar and which display complex word order, with complex and obscure kennings. I begin with the first group. While hendingar will be the main parameter in the following, onset (unstressed syllable(s) before the alliterative stave of even lines) will also be mentioned. The saga contains two clusters of archaic/archaizing stanzas. The first of these consists of stanzas 12–14. Stanza 12, where Egill exhorts his men to attack Lund, has a clear stylistic rupture in the middle (hendingar marked in italics, onset in bold): Upp skulum órum sverðum, ulfs tannlituðr, glitra, eigum dǫ́ð at drýgja í dalmiskunn fiska; leiti upp til Lundar lýða hverr sem bráðast, gerum þar fyr sjǫt sólar seið ófagran vigra.96

contributions include: Finnur Jónsson, ‘Sagaernes lausavísur’; Egils saga, ed. Sigurður Nordal; Jón Helgason, ‘Höfuðlausnarhjal’; Dietrich Hofmann, ‘Das Reimwort giǫr in Egill Skallagrímssons Hǫfuðlausn’; Bjarni Einarsson, Litterære forudsætninger for ‘Egils saga’ (Reykjavík: Stofnun Árna Magnússonar, 1975); Baldur Hafstað, Die ‘Egils saga’ und ihr Verhältnis zu anderen Werken des nordischen Mittelalters; Torfi H. Tulinius, Skáldið í skriftinni. Snorri Sturluson og ‘Egils saga’ (Reykjavík: Hið íslenska bókmenntafélag, 2004); Jónas Kristjánsson, ‘Kveðskapur Egils Skallagrímssonar’; Haraldur Bernharðsson, ‘Göróttur er drykkurinn’. 95 Guðrún Nordal has studied the use of irregular hendingar in the saga, but from a dramaturgical rather than a chronological perspective (‘Ars Metrica and the Composition of Egils saga’, in Scandinavia and Christian Europe in the Middle Ages. Papers of the 12th International Saga Conference, Bonn, Germany, 28th July – 2nd August 2003, ed. Rudolf Simek and Judith Meurer (Bonn: Hausdruckerei der Universität Bonn, 2003), pp. 179–86). 96 Egils saga, ed. Sigurður Nordal, p. 119; Skj B 1, p. 43; A 1, p. 50.

4.3 Metrical Archaization in Egils saga

221

Wolf’s tooth-stainer [warrior], let our swords now glisten in the valley-fishes’ [snakes’] mercy [summer]. Let every man make his way to Lund as soon as possible. Let us make there, before the setting of the sun, the spears’ frightful incantation [battle].

Here, the first half stanza is regular apart from the lack of hending in line 1 (not an uncommon pattern in early regular poetry),97 displays intertwined word order and is attested elsewhere (Skáldskaparmál).98 Onset in line four may be an archaic feature. Its warlike content is typical of the skaldic corpus at large and could have served equally well at countless points in the sagas. The second half is unique to the saga. In it, odd lines lack hending and even lines have skothending, the word order is straightforward, and only this part of the stanza ties it specifically to the attack on Lund as described in the prose.99 Stanzas 13 and 14 follow the second half of stanza 12 in displaying a similar pattern of highly irregular hendingar and simple word order, and stanza 13 is ascribed to the anonymous daughter of the earl Arnfiðr.100 That anonymous stanza, tightly knit to the prose of the saga, is unlikely to have been transmitted orally for very long. Seeing that this stanza deviates from the typical, regular stanzas of the saga in the same way as the second half of Egill’s stanza 12 and all of his stanza 14, it would seem that it is not the poet, but rather the grouping of the stanzas, that accounts for their deviant features. Pseudonymous composition of the second half of stanza 12 and the entire stanzas 13 and 14 is thus supported by the nexus of: grouping/attribution; metre/word order/kenning complexity; independent attestation vs. unique occurrence; stylistic similarity based on grouping rather than poet; connection or lack thereof to the prose. It is thus highly probable that the entire two and a half stanzas have been composed for insertion into the saga. The example also shows that if older building blocks were available – in this case the first half of stanza 12, differing from the rest in style, content and independent attestation – they would be used in the construction of the prosimetrum. The next cluster of archaic/archaizing verse is found in stanzas 48–50 of the saga.101 Again, there is independent attestation for the first half stanza in the cluster, this time on a runic stick from Trondheim (c. 1200). In the saga, Egill is on his way to collect taxes in Vermaland (Värmland) in modern-day Sweden for the Norwegian king. While visiting 97 ór: er do not rhyme here. -r-: -rð- generally do not rhyme, and the position of the rhymes, if such they were, would break with Kuhn’s Zäsurgesetz, which has been further elaborated in Þorgeir Sigurðson’s milliregla (this rule postulates that an alliterating stave must come between the skothendingar) (Hans Kuhn, Das Dróttkvætt, p. 89; Þorgeir Sigurðson, ‘Þróun dróttkvæða’, p. 10). 98 Snorri, Skáldskaparmál, ed. Faulkes, p. 40. 99 Frank, ‘Snorri and the Mead of Poetry’, p. 167–68 n. 31, has noted that the focus on Lund gives a probable post quem of the establishment of the archbishopric in 1104, which may further support that it is the work of the saga author. 100 Egils saga, ed. Sigurður Nordal, p. 121; Skj B 1, pp. 603 (3), 43–44 (7); A 1, pp. 603–04, 50. 101 Egils saga, ed. Sigurður Nordal, pp. 230–31, 268–69; Skj B 1, p. 51 (38–40); A 1, p. 58.

222

4 Prosimetrical Narrative

the farmer Þorfinnr, he finds out that his daughter Helga is sick, and when he investigates further into the matter, he finds a baleen (a blade from the filtering system inside the mouth of a baleen whale) with runes inscribed on it under her bed. He scrapes them off, burns the baleen, and carvs new runes for her health. Stanza 48 reads: Skalat maðr rúnar rísta, nema ráða vel kunni, þat verðr mǫrgum manni, es um myrkvan staf villisk; sák á telgðu talkni tíu launstafi ristna þat hefr lauka lindi langs oftrega fengit.102 You should not carve runes unless you are well acquainted with their interpretation. It happens to many a man that he goes astray among the obscure staves. I saw ten secret runes carved on a baleen plate. That has caused the leeks’ linden [woman] enduring trouble.

The runic stick from Trondheim matches the first three lines and the first word of the fourth but reads somewhat differently: Sá skyli rúnar rísta, er r[áða ve]l kunni, þat verðr mǫrgum manni at. . .103 He shall carve runes, who is well acquainted with their interpretation. It happens to many a man that [. . .]

In this instance, the first half of the stanza seems to have been a proverbial quatrain of general import to the carving of runes and one which the author found useful for the narrative in Egils saga. Independent half stanzas seem to have circulated widely, but in Egils saga, like most sagas, full stanzas are the norm. It is only the second half that ties the stanza to the prose, apart from the general topic of runes. In the second half we find the sick girl, the ten runes and the baleen. The baleen is of particular interest, since it betrays that this stanza can hardly have been composed under anything like the circumstances described in the saga, or even by someone who later reported his experiences under those circumstances; while baleens are common objects in Iceland, they are, and have always been, absent from the deep forests of the SwedishNorwegian border regions. Rather, the baleen is an Icelandic projection onto the distant and adventurous past.

102 Skj B 1, p. 51 (38) (I have removed the emendation of um to of); A 1, p. 58. 103 For a runological commentary, see James E. Knirk, ‘Runes from Trondheim and a Stanza by Egill Skalla-Grímsson’, in Studien zum Altgermanischen, ed. Heiko Uecker (Berlin: de Gruyter, 1994), pp. 411–20.

4.3 Metrical Archaization in Egils saga

223

In sum, there can be little doubt that the second half was composed for the saga, whereas the first half was not. The ratio of classical skaldic features is here the opposite one as compared to stanza 12: the older, independently attested stanza has no hendingar, two onsets and no kennings, whereas the younger half has skothending in the last line, no onsets, and one kenning, albeit a simple one. The features of the second halves of stanza 12 and stanza 48 are quite similar; the inverted ratio is due to the profound difference between the two first halves. The simplicity of the first half of stanza 48 is so extreme as to suggest that we are not here dealing with archaic features so much as with a folkish, proverbial register, and it is unlikely that this simple composition was ever attributed to the accomplished and fiercely individualistic Egill before the composition of the saga.

The first side of the runic stick from Trondheim, showing the first line and the damaged second (the l in ‘vel’ barely visible to the right). Note the peculiar shape of the fifth rune, , a possible intermediary stage between Norwegian and Icelandic ↕ (see above p. 183). Image: NTNU Vitenskapsmuseet. Photo: Ole-Aleksander Ulvik.

As with stanza twelve, the two stanzas that follow are composed in the same pattern, with relatively straightforward word order and irregular hendingar. In stanzas 49 and 50, furthermore, the wording ‘hlífar hneitiknífum’ / ‘hlífar skelfiknífum’ suggests that one of them has influenced the other.104 The second half of stanza 50, where this expression occurs, is difficult and nearly regular and may constitute an older building block.105 If so, Pseudo-Egill has paraphrased the older stratum in his composition of stanza 49. Whatever the precise status of the second half of stanza 50, however, grouping, hendingar and syntax, and the fact that the parts of the stanzas that relate to the prose are the ones that lack independent attestation, provide signs that all three stanzas 48–50 were reworked for the saga. The mention of the baleen in stanza 48 is a particularly strong indication, but it has bearing on the other two stanzas only when additional features are taken into account. These two clusters of stanzas, 12–14 and 48–50, bring us close to the workshop of the prosimetrical saga author. A few conclusions can be drawn. Skaldic stanzas were

104 The variants in stanza 49 in Möðruvallabók are obvious scribal errors (‘hveíti kívfum’ (fol. 92v b22)). 105 Egils saga, ed. Sigurður Nordal, pp. 268–69; Skj B 1, p. 51 (40); A 1, p. 58. The difficult wording of line 7–8 may be a sign of great age, though corruption is another option. In the second half stanza, only line 7 lacks hendingar, a low ratio that seems more archaic than archaizing.

224

4 Prosimetrical Narrative

obviously seen as adding to the telling, and if no stanzas existed for a passage where they were thought desirable, the author might compose them himself. If any preexisting material was available, such as stanza 12.1–4, 48.1–4 and perhaps 50.5–8, it would be used and adapted to its new function by tying it to the narrative and/or by composing what was missing to get a full stanza.106 The first half of 48 reads more as a proverb than as a poetic achievement by a famous poet, and it would thus seem that the building blocks did not necessarily have to be already attributed to the poet in question to serve this function. One point that remains to be accounted for in this description is the choice of metrical features. I return to this issue below. Three other stanzas which share similar features of irregular hendingar, simple word order and few kennings are spread across the saga, namely stanzas 7, 9 and 44.107 While we saw that stanza 13 is attributed to an anonymous daughter of the earl Arnfiðr, stanza 44 was purportedly composed by the likewise anonymous daughter of a certain Armóðr and stanza 7 by a seven-year-old Egill.108 These two stanzas, furthermore, show affinity through their opening lines: Þat mælti mín móðir [. . .] (My mother said [. . .]) (stanza 7.1) and Því sendi mín móðir [. . .] (Therefore, my mother sent [. . .]) (stanza 44.1), respectively. Their general similarities to the pseudonymous group are thus supported by circumstantial evidence, suggesting that they were composed for the saga. Only stanza 9 arouses no obvious suspicions in addition to the features it shares with the other stanzas. Stanzas 8 and 9 both deal with a feast at Bárðr’s, the vasall of Egill’s enemy king Eiríkr. Egill and his men had arrived there and been told that only few and simple provisions were available, whereas Bárðr was in fact hosting a lavish feast for king Eiríkr, who was then visiting. In stanza 8, Egill accuses Bárðr of having lied to him about having few provisions, whereas in stanza 9 he describes how he carves runes on a drinking horn whose contents he suspects of being poisoned. The horn then breaks and danger is averted. Stanza 8 has complex word order and kennings and displays classical metre, whereas stanza 9 has straightforward word order,

106 We may see the traces of expectations of full stanzas in the manuscript evidence of stanza 23 (Skj B 1, p. 45 (14); A 1, p. 52). The full stanza is found only in WolfAug 9 10 4to (c. 1330–1370). In AM 162 A θ fol (c. 1250) its second half is found as the first half of a stanza, with empty space left for the second half; presumably, the scribe could not procure that half (which may not have existed at the time). In Möðruvallabók (AM 132 fol, c. 1330–1370) the second half is written down with no empty space, and was thus probably perceived as complete (Clunies Ross, ‘Verse and Prose in Egils saga Skallagrímssonar’, p. 203). 107 Skj B 1, pp. 42 (1), 43 (3), 603 (4); A 1, pp. 48, 49, 604. 108 Sigurður Nordal observes that both metre and content of this stanza are childish (Egils saga, ed. Sigurður Nordal, pp. 100–101). He notes that ships of the type knǫrr rarely had oars, and that a ship is not steered from the prow, as the stanza suggests. The childish perspective, however, suits the narrative well and thus has little bearing on the question of authenticity. Some possible flaws, such as what may be over-alliteration in the first line and fronted hendings in the sixth, may also have suited the perceived or actual style of a child.

4.3 Metrical Archaization in Egils saga

225

one kenning and highly irregular hendingar. There is one more relevant difference between the stanzas. Both contain the name Bárðr in the last line, but whereas stanza 8 has it in the archaic disyllabic form Bárøðr, stanza 9 has it as Bárðr (this is corroborated by all four manuscripts where the stanza occurs). Editors have emended the words Bárðr of to Bárøðr in stanza 9, in analogy to stanza 8 and in accordance with what we know to be the older form. In the Middle Ages, however, syncopated and unsyncopated forms of names were seen as synchronic variants, whereas of, as we shall see, was considered to be an archaic feature.109 While Bárðr of is clearly the lectio facilior, there is no reason to posit scribal alteration when the old form has been preserved in a stanza only a few lines before. Rather, it seems likely that stanza 8 has Bárøðr because that stanza is old, whereas the default form Bárðr is found in stanza 9 even though, as modern emendations show, there would have been room for the form Bárøðr here. The narrative centres on a person named Bárðr, and it would thus make sense to use that form in a stanza composed for the narrative. This might explain the formal discrepancy of the names in the two stanzas. Given that stanza 9 displays formal features that are otherwise only found in stanzas which are for many reasons likely to have been composed for the saga, I believe that this is the right explanation of the form Bárðr. Stanza 9 not only shares the general characteristics of the other stanzas considered above, but it also, more specifically, expresses the fascination with runes that we also find in stanza 48. This is a recurrent theme in Egils saga; famous instances are the episode where Egill raises an insult-pole and carves a curse on it and the one where Egill’s daughter says that she will carve the poem Sonatorrek on a stick.110 Of the poetry, however, only these two stanzas deal with the topic, and the runic presence does nothing for the plot; it could easily have been cut out or – to get to the point – never have been inserted. The fact that the topic of runes has been added to whatever stories about Egill that the author had inherited from tradition is further underlined by a comparison of the episode where stanza 9 occurs and a passage in Gregory’s Dialogues. This text was translated into Old Norse in the twelfth century (and later reworked into Benedikts saga).111 In the Dialogues (and in Benedikts saga), some monks asked

109 With regard to syncopated and unsyncopated forms, both Þorkell/Þorketill are used in the poem Jómsvíkingadrápa, obviously for metrical reasons (SkP 1, pp. 966, 970). Neither 3GT nor Háttatal makes any connection between long forms and age, but they both construe them as due to poetic licence (Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, pp. 88–89; Snorri, Háttatal, ed. Faulkes, p. 7). 110 Egils saga, ed. Sigurður Nordal, pp. 171, 245. 111 This influence was noted in passing by Anne Holtsmark, review of Odd Nordland, Hǫfuðlausn i Egils saga. Ein traditionskritisk studie (Oslo: Samlaget, 1956), Maal og Minne (1956), p. 131; cf. Klaus von See, ‘Der Skald Torf-Einar’, Beiträge zur Geschichte der Deutschen Sprache und Literatur, 82 (1960), p. 32. Holtsmark (and von See) writes that the passage is influenced by Benedikts saga, but it might be safer to say that it has been influenced either by the translation of Gregory’s Dialogues

226

4 Prosimetrical Narrative

Benedict to come to govern them, but his rule turned out to be stricter than they had anticipated and they decided to kill him: En es þeir sá sér bannaða vera alla ranga hluti, ok váru ófúsir at fyrláta vanða sinn, þá réðu þeir fjǫrráðum umb hann ok blendu drykk hans eitri. En es hánum vas sjá dauðadrykkr fœrðr í glerkeri, þá gǫrði hann þegar krossmark á mót, áðr hánum væri glerkerit selt, sem hann var opt vanr. En þá sprakk í sundr glerkerit, svá sem hann lýsti steini á þat. Þá skilði Guðs maðr, at þat ker hafði dauðadrykk, er eigi mátti standask lífs mark.112 But when they realized that all illicit behavior was prohibited to them, and they were reluctant to abandon their habits, then they plotted against his life and mixed poison in his drink. And when the lethal drink was brought to him in a glass vessel, then he made the sign of the Cross in that direction before the glas vessel was given to him, as he was often wont to do. And then the glass vessel split apart as if he had hit it with a stone. Then the man of God understood that that vessel held a lethal drink that could not withstand the sign of life.

In Egils saga, Bárðr conspires with the queen to take Egill’s life: Dróttning ok Bárðr blǫnduðu þá drykkinn ólyfjani ok báru þá inn; signdi Bárðr fullit, fekk síðan ǫlseljunni; fœrði hon Agli ok bað hann drekka. Egill brá þá knífi sínum ok stakk í lófa á sér; hann tók við horninu ok reist á rúnar ok reið á blóðinu. Hann kvað: ‘Rístum rún á horni | rjóðum spjǫll í dreyra [. . .].’ Hornit sprakk í sundr, en drykkrinn fór niðr í hálm.113 The queen and Bárðr then mixed poison into the drink and carried it inside. Bárðr blessed the drink and gave it to the waitress; she took it to Egill and bade him drink. Egill then drew his knife and thrust it into his wrist. He received the horn and carved runes onto it and smeared his blood

or by Benedikts saga, since the second may be of later date than Egils saga (on Benedikts saga, see Heilagra manna søgur. Fortællinger og legender om hellige mænd og kvinder efter gamle haandskrifter, ed. Christian R. Unger, 2 vols (Christiania: Bentzen, 1877), 1, p. x; Kirsten Wolf, ‘Gregory’s Influence on Old Norse-Icelandic Religious Literature’ in Rome and the North. The Early Reception of Gregory the Great in Germanic Europe, ed. Rolf H. Bremmer Jr, Kees Dekker, and David F. Johnson (Paris: Peeters, 2001), p. 267; Lives of Saints. Perg. Fol. Nr. 2, Royal Library, Stockholm, ed. Peter Foote. Early Icelandic Manuscripts in Facsimile, 4 (Copenhagen: Rosenkilde and Bagger, 1962), p. 25; for manuscript date, see ONP Registre, p. 34). 112 Heilagra manna søgur, ed. Unger, 1, p. 203. The wording in Benedikts saga is similar: ‘En er þeir sá at hann lofaði þeim ekki þess at gera sem mótstaðligt var munka lǫgum, þá kunnu þeir illa því ok vildu ráða dauða hans, ok bundu þat ráð saman, at þeir blǫnduðu eitri drykk hans. En er hánum var ker þat fœrt sem dauðadrykkr var í, þá gerði hann at vanða sínum krossmark á móti eitrinu ok blezaði, áðr hann tœki við kerinu, en þá sprakk í sundr kerit fyrir krossmarkinu, svá sem hann lýsti steini á. Þá skilði Guðs maðr at dauðadrykkr var í kerinu ok mátti ekki standask lífs mark.’ (Heilagra manna søgur, ed. Unger, 1, p. 161) (But when they realized that he did not allow them to do things that were against the monastic rules, then they disapproved of that and wanted to kill him, and together they took that course of action that they mixed poison in his drink. And when the vessel containing the lethal drink was brought to him, he followed his custom and made the sign of the Cross against the poison and blessed the drink before he received the vessel, and then the vessel split apart for the sign of the Cross, as if he had hit it with a stone. Then the man of God understood that there was a lethal drink in the vessel and that it could not withstand the sign of life.) 113 Egils saga, ed. Sigurður Nordal, p. 109.

4.3 Metrical Archaization in Egils saga

227

on them. He proclaimed: ‘We carve runes on the horn | we redden the words with blood [. . .].’ The horn split apart and the drink fell down into the hay.

There are obvious differences between the two passages, most notably that the setting in Egils saga is completely pagan and that the blessing is performed by the villain rather than the hero. Nonetheless, the common elements remain so many and so conspicuous that they suggest direct influence: there is a conspiracy to kill the hero with a poisoned drink; the blessing is present, albeit moved; the hero becomes aware of the ruse by supernatural means; the vessel splits apart. In the last instance, the same words sprakk í sundr are used, supporting the hypothesis on the verbal level. The text from Gregory’s Dialogues is attested in a manuscript of c. 1200–1225 and the translation is likely to have been made before c. 1150.114 Influence from Egils saga on the Dialogues is thus ruled out. Given the cultural importance of the Dialogues and the early date of translation, the opposite direction of influence is plausible.115

114 Wolf, ‘Gregory’s Influence’, pp. 256–57, 266. The main manuscript is AM 677 4to, dated c. 1200–1225 (ONP Registre, p. 258). On the manuscripts generally, see Heilagra manna søgur, ed. Unger, 1, pp. x–xi; ONP Registre, p. 258; The Life of St. Gregory and his Dialogues. Fragments of an Icelandic Manuscript from the 13th Century, ed. Hreinn Benedikttson. Editiones Arnamagnæanæ, B 4 (Copenhagen: Munksgaard, 1963). The Latin text is slightly simplified and abbreviated in the translation, but apart from that, the translation stays true to its original (Grégoire le Grand, Dialogues, 2, ed. Adalbert de Vogüé, trans. Paul Antin. Sources Chrétiennes, 206 (Paris: Éditions du Cerf, 1979), pp. 140–142) 115 Sigurður Nordal has suggested influence from the Dialogues on another another portion of Egils saga (Egils saga, ed. Sigurður Nordal, p. 183 n. 1). The similarities there, however, are of such a general nature that coincidence cannot be ruled out. In Egils saga, Egill has on pain of death been assigned to compose a poem, but a swallow disturbs him so that he cannot concentrate. His friend Arinbjǫrn helps him out, chasing the swallow away, and when he does so he sees that another creature has taken on the form of the swallow – presumably, though we are not told so, the wicked queen Gunnhildr. In the Dialogues, we read: ‘Quadam uero die, dum solus esset, temptator adfuit. Nam nigra paruaque auis, quae uulgo merola uocatur, circa eius faciem uolitare coepit, eiusque uultui inportune insistere, ita ut capi manu posset, si hanc uir sanctus tenere uoluisset. Sed signo crucis edito, recessit auis.’ (Grégoire le Grand, Dialogues, 2, ed. Adalbert de Vogüé, p. 136). (One day, when he [Benedict] was alone, the tempter was there. For a small, black bird which is called merola [blackbird] in the vernacular began to fly about his face and violently attack it, so that, if the holy man would have wished to hold it, he could have grabbed it with his hand. But after he had made the sign of the Cross, the bird receded.) The Norse translations are true to the original, except that in the saga, the bird has been turned into a fly (Heilagra manna søgur, ed. Unger, 1, pp. 160, 202). After this, a woman appears in Benedict’s mind and tempts him. Sigurður Nordal has made the parallel somewhat more conspicuous by saying that Benedict was at prayer, which could perhaps be seen as matching Egill’s poetry, but this is found neither in the Latin nor in the Norse. Siân Grønlie has retained Sigurður’s addition in her analysis of the passage (‘Saint’s Life and Saga Narrative’, p. 15). In general on influence from the Dialogues in Old Icelandic literature, see Wolf, ‘Gregory’s Influence’, pp. 268–69; Eugene J. Crook, ‘Gregory’s Dialogi and the Old Norse Sagas: Njáls Saga’, in Rome and the North. The Early Reception of Gregory the Great in Germanic Europe, ed. Rolf H. Bremmer Jr, Kees Dekker, and David F. Johnson (Paris: Peeters, 2001),

228

4 Prosimetrical Narrative

Since stanza 9 is found in this passage, it would seem that our poet was not loath to avail himself of what hagiography had to offer in his construction of the past, with its uncouth warriors, magic, runes and rough poetics. Furthermore, the stanza describes how Egill carved runes on the horn to test whether it would harm him, and is thus closely aligned with the prose. Pseudo-Egill was thus more than a poetic interpolator: he also had a hand in shaping the prose narrative. Because of this, and since all poetic attestation of interest in runes is attributable to him and he shares this interest with the saga author, there is a strong case to be made that he was indeed none other than the author. Few scholars today would take saga accounts at face value, but since the connection between Egill and runes is borne out by the poetry, it has so far appeared that this feature is older than the saga.116 As it turns out, however, the interest in runes is attributable – exclusively, as it would seem – to Pseudo-Egill, the author of the saga. I return to the authorship of Egils saga below. The nine stanzas considered above indicate that their author thought that Egill composed with irregular hendingar, and Egils háttr in Háttatal betrays a similar perception.117 In the case of Egils háttr, Snorri’s specific model is probably the half stanza beginning Upp skulum órum sverðum quoted in Skáldskaparmál (the first half of the saga’s stanza 12 discussed above). The defining features of Egils háttr are that even lines should be composed in riðhent (only one syllable between the hendingar) and that there should be no hendingar in odd lines. This is a rare pattern to which Egill’s half stanza fully conforms apart from the skothending in line three.118 Still, Snorri’s sample stanza suggests that he may have known poetry by Egill with even more irregular hendings, since he opted for lack of hendingar in all odd lines, and the irregular stanzas in Egils saga point in the same direction. The background to this perception is not entirely clear, since all independently attested lines from Egill’s lausavísur – twelve in all in Skáldskaparmál and 3GT – conform to the classical pattern (apart from the missing hending in the first line of stanza 12 of the saga).119 There may, however, be internal evidence in the saga. Two stanzas, 36 and 54, display a rare hending distribution.120 This is the second group of stanzas with irregular hendingar to which I referred at the outset.

pp. 275–85; Joseph Harris and Tom D. Hill, ‘Gestr’s “Prime Sign”: Source and Signification in Norna-Gests þáttr’, Arkiv för nordisk filologi, 104 (1989), pp. 103–22. 116 See Egils saga, ed. Sigurður Nordal, pp. xviii–xix; Björn M. Ólsen, Runerne i den oldislandske literatur (København: Gyldendal, 1883), p. 6. 117 Snorri, Háttatal, ed. Faulkes, p. 25. 118 I thank Haukur Þorgeirsson for this observation. 119 Skj B 1, pp. 43 (6.1–4), 45 (13.5–6), 52 (44.1–4), 53 (47.1–2); A 1, pp. 50, 51, 59, 60. 120 Egils saga, ed. Sigurður Nordal, pp. 200, 272–73; Skj B 1, pp. 42 (5.1–4, 7–8), 48 (27.1–6); A 1, pp. 48, 55. Stanza 57 has one couplet conforming to the same pattern (Egils saga, ed. Sigurður Nordal, p. 293; Skj B 1, p. 52 (43.7–8); A 1, p. 59).

4.3 Metrical Archaization in Egils saga

229

In three instances in each stanza, the poet makes up for a missing hending with a hending to the next line, tying the lines together in a way reminiscent of the alliterative pattern. The remaining two lines of each stanza have regular hendingar. As a result, no lines in these stanzas lack hendingar entirely. (Note here an instance of cohesion in the saga’s stanza 36.5–6: ‘þar˯nautk enn sem optar | arnstalls sjǫtulbjarnar’.121 Note also that editorial intervention has somewhat obscured the pattern in 54.5.)122 Far from being random, the distribution of hendingar in these lines is clearly the result of conscious elaboration. The technique is realized in two ways (stanza 36.1–4): Urðumk leið123 en ljóta landbeiðaðar reiði. Sígrat gaukr ef glamma gamm veit of sik þramma.124 I grew tired of the foul anger of the land-craver [king]. The cuckoo will not settle down if he knows that the shriek’s griffin [eagle] is winging heavily about him [meaning of line 3–4 insecure].

121 Egils saga, ed. Sigurður Nordal, p. 200; Skj B 1, p. 48 (27); A 1, p. 55; cf. Snorri, Háttatal, ed. Faulkes, pp. 83–84. This instance of cohesion has not been noted previously. 122 As it stands in normalized editions, 54.5 lacks hendingar without compensation. Lines 5–6 (plus a syntactically necessary word from line 8) read: skalat at grundar Gylfa | glaums misfengnir taumar [. . .] erðgróins (I shall not wrongly wield the reins of Gylfi’s [a sea-king’s] land’s [the sea’s] horse [ship] of the earth-grown one [dwarf; dwarf’s ship > poetry]) (Egils saga, ed. Sigurður Nordal, p. 273; Skj B 1, p. 42 (5; substituting of for at)). The problem here is that /l/ does not rhyme with /lf/ (Myrvoll, Kronologi i skaldekvæde, pp. 62–63). This metrical flaw is not, however, attributable to Egill or even to the scribe of the only manuscript witness, Möðruvallabók, which does not read Gylfa but gilia (Skj A 1, p. 48). Rather, the rhyme was unwittingly removed by Finnur Jónsson through a ‘graphically speaking very insignificant change’ (Kritiske studier, p. 118). Kock retains Finnur Jónsson’s emendation (see Den norsk-isländska skaldediktningen, ed. Kock, 1, p. 27 (5)). The problem with reading Gilja, genitive of the giant’s name Gili, is that while it gives a good kenning for ‘mountian’ (giant’s land), it is unclear how this would fit into the context. There are, however, also other problems with this half stanza. The participle erðgróins or jarðgróins (earth-grown) should preferably not be an independent, nominalized heiti for ‘dwarf’, as the editors have supposed, but one would rather have expected it to serve as an attribute (Sigurður Nordal expresses some doubts on this point). It would, in fact, fit quite well with Gilja. Furthermore, as Kock has noted, skalat should really have to read skulut in the plural and, as Sigurður Nordal notes, at ought not to govern the genitive. Given all of these difficulties with the reading proposed by Finnur Jónsson and other editors, there is little reason to suppose that the problem lies in the word Gilja. Not only is this reading the only one with manuscript support; it is furthermore corroborated internally by the overall pattern of the stanza of either having hendingar or replacing them – in this case having them. As things stand, it may be advisable to consider the overall reading of the half stanza as insecure, but not so the reading of Gilja. This means that there is no sea-king (Gylfi), and therefore no ship, and thus no ‘dwarf’s ship [poetry]’ (cf. above pp. 161–64). 123 Here, the hending is fronted to the third, rather than the fifth, position. Fronted hendingar are generally considered an archaic feature. 124 Normalization based on Skj A 1, p. 55, but adopting the emendation of of to ef in line 3 (cf. Egils saga, ed. Sigurður Nordal, p. 200; B 1, p. 48).

230

4 Prosimetrical Narrative

Here, stressed syllables in line 1 and 3 form hendingar with the hending pairs in line 2 and 4. The other possibility is that the hendingar between the lines fall outside of the hending pair to follow (stanza 54.1–2): Mál er lofs at lýsa Ljósgarð, es þák, barða125 It is time to praise the ship’s bright fence [shield], which I have received.

This kind of compensation for lack of hendingar – in both variants – vanished almost completely during the first half of the eleventh century.126 Snorri, however, knew of the possibility and regularized it in Fleins háttr and Braga háttr. That section of Háttatal deals with the fornskáld (ancient poets), indicating that Snorri thought of this as an archaic feature, which supports the chronology of the feature as found in the extant corpus of skaldic poetry. The trait otherwise occurs, above all, in Ragnarsdrápa (mid–late ninth century) and in Vellekla, which was composed in the late tenth century and which is marked by experiments with hendingar.127 Both of these poems are quoted extensively in Snorri’s Edda, and it is thus likely that Snorri’s awareness is at least partly based on the study of them.128 Egils saga’s stanzas 36 and 54 have none of the straightforward word order or restricted kenning use of Pseudo-Egill’s stanzas, and their inaccessibility may indicate that some keys to their understanding have been lost over time. Because of this and the general disappearance of hending compensation during the first half of the eleventh century, I concur with the opinion of Hans Kuhn that stanzas 36 and 54 are ‘surely authentic’.129 It may be further noted that according to the saga stanza 54 was composed in thanks for a shield given to Egill by Einarr skálaglamm, composer of Vellekla – the one skaldic poem that is most reminiscent of stanzas 36 and 54 in its rhyme structure.130 This remarkable convergence of features suggests that the connection between the two poets as described in the saga is not wholly fictitious.131

125 Skj B 1, p.42. The second line reads ‘ljós garð ef þa er ek barða’ in the only manuscript, Möðruvallabók, but this is syntactically untenable (Skj A 1, p. 48). The choice of emended or manuscript reading does not affect the present analysis. 126 Kuhn has identified three instances in Hallfreðr, two in Sighvatr (Myrvoll, Kronologi i skaldekvæde, p. 128, has added one more), two in Þormóðr kolbrúnarskáld, a doubtful one in Þormóðr Kolbeinsson and one in Grímr Droplaugarson. In later poetry, he has found only one instance in Plácitúsdrápa (Kuhn, ‘Vor tausend Jahren’, pp. 306–07; see also Myrvoll, ‘The Authenticity of Gísli’s Verse’). 127 Kuhn, ‘Vor tausend Jahren’, pp. 298–305. 128 Ragnarsdrápa is not found in U and may have been included after Snorri’s time. Vellekla is quoted in all manuscripts, however. 129 Kuhn, ‘Vor tausend Jahren’, p. 301. 130 Kuhn, ‘Vor tausend Jahren’, p. 304. 131 Kuhn, ‘Vor tausend Jahren’, is decidedly optimistic in this regard.

4.3 Metrical Archaization in Egils saga

231

Taken together, the evidence that these stanzas are old is substantial, and on that supposition, it is likely that they served as models and inspired the use of irregular hendingar in the stanzas of Pseudo-Egill. He did not attempt, however, to imitate the feature of compensatory hendingar. Rather, he seems to have analysed the stanzas as primarily irregular. This impression is readily understandable and was to some extent shared by scholars like Finnur Jónsson and Hans Kuhn (see p. 229 nn. 121–22): 36.1 has a hending in the third rather than the fifth position, 36.5–6 has cohesion, hendingar in odd lines tie into the next line in two ways – either into the hending pair or to another syllable – and both stanzas have conventional hendingar in one couplet. Both stanzas would thus have appeared highly irregular from a later, normative perspective, and taking them as models for irregular composition therefore makes a certain amount of sense. Four stanzas in the saga do not have direct bearing on the pseudonymous stanzas but may merit a brief comment on account of the general topic of hending patterns. Stanzas 10, 25, 26 and 30 display interlinear hendingar in three out of four couplets, just like stanzas 36 and 54, but unlike these stanzas, they do not lack internal hendingar in odd lines.132 Stanza 26.1–2 may serve as an example: Erfingi réð arfi Arflyndr fyr mér svarfa.133 The inheritance-eager son [of Þorgeir þyrnifótr] undid my inheritance claim.

The use of hendingar in these stanzas is more elaborated than in classical dróttkvætt and cannot therefore explain Pseudo-Egill’s choice of irregular hendingar. When viewed in tandem with stanzas 36 and 54, however, they may be of some interest for developments in the tenth century, since they share the feature of interlinear hendingar with them, as well as the occurrence of this feature in three out of four couplets, the fourth remaining regular. Egill’s contemporary and, according to the saga, personal acquaintance Einarr skálaglamm composed Vellekla using both compensatory hendingar, as in stanzas 36 and 54 of Egils saga, and interlinear hendingar of the kind found in the four just mentioned.134 The convergence of these features in the work of a poet likely well known to Egill may indicate that the four stanzas in question, or at least considerable portions of them, are authentic, and that we may discern here a small milieu of poets undertaking experiments with hendingar in the second half of the tenth century.135

132 Exceptions only in lines 5–6 in stanzas 10, 25, and 26 and lines 1–2 in stanza 30 (Egils saga, ed. Sigurður Nordal, pp. 110, 156, 159, 169; Skj B 1, pp. 43 (4), 46 (16, 17), 47 (21); A 1, pp. 49, 52–54). 133 Normalization based on Skj A 1, p. 53 (17). 134 Hans Kuhn, ‘Vor tausend Jahren’, pp. 303–04. 135 Another poet who experimented with hendingar at this time was Gísli, the protagonist of Gísla saga, who also exhibits frequent interlinear hendingar. The poetry of Gísla saga has for a long time been considered to be late, but arguments apart from those of Fredrik Paasche have been based on

232

4 Prosimetrical Narrative

The above analysis has shown that Egill’s hending use, often referred to as irregular in scholarly literature, can actually be broken down into four styles: one regular, one with interlinear hendingar, one with compensatory hendingar and one truly irregular. The first three are either as elaborate as ordinary dróttkvætt or more so, whereas the last one lies at the opposite end of the scale. The output of the historical Egill was probably not quite as bewildering as that, since only the first three styles are likely to be his. Pseudo-Egill, however, was prone to look for irregularities, presumably because of his exposure to a didactic system where these were to be identified and corrected. When composing in Egill’s style, therefore, he took irregularity as his guideline. This resulted in stanzas that were alien to the principles that guided early composition, or at least Egill’s composition. Perhaps no other saga displays such a clear poetic stratigraphy as Egils saga, where it is possible not only to extract a coherent corpus of stanzas composed for the saga, but also to discern older poetic building blocks and the means by which these were expanded and adapted to fit the prose.136

questionable criteria, such as claiming that poets would archaize withouth showing this to be the case or that a perceived opposition between good and evil would not fit the mental landscape of the tenth century (see Fredrik Paasche, ‘Esras aabenbaring og pseudo-Cyprianus i norrön litteratur’, in Festskrift til Finnur Jónsson, 29. Maj 1928, ed. Johannes Brøndum-Nielsen et al. (København: Levin & Munksgaard, 1928), p. 201; Foote, ‘An Essay on the Saga of Gisli’; Turville-Petre, ‘Gísli Súrsson and his Poetry: Traditions and Influences’; Emily Lethbridge, ‘Dating the Sagas and Gísla saga Súrssonar’, in Dating the Sagas. Reviews and Revisions, ed Else Mundal (Copenhagen: Museum Tusculanum Press, 2012), pp. 77–113). Recently, Klaus Johan Myrvoll has reassessed the question, and one of his findings is that stanzas containing interlinear hendingar typically display other diagnostic features indicative of great age (hiatus forms, expletive of, aðalhending in a: ǫ etc.) (‘The Authenticity of Gísli’s Verse’). As in Egils saga, formal criteria are supported by the grouping of stanzas. Although the attribution of individual stanzas will probably remain open to debate, it now seems clear that the prosimetrical form of Gísla saga is the product of a process comparable to that of Egils saga: the author had access to a considerable body of poetry by the protagonist, but opted for expanding it further to achieve a full prosimetrical form. 136 In Njáls saga, the earliest manuscripts contain many additional stanzas that have been taken to represent a younger stratum. If so, Njáls saga is a valuable parallel to Egils saga. Guðrún Nordal has conducted an interesting analysis of the interplay between these stanzas and the prose, but for the dating she relies on Finnur Jónsson. His arguments are not exhaustive, but they do involve some strong linguistic criteria (Guðrún Nordal, ‘The Dialogue between Audience and Text: The Variants in Verse Citations in Njáls saga’s Manuscripts’, in Oral Art Forms and their Passage into Writing, ed. Else Mundal and Jonas Wellendorf (Copenhagen: Museum Tusculanum Press, 2008), pp. 185–202; Finnur Jónsson, ‘Om Njála’, Aarbøger for nordisk oldkyndighed og historie, 2: 19 (1904), pp. 93–97; idem, ‘Sagaernes lausavísur’, Aarbøger for nordisk oldkyndighed og historie, 3: 2 (1912), pp. 4–7; idem, Den oldnorske og oldislandske litteraturs historie, 2, p. 166).

4.4 The Authorship of Egils saga

233

4.4 The Authorship of Egils saga The question reports itself whose method we have here reconstructed. Many scholars believe that Snorri composed Egils saga, but their studies do not address the problem that Snorri seems not to have composed pseudonymous poetry in his known works, whereas the author of Egils saga did. The matter therefore needs to be reassessed in light of the preceding analysis. For reasons that will become evident below, I assume that previous scholarship is right in that the author must either be Snorri or someone close to him and sharing his interests. Óláfr Þórðarson is probably the best candidate after Snorri, not least since he demonstrably shared the author’s unusually strong interests in runes (see pp. 182–86). I therefore begin with a discussion of whether Óláfr is more likely to have composed pseudonymously than Snorri.137 In order to address this question, it is necessary to have a saga of his for comparison, and the evidence that Óláfr composed Knýtlinga saga is strong. Knýtlinga saga treats the Danish kings from Haraldr blátǫnn (r. 958-986) to Knútr Valdimarsson (r. 1182-1202). Towards the end of the saga, the author digresses in order to mention Knútr’s brother Valdimarr (r. 1202–1241). About him we read: [Valdimarr] er síðan var konungr í Danmǫrk, er einhverr hefir verit ágætastr konungr hingat á Norðlǫnd. Með honum var Óláfr Þórðarson ok nam at honum marga frœði, ok hafði hann margar ágætligar frásagnir frá honum.138 [Valdimarr] who was later king in Denmark, who was one of the greatest kings here in the North. With him was Óláfr Þórðarson and learned much lore from him, and he got many excellent stories from him.

After this, the narrative about Knútr resumes, and Valdimarr never reappears. As the text reads, the most excellent thing about Valdimarr seems to be that Óláfr was with him. This sounds very much like the Óláfr we know from 3GT, who is so proud of having discussed runes with Valdimarr.139 This passage is the main reason that scholars have thought that Óláfr is the author, and it is so out of place and so reminiscent of his remarks in 3GT that this seems like the most plausible explanation of its presence. It is not, however, an explicit attribution, and the hypothesis therefore needs further evaluation. Knýtlinga saga is composed as a complement to Heimskringla, focusing on episodes and quoting poetry that are bypassed there.140 I discuss Heimskringla’s

137 In ‘Er Ólafur Þórðarson höfundur Eglu?’, Són, 13 (2015), pp. 173–79, I came down in favour of Óláfr, since Snorri does not appear to have composed pseudonymously in his other works. At the time, however, I had not conducted a comparative analysis of Knýtlinga saga. 138 Danakonunga sǫgur, ed. Bjarni Guðnason. Íslenzk fornrit, 35 (Reykjavík: Hið íslenzka fornritafélag, 1982), p.315. 139 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 6. 140 Danakonunga sǫgur, ed. Bjarni Guðnason, pp. lxxi, lxxiv, lxxvi, xcvii, cxxiv–v, cxxxiv, cxxxix; a separate *Knúts saga is convincingly rebutted on pp. cxii–cxxxiv.

234

4 Prosimetrical Narrative

authorship below and conclude that the arguments in favour of the traditional attribution to Snorri are many and strong. The relationship between Knýtlinga saga and Heimskringla therefore reminds us of the Óláfr who wrote 3GT, a Latinate alternative to Snorri’s Edda. Even the Latin aspect is comparable, since the author of Knýtlinga saga drew much material from Danish sources in Latin, and most notably Saxo, which was an option only to those with a high level of Latinity. Again, this fits Óláfr, whom 3GT reveals as a good Latinist by Icelandic standards. Furthermore, the Danish sources used by the author are many, and none of them appears to have made it to Iceland.141 He must therefore have spent a period at a good library in Denmark, which suits Óláfr’s biography, since he must have spent at least the winter of 1240–1241 there.142 Knýtlinga saga was probably finished after 1252, since that is when Hákon and Ríkiz got married, and the saga mentions them as married.143 Óláfr died in 1259. In sum, we have here an unusually good Latinist who has spent time at a library in Denmark and composes his text as a complement to Heimskringla. When this is viewed in tandem with the mention of Óláfr’s learned discussions with Valdimarr, 3GT’s mention of the same and the lack of any indications to the contrary, I believe that the case for Óláfr’s authorship is strong. This attribution allows us to study Óláfr’s treatment of poetry in a saga context. Knýtlinga saga contains 60 stanzas, and all but the last of them are authenticating. Even the situational stanza 60 raises no particular suspicions: it is attributed to Einarr Skúlason and it is a complaint about the lack of payment for poetry by king Sveinn.144 Stanza 31 is anonymous and stanzas 16–17 have another attribution in other sources.145 The author appears to have found stanzas 16–17 in a text related to Styrmir’s book (on this book, see pp. 72–75), and his recourse to this somewhat unreliable line of transmission is probably due to his ambition to complement, rather than overlap with, Heimskringla.146 This hardly indicates a sloppier standard than Heimskringla, but rather that the massive corpus of poetry presented there forced the author of Knýtlinga saga to go further afield. Overall, Óláfr’s treatment of poetry in Knýtlinga saga is about as rigorous as Snorri’s in Heimskringla, or even more so, if one considers that Knýtlinga saga has much fewer situational quotations than Heimskringla. This means that Óláfr is not a better candidate for authorship of Egils saga than Snorri. True, Óláfr had a demonstrable predilection for runes, but given Snorri’s obsession with things old and Nordic, it is more likely than not that Óláfr shared these views with his uncle. In Háttatal Snorri mentions runes as a point of comparison, saying that

141 142 143 144 145 146

Danakonunga sǫgur, ed. Bjarni Guðnason, pp. lxxxv, cxxv–vi, cxxix, cli, clvii–clxvii, clxxvi–vii. Danakonunga sǫgur, ed. Bjarni Guðnason, pp. clxxi–iv. Danakonunga sǫgur, ed. Bjarni Guðnason, p. clxxx. Danakonunga sǫgur, ed. Bjarni Guðnason, p. 275. Danakonunga sǫgur, ed. Bjarni Guðnason, pp. 116, 144. Danakonunga sǫgur, ed. Bjarni Guðnason, p. 116.

4.4 The Authorship of Egils saga

235

‘Þessi er upphaf allra hátta sem málrúnar eru fyrir ǫðrum rúnum’147 (this [dróttkvætt] is the origin of all metres as runes with a sound value are for other runes). He uses the word rún a few additional times, but it then appears to mean ‘wisdom’, ‘proverb’ or ‘mystery’ rather than ‘rune’.148 Nonetheless, the quotation from Háttatal shows that he was acquainted with their use. Other grammarians remained sceptical of or indifferent to runes, and while they are mentioned in several sagas, they never assume centre stage as they do in Egils saga.149 Much has been written about Snorri’s possible authorship of Egils saga. First, it should be noted that Oddaverja annáll says that Snorri ‘samsetti Eddu ok margar aðrar frœðibœkr, íslenzkar sǫgur’ (composed the Edda and many other learned books, Icelandic sagas). The manuscript is from the fifteenth century, but here, an older chronicle reaching to 1313 has been used.150 Furthermore, Icelanders did not know that Snorri composed other works than the Edda in the late fifteenth century.151 The reference therefore clearly dates to a period close to Snorri, and ‘Icelandic sagas’ probably refers to sagas of Icelanders, although it could conceivably also refer to contemporary sagas. It is therefore likely that Snorri composed some saga of Icelanders, and the obvious choice is Egils saga, the saga about Snorri’s relative who lived at Borg, Snorri’s first possession, and who like Snorri was a famous poet. Furthermore, the saga author has a very good knowledge about Rangárvellir, where Snorri was raised, and Borgarfjörður, which later became his home and power base. The great focus on Ketill hœingr’s settlement and thus on Rangárvellir does not seem motivated in a saga about Egill, but makes sense from Snorri’s geographical horizon.152 This horizon, it should be noted, fits Snorri, but not Óláfr. Björn M. Ólsen argued that Snorri used Egils saga in Heimskringla. Like G. A. Gjessing before him and scholars since, Björn M. Ólsen assumed that most similarities between Egils saga and Heimskringla were due to their use of the same sources.153 He thought, however, that some of the overlaps which he presented could not come from

147 Snorri, Háttatal, ed. Faulkes, p. 5. 148 Snorri, Prologue and ‘Gylfaginning’, ed. Faulkes, p. 53; Snorri, Skáldskaparmál, ed. Faulkes, p. 3; Heimskringla, ed. Bjarni Aðalbjarnarson, 1, p. 19 (the last of these occurrences should perhaps be understood as ‘rune’). 149 Males, ‘Er Ólafur Þórðarson höfundur Eglu?’; see also Tarrin Wills, ‘The Thirteenth-century Runic Revival in Denmark and Iceland’, North-Western European Language Evolution, 69: 2 (2016), pp. 114–29. 150 Egils saga, ed. Sigurður Nordal, p. xcii. 151 Heimskringla, ed. Bjarni Aðalbjarnarson, 1, p. viii. 152 Björn M. Ólsen, ‘Landnáma og Egils saga’, Aarbøger for nordisk Oldkyndighed og Historie, 2: 19 (1904), pp. 167–247; Egils saga, ed. Sigurður Nordal, p. xc. Sigurður Nordal shows that Björn M. Ólsen’s hypothesis that the author considerably expanded Ketill’s and Skalla-Grímr’s settlement is not valid, but the focus on Rangárvellir remains (Egils saga, ed. Sigurður Nordal, pp. xxxii–iv, lxxiii). 153 Gustav Antonio Gjessing, ‘Egils-saga’s Forhold til Kongesagaen’, Arkiv för nordisk filologi, 2 (1885), pp. 289–318.

236

4 Prosimetrical Narrative

those sources, but must be due to Heimskringla’s use of Egils saga, or that the author remembered them from his earlier work on Egils saga.154 This seems likely, but it is still possible that they are due to direct recourse to a longer version of the þáttr about Hálfdan svarti and Haraldr harðráði, as well as the saga about Hákon góði to which Egils saga refers.155 At least, both authors used the same sources, but in Heimskringla, the author has revised many facts in light of the greater historiographic undertaking.156 The use of the same sources lends some support to the attribution to Snorri, but the decisive evidence must be sought elsewhere. If his authorship of Egils saga can be demonstrated, however, the reworking in Heimskringla means that Egils saga must be the earlier of the two texts. There is thus much circumstantial evidence for Snorri’s authorship of Egils saga, but scholars have also taken a stylistic approach. In 1962 and 1968, Peter Hallberg studied the use of certain words and other features in Egils saga, Heimskringla and many other texts.157 Haukur Þorgeirsson has later revisited Hallberg’s analysis of verbal matches using digital tools and little interference of active choice on the part of the scholar.158 Both find that Heimskringla and Egils saga are the closest matches among comparable saga texts. Haukur Þorgeirsson finds that Knýtlinga saga is the closest match to Heimskringla after Egils saga, and this is not surprising, seeing that Knýtlinga saga is composed as a complement to Heimskringla.159 Nonetheless, Egils saga remains the closest match, and this is somewhat remarkable given that Knýtlinga saga is generically closer to Heimskringla and that it used Heimskringla as a model. This speaks in favour of Snorri’s authorship of Egils saga. Importantly, Haukur Þorgeirsson treats Heimskringla 2 separately, since it is indebted to the Separate Saga of Saint Óláfr, and Jonna Louis-Jensen has contended that it may have another author than Heimskringla 1 and 3.160 Nonetheless, Egils saga remains

154 Björn M. Ólsen, ‘Landnáma og Egils saga’, pp. 203–09. 155 Heimskringla, ed. Bjarni Aðalbjarnarson, 1, pp. liv, lviii, lxxxi–v; Egils saga, ed. Sigurður Nordal, p. 239. 156 Heimskringla, ed. Bjarni Aðalbjarnarson, 1, pp. liv, lviii, lxxxi–v; Margaret Cormack, ‘Egils saga, Heimskringla, and the Daughter of Eiríkr blóðøx’, Alvíssmál, 10 (2001), pp. 61–68. 157 Peter Hallberg, Snorri Sturluson och ‘Egils saga Skallagrímssonar’. Ett försök till språklig författarbestämning (Reykjavík: Heimspekideild Háskóla Íslands, 1962); idem, Stilsignalament och författarskap i norrön sagalitteratur. Synpunkter och exempel (Göteborg: Acta Universitatis Gothoburgensis, 1968). 158 Haukur Þorgeirsson, ‘How Similar are Heimskringla and Egils saga? An Application of Burrow’s Delta to Icelandic Texts’, European Journal of Scandinavian Studies, 48 (2018), pp. 1–18. 159 Haukur Þorgeirsson, ‘How Similar are Heimskringla and Egils saga?’, pp. 14–15. 160 Jonna Louis-Jensen, ‘Heimskringla – Et verk af Snorri Sturluson?’, Nordica Bergensia, 14 (1997), pp. 230–45, argues that Kringla is the only witness to Heimskringla 2, and that the revision of this part need not have been undertaken by Snorri, but is perhaps an ad hoc-adaptation of the Separate Saga of Saint Óláfr to Kringla itself. She suggests that this revision may have been undertaken by Óláfr Þórðarson in the last year of his life. Later, Louis-Jensen has noted that the attribution of the Separate Saga is based on the assumption that Heimskringla 2 is by Snorri, and that if Heimskringla 2 is simply an ad hoc-revision of the Separate Saga, there is no particular

4.4 The Authorship of Egils saga

237

the closest match to both Heimskringla 1 and 3 and Heimskringla 2. Jón Karl Helgason, Sigurður Ingibergur Björnsson and Steingrímur Páll Kárason have compared the various parts of Heimkringla with Egils saga, using a similar method.161 Interestingly, the texts that are most similar to Heimskringla 2 are Egils saga, Óláfs saga Tryggvasonar from Heimskringla 1 and Haraldar saga Sigurðssonar from Heimskringla 3. This suggests that the three parts of Heimskringla are by one and the same author, or that the author of Heimskringla 1 and 3 also composed the Separate saga of Saint Óláfr, subsequently reworked into Heimskringla 2, whether this revision was undertaken by him or someone else.162 Furthermore, as will be discussed below, the treatment of sources is remarkably similar in Heimskringla 1 and 2, and this also speaks in favour of shared authorship. Haukur Þorgeirsson also addresses the observation made by Jonna Louis-Jensen that some archaic features (unz, later til þess er and of, later um) that were found in the archetype of Egils saga appear not to have been present in the archetype of any of the parts of Heimskringla.163 He notes that these changes occurred in Snorri’s adult lifetime and that such stylistic changes are common in one and the same author.164 A remaining weakness in the argument, however, is that the studies reviewed above do not compare Egils saga and Heimskringla to the prose in Snorri’s Edda, which is the text that is most securely attributable to Snorri. In an earlier study of the distribution of ok er versus en er (and when), Haukur Þorgeirsson conducted such a comparison, and he found that Snorri’s Edda had a very high ratio of en er (83%) as compared to most sagas, although not quite as high as Egils saga (89%) or Heimskringla (94%).165 It is possible that some additional parameters used by Hallberg are applicable to the narrative prose in Snorri’s Edda. I have therefore read the prose of the prologue, Gylfaginning and Skáldskaparmál with an eye to such features. Some of the technical prose of Skáldskaparmál is irrelevant for a comparison with saga text, and coming up with an exact word count for relevant prose is not easy, but the number 26.000 is certainly not too high (a high number would skew the results slightly in favour of

reason to assume that the latter was composed by Snorri (‘Dating the Archetype. Eyrbyggja saga and Egils saga Skallagrímssonar’, pp. 140–41). 161 ‘Fingraför fornsagnahöfunda. Fráleiðsla í anda Holmes og stílmæling í anda Burrows’, Skírnir, 191 (2017), pp. 288–91. 162 See previous note. 163 Jonna Louis-Jensen, ‘Heimskringla og Egils saga – samme forfatter?’, in Studier i Nordisk. Selskab for Nordisk Filologi. Foredrag og årsberetning, ed. Dorthe Duncker (Selskab for Nordisk Filologi: København, 2009), pp. 103–11; eadem, ‘Dating the Archetype. Eyrbyggja saga and Egils saga Skallagrímssonar’, in Dating the Sagas. Reviews and Revisions, ed. Else Mundal (Copenhagen: Museum Tusculanum Press, 2013), pp. 133–47. 164 Haukur Þorgeirsson, ‘How Similar are Heimskringla and Egils saga?’, p. 16. 165 Haukur Þorgeirsson, ‘Snorri versus the Copyist. An Investigation of a Stylistic Trait in the Manuscript Traditions of Egils Saga, Heimskringla and the Prose Edda’, Saga-Book, 38 (2014), pp. 61–74.

238

4 Prosimetrical Narrative

Snorri, resulting in confirmation bias). I have used Faulkes edition, which is based on R and is therefore a reasonable approximation of Snorri’s style for the present purposes (see pp. 109–10, 116–18) Three features in particular appear promising: Hallberg notes that Egils saga and Heimskringla display an extremely low ratio of epic nú (‘now he rode’ etc.), and this is true also of Snorri’s Edda. I have found only ten instances, although other uses of nú abound, partly due to the complex layering of mythical time, internal narrator’s time and external narrator’s time.166 Ten instances in a body of 26.000 words gives 3,8 instances per 10.000 words, which is very close to Heimskringla’s 3,5 and Egils saga’s 3,1.167 This is far below any other kings’ sagas studied by Hallberg, except Hákonar saga Hákonarsonar with 3,8. Next comes Orkneyinga saga with 8,1, thereafter Knýtlinga saga with 15. Apart from Egils saga, other sagas of Icelanders have 10,0 occurrences or more. Heimskringla, Egils saga and Snorri’s Edda are thus all on the extreme low end of the use of epic nú. Another peculiar feature observed by Hallberg is that while most saga authors use the verbs venda (turn) and snúa interchangeably, Egils saga and Heimskringla avoid venda in favour of snúa, halda aptr (turn back) or fara (travel). As it turns out, venda is absent in both Egils saga and Snorri’s Edda. In Heimskringla venda is used nine times, but this seems to be due to the sources used, since it is possible to verify that Snorri has changed venda to other verbs on at least seven occasions.168 The occurrences of venda listed in ONP do not suggest any clear chronological distribution, and Snorri’s dislike for the verb therefore appears to have been idiosyncratic. Hallberg also stresses the rare occurrence of historic present in Egils saga and Heimskringla.169 I have opted for reading the text and taking note of the overall use of historical present, rather than adding to Hallberg’s complex statistical analysis of individual words. Like Hallberg, I have excluded words of saying, since these are often found in the present in saga texts generally. As it turns out, Snorri’s Edda is generally very restrictive with the use of historic present, except that a short chain of such forms may be triggered by an utterance, due to tense interference from direct speech (‘he says/said that he wants’ etc.). There are, however, two noteworthy exception to this rule. In the second half of Gylfaginning and in the beginning of Skaldskaparmál, we find stretches of historic present in Þórr’s journey to Útgarðaloki and, somewhat less pronounced, in Ægir’s report to Bragi.170 The reason for this stylistic breach is not clear. Snorri does not quote poetry in

166 These are found in Snorri, Prologue and‘Gylfaginning’, ed. Faulkes, pp. 39 (2), 40 (8), 41 (12, 16 [twice], 26), 48 (36); Snorri, Skáldskaparmál, ed Faulkes, pp. 42 (9, 15), 45 (15); cf. Hallberg, Stilsignalament och författarskap i norrön sagalitteratur, pp. 31–35, 195–97. 167 Hallberg, Stilsignalament och författarskap i norrön sagalitteratur, pp. 195–96. 168 Cf. Hallberg, Stilsignalament och författarskap i norrön sagalitteratur, pp. 22–24. 169 Cf. Hallberg, Stilsignalament och författarskap i norrön sagalitteratur, pp. 61–79, 206–08. 170 Snorri, Prologue and ‘Gylfaginning’, ed. Faulkes, pp. 38–43. Note that this is not the case with the beginning of the narrative (p. 37); Snorri, Prologue and‘Gylfaginning’, ed. Faulkes, pp. 1–4.

4.4 The Authorship of Egils saga

239

the section on Útgarðaloki, which is at odds with his normal way of substantiating his narrative by quoting older sources, and the dialogue between Ægir and Bragi is an awkward double to the elaborate dialogue setting of Gylfaginning. It is possible that he in these cases drew on earlier texts and that this has resulted in the methodological and compositional incoherence, as well as the atypical use of historic present. Whatever the case may be, these two sections are at odds with the overall style of Snorri’s Edda, which may be assumed to be that of its author. This restrictive use of historic present is in accordance with Hallberg’s analysis of Heimskringla and Egils saga. These observations on the use of epic nú, venda and historic present complement previous analyses by comparison with the most securely attributed text, and it turns out that several parameters mark off Snorri’s Edda, Heimskringla and Egils saga as a distinct stylistic group within the Old Icelandic corpus. While Knýtlinga saga shares the very low ratio of historic present with these three texts, it has a much higher incidence of epic nú (15 per 10.000 words) and uses venda and snúa interchangeably, albeit with a preference for snúa (ten snúa, four venda).171 Three parameters thus corroborate the stylistic affinity of Heimskringla, Egils saga and Snorri’s Edda, and two of these indicate that the author of Knýtlinga saga had slightly different stylistic preferences than those represented by the other three, in spite of the fact that Knýtlinga saga is the closest generic match to Heimskringla among these texts. Overall, many mutually independent factors speak in favour of Snorri’s authorship of Egils saga. As noted above, the author’s re-evaluation of many historical facts, based on further historical research, indicates that Snorri composed Egils saga some time before Heimskringla. There has been some debate about when he might have done so, but his detailed knowledge of Norwegian topography makes it likely that the saga should be dated after Snorri’s first sojourn in Norway 1218–1220, and thus probably to the 1220s, perhaps early in that decade.172 In light of the preceding discussion, it is necessary also to revisit the question of the authorship of Heimskringla, in particular since we have no medieval attribution of the text. The first attribution is found in the Danish translation by the Norwegian Laurents Hanssøn from the middle of the sixteenth century.173 At the time, Snorri was

171 Hallberg, Stilsignalament och författarskap i norrön sagalitteratur, pp. 195–97, 206–08. My count of venda and snúa. 172 ‘Egils saga Skallagrímssonar’ nebst den grösseren Gedichte Egils, ed. Finnur Jónsson, 2nd ed. (Halle: Niemeyer, 1924), pp. xxiii; Björn M. Ólsen, ‘Landnáma og Egils saga’, p. 224 n. 3, notes that only two of the topographic inaccuracies listed by Finnur Jónsson are really inaccurate (the reference is to the first edition of 1894). Further studies on Snorri’s authorship of Egils saga include: Torfi H. Tulinius, Skáldið í skriftinni. Snorri Sturluson og ‘Egils saga’, pp. 167–77; Jónas Kristjánsson, ‘Var Snorri Sturluson upphafsmaður íslendingasagna?’, in Sagnalíf. Sextán greinar um fornar bókmenntir, ed. Þórður Ingi Guðjónsson (Reykjavík: Stofnun Árna Magnússonar í íslenskum fræðum, 2015 [1990]), pp. 250–74; Vésteinn Ólason, ‘Er Snorri höfundur Egils sögu?’, Skírnir, 142 (1968), pp. 48–67; Sigurður Nordal, ‘Sagalitteraturen’, in Litteraturhistorie, B: Norge og Island, ed. Sigurður Nordal. Nordisk kultur, 8: B (Stockholm: Bonniers, 1953), pp. 241–44. 173 Heimskringla, ed. Bjarni Aðalbjarnarson, 1, p. vi.

240

4 Prosimetrical Narrative

known in Iceland as the author of his Edda, but not of historiography, and Snorri was in no way the towering literary figure that he would later become.174 Laurents Hanssøn would therefore have had no contemporary reason to attribute the text to Snorri, but must have drawn on medieval evidence. Some scholars have argued that the information about Snorri’s authorship was drawn from the lost prologue of Kringla (c. 1258–64) – the best witness to Heimskringla, which is now lost but for one page (Lbs fragm 82), but which exists in what appears to be very good transcripts.175 As Jon Gunnar Jørgensen has shown, however, Kringla had no prologue in the early seventeenth century and may very well never have had one.176 Laurents Hanssøn’s main exemplar was Codex Frisianus (AM 45 fol), but he also consulted another manuscript, which appears to have contained the prologue. This manuscript cannot have been Kringla, but rather a manuscript with close affinities to the one used in the compilation of the Greater Saga of Óláfr Tryggvason.177 Its prologue could possibly have contained Snorri’s name, but this seems unlikely, since this is not the case in preserved manuscripts, and since kings’ saga compilations are generally not attributed to authors. Jon Gunnar Jørgensen therefore argues that the information about the author stems from the activities of the humanists in Bergen in the 1530s and 1540s.178 These scholars had a great interest in both Bergsbók and Orkneyinga saga, where Snorri is mentioned as an authority on Norwegian royal history. If Jørgensen’s hypothesis is correct, our guess is as good as that of the Bergen humanists, and we must look to Orkneyinga saga and the Greater Saga of Óláfr Tryggvason (found in Bergsbók and elsewhere). In these texts, we find references to what ‘some say’ versus what ‘Snorri Sturluson says’. As Ólafur Halldórsson has shown, the reference in Orkneyinga saga could be to either Magnúss saga berfœtts in Heimskringla 3, Fagrskinna or Morkinskinna, but since it can on other grounds be shown that the person who revised Orkneyinga saga used Óláfs saga helga as found in Heimskringla 2, it would seem that he is referring to Heimskringla 3.179 The references in the Greater Saga of Óláfr Tryggvason are clearly to

174 Heimskringla, ed. Bjarni Aðalbjarnarson, 1, p. viii. 175 Heimskringla. Nóregs konunga sǫgur, ed. Finnur Jónsson, 3 vols (København: Samfund til udgivelse af gammel nordisk litteratur, 1893–1900), 1: ix–x; Heimskringla, ed. Bjarni Aðalbjarnarson, 1, pp. vi–ix; Sverrir Tómasson, Formálar íslenskra sagnaritara á miðöldum, pp. 382–83. On the date of Kringla, see Stefán Karlsson, ‘Kringum Kringlu’, in Stafkrókar, ed. Guðvarður Már Gunnlaugsson (Reykjavik: Stofnun Árna Magnússonar, 2000 [1977]), 251–73. On the quality of the transcripts, see Jon Gunnar Jørgensen, ‘Om verdien av sagaavskrifter fra 1600-tallet’, Collegium medievale, 11 (1998), pp. 87–103; idem, The Lost Vellum Kringla (Copenhagen: Reitzel, 2007), pp. 289–311. 176 Jon Gunnar Jørgensen, ‘“Snorre Sturlesøns fortale paa sin chrønicke”. Om kildene til opplysingen om Heimskringlas forfatter’, Gripla, 9 (1995), pp. 43–62 (p. 49). 177 Jakob Benediktsson, ‘Hvar var Snorri nefndur höfundur Heimskringlu?’, Skírnir, 129 (1955), pp. 118–27 (pp. 123–24; Jørgensen, ‘“Snorre Sturlesøns fortale paa sin chrønicke”’, pp. 50–54. 178 Jørgensen, ‘“Snorre Sturlesøns fortale paa sin chrønicke”’, pp. 57–60. 179 Ólafur Halldórsson, ‘Sagnaritun Snorra Sturlusonar’, in Snorri. Átta alda minning, ed. Gunnar Karlsson (Reykjavik: Sögufélag, 1979), pp. 122–23

4.4 The Authorship of Egils saga

241

Óláfs saga Tryggvasonar as found in Heimskringla 1, often verbatim.180 In these instances, then, ‘Snorri Sturluson’ appears to mean Heimskringla, or at least Heimskringla 1 and 3. Another factor to take into account is that the critically acute discussion of poetry in Heimskringla’s prologue looks like nothing if not the author who made detailed comments on how poetry has changed over time and how its formal integrity must be respected in Snorri’s Edda. I would also mention two factors that have not previously entered the discussion. First, as noted above, the only previous stylistic comparison of Snorri’s Edda and Heimskringla is Haukur Þorgeirsson’s study of ok er and en er, but also other of Hallberg’s parameters combine to suggest that Snorri’s Edda, Heimskringla and Egils saga form a distinctive stylistic group. Second, we know that Óláfr composed a Latinate alternative to Snorri’s Edda in 3GT. In Knýtlinga saga, he similarly composed a complement to Heimskringla based on Latin sources, and it seems plausible that this is yet another attempt at emulating the literary achievements of his uncle. Above I used the same argument in favour of Óláfr’s authorship of Knýtlinga saga, and the reasoning may therefore appear circular. This is not the case, however. The authorship of Snorri’s Edda and 3GT is known (see pp. 116–18, 178 n. 268, 235), and the relationship between these two texts is comparable to that between Heimskringla and Knýtlinga saga. The argument is based on a comparison of a pair of authors and texts to another pair of texts and the possibility that the similarities between the two pairs of texts are due to the fact that they were written by the same pair of authors. The analogous relationship between the two sets of texts need not be due to the same cause, that is, the same authors, but it is a plausible hypothesis that it is, and this observation may therefore be added to the other factors that speak for Snorri’s authorship of Heimskringla (and Óláfr’s authorship of Knýtlinga saga). This is certainly not an observation that can stand on its own merits, but it supports the attribution made on other grounds. Based on these observations – not least the stylistic affinity of Snorri’s Edda and Heimskringla – I would suggest that Snorri’s authorship of Heimskringla is more secure than scholars sometimes assume.181 To return now to Egils saga, we must conclude that if Snorri is its author, he held himself up to different standards in different contexts, since he clearly did not compose pseudonymously in his Edda (see above p. 174). In Heimskringla, several factors show that he did not add any poetry of his own, and I turn now to these. As Jon Gunnar Jørgensen has observed, no poetry about Óláfr Haraldsson in Heimskringla portrays him as holy before his death. Such poetry appears only after his death and indeed, from then on, this is true of all poetry about him. The difference 180 Ólafur Halldórsson, ‘Sagnaritun Snorra Sturlusonar’, pp. 123–27 181 Additional studies of the authorship of Heimskringla include: Patricia Pires Boulhosa, Icelanders and the Kings of Norway. Mediaeval Sagas and Legal Texts (Leiden: Brill, 2005), pp. 6–21; Sigurjón Páll Ísaksson, ‘Ólafs saga helga’ og ‘Heimskringla’ (Reykjavík: Skúmhöttur, 2014).

242

4 Prosimetrical Narrative

could hardly have been more striking. Snorri, by contrast, emphasizes Óláfr’s Christian behavior in life as well as in death, and it is difficult to think of a more compelling reason to fabricate ‘old’ skaldic poetry than that of verifying Óláfr’s saintly credentials with contemporary poetry. This Snorri did not do, and his claims regarding Óláfr’s holy life thus remain unsubstantiated by poetry.182 It is also noteworthy that the poets who could have commemorated the battle of Stiklarstaðir died in the battle, and because of this, Heimskringla contains almost no poetry by people who were there, except the stanzas composed before the battle, a stanza by the young Haraldr Sigurðarson, and some additional stanzas by Þormóðr. Instead, Snorri quotes poets who were absent, such as Sighvatr, Hofgarða-Refr and Bjarni gullbrárskald.183 In Óláfs saga Tryggvasonar in Heimskringla 1 there is no poetry addressed to Óláfr attesting to his activities in Norway, except that he gave a sword to Hallfreðr.184 It would seem that Snorri did not have access to such poetry, or he would presumably have backed up his description of Óláfr’s missionary activities with contemporary evidence. The poetry found in Hallfreðar saga is of little use in this regard, and Hallfreðr’s praise poems treat Óláfr’s youthful exploits and his fall at Svǫlðr, but not his activities in Norway, although parts of these poems have been lost.185 By all appearances, there had never been much poetry addressed directly to the living Óláfr and possibly nothing about his rule of Norway, and Snorri’s account reflects this state of affairs. Snorri favoured contemporary evidence, and he therefore did not quote twelfth-century poems about Óláfr Tryggvason like Rekstefja. Nonetheless, it seems that Snorri wanted to insinuate that his narrative was based on contemporary sources, since he writes: ‘Af Hallfrøðar kvæðum tǫkum vér vísendi ok sannendi, þat er þar er sagt frá Óláfi konungi Tryggvasyni’ (We get knowledge and confirmation [of historical facts] from the poems of Hallfreðr, [based on] what is said there about king Óláfr Tryggvason).186 This sounds reassuring, but Snorri’s habit in Heimskringla is to corroborate the kings’ activities with contemporary poetry, preferably addressed to the kings themselves, and none is forthcoming. His statement says nothing about exactly what is known or corroborated by Hallfreðr’s poetry, so that while it sounds as if Snorri had reliable sources for Óláfr’s Norwegian activities, this is not what he actually says. We see here that he allowed

182 Jon Gunnar Jørgensen, ‘Óláfr Haraldsson, King, Warrior and Saint: Presentations of King Óláfr Haraldsson the Saint in Medieval Poetry and Prose’, in King and Warrior in Early North-West Europe, ed. Jan Erik Rekdal and Charles Doherty (Dublin: Four Courts Press, 2016), pp. 345–98. 183 Heimskringla, ed. Bjarni Aðalbjarnarson, 2, pp. 358–94. 184 Heimskringla, ed. Bjarni Aðalbjarnarson, 1, pp. 291–336; the stanza is on p. 331. It is also quoted in Hallfreðar saga (Vatnsdœla saga, ed. Einar Ól. Sveinsson, pp. 161–62). Three stanzas addressed to Hákon are found on pp. 298–300. 185 SkP 1, pp. 386–441. 186 Heimskringla, ed. Bjarni Aðalbjarnarson, 1, p. 332.

4.4 The Authorship of Egils saga

243

himself some rhetorical posturing in order to bolster the impression of reliability of his narrative, but he stopped short of claiming that he had access to poems that did not exist, and he certainly did not make up for the lack of sources by composing them himself. These observations about Snorri’s treatment of the two Óláfar do not necessarily indicate that all poetry in Heimskringla is authentic, but that Snorri did not compose any additional stanzas himself, not even when he would have had every reason to do so. Furthermore, the similar treatment of sources in Ólafs saga Tryggvasonar (Heimskringla 1) and Ólafs saga helga (Heimskringla 2) also serve to strengthen the hypothesis that they were composed by the same author, and Óláfs saga Tryggvasonar is the one text in Heimskringla which is most securely attributable to Snorri based on medieval references (see above). In Egils saga, by contrast, Snorri apparently had no qualms about composing some of the poetry himself. In so doing, he even broke the rule he himself gives in Háttatal, namely that contemporary poets may not compose in the irregular metres of the old skalds (see pp. 35–37, 119–20). It is perhaps not surprising that he thought that he could take liberties in the presentation of a colourful, local poet that he would not do when writing the history of kings or teaching aspiring poets about composition, but the contrast is certainly stark. Egils saga is therefore not only the most informative source to the methods of a pseudonymous poet, but it also shows us just how much the methodology of the same author might change depending on context. Furthermore, it is noteworthy that Snorri did not attempt to imitate Egill’s poetry closely. He was an extremely acute observer of poetic patterns, as his construction of Egils háttr based on Upp skulum órum sverðum shows (see above p. 228). He is also the only poet after c. 1000 who can be shown to have understood the system of compensatory hendingar, as in the saga’s stanzas 36 and 54 (see above pp. 35–37). One would therefore assume that he could have produced stanzas that were more similar to 36 and 54 than the chaotic ones that he seems to have composed. The main reason for why he did not can probably be inferred from Háttatal. His commentaries there indicate that he saw irregularity as the most conspicuous feature of the early poets, including Egill (Egils háttr). In line with normative grammatica, he notes that ‘we should not compose in that manner’, and he therefore had to extract regular metres from common features or suitable couplets and half stanzas found in early poetry. In Egils saga, by contrast, he composed as Egill, and he could therefore let the primary characteristic of irregularity remain intact, since this ‘is not seen as unbefitting in old poems’. He therefore had no reason to undertake the kind of detailed analysis that was necessary in order to construct new metres, but could instead indulge in his intuitive approximation of how one composes in an irregular manner. This is probably what accounts for the startling difference in analytical rigour between Snorrias-Egill in Háttatal and Snorri-as-Egill in Egils saga. In conclusion, Egils saga is an extremely valuable source for understanding how a pseudonymous poet might work, for seeing how the methodology of the same author

244

4 Prosimetrical Narrative

might change under different circumstances, and for fleshing out the image of the most important author of Old Norse literature.

4.5 Metrical Archaization in Other Sagas Another saga where some groups of stanzas betray archaizing tendencies is Kormáks saga. In 1961, a heated debate on the dating of the poetry in Kormáks saga arose with the publication of Bjarni Einarsson’s Skáldasögur. Um uppruna og eðli ástaskáldasagnanna fornu (Reykjavík: Bókaútgáfa menningarsjóðs, 1961). Bjarni Einarsson argued that the poetry in the saga (as well as that of Hallfreðar saga, Bjarnar saga hítdœlakappa and Gunnlaugs saga) was influenced by chivalric literature and that it was late. His claims were countered by Einar Ólafur Sveinsson, Theodore Andersson and, much later, by Alison Finlay.187 Bjarni Einarsson issued a defence of his views in 1971.188 It now seems clear that Bjarni Einarsson went too far: linguistic and other features in some of the stanzas strongly indicate an early date. His opponents, however, defend all or nearly all of the poetry in the saga, but there are strong indications that some of it may be pseudonymous. I present here a group of stanzas which I believe bears out the possibility of spuriousness particularly well. Here the use of onset is the most conspicuous feature. Stanza 77 of the saga reads: Vísu munk of vinna áðr vér til skips gangim senda sǫrva Rindi til Svínadals mína; koma skulu ǫll til eyrna orð mín Skǫgul borða, betr ann ek siglis Ságu an sjǫlfum mér hǫlfu.189 I shall see to it that my stanza is sent to Svínadalr, to the necklaces’ Rindr [woman], before we go to the ship. All my words shall reach the ears of the tapestry’s Skǫgul [woman]. I love the trinket’s Sága [woman] better by half than myself.

187 Einar Ól. Sveinsson, ‘Kormákr the Poet and his Verses’, Saga-Book, 17 (1966), pp. 18–60; Theodore Andersson, ‘Skalds and Troubadours’; Alison Finlay, ‘Skalds, Troubadours and Sagas’, Saga-Book. Heather O’Donoghue, The Genesis of a Saga Narrative. Verse and Prose in Kormáks saga (Oxford: Clarendon Press, 1991), pp. 173–77, identifies thematic groups of stanzas, but apart from stating that all of the poetry is older than the saga, she refrains from providing absolute or relative dates. 188 ‘The Lovesick Skald: A Reply to Theodore Andersson’, Mediaeval Scandinavia, 4 (1971), pp. 21–41. 189 Vatnsdœla saga, ed. Einar Ól. Sveinsson, p. 292; Skj B 1, p. 83 (56); A 1, p. 89.

4.5 Metrical Archaization in Other Sagas

245

This rich use of onset, two–three times per stanza, is found in stanzas 77–79 and in one stanza shortly thereafter (82).190 The occurrence of onset in these stanzas is 25 times higher than the average of the other stanzas in the saga. Only in stanza 18 is it otherwise used twice. Stanza 77 and 82, furthermore, exaggerate the use of irregular hendingar that may be found in much of the poetry in the saga. They display irregular hendingar in five lines each, whereas the typical stanza in the saga has irregular hendingar in up to two lines, rarely in three. The entire group of stanzas is syntactically straightforward and has few and simple kennings, which, much like in Egils saga, marks it out against the bulk of the poetry in the saga. Stanza 78 and 79 also exhibit young linguistic features.191 In sum, there are many indications that this group of stanzas was composed for the saga. This analysis runs some risk of skewing the data, since stanza 82 is divided from the other three by two stanzas and seeing all four as a group is therefore somewhat questionable. If it were taken out, the contrast between stanzas 77–79 to the rest of the poetry would be less stark, since stanza 82 would then belong to the latter corpus. In light of the three features shared with 77–79 (onset, word order and simple kennings) and the one shared with stanza 77 (hending use), however, I believe that the collected evidence is strong enough to justify the analysis, and a possible scenario may be, for instance, that the two intermediate stanzas were already available and that the saga author composed the group around them. The author’s desire for having as much poetry as possible in the saga is obvious, since no other saga of Icelanders contains as much, and with its meagre storyline, it comes through as more of a poetic composition than a prose narrative. As for his method, the author would have found the relevant features – onset and irregular hendingar – occasionally in the poetry he used when compiling the saga, and he then exaggerated them when trying to compose in the archaic style of Kormákr. His method was similar to that of Pseudo-Egill, although older building blocks and specific model stanzas are in this case not clearly discernable.192

190 Vatnsdœla saga, ed. Einar Ól. Sveinsson, pp. 292, 294–95, 299; Skj B 1, pp. 83–84 (56–58, 61); A 1, pp. 89–90. 191 Vatnsdœla saga, ed. Einar Ól. Sveinsson, pp. 294–95; idem, ‘Kormákr the Poet and his Verses’, p. 35. 192 In my article ‘Egill och Kormákr’ (pp. 135–36), I singled out another group of stanzas, 69–71, which displays two examples of expletive of per stanza – nine times as much as in other poetry in the saga, twelve times as much as in other poetry by Kormákr in the saga (Vatnsdœla saga, ed. Einar Ól. Sveinsson, pp. 284–86; Skj B 1, p. 81 (48–50); A 1, pp. 88–89). I argued that this extreme ratio is an indication of archaization, like in the group discussed above. This may be correct, but since hendingar, onset, and style in stanzas 69–71 are not conspicuous and late composition cannot therefore be corroborated by further evidence, the hypothesis remains somewhat weak. Kari Ellen Gade contends that the overall metrical features of the poetry in Kormáks saga suggest that the bulk of the poetry is authentic (‘The Dating and Attribution of Verses in the Skald Sagas’, pp. 73–74).

246

4 Prosimetrical Narrative

Since archaic features generally entail a relaxation of the strict rules of dróttkvætt, it could be argued that the authors used them simply to facilitate composition, rather than as an attempt to archaize. Irregular hendingar and onset are, however, used concurrently in the pseudonymous stanzas of both Egils saga and Kormáks saga, and this, rather than a more random distribution, betrays an active effort at archaization (Egils saga has onset in stanzas 7.2, 12.4, 13.6, 14.2, 14.8, 48.2, 48.4).193 Among other poetry to which scholars unanimously attribute a late date, the same blend of archaizing features is also found in the poetry in Ragnars saga loðbrókar and in the drápa in StjörnuOdda draumr.194 In these sagas, the choice of archaic features is unlikely to be based on poetry that was already attributed to the poets in question, since we have no traces of such poetry and saga authors seem to have preferred to draw on older poetry when possible. Rather, the choice of features is due to chronology. Ragnarr was supposedly active at the beginning of the skaldic tradition.195 Stjörnu-Oddi’s drápa is a somewhat more complex matter. It was probably composed in the thirteenth or fourteenth century and was placed in the mouth of the twelfth-century character Stjörnu-Oddi. He, however, is said to have composed the drápa in a dream about long-gone days in mainland Scandinavia, and for that reason, one may presume, the poem displays archaic features.196 Egils saga, Kormáks saga, Ragnars saga and Stjörnu-Odda draumr indicate that the combination of irregular hendingar and onset was the main archaizing style, based on the observation, made also by modern scholars, that these features tend to come together in archaic poetry.197 Ragnars saga is of particular interest in several respects.198 For one thing, it is the only fornaldarsaga to be treated here, since it contains skaldic rather than eddic poetry,

193 Males, ‘Egill och Kormákr’, p. 140. Note that onset in stanzas 12 and 48 is not the work of Pseudo-Egill. 194 On Ragnarr’s poetry, see Finnur Jónsson, Den oldnorske og oldislandske litteraturs historie, 2, p. 142. On Stjörnu-Odda draumr, see Harðar saga, ed. Þórhallur Vilmundarson and Bjarni Vilhjálmsson. Íslenzk fornrit, 13 (Reykjavík: Hið íslenzka fornritafélag, 1991), pp. 477–81. 195 See Guðrún Nordal, Tools of Literacy, p. 314. 196 See Harðar saga, ed. Vilmundarson and Vilhjálmsson, pp. ccxxii–ccxxiv, though note that a twelfth-century date is there proposed for the poetry. 197 See, for instance, Kuhn, Das Dróttkvætt, p. 277. 198 Ragnars saga, including the poetry in it, seems to date from the thirteenth century. This dating is based on Ragnarssona þáttr, which is found in a section of Hauksbók datable to 1302–1310 (ONP Registre, p. 359). The þáttr seems to be an abbreviated version of Ragnars saga, containing some of the poetry found in the saga (all dróttkvætt poetry in the þáttr except one stanza is also found in the saga). The saga is also mentioned in the þáttr, and some form of it thus probably existed in the thirteenth century (‘Hauksbók’ udgiven efter de arnamagnæanske håndskrifter no. 371, 544 og 675, 4º, ed. Finnur Jónsson (København: Det kongelige nordiske oldskrift-selskab, 1892–96), p. 458). The most extensive study of Ragnars saga is Rory McTurk, Studies in ‘Ragnars saga loðbrókar’ and its Major Scandinavian Analogues (Oxford: Society for the Study of Medieval Languages and Literatures, 1991), but McTurk does not discuss the dating of the poetry in the

4.5 Metrical Archaization in Other Sagas

247

and this chapter’s parameters are relevant to it. Furthermore, even if Pseudo-Ragnarr had for the most part made observations similar to those of other archaizing poets, he patently failed on one account. For the reconstruction of the method of the archaizers, this is a significant point and merits some elaboration. Pseudo-Ragnarr has made at least three sound observations of early poetry. Thus, the 37 full and two half stanzas in dróttkvætt in Ragnars saga display highly irregular hendingar, extremely frequent onset (49 instances) and frequent use of expletive of (a metrical filler used in some early poetry; seven times in the saga).199 Although Pseudo-Ragnarr was prone to exaggeration, above all of onset, the features in question are actually present in early poetry. He failed, however, to make one crucial observation with regard to irregular hendingar. In classical dróttkvætt, even lines are more marked by hendingar than odd lines, since even lines have aðalhending whereas odd lines have skothending. The same principle applies to the lax, early type of dróttkvætt, and can be summarized in the rule that even lines shall be as marked by hendingar as odd lines or more so: a hierarchical principle is here at work. Thus, for instance, the opening stanza of Ragnarsdrápa: Vilið Hrafnketill heyra, hvé hreingróit steini Þrúðar skalk ok þengil þjófs ilja blað leyfa.200 Will you, Hrafnketill, hear how I praise the king and Þrúðr’s thief’s [Hrungnir’s] foot-soles’ leaf [the shield], grown with bright colour.

This stanza has no hending–aðalhending–no hending–skothending. Other patterns are possible, but there are almost no exceptions to the general hierarchical principle.201 In Pseudo-Ragnarr’s poetry, by contrast, we find examples like: Vammlausa skaltu vísi ef viltu griðum þyrma202 You shall, lord, [let me travel hence] without dishonour, if you wish to keep the truce.

saga. On the thirteenth-century date of the poetry, see Finnur Jónsson, Den oldnorske og oldislandske litteraturs historie, 2, p. 142; SkP 8, p. 623. 199 Onset: SkP 8, pp. 629 (2), 630 (4), 631 (2), 632 (2), 633 (2, 6, 8), 635 (2, 4, 8), 639 (4, 8), 642 (4), 644 (2), 648 (2), 649 (2, 4, 6), 651 (2, 6), 655 (6), 661 (6), 663 (2, 6), 665 (2), 666 (2), 668 (2, 4, 8), 670 (2, 8), 675 (2, 4, 6, 8), 676 (2, 8), 678 (2, 8), 684 (2), 687 (6, 8), 689 (4), 692 (2, 4), 695 (2), 697 (2, 4, 6). Expletive of: SkP 8, pp. 632 (3), 633 (3, 7), 635 (6), 639 (1), 680 (4), 685 (1). 200 SkP 3, p. 28. 201 See Myrvoll, Kronologi i skaldekvæde, pp. 109–32. 202 SkP 8, p. 631.

248

4 Prosimetrical Narrative

Here, the first line has hending but the second does not. This is not a slip of the pen on the part of the author, since such breaks to the hierarchical principle occur thirteen times in the poetry in Ragnars saga.203 Pseudo-Ragnarr thus displays the sensitivity, or lack thereof, of the new poets to different kinds of archaic features. Their grammatical, normative mindset predisposed them to look for fixed patterns, and this approach was well suited to extract rules from earlier licence, but it was less apt to analyse the principles that posed limits to this licence and created a kind of order in the preliterate chaos. Onset, irregular hendingar and expletive of were conspicuous features of archaic poetry and could be reproduced at will, but their distribution was less obvious. Metres, within a grammatical framework, could not abide by such vague rules as the hierchical principle of hending distribution: one finds nothing comparable, for instance, in Servius’s De centum metris or in Isidore’s description of metres. In Snorri’s analysis of the metres of the early poets, he mentions their recurrent metrical flaws, obviously referring above all to hending distribution, and it is unlikely that a person with grammatical training could have perceived of archaic poetry otherwise. Accordingly, pseudo-Ragnarr has produced a body of poetry full of metrical flaws, and from that perspective his endeavour was a success. His source of inspiration may have been a more specific one than the corpus of archaic poetry generally, namely Krákumál. This is a poem placed in the mouth of the dying Ragnarr and preserved as an appendix in one version of the saga and within the saga in another.204 It is similar to pseudo-Ragnarr’s poetry in so far as it is composed in dróttkvætt with very few hendingar. Krákumál has a lower ratio of onsets, however, than the poetry of Pseudo-Ragnarr, 27 instances in all.205 It does not contain expletive of, and it breaks the hierarchical principle only two or possibly three times.206 This poet thus has a different way of producing archaic poetry than that of Pseudo-Ragnarr, and the continuous form suggests that he was not composing for a saga. Rather, this poet has composed a historical poem comparable to, for instance, Jómsvíkingadrápa, althought that poem exhibits a regularized rather than irregular form of metrical archaization (see above pp. 34–35). Historical poems seem to have

203 SkP 8, pp. 629 (5–6), 630 (3–4), 631 (1–2), 632 (3–4), 633 (3–4), 635 (5–6), 639 (3–4), 644 (3–4), 648 (1–2), 649 (5–6), 660 (3–4), 665 (1–2), 670 (1–2). I have included fronted hendingar – when the second hending falls before the fifth position – since I believe that these were intended as another archaic feature. This occurs in two instances: SkP 8, pp. 639 (3–4), 665 (1–2). I have also included somewhat unorthodox rhymes, such as one on oð: óð (with different quantity of vowels; this example is found in SkP 8, p. 648), since they seem to be intentional (this one is placed in the penultimate syllable, as one would normally expect). 204 SkP 8, pp. 707–08. 205 SkP 8, pp. 717 (4), 722 (4), 724 (4), 726 (6), 728 (8), 732 (8), 734 (4), 736 (4), 738 (4, 8), 743 (4, 6), 745 (4), 749 (6), 753 (6, 10), 755 (4, 8), 757 (4, 10), 760 (6), 762 (6), 765 (6, 8), 767 (6), 770 (6), 774 (6). 206 SkP 8, pp. 738 (9–10), 749 (7–8; this instance should probably not count, but rather, the hending should be understood as a(r)st: ast), 774 (5–6).

4.5 Metrical Archaization in Other Sagas

249

been a precursor to saga literature, and the date c. 1150–c. 1200 suggested by editors appears likely.207 Here was a model for Pseudo-Ragnarr to follow, but he added the feature of expletive of, presumably based on his own observations. Furthermore, the marginal breaks to the hierarchical principle in Krákumál became rampant in Pseudo-Ragnarr’s poetry. The other poets studied here – the Krákumál poet, Pseudo-Egill, Pseudo-Kormákr and Pseudo-Stjörnu-Oddi – on the whole respected the hierarchical principle, but this need not imply that they were aware of it in a conscious way. It is unlikely that they could have formulated it as a rule, since their schooling would not have provided them with the means to do so, as is evident from Snorri’s remark about the ‘flaws’ of early poets (see pp. 35–37, 119–20). Rather, they saw what their intellectual background had predisposed them to see, namely rules and faults. Indeed, throughout the Old Norse tradition, the hierarchical principle may never have been formulated, since before the arrival of Latin learning, skaldic poetry was probably largely a matter of learning by doing. It is important to note that the archaizing features used by saga authors are few – namely, irregular hendingar and metrical fillers (onset and expletive of) – and that they are of a metrical rather than a linguistic character. (Modern scholars analyse expletive of from the perspective of historical linguistics, but thirteenth-century Icelanders probably understood it as a metrical filler.) There are no indications that pseudonymous poets aimed for phonological or morphological archaization, as as often been claimed but never demonstrated (see p. 215). I have found no instances of, for instance, hiatus forms that never return or aðalhendingar in a: ǫ in stanzas that may for other reasons be deemed spurious, even though it is clear that the poets in question actually aimed to archaize. A case in point is that the linguistically acute archaization in choosing the form Bárøðr for later Bárðr was made by modern editors, not by Pseudo-Egill (see p. 225). Furthermore, Haukur Þorgeirsson has been able to rebut the claim that alliteration in vr- may be due to archaization, since scribes evidently did not even understand the phenomenon (see pp. 166–67). Similarly, Snorri had no understanding of the historical underpinnings of hiatus and erroneously treated ‘Vindhlés’ as a hiatus form in Háttatal 7. It would therefore seem that pseudonymous poets did not engage in linguistic archaization, but that they might archaize by use of a few and simple metrical features. As we have seen, pseudonymous composition in the early, lax style seems mainly to have taken place when some poetry traditionally attributed to the poet in question displayed such features or when the poet was perceived as composing at the very beginning of the skaldic tradition, as king Ragnarr loðbrók.208 Orms þáttr Stórólfssonar has a curious hybrid form. Its narrative spans the registers of sagas of Icelanders (or kings’ sagas) and fornaldarsögur, beginning and ending in Iceland but proceeding by

207 SkP 8, pp. 710–11. 208 Males, ‘Egill och Kormákr’, pp. 133–34, 137.

250

4 Prosimetrical Narrative

way of two monster fights in Norway.209 During the first of these, the dying bloodbrother of the hero, Ásbjörn, composes a set of stanzas while a giant is winding his entrails around a pole. In so doing he progresses from the simple eddic metre fornyrðislag (stanza 3) to nearly hending-less dróttkvætt (4) to sporadic hendingar (5 and 6) to classical dróttkvætt (7 and 8) and finally back to fornyrðislag (9 through 12).210 The action is set in the late 990s, and the date cannot account for the archaic features. Rather, it is probably the case that the metre, just like the narrative, exhibits the typical features of both fornaldarsögur (fornyrðislag) and sagas of Icelanders (dróttkvætt). In this regard, Orms þáttr is reminiscent of Jómsvíkinga saga and Jómsvíkingadrápa, whose narrative and metre would normally pertain to heroes of a pre-settlement date, even though the date of action is the second half of the tenth century (see above pp. 34–35). Orms þáttr and Jómvíkingadrápa are interesting for showing us how sensitive saga authors and poets were not only to chronology, but also to what kind of metre was appropriate to a legendary versus a realistic type of plot. In other cases, composition was probably most often performed in classical dróttkvætt, and we shall see several examples of this in the next two sections. Due to the lack of distinctive features of classical dróttkvætt, however, it is difficult to assess how extensive such composition was, at least with the parameters employed here. Grettis saga is a rare case, where most of the poetry was composed so late that it can be dated to the fourteenth century on linguistic grounds, and the same could be said of Harðar saga, although the latter will not be studied here.211 A combination of stylistic and linguistic features indicate a fourteenth-century origin for the stanzas in Víglundar saga as well.212 Such clear-cut cases are rare, but at least, they show that saga authors would indeed compose poetry in the name of the early poets.

209 For a concise description of this þáttr within Flateyjarbók, see Elizabeth Ashman Rowe, The Development of Flateyjarbók. Iceland and the Norwegian Dynastic Crisis of 1389 (Odense: The University Press of Southern Denmark, 2005), pp. 65–68. 210 Harðar saga, ed. Vilmundarson and Vilhjálmsson, pp. 410–414. 211 See the overview in Grettis saga, ed. Guðni Jónsson. Íslenzk fornrit, 7 (Reykjavík: Hið íslenzka fornritafélag, 1936), pp. xxxi–xlii; Janus Jónsson, ‘Um vísurnar í Harðar sögu Grímkelssonar’, Tímarit hins íslenzka bókmenntafjelags, 13 (1892), pp. 259–75; Harðar saga, ed. Vilmundarson and Vilhjálmsson, p. xix. 212 Kjalnesinga saga, ed. Jóhannes Halldórsson. Íslenzk fornrit, 14 (Reykjavík: Hið íslenzka fornritafélag, 1959), p. xxiv. The two linguistic features adduced there are the pronunciation kallmann for karlmann to achieve aðalhending with snjallri and the pronunciation það for þat to achieve skothending with eiði. While the first of these is surely intended, the stanzas in the saga quite often lack skothending, which may cast some doubt on the second. The editor takes the repeated lack of skothendingar to indicate a young date, but this is no clear dating criterion on its own. There is, however, one strong criterion that the editor does not note. In stanza 9.3–4 of the saga there is a break of the hierarchical principle (þó renni ek til hennar / hugreik, vinaraugum ‘yet I let my mind run to her, [I look thither] with friendly eyes’; the interpretation of -reik is problematic, but does not affect the hendingar; see Kjalnesinga saga, ed. Jóhannes Halldórsson, p. 104–05). The poetry in the saga thus exhibits at least two, and probably three, strong indications of a young date. There are no forms which indicate great age and there is no

4.5 Metrical Archaization in Other Sagas

251

Most scholars agree that many stanzas in Njáls saga, Svarfdœla saga and Þorleifs þáttr jarlsskálds, and most or all in Gunnlaugs saga, Bárðar saga, Þórðar saga hreðu, Bergbúa þáttr and Draumr Þorsteins Síðu-Hallssonar are composed by the author of the saga or some later redactor (on Bandamanna saga, see below). In most of these cases, however, the arguments have never been made explicit or, when they are, they are open to a number of different interpretations: the direction and date of intertextual influence is often uncertain and some linguistic phenomena may be older than we assume.213 In other cases, indications point both ways, and their significance for the overall poetic corpus contained in the saga remains unclear.214 There is an urgent need for a clearer methodology for the dating of poetry in the sagas. Gísla saga is a case in point. For decades, scholars have assumed that the bulk of the poetry in the saga is spurious, but in a forthcoming publication, Klaus Johan Myrvoll reviews the empirical evidence and suggests that factors that cannot be measured have been given precedence over those which can and that if the tables were turned, most of the poetry in the saga would appear to be authentic.215 This problem cannot be addressed here, where focus lies not on our critical methods but on those of the pseudonymous poets, but I take it into account in so far as I attempt to avoid

independent attestation. Furthermore, since one of the irregular hendingar can be attributed to a late date, it is likely that most if not all of the missing skothendingar are the product of the same effort at archaization and that the editor, in the end, was right in seeing them as an indication of lateness (but they are too infrequent to be included in the survey above). It cannot, however, be ruled out on these criteria that there may be some remnants of early poetry in the saga. 213 Implicit arguments, as well as dating based on taste, were common in early scholarship (see, for instance, Den oldnorske og oldislandske litteraturs historie, 2, p. 166). One saga that has continually been subject to suspicions, but where the many possible scenarios make firm conclusions difficult to draw, is Gunnlaugs saga. Björn M. Ólsen argued that nearly all of the poetry in the saga was young, but on somewhat flimsy grounds (Om Gunnlaugs saga ormstungu. En kritisk undersøgelse (København: Andr. Fred. Høst & sønn, 1911), pp. 46–49). Finnur Jónsson strongly rebutted this, but his arguments are no more solid than those of Björn M. Ólsen (Finnur Jónsson, ‘Sagaernes lausavísur’, pp. 19–27). Russell Poole has identified a thematic group of stanzas in the saga, but with regard to dating, the brunt of the argument is carried by the form yngri for œri (younger) alone, found in just one stanza (Russell Poole, ‘Compositional Technique in some Verses from Gunnlaugs saga’, Journal of English and Germanic Philology, 80 (1981), p. 480). As far as we know, however, this form may have been in use already in the lifetime of the historical Gunnlaugr (Finnur Jónsson, ‘Sagaernes lausavísur’, p. 21). Kari Ellen Gade has noted the many verbal similarities between the poetry in Gunnlaugs saga and other early poetry (‘The Dating and Attribution of Verses in the Skald Sagas’, p. 73). The parallels are so many that it is tempting to see the poetry in Gunnlaugs saga as a product of scholarly eclecticism. Gade’s argument is probably the strongest that has been put forward to date, but I hesitate to consider it conclusive. 214 In Svarfdœla saga young and old features both occur, suggesting perhaps a mixture of old and new stanzas, but the poetry has been so corrupted in the manuscript transmission that it is difficult to evaluate the significance of the young forms for the original composition of the stanzas (Finnur Jónsson, ‘Sagaernes lausavísur’, pp. 8–9; Eyfirðinga sǫgur, ed. Jónas Kristjánsson. Íslenzk fornrit, 9 (Reykjavík: Hið íslenzka fornritafélag, 1956), pp. xc–xcii, 167–68). 215 Myrvoll, ‘The Authenticity of Gísli’s Verse’.

252

4 Prosimetrical Narrative

unclear dating criteria. Since unclear criteria have often prevailed in scholarship on the dating of the poetry in the sagas mentioned above, but also because that poetry is on the whole not susceptible to the parameters employed here, those sagas have been left out of the study.216

4.6 Archaizing Diction The foregoing presentation of pseudonymous poetry suggests that the role of poet and saga author was sometimes combined. In some cases, these author-poets displayed an awareness of the historical development of metrics. In a similar vein, poets composing under the name of saga heroes partook of the revival of archaic diction that had been going strong since the twelfth century. This section and the next are devoted to identifying and analyzing such instances. Unlike metrics, diction does is not describable in terms of a few recurrent features, and determining which stanzas are likely to be pseudonymous requires other methods. Poets composing with archaic diction in the thirteenth and fourteenth centuries would occasionally use anachronistic language, and some cases of this will be discussed based on the chronological patterns described in Chapter Two, but most of the time there is no way of telling if the poetry is pseudonymous on purely formal grounds. In some instances, however, a particular circumstance comes to the scholar’s aid. After the writing of Skáldskaparmál, some authors appear to have turned to that work as a guide to archaic diction. As we have seen in Chapter Three, Snorri was not a passive recorder of tradition, and he seems sometimes to have altered the interpretation of a stanza or added narrative without basis in poetry. In the poetry of some sagas, there are indications that a stanza may be derived not from tradition in a general sense, but rather from Snorri’s rendition of it. In the following, dependence on Snorri will be the main dating criterion, but

216 See nn. 213–14 above and Finnur Jónsson, ‘Sagaernes lausavísur’, pp. 4–10; idem, Den oldnorske og oldislandske litteraturs historie, 2, p. 166; idem, ‘Om Njála’, pp. 93–97; Guðrún Nordal, ‘The Dialogue between Audience and Text’. Russell Poole, ‘The Origins of the Máhlíðingavísur’, Scandinavian Studies, 57 (1985), pp. 244–85, argues convincingly that the 17 stanzas in Eyrbyggja saga attributed to Þórarinn originated as a single long poem. His conclusion is that they originated in the late eleventh or twelfth century, as a poem to accompany the saga. He discusses many archaisms in the stanzas, and even though Poole observes that none of these, when taken individually, can serve as a dating criterion, they are so numerous and varied that they collectively lend strong support to an early date. This may be cause for concern with regard to Poole’s hypothesis, and as noted on p. 218, pseudonymous composition appears to pertain to the period of saga writing. Taken together, these factors probably indicate that the stanzas are authentic (but see Poole, ‘The Origins of the Máhlíðingavísur’, pp. 256–63). Ursula Dronke posits a similar background to the poetry in Víga-Glúms saga, based on a supposed allusion to Rígsþula (‘Sem jarlar forðum. The Influence of Rígsþula on two saga-episodes’, in Speculum Norroenum: Norse Studies in Memory of Gabriel Turville-Petre, ed. Ursula Dronke et al. (Odense: Odense University Press, 1981), pp. 57–64).

4.6 Archaizing Diction

253

anachronistic diction identified through the analysis in Chapter Two will also enter the discussion. This section is then followed by a case study of poetic diction in Grettis saga. It bears repeating that pseudonymous stanzas in all likelihood make up only a relatively small portion of the poetry in skaldic prosimetra. They are an asset for showing us how saga authors received earlier traditions and contributed to their literary manifestation as well as how a work like Snorri’s Edda could interfere in that process. They bring us close to the methods of the authors and to their views on how the early literary history ought to have played out. They are more useful than authentic poetry for understanding all of these aspects, and their evidence is in some regards easier to use than that of the prose, since authorial intervention can here be clearly discerned. Pseudonymous stanzas, however, can neither account for the origin of poetic quotation in saga literature at large, nor indeed shed very much light on preceding developments, since they appear to pertain almost exclusively to the age of saga writing, as we saw on pp. 218, 231 n. 135, 232 n. 136 and 252 n. 216. They have synchronic value for explaining how Icelanders set about writing their history, but not for why they would have done so in the first place. With these restrictions in mind, let us now consider the impact of Snorri’s Edda on pseudonymous composition. Roberta Frank’s 1981 article ‘Snorri and the Mead of Poetry’ was discussed in Chapter Three. As noted there, some of her arguments are problematic, but I wish here to draw attention to what she argues to be an instance of later use of a part of Snorri’s narrative which I believe to be particularly convincing and which may be further elaborated. Frank has identified a unique kenning type with reference to the Skáldskaparmál version of the myth of the mead of poetry. It is found in one stanza only, in Bandamanna saga. The kenning in question is Áms ok Austra sáttir (Ámr’s [a giant’s] and Austri’s [a dwarf’s] agreement [mead of poetry]) in a situational quotation put in the mouth of Ófeigr.217 In Old Icelandic literature, the word sátt, sætt (agreement) is used only on one other occasion in connection to the mead of poetry, namely in the prose of Skáldskaparmál describing the incident in question.218 There this word is not used in the description of ways to refer to poetry, but in the narrative immediately preceding it. This indicates that Snorri did not see ‘agreement’ as one of the traditional ways of referring to poetry, but when his text was used as a guide to composition, his descriptions could give rise to new, quasi-traditional modes of expression. This appears to be what has happened in the case of Áms ok Austra sáttir. Of the six stanzas in the saga, at least five (Ófeigr’s stanzas) have been seen as spurious, including the one in question.219 Apart from the suspect kenning, several

217 Frank, ‘Snorri and the Mead of Poetry’, p. 166; Grettis saga, ed. Guðni Jónsson, p. 356; Skj B 1, p. 376; A 1, p. 406. 218 Snorri, Skáldskaparmál, ed. Faulkes, pp. 3–4. 219 Hallvard Magerøy, Studiar i Bandamanna saga: Kring gjerd-problemet, Bibliotheca Arnamagnæana, 18 (København: Munksgaard, 1957),pp. 108–9; Bandamanna saga, ed. Hallvard Magerøy (Oslo: Dreyers forlag, 1981), pp. lii–liii.

254

4 Prosimetrical Narrative

more arguments support of this view. First, these five stanzas are found in one manuscript only (Möðruvallabók), indicating that they may be a later addition to the saga.220 Second, they possess several echoes of poetry from Gunnlaugs saga.221 Third, Ófeigr is not known to have been a poet nor is any of his verse quoted outside of the saga.222 Finally, the dramatic date of Bandamanna saga is shortly after the middle of the eleventh century, when specific mythological references were generally not in use, and another of Ófeigr’s stanzas has the kenning Iðja hlátr (Iði’s [a giant’s] laughter [gold]).223 While none of these arguments is conclusive in its own right, when they are combined and viewed in tandem with a unique kenning which has its nearest analogy in Snorri’s prose, the case for pseudonymous composition of the five stanzas is strong.224 We may probably conclude that Bandamanna saga’s settlement has little to do with earlier poetry or mythological lore but much to do with Snorri’s ingenious rendering of tradition. Another kenning relating to the mead of poetry might rouse suspicion, since it refers to the eagle vomiting forth the mead of poetry, a story known only from Skáldskaparmál and one which Snorri does not support with poetic quotation. In stanza 56 of Egils saga we find the kenning arnar kjapta ǫrð (‘the eagle’s jaws’ [the beak’s] seeds [vomit = poetry]’).225 According to the saga, this is the first stanza in Berudrápa (drápa about a shield), a poem composed by Egill in response to the gift of a shield sent to him by Þorsteinn Eiríksson.226 This is somewhat conspicuous, since we have just been presented with another stanza from another drápa in thanks for another shield, this one given to Egill by Einarr skálaglamm. That stanza, number 54 of the saga (Mál es lofs at lýsa), has been discussed above, and several indications point to its authenticity. One might therefore be tempted to conclude that stanza 56 was composed for the saga under the influence of the preceding chapter. It is, however, only the prose that connects stanza 56 to a shield: nothing of the sort is found in the

220 Grettis saga, ed. Guðni Jónsson, p. xcvii. 221 Bandamanna saga, ed. Magerøy, p. liii. 222 Bandamanna saga, ed. Magerøy, p. liii. 223 Grettis saga, ed. Guðni Jónsson, pp. 333; Skj B 1, pp. 375; A 1, p. 406. 224 With regard to the kenning in question, the opposite relation is possible in principle, that is, that Snorri extracted the story out of this pre-existent stanza. Two factors, however, make this unlikely. First, although unique kenning types referring to otherwise unknown narrative elements do exist (Bragi’s Þrúðar þjófr (Þrúðr’s thief [Hrungnir]) is a famous example), they should not be posited out of hand, since the kenning system is dependent on coherence to be intelligible. Second and more importantly, Snorri’s method is generally to extract names from poetry, and if he drew on this kenning, the names in it would probably have entered Snorri’s prose. Guðni Jónsson singles out this kenning as spurious in his edition (Grettis saga, ed. Guðni Jónsson, p. xcvii). 225 Egils saga, ed. Sigurður Nordal, pp. 275–76; Skj B 1, p. 42; A 1, p. 48. 226 Drápur in thanks for shields is a recurrent theme in skaldic poetry. Most famous are Bragi’s Ragnarsdrápa and Þjóðolfr’s Haustlǫng, but the stanzas by Einarr Skúlason gathered under the name Øxarflokkr, though probably about an axe, deserve mention as well. On the shield poems and other ekphrastic poetry, see Margaret Clunies Ross, ‘Stylistic and Generic Identifiers of the Old Norse Skaldic Ekphrasis’; Poole, ‘Scholars and Skalds’.

4.6 Archaizing Diction

255

stanza. It would thus appear that a function has been given to the stanza which it is not really apt to fulfill. Furthermore, the stanza is regular and bears none of the hallmarks of Pseudo-Egill discussed in the previous section. Finally, it would seem that Egill drew on the same image also in another instance: on p. 90, I discussed the kenning Hǫ́ars þegna regn (Hǫ́ars [Óðinn’s] men’s [the gods’/the poets’] rain [poetry]).227 The word regn (rain) is deviant as a base-word in a kenning for ‘poetry’, but makes sense as pertaining to the image of Óðinn ‘raining’ (‘getr at rigna’) down the mead of poetry. This is stanza 10 of the saga, exhibiting a structure with interlinear hendingar which appears to have been typical of the historical Egill (see above pp. 229–31). Later poets did not understand the kenning Hǫ́ars þegna regn as a reference to a mythical motif, however (see p. 90). It is therefore unlikely to have inspired a pseudonymous poet to use the kenning arnar kjapta ǫrð. Rather, the repeated yet varied references to the motif of Óðinn vomiting the mead of poetry suggest that it was an important one for Egill. I would therefore conclude that stanza 56 was indeed composed by Egill and later inserted into a narrative frame. At that point, the topic of the preceding chapter may have affected its presentation, perhaps because the author was not acquainted with any narrative context for the stanza. So far, this leaves us with Áms ok Austra sáttir as the sole example of a kenning which draws on the prose narrative of Skáldskaparmál.228 In this section and the following on pseudo-Grettir, I shall argue for further likely instances. In Sneglu-Halla þáttr in the kings’ saga compilation Morkinskinna the haughty royal poet Þjóðolfr Arnórsson is taunted by King Haraldr harðráði into composing a stanza about a street brawl between a tanner and a smith. When Þjóðolfr protests at the lowly theme, Haraldr tells him that the assignment is more difficult than he might think: he is to have the tanner be the giant Geirrøðr and the smith be the god Þórr.229 At the king’s insistence, Þjóðolfr obeys and composes a stanza, focusing on the scene where Þórr and Geirrøðr are throwing a lump of red-hot iron back and forth at each other. Þjóðolfr acquits himself well and the stanza is rightly famous, as is the next one, where the fight stands between Sigurðr and Fáfnir. Only the first stanza concerns us here, and I render it according to its oldest reading, from Morkinskinna (c. 1275)230:

227 Egils saga, ed. Sigurður Nordal, p. 110; Skj B 1, p. 43; A 1, p. 49. 228 Snorri, Skáldskaparmál, ed. Faulkes, pp. 3–5. 229 The encounter between the two is principally known from Skáldskaparmál, though there is also an eerie rendition of it in Saxo and an apocalyptic one in Þorsteins saga bœjarmagns (Snorri, Skáldskaparmál, ed. Faulkes, pp. 24–30; Saxo Grammaticus, Gesta Danorum, ed. Friis-Jensen and trans. Zeeberg, pp. 568–69; Fornaldar sögur norðurlanda, ed. Guðni Jónsson, 4 vols (Reykjavik: Íslendingasagnaútgáfan, 1950), 4, pp. 319–44). 230 Due to the complexity of this and the following stanza, I have as far as possible retained the reading of manuscripts that are generally held to be good, rather than producing eclectic stanzas. It should be noted, however, that variant readings and textual difficulties have little bearing on the present argument.

256

4 Prosimetrical Narrative

Varp ór þrætu þorpi Þórr smiðbelgja stórra hvatt eldingum hǫldnum hafra kjǫts at jǫtni; hljóðgreipum tók húða hrǫkkviskafls ór afli glaðr við galdra smiðju Geirrøðr síu231 þeiri.232 The great bellows’ Þórr [the smith] quickly threw the held thunderbolts [insults; cf. that Þórr threw a red hot rod/lump of iron at Geirrøðr] at the goat-flesh’s giant [the tanner] from the quarrel’s hamlet [mouth]. With the sound-grips [ears], the hide-scraper’s (?) Geirrøðr [the tanner] happily received that incantations’ smithy’s [mouth’s] glowing iron rod/lump [insult] from the furnace [mouth].

While this close translation may not be apt to convince the reader of the poetic qualities of the stanza, it brings out evocative features of its background.233 A comparison to Eilífr Goðrúnarson’s Þórsdrápa 17, with which Þjóðolfr’s stanza shares a number of similarities, makes this apparent. That stanza reads in R: Svá at hraðskyndir handa hrapmunnum svalg gunnar syptisylg á lopti síu langvinr þrǫngvar234 þá er ?aurþursis eisu? ós235 Hrímnis fló drósar

231 Written ‘ſiǫ’ where one would expect *‘ſio’ (normalized síu). This is probably due to the preceding in ‘geirrǫðr’ (here normalized to Geirrøðr). The spelling has in all likelihood not caused problems for understanding, as the gender of the word is marked by ‘þeiri’ (sjó (sea) would thus be ungrammatical). 232 GKS 1009, fol., 15v ll. 20–21; Morkinskinna, ed. Finnur Jónsson. Samfund til Udgivelse af Gammel Nordisk Litteratur, 53 (København: Samfund til Udgivelse af Gammel Nordisk Litteratur, 1932), p. 235; SkP 2, p. 169. Dating according to ONP Registre, p. 471. 233 It may be noticed that the stanza as it reads in Morkinskinna flies in the face of editorial stylistics; both the thunderbolts and the furnace (or possibly the glowing iron rod/lump) lack definers and function as pure metaphors understandable from the context, rather than as parts of kennings. Cf. the commentary in SkP 2, pp. 170–71. The new edition of Morkinskinna retains the manuscript reading (Morkinskinna, ed. Ármann Jakobsson and Þórður Ingi Guðjónsson, 1, p. 271). 234 Other manuscripts read (normalized): ‘lyptisylg á lopti | langvinr síu þrǫngvar’ ‘the raised draught . . . ’ (SkP 3, p. 115; Skj A 1, p. 151). This is the reading preferred by editors, presumably because the first syllable in síu is short, which makes it apt to fill the third and fourth positions of an A2k line, but not the first and second (see, for instance, Kuhn, Das Dróttkvætt, pp. 53–54). It thus seems likely that the R reading is secondary, originating in scribal confusion of ſ and l (lyptisylg > syptisylg). This gave rise to a lack of alliteration in the first syllable of the next line, which prompted the scribe to move the word síu to the beginning of the line. 235 Reading following T and supported by aðalhending. R has ‘aſ’.

4.6 Archaizing Diction

257

til þrámóðnis Þrúðar þjóst af greipar brjósti.236 So that the hasty battle-inciter [Þórr], the tight spot’s [Old Norse þrǫng > þjalfi (enclosure) > Þjalfi’s]237 old friend [Þórr], swallowed the spark’s snatched draught [glowing iron rod/lump] with his quick hands’ mouths [palms], when the giant’s woman’s [giantess’s] giant’s (?) [giant’s] fire-beam [glowing iron rod/lump]238 flew fiercely at Þrúðr’s desirer [Þórr] from the grip-breast [palm].

As in Þórsdrápa generally, there are some textual difficulties here, but they are not important for the present analysis and I will not focus on them here. The two stanzas both use two words that are otherwise rare, namely greip (grip [of the hand]) and sía (glowing iron rod/lump; spark). Most striking, however, is the similarity in corporeal imagery, that is, the transposition of functions from one body part to another: hljóðgreipar (sound-grips [ears]) in Þjóðolfr’s stanza, handa munnar (hands’ mouths [palms]) and greipar brjóst (grip-breast [palm]) in Þórsdrápa. Such transpositions are otherwise rare in the extreme.239 These factors indicate influence from Þórsdrápa on Þjóðolfr’s stanza, and this observation is not new.240 The word sía, however, has more to add to the story. In Þórsdrápa, sía has its ordinary meaning ‘spark’ and qualifies sylgr (draught), showing that the draught is of a kind that has to do with sparks,241 that

236 GKS 2367, 4to, 25r, 15–17; SkP 3, p. 115. 237 I adopt here the interpretation of SkP 3, pp. 115–16, which respects the reading of the manuscripts and presupposes ofljóst, a feature also found at another point in the poem (see above p. 50). 238 The word áss (beam [in a house]) would point to an iron rod instead of a lump. 239 Arinbjarnarkviða 6, 8 (Skj B 1, p. 38; A 1, p. 44) and Íslendingadrápa 1, 6 (Skj B 1, p. 539; A 1, p. 556) both have hlusta munnum ([drink the mead of poetry] with hearing’s mouths [ears]). The last stanza in Grettis saga has axlarfótr (shoulder-leg [arm]), a kenning also found in the prose of Litla Skálda and Skáldskaparmál (see above pp. 140–41 and below p. 269 n. 281). Litla Skálda has ‘Eyru manns er kallat hlustar auga eða sjónir’ (a person’s ears are called the eye or sight of hearing), and Skáldskaparmál has a similar expression (Edda Snorra Sturlusonar, ed. Finnur Jónsson, pp. 190, 256; see above p. 141). I am not aware of other occurrences. These transpositions are reminiscent of topical expressions in the Christian tradition, such as the ‘gum’ or ‘ear of the heart’ (palatum cordis, auris cordis) in Gregory the Great and particularly of later and more uninhibited imagery, such as ‘the hands of the tongue’ (manus linguae) in Geoffrey of Vinsauf’s Poetria nova (see Jean Leclercq, The Love of Learning and the Desire for God: A Study of Monastic Culture, trans. Catharine Mishrahi (New York: Fordham University Press, 1961 [1957]), p. 30; Les arts poétiques du xiiᵉ et du xiiiᵉ siècle: Recherches et documents sur la technique littéraire du moyen age, ed. Edmond Faral. Bibliothèque de l’Ecole des hautes études, Sciences historiques et philologiques, fasc. 238 (Paris, 1924), p. 260 l. 2060). Monastic influence on Þórsdrápa seems highly unlikely, however, and these expressions have thus probably developed independently in skaldic poetry. 240 See Roberta Frank, Old Norse Court Poetry, p. 114, whereas Diana Whaley refrains from drawing conclusions from verbal similarities (SkP 2, p. 170). 241 LP, s.v. sía, translates it in Þórsdrápa as ‘iron rod’ and Anthony Faulkes as ‘molten or glowing lump (of metal)’ (Snorri, Skáldskaparmál, ed. Faulkes, p. 385), which is how it may have been interpreted after the composition of Skáldskaparmál. This, however, is almost certainly wrong for earlier times, as will become evident below. LP (1860) does not give the occurrence in Þórsdrápa, and the translation strictura (glowing iron mass) is valid for Þjóðolfr’s stanza, though probably not for the

258

4 Prosimetrical Narrative

is, the glowing iron rod/lump. The rod/lump is in turn designated by ‘draught’ because of the daring corporeal imagery in the stanza, where the hand-mouths swallow it. In Þjóðolfr’s stanza the situation is different. The referential web of this stanza is staggeringly complex, but on a basic, syntactic level, we read that Geirrøðr tók við síu þeiri (received that sía).242 A spark can hardly be the object of [throwing and] catching, and the intended meaning must therefore rather be ‘glowing iron rod/lump’.243 The inspiration for this use of sía is not to be found in Þórsdrápa. To judge by the three occurrences of sía in poetry and the nine in prose that I have found,244 the meaning ‘glowing iron rod/lump’, as opposed to ‘spark’, is only found in one other place: namely, in the prose narrative which serves as an introduction to Þórsdrápa in Skáldskaparmál.245 There it occurs first as járnsía and then twice without prefix.246 It would seem that Snorri has altered the meaning of the word to help the reader interpret a very inaccessible poem, since this is not the meaning normally ascribed to this rare word even by him: in Gylfaginning he uses it three times to mean ‘spark’.247 More precisely, he seems to have conflated the imagery of the preceding stanza, where the iron lump is referred to as a piece of meat, with that of the drink of the stanza quoted above.248 In the preceding stanza, the piece or lump (segi) is the base-word in the kenning, which is how Snorri treats the definer sía, thus retaining relevant semantic aspects of both words (‘lump’ and ‘fire’), the form of one of them (sía), and the base-word function of the other (segi). In so doing, he used as few words as possible to clarify the content of both stanzas, and by altering the semantics of sía, he created the preconditions for the new kenning formation in Þjóðólfr’s

other occurrence given, that of a variant reading of Vafþrúðnismál 31 (s.v. sía; see also Norrœn fornkvæði, ed. Sophus Bugge, p. 70). 242 This observation is equally valid whether sía forms part of a kenning or not. 243 In the stanza, sía works on at least three levels: That of the smith, that of the kenning and thus the smithy, and that of Þórr and Geirrøðr. It is possible that the meaning ‘spark’ is alluded to on the first two levels, but sía, nonetheless, remains the object of [throwing and] catching, which presupposes the meaning ‘glowing iron rod/lump’. 244 See Fritzner, Ordbog, 2, s.v. jarnsía and 3, s.v. sía, to which ONP has added two (http://www. onp.hum.ku.dk/). Also LP, s.v. sía, p. 497; Beatrice La Farge and John Tucker, Glossary to the Poetic Edda Glossary to the Poetic Edda: Based on Hans Kuhn’s Kurzes Wörterbuch, Skandinavistische Arbeiten, 15 (Heidelberg, 1992), s.v. sía. 245 In U, Þórsdrápa is mentioned but not quoted. It is possible that Þórsdrápa was added to Skáldskaparmál after Snorri’s time. If so, we are not dealing with an introduction to the poem, but rather with allusions to it in U. 246 Snorri, Skáldskaparmál, ed. Faulkes, 1, p. 25. In U only the last occurrence lacks the prefix. 247 This can be inferred since Snorri speaks of síur as flying out of a fire (Snorri, Gylfaginning, ed. Faulkes, pp. 10, 12, 13, 136). 248 SkP 3, pp. 111–13. Similar imagery is also found in another stanza by Eilífr, quoted earlier in Skáldskaparmál (SkP 3, p. 113).

4.6 Archaizing Diction

259

stanza. The observation of influence from Þórsdrápa on Þjóðolfr’s stanza can thus be expanded to include the prose introduction to that poem and the date of such influence must be postponed by some 150–200 years. The following þáttr in Morkinskinna consists of a poetic exchange between Haraldr, Þjóðólfr and a mysterious fisherman, who reveals his name to be Þorgils at the þáttr’s end. As in the preceding þáttr, Haraldr commands Þjóðolfr to compose a stanza. In this stanza, Þjóðolfr extolls Haraldr’s victory over the Danes with the words ‘goð vǫ́ru grǫm Dǫnum’ (the gods were angry with the Danes). Ascribing such agency to the gods is, as far as I am aware, unique for a poet of the post-pagan Middle Ages. The last instance is found in a poem by Eyjólfr dáðaskáld in the beginning of the eleventh century (see above pp. 63–64). Although it is tempting to conclude that the stanza is spurious, I shall argue that it probably is not. This þáttr, including the stanza, is found not only in Morkinskinna and in manuscripts from its tradition, Hulda-Hrokkinskinna and Flateyjarbók. The þáttr and the stanza are also found in Fríssbók, where both prose and poetry differ considerably from what is found in Morkinskinna, both in the prose and in the poetry. In Fríssbók, Haraldr and Þjóðólfr engage in poetic dialogue with three men, rather than just one, and there is very little lexical overlap in the prose of the two versions. Furthermore, each version contains one and a half stanzas that the other does not, and the ones that do overlap display variants that appear to be oral rather than written.249 Up to the beginning of this þáttr, Fríssbók follows the Heimskringla tradition, but at the end of the saga of Haraldr harðráði, the scribe has left two blank lines and then introduces the þáttr with a large ornamented initial. After the þáttr follow two and a half columns of blank space before the saga of Óláfr kyrri begins. It would appear that the scribe intended to collect further material about Haraldr that did not belong to the Heimskringla tradition, but in the end, he only wrote down this þáttr.250 The differences between this and the Morskinskinna version of the þáttr are so conspicuous that the easiest way to explain them would be that they are due to two independent recordings from oral tradition. This is what Bjarne Fidjestøl concluded in a study of the two versions of the þáttr.251 Jonna LouisJensen was of the same opinion, but her discussion is much less detailed.252

249 See Bjarne Fidjestøl, ‘The Tale of Haraldr harðráði and Þorgils the Fisherman’, in Bjarne Fidjestøl, Selected Papers, ed. Odd Einar Haugen and Else Mundal (Odense: Odense University Press, 1997 [1971]), pp. 277–93. 250 See Codex Frisianus. En samling af norske konge-sagaer, ed. C. R. Unger (Christiania: Mallings, 1871), p. iv. 251 Fidjestøl, ‘The Tale of Haraldr harðráði and Þorgils the Fisherman’. 252 Jonna Louis-Jensen, Kongesagastudier. Kompilationen Hulda-Hrokkinskinna. Bibliotheca arnamagnæana, 32 (Hafniæ: Reitzel, 1977), pp. 85–87.

260

4 Prosimetrical Narrative

There is one strong indication to the contrary, however. In Fríssbók, the þáttr opens with the words ‘Þorgisl hét maðr ok var kallaðr hástaði. Hann var kynjaðr af þeim bœ er heitir Hástǫðum í Noregi’ (a man was named Þorgisl and was called hástaði. His family came from the farm called Hástaðir in Norway).253 This introduction is lacking in Morkinskinna, Hulda-Hrokkinskinna and Flateyjarbók, where the reader is left in suspense until Þorgils’ name is revealed in the end of the þáttr (Þorgisl and Þorgils are variants of the same name). The Fríssbók version of the þáttr explains the nickname hástaði as a derivation of Hástaðir, the name of a farm, but there are no clear parallels to this formation. Nicknames referring to a person’s origin and lacking a suffix, such as sk or -ing, typically denote a region or a country, not a farm.254 The exception is nicknames in -dœll, a derivation of dalr (valley) referring to people coming from farms whose names end in -dalr. In the name of a farm ending in -staðir, one would rather have expected a compound *Hástaða-Þorgils, like Staðar-Bjarni (and Staðar-Bǫðvarr, Staðar-Kolbeinn) or Óláfr Geirstaðaálfr.255 The nickname hástaði thus requires an explanation, and it would seem that one may be found in the Morkinskinna tradition. In the end of the þáttr, where Þorgils tells the king his name according to Morkinskinna and Fríssbók alike, manuscripts in the Morkinskinna tradition read ‘Heiti ek Þorgils. Kastaði [thus Flateyjarbók; Morkinskinna has ‘kasti’, an error for kastaði, and Hulda-Hrokkinskinna have ‘varp’, synonymous to kastaði] síðan fiskikuflinum [. . .]’ (My name is Þorgils. He then threw off his fishing coat [. . .]).256 Fríssbók here reads ‘Ek heiti Þorgils hástaði’ (My name is Þorgils hástaði). I have here added the length marker according to the interpretation presented in Fríssbók, that is, that the name is derived from Hástaðir, but the word is written ‘hastaði’ in the manuscript. It is thus very graphically similar to kastaði with a short vowel.257 The most likely explanation of the peculiar formation hástaði is therefore that it is based on a creative misreading of the words Þorgils kastaði (Þorgils threw) in the end of the þáttr, following a manuscript with the reading of Flateyjarbók and Morkinskinna (but without the error in Morkinskinna). Perhaps the misreading was due to that the following words were illegible.258

253 Codex Frisianus, ed. Unger, p. 254. 254 Finnur Jónsson, ‘Tilnavne i den islandske oldlitteratur’, Aarbøger for nordisk oldkyndighet og historie, 2: 22 (1907), pp. 173–74. Both Finnur Jónsson and Lind accept the explanation of the nickname found in Fríssbók (Finnur Jónsson, ‘Tilnavne’, p. 173; E. H. Lind, Norsk-isländska personbinamn från medeltiden. Samlade och utgivna med förklaringar (Uppsala: Lundequistska bokhandeln, 1920–21), p. 137. 255 Finnur Jónsson, ‘Tilnavne’, pp. 187, 189. 256 Morkinskinna, ed. Ármann Jakobsson and Þórður Ingi Guðjónsson, 1, p. 288. 257 AM 45 fol 54 va l. 12, vb l. 19. 258 Fidjestøl, ‘The Tale of Haraldr harðráði and Þorgils the Fisherman’, p. 281.

4.6 Archaizing Diction

261

‘Þorgils kastaði’ and Þorgils hástaði: GKS 1005 fol (Flateyjarbók) 202rb l. 6 ‘kastadi’ (Image: Reykjavík, Stofnun Árna Magnússonar); GKS 1009 fol (Morkinskinna) 17r l. 22 ‘kasti’ (Image: Det Kgl. Bibliotek, Copenhagen); AM 45 fol 54vb 19 (Fríssbók) ‘hastaði’ (Image: Copenhagen, Den Arnamagnæanske Samling. Photo: Suzanne Reitz).

This misreading implies that even though the form of the þáttr in Fríssbók differs substantially from that of the Morkinskinna tradition, it is not likely to have entered Fríssbók based on oral tradition alone. A further detail supports a partially written transmission. In Morkinskinna, the king asks the fisherman to produce a stanza, and the fisherman then complains that it is difficult to do so after Þjóðólfr and the king ‘at upphafi’ (to begin with).259 In the Morkinskinna version, the fisherman, not the king, proclaims the first stanza, but in Fríssbók, the king is first. The words ‘at upphafi’ have apparently been retained from an exemplar where the first three stanzas came in the order they do in Fríssbók. Because of the misreading and the remaining trace of an older order of stanzas corroborated by Fríssbók, it is unlikely that we are dealing with two recordings from oral tradition, but rather with written transmission behaving in the way we would normally expect oral transmission to do. A number of factors may have contributed to this outcome. The words ‘Þorgils kastaði’ are found in the end of the þáttr in Morkinskinna, where the revelation of the fisherman’s name concludes the narrative. This was probably the case in the exemplar that has been misread, meaning that a scribe in the Fríssbók line of transmission read through the whole þáttr before he began to write it down. Such a procedure suggests that he may have intended to adapt it all along, rather than to just follow his exemplar. In Morkinskinna, unlike Fríssbók, the king makes a point of teasing Þjóðólfr, just as he does in the preceding þáttr, and this thematic coherence suggests that the prose has been adapted to the tenor of Morkinskinna. At various points in both versions, only a few words introduce a stanza, such as ‘now you compose, Þjóðólfr’, which suggests that the prose frame in the common ancestor was minimal and that the adaptations partly consisted in transforming this frame into a format more in keeping with written saga style.

259 Morkinskinna, ed. Ármann Jakobsson and Þórður Ingi Guðjónsson, 1, p. 286.

262

4 Prosimetrical Narrative

Some variants in the stanzas may be due to the scribes’ recollection of oral tradition. The three men in the Fríssbók version may also reflect variation that goes back into oral transmission. If the frame was as minimal as suggested here, however, it may not be very useful to think of the number of protagonists as variation in oral storytelling, since there was no more story than what was needed to hold the stanzas together. Rather, we see variation in the factors which set the scene and little else. As argued above on pp. 210–12, this probably reflects the transmission of situational stanzas in an oral setting and the character of the oral background of skaldic prosimetra. The bulk of the stanzas in the þáttr appear to antedate the age of prosimetrical saga writing. The fourth line of each stanza ends in the word skǫmmu (a short while ago), whereas the eighth ends in lǫngu (long ago) or fyr lengra (longer ago). In three instances (one in Fríssbók), skǫmmu forms aðalhending with -am-, and in one instance in both versions we find the aðalhending ang: ǫng. Such aðalhendingar are not attested after c. 1200. In the stanza containing the words ‘goð vǫ́ru grǫm Dǫnum’ (the gods were angry with the Danes), the aðalhending is grǫm: skǫmmu, which provides no dating criterion. This stanza is, however, found in both versions, and it is therefore reasonable to assume that it had been transmitted with the others for a long time. If so, it may well be authentic, particularly since no Norwegian king during the eleventh century seems to have appreciated mythological references more than Haraldr (see above pp. 70–71, 74–75). If authentic, the stanza is a remarkably late expression of agency of the gods in a clearly Christian context, albeit a secular one. This might in turn make it doubtful whether Eyjólfr’s expression of such agency may be taken as an indication of Eiríkr jarl’s religious affiliations, even if it must at least indicate that Eiríkr was tolerant of other religious expressions than Christian ones. Perhaps it may be right, as Heimskringla says, that Eiríkr was baptized but allowed everyone to do as they pleased in relation to Christianity.260 We are thus left with one stanza attributed to Þjóðolfr in Morkinskinna which can with some confidence be assigned to Pseudo-Þjóðólfr: namely, the stanza containing the word sía. This is of some importance since Morkinskinna is a compilation of kings’ sagas. Most other examples of pseudonymous composition are found in sagas of Icelanders and Ragnars saga. It is probably the case that pseudonymous composition was considerably more common in sagas of Icelanders than in kings’ sagas, but three more likely instances of pseudonymous composition in a king’s saga were identified in Chapter Two: these are Sighvatr’s kennings Dáins mun- (var. munn-) vágr (Dáinn’s [a dwarf’s] delight- (var. mouth-) wave [poetry]) and hella mildings mál (caves’ lord’s [giant’s] speech [gold]) and Óláfr’s Mardallar grátr (Mardǫll’s [Freyja’s] weeping [gold]).

260 Heimskringla, ed. Bjarni Aðalbjarnarson, 1, pp. 554–57.

4.6 Archaizing Diction

263

As noted on pp. 72–75, it is highly unlikely that Óláfr and his chief poet would have composed using specific mythological references, and as it turns out, all three occurrences can be traced to the so-called ‘Styrmir’s book’, that is, Styrmir’s saga of Óláfr. They are thus probably all attributable to Styrmir (a linguistic feature indicates that they are not older than his day), rather than to poets active two hundred years earlier. These four occurrences of likely pseudonymous stanzas in the kings’ sagas is a small corpus indeed, and in a sense, its paucity may offer some confirmation of an overall difference between kings’ sagas and sagas of Icelanders with regard to pseudonymous composition. I would suggest, however, that these findings primarily indicate that pseudonymous compositions in the kings’ sagas are to be found in þættir about eleventh-century kings. It is very difficult to tell how common such preudonymous composition was. Very few of the pseudonymous compositions treated in this book or elsewhere, except for those in Grettis saga and a few in other sagas, are late enough to be corroborated on linguistic grounds. The criteria reviewed in the previous section apply to purported early poetry, where saga authors would have striven to create pre-classical structures with which they were not actively familiar. The eleventh century was the age of classical dróttkvætt, the age of the hǫfuðskáld (main poets) whose style Snorri said that present poets should emulate, rather than learn in order to do otherwise as in the case of earlier poets. Eleventh-century poetry is thus on the whole formally indistinguishable from thirteenth-century poetry. The morphological and phonological changes between the eleventh and early thirteenth century are so few that it requires some luck to find examples which pertain to the later period. Thus, for instance, in the pseudonymous corpus identified in this chapter, there are no examples of aðalhendingar in older ø: ǫ nor of contracted variants of earlier hiatus forms (e.g. blán for bláan), two features that could possibly have betrayed pseudonymous composition. This should warn us against assuming that their absence in the relevant portions of Morkinskinna or the articles of Styrmir means that the poetry is trustworthy. For linguistic and cultural reasons alike, then, the þættir in the kings’ sagas are not likely to provide us with formal criteria for dating their poetry. In light of these considerable restrictions, the four stanzas singled out as probably spurious may have implications far beyond their number, suggesting that the methods for achieving a prosimetrical style in the þættir may have more in common with sagas of Icelanders than with the rest of the kings’ saga corpus. This is all the more noteworthy since the þættir were not paired with sagas of Icelanders in medieval transmission, as is common in modern editions and translations.261

261 Ármann Jakobsson, ‘The Life and Death of the Icelandic Short Story’, Journal of English and Germanic Philology, 112 (2013), pp. 257–91.

264

4 Prosimetrical Narrative

4.7 Pseudo-Grettir One fourteenth-century poet shows more clearly than others how Snorri’s Edda was used for the composition of pseudonymous poetry, but also how this learned endeavour went hand in hand with a predilection for the poetic figures that were the trademark of the learned poets, including the Sturlungs: specific mythological references, wordplay and extended metaphor (as described in Chapter Two). The poet in question is Grettir, the hero of Grettis saga. The historical Grettir lived c. 996–c. 1031,262 but the scholarly consensus, including the conservative Finnur Jónsson, is that the bulk of the stanzas in Grettis saga can be dated to the time of the composition of the saga in the first half of the fourteenth century.263 Here, we are in the rare position of having massive linguistic evidence for dating, and this is succinctly overviewed by Guðni Jónsson in his edition.264 Five stanzas have aðalhendingar in æ: œ (stanzas 20, 21, 27, 29, 65), and these phonemes do not blend before the second half of the thirteenth century; three stanzas have aðalhendingar in ǫ: ø (stanzas 3, 20, 28), which blend around 1200; two stanzas have aðalhendingar in vá: ó (stanzas 6, 60), indicating that the shift from vá > vo, beginning in the early fourteenth century, was under way265; nominative masculine -r must be read as -ur in two stanzas (stanzas 38, 67), a change which begins in the second half of the thirteenth century; t must be read as ð in two stanzas (stanzas 27, 71), a change which occurs in the thirteenth century. Several additional features which occur only in one stanza each could be added. In the following, I add some stylistic indications of a late date, including some for Grettir’s poetry in kviðuháttr. Since this metre does not have rhyme, it is not susceptible to such dating criteria as the above, and most scholars have refrained from assigning a date to these stanzas. Perhaps because of its apparent inauthenticity, Grettir’s poetry has received limited scholarly attention, the most noteworthy exceptions being a study of the stanzas in their saga context by Rolf Stavnem in 2000 and one by Heather O’Donoghue in 2005.266 While it may be true that Grettir merits little study as an eleventh-century

262 Finnur Jonsson, Den oldnorske og oldislandske litteraturs historie, 1, p. 512. 263 Finnur Jónsson, Den oldnorske og oldislandske litteraturs historie, 1, pp. 513–14; Grettis saga, ed. Guðni Jónsson, pp. lxix–lxx. 264 Grettis saga, ed. Guðni Jónsson, pp. xxxi–xlii. Kathryn Hume, ‘Thematic Design of ‘Grettis saga’’, Journal of English and Germanic Philology, 73 (1974), pp. 469–86, argues not only that this is valid for the stanzas but also that the narrative into which they are embedded is to a great extent purely fictive and dependent on the author’s views of society. 265 The details of this change could be elaborated; see Hreinn Benediktsson, ‘Relational Sound Change: vá > vo in Icelandic’, in Lingustic Studies, Historical and Comparative, ed. Guðrún Þórhallsdóttir et al. (Reykjavík: Institute of Linguistics, 2002 [1979]), pp. 227–42. 266 Rolf Stavnem, Stroferne i ‘Grettis saga’: Deres funktion og betydning (København: Det armnamagnæanske Institut, 2000); Heather O’Donoghue, Skaldic Verse and the Poetics of Saga Narrative (Oxford: Oxford University Press, 2005), pp. 180–227.

4.7 Pseudo-Grettir

265

poet, however, pseudo-Grettir is the single most promising poet of the fourteenth century for the study of how the stylistic legacy of the preceding two centuries was upheld and adapted to saga writing. Before moving on to the analysis of the poetry in the saga, it is worth considering the implications of the linguistic evidence in the poetry of Grettis saga as compared to the rarity of such evidence in other sagas. The parameters involved are so many and so clear that we must probably conclude that massive interpolation of spurious stanzas into other sagas rarely took place in the fourteenth or even at the end of the thirteenth century, or we would have many more cases like Grettis saga. This means that the thirteenth century was probably the most productive period for pseudonymous poetry, with a very limited production before and a relatively small production also afterwards (see p. 218). The poetry in Harðar saga and Víglundar saga shares some of these features, but there is less poetry in these sagas. The same cannot necessarily be said of the beginnings of saga writing, since Old Norse-Icelandic remained very phonemically and morphologically stable until the second half of the thirteenth century. The most noteworthy exception for our purposes is the merger of /ǫ/ and /ø/ to /ø/, which took place around 1200, and both Snorri and Óláfr take it for granted. Before the merger, /ǫ/ could serve as aðalhending to /a/, but no stanza with an aðalhending of this kind has as yet been shown to have been composed after 1200. Such absolute criteria are rare, and a detailed evaluation of linguistic and metrical dating criteria is beyond the scope of this book. In fact, comprehensive discussions of these criteria have been rare since the early twentieth century and the matter has only recently been reopened by Klaus Johan Myrvoll.267 In the intervening period, literary and historical criteria which are difficult or impossible to test have dominated, and most studies have focused on individual sagas, rather than on the saga corpus at large. In many cases, scholars have opted for an ‘all-or-nothing’ approach, questioning or defending the entire corpus of the individual saga poets, often based on a close study of only a few stanzas. A study of more measurable features, such as that of Myrvoll or that attempted in the chapter on metrics above suggests, however, that it was rare that saga authors composed the bulk of the poetry themselves, at least in the thirteenth century. Saga authors would use the poetry they knew, and in some cases that poetry may have served as the original impetus to the writing of the saga (this might, for instance, have been the case with Kormáks saga). Some authors would add what they found to be missing, in particular since the poetry they had was not calculated for a prosimetrical saga account. Only a few would do as the author of Grettis saga did and compose almost all of a rich prosimetrum themselves, at least in dróttkvætt-based genres. The only plausible instance of this from the preceding thirteenth century is PseudoRagnarr, and although he composed in a kind of dróttkvætt, the saga belongs to a

267 Myrvoll, Kronologi i skaldekvæde; idem, ‘The Authenticity of Gísli’s Verse’.

266

4 Prosimetrical Narrative

genre which normally features fornyrðislag. In general, the creative impulse appears to have come from pre-existent poetry, and most authors would cherish and sometimes supplement the poetic past, but they would not cut it out of whole cloth. The few who did offer us a rare opportunity of seeing what aspects of that past were seen as particularly interesting and important. The author of Grettis saga was among the most gifted of these. As we shall see, his interests were entirely consonant with those of his learned forebears – poets and saga authors – whose focus lay in advanced diction. What had changed by this point, however, was that Snorri’s Edda had become the canonical expression of what the poetic heritage was and how it should be used. I begin, therefore, with two examples of how the use of Snorri’s Edda influenced the composition of Pseudo-Grettir. The first occurs in the so-called Ævikviða II, a group of four stanzas in kviðuháttr.268 Since Ævikviða I and Ævikviða II both share their peculiar treatment of proper names with stanzas in the saga which can on linguistic grounds be dated to the fourteenth century, it is likely that Ævikviða I and Ævikviða II are also late.269 Another possibility is that Ævikviða I and Ævikviða II are early, and that the stylistic features in the other stanzas are modelled on them. If, however, influence from Skáldskaparmál can be demonstrated in either Ævikviða, this explanation becomes unlikely (although the faint possibility that one Ævikviða is authentic and the other pseudonymous cannot be conclusively ruled out). In Ævikviða II, two kennings refer by means of ofljóst to the formidable woman Þórbjǫrg, who has just saved Grettir from hanging. The kennings are: reynirunnr (rowan bush) > Þórs bjǫrg (Þórr’s salvation) > Þórbjǫrg and Sifjar vers beggja handa hjǫlp (Sif’s husbands’ [Þórr’s] both hands’ help) > Þórs bjǫrg > Þórbjǫrg.270 These mysterious references are explained by a passage in Skáldskaparmál and only there: Þórr is on his way to an encounter with the giant Geirrøðr and is nearly drowned when crossing a river, swelled with the urine of one of Geirrøðr’s daughters (R-text): Ok í því bili bar hann at landi ok fekk tekit reynirunn [T: runn; U: rísrunn] nokkvorn ok steig svá ór ánni. Því er þat orðtak haft, at reynir er bjǫrg Þórs.271 And at that point he [Þórr] came close to land and got hold of a rowan bush [T: bush; U: bush] and thus climbed out of the river. Thence we have the saying that the rowan is the salvation of Þórr.

The saying ‘reynir er bjǫrg Þórs’ (the rowan is the salvation of Þórr), or variants thereof, is not found elsewhere, but that does not rule out its oral pre-existence. The collocation of the saying with the word reynirunnr, however, is another matter. This word appears

268 The division into Ævikviða I and II is in accordance with the coming edition (SkP), as well as the saga, where the stanzas appear in two groups (Grettis saga, ed. Guðni Jónsson, pp. 86–87, 170–72). Finnur Jónsson combines the two groups under the heading Ævikviða (Skj B 1, pp. 288–89; A 1, pp. 309–11). On the age of Ævikviða, see Grettis saga, ed. Guðni Jónsson, pp. xxxvii–xxxviii. 269 Grettis saga, ed. Guðni Jónsson, p. xxxvii. 270 Grettis saga, ed. Guðni Jónsson, pp. 171–72; Skj B 1, pp. 287–88; A 1, p. 310. 271 Snorri, Skáldskaparmál, ed. Faulkes, p. 25.

4.7 Pseudo-Grettir

267

in both Grettis saga and in the R and W manuscripts of Snorri’s Edda. Elsewhere in Old Norse literature, the word appears to be found only in a passage in Geirmundar þáttr heljarskinns, considerably postdating Snorri’s Edda.272 A fortuitous collocation of this extremely unusual word for ‘rowan’ with the saying in two separate instances defies all probability (albeit the saying is only found in the subtext of the stanza). Textual influence must thus have gone one of two ways: from Ævikviða to Snorri’s Edda or the way around. The decoding of the name of the woman in the stanzas presupposes the saying, so as to arrive at the exact wording Þórs bjǫrg. The saying, in turn, is only present in Snorri’s Edda. Since influence must in all likelihood have gone from one text to the other, the one which has both the saying and the word reynirunnr is likely to be the giver: that is, Skáldskaparmál.273 Ævikviða II, then, appears to postdate Snorri’s Edda and to be composed for Grettis saga.

272 The largest dictionaries list no other occurrences (Fritzner, Ordbog, 3, s.v.; Dictionary of Old Norse Prose: http://dataonp.ad.sc.ku.dk/wordlist_d.html (viewed on 2 June 2014)). Geirmundar þáttr serves as an introduction to the Sturlunga saga compilation. In Geirmundar þáttr, the word reynirunnr occurs in a context that indicates that here, too, the word may be part of an allusion to Snorri’s Edda. We read that the pagan settler Geirmundr had on his lands a hillock where there grew a reynirunnr, the sight of which he could not stand because of the strange light it emitted. The þáttr further reports that the tree stood on the spot where the church at Skarð would later be raised and ends by tracing the descendants of Geirmundr to Skarð-Snorri who lived there (Sturlunga saga, ed. Kålund, pp. 5–6). The þáttr thus presents a settlement background to Conversion, and more particularly to the church at Skarð. The choice of the reynirunnr as a pre-Christian token of Christianity suggests that one should perhaps follow Snorri’s Edda and understand the tree as connected to Þórr, since he in various contexts, including saga writing (Eyrbyggja saga), seems to have served as a prefiguration of Christ (see Janson, ‘Edda and “Oral Christianity”’; Gschwantler, ‘Christus, Thor und die Midgardschlange’; Males, ‘Allegory in Old Norse Secular Literature’, pp. 115–23). The likely connection of Geirmundar þáttr to the Sturlungs makes this assumption attractive (see ‘Sturlunga Saga’ Including the ‘Islendinga Saga’ of Lawman Sturla Thordsson and Other Works, ed. Gudbrand Vigfusson, 2 vols (Oxford: Clarendon Press, 1878), 1, pp. ciii–cv; Jón Jóhannesson, Gerðir Landnámabókar, pp. 169–70; Sturlunga saga, ed. Jón Jóhannesson, Magnús Finnbogason, and Kristján Eldjárn, 2 vols (Reykjavík: Sturlunguútgáfan, 1946), 2, pp. xvi – xix; xxi–xxii. On the function of Geirmundar þáttr in the compilation generally, see Marlene Ciklamini, ‘Divine Will and the Guises of Truth in Geirmundar þáttr heljarskinns’, Skandinavistik, 11 (1981), pp. 81–88; Thomas Krömmelbein, ‘Die Spitzenstillung des Geirmundar þáttr heljarskinns innerhalb der Kompilation Sturlunga saga’, Alvíssmál, 4 (1994), pp. 33–50). 273 I can think of two other, albeit unlikely, explanations. For Snorri to have derived his wording from Ævikviða, he would have had to know the saying independently and drawn the word reynirunnr from the poem (bearing in mind that textual influence still remains necessary). Alternatively, he could have derived the ‘saying’ from the poem itself, only claiming that it was a ‘saying’. Snorri, however, seems not to have operated in that manner; rather, when he quotes or paraphrases poetry, he does so without comments (cf. gómsparri p. 87; similarly arnsúgr from Haustlǫng and ísarnkol from Grímnismál), but when he refers to ‘sayings’ (orðtǫk), he does so in order to comment on etymologies or quaint expressions. In these instances, his sources do not appear to belong to the poetic canon. While neither of these two options is impossible, both remain implausible. Furthermore, as we shall

268

4 Prosimetrical Narrative

I have found no independent dating criteria for Ævikviða I, which contains all stylistic markers of Pseudo-Grettir: mythology, complex wordplay and extended metaphor. In principle, then, the use of these features by Pseudo-Grettir could be inspired by Ævikviða I, even though Ævikviða II is pseudonymous. This is, however, unlikely for two reasons: both Ævikviður are extremely similar in style, and nothing quite like them can be found in the entire skaldic corpus. They thus appear to be the product of a single creative act. Furthermore, the wordplay in Ævikviða II is primarily predicated on Skáldskaparmál, rather than on Ævikviða I, suggesting that no direct influence from Ævikviða I need have been involved. There is thus some circumstantial evidence to support that we have the work of Pseudo-Grettir also in Ævikviða I. In the unlikely event that Ævikviða I is composed by the historical Grettir, however, its three stanzas would be a meagre source of inspiration for the corpus of Pseudo-Grettir, and his creativity would still be out of the ordinary. Pseudo-Grettir seems also to have been acquainted with another section of Snorri’s Edda, namely Háttatal. This is the only instance known to me where Háttatal has influenced the content, rather than metrical form, of a stanza.274 In this case, unlike Ævikviða, the stanza contains a metrically secure young form (komk for komumk, kvámumk), providing an independent indication that the stanza is not authentic.275 In stanza 6 of Háttatal, Snorri gives examples of extended metaphor, which he then explains in the commentary. In the Háttatal stanza, a sword is referred to as sóknar naðra (snake of battle), and in the commentary Snorri clarifies that this snake ‘skríðr hugar stígu, þat er brjóst manna’ (slithers along the paths of the mind, that is, the chest of men).276 Stanza 47 of Grettis saga has it that ‘vápnhríðar naðr varð fram skríða á benstíga’ (the weapon-storm’s [battle’s] snake [sword] had to slither forward on the wound-paths [wounds]).277 Not only is similar imagery used, but the same words recur: naðr/naðra, skríða and stígr.278 These words occur together only in these two places, and extended metaphor based on ‘sword’ as ‘snake’ is rare.279 The closest parallel is found in a stanza attributed to Grímr Droplaugarson, but none of the relevant words occur there: ‘hræmána úrfræningar rendu fjǫrbrautir’ (the corpse-moons’ [shields’] liquid- [blood-] snakes [swords] ran along the life-ways [chests]).280

see, the other instance of mutual agreement between Snorri’s Edda and poetry in Grettis saga exhibits a young form in the latter, clarifying the line of influence. 274 See Snorri, Háttatal, ed. Faulkes, pp. xvii–xviii. 275 Grettis saga, ed. Guðni Jónsson, p. xxxix. This is the reflexive, not the verb with enclitic pronoun, which would have been komk also in an earlier period. 276 Snorri, Háttatal, ed. Faulkes, pp. 6–7. 277 Grettis saga, ed. Guðni Jónsson, p. 185; Skj B 2, p. 471; A 2, p. 440. 278 Snorri, Háttatal, ed. Faulkes, pp. 6–7. 279 Based on a reading of the stanzas referred to by Meissner, Die Kenningar, p. 154. 280 Skj B 1, p. 183; A 1, p. 193. Two of the relevant words occur in a comparable image in KrókaRefr’s stanza 1, but it is likely to be pseudonymous (only the second of Króka-Refr's three stanzas has independent attestation in Laufásedda [stanza 146/384]): ‘hefik kannat hjarta stíg unda

4.7 Pseudo-Grettir

269

Wording and imagery in stanza 47 of Grettis saga thus indicate that the poet has been influenced by Háttatal, and indeed by both stanza and commentary there. Only the stanza has the word naðra and defines it with ‘battle’. Both stanza and commentary have skríða, but only in the commentary does the snake slither across the breast of men (in the stanza it slithers along the path of the scabbard). Both have stígr in comparable surroundings. This, together with the use of Skáldskaparmál for the composition of Ævikviða II, reveals the poet as an avid student of Snorri’s Edda.281 Pseudo-Grettir was an extraordinary artist, thoroughly steeped in the intricacies of skaldic diction not only through practice, but apparently also through extensive study of Snorri’s Edda. In the second chapter of this book, I discussed the stylistic nexus of wordplay, extended metaphor and complex mythological references, and the attraction they exerted on grammatically minded poets. No single saga shows more clearly than Grettis saga how these served as an asset in the construction of the poetic past. As we saw, the words reynirunnr (rowan tree [Þórr]) and hjǫlp beggja handa Sifjar vers (Sif’s husbands’ [Þórr’s] both hands’ help [Þórr]) betray influence from Skáldskaparmál, but they also indicate interest in both wordplay and mythological lore. Ævikviða II contains two additional instances of wordplay: marþaks fjǫrðr (sea-roof’s [ice-] fjord [Ísafjǫrðr]) and Þundar beðju þvengr (Þundr’s (Óðinn’s) bedfellow’s [Jǫrð’s = the earth’s] thong [snake, normally ormr, but here the synonym grettir = the hero Grettir]).282 The word reynirunnr, furthermore, forms part of an extended metaphor: this rowan is lofgróinn laufi sœmðar (praise-grown with the foliage of honour). In Ævikviða II, wordplay,

naðri’ (I have probed the heart’s path [breast] with the wounds’ snake) (Skj B 2, p. 487; A 2, p. 454). Other extended metaphors based on the sword as a snake are more restricted and use different images, but they are also late: Háttalykill 25b (Skj B 1, p. 499; A 1, p. 522) has alstiðr hjaldrliðr beit hausa (the very stiff battle-snake [sword] bit the heads), Íslendingadrápa 5: sónar ófnir varð saddr (the blood’s snake became full) (Skj B 1, p. 540; A 1, p. 556). 281 Another instance of overlap with Skáldskaparmál (and Litla Skálda) is found in the last stanza of the saga, attributed to Þorsteinn drómundr (Grettis saga, ed. Guðni Jónsson, p. 275). There, the kenning axlarfótr (shoulder-leg [arm]) is used, and this kenning is otherwise found only in the prose of Litla Skálda and Skáldskaparmál (Edda Snorra Sturlusonar, ed. Finnur Jónsson, pp. 192, 256). The substitution of one body-part for another in a kenning is extremely rare and otherwise occurs only in the learned poets or in such poetry as they used as models (Íslendingadrápa, Pseudo-Þjóðolfr, Arinbjarnarkviða, Þórsdrápa; see above p. 257 n. 239). Since no other poetry by Þorsteinn is attested and the stanza is found in connection with other pseudonymous poetry, one might conclude that it is probably spurious. The last line exhibits aðalhending in a: ǫ, however, which is otherwise not found after c. 1200 (see pp. 60–61). Finnur Jónsson, who with some hesitation believed the stanza to be late, solved that problem by emendation, but without manuscript support (Skj B 2, p. 476; A 2, p. 445). Furthermore, his solution could be criticized for what it does to the word order of the stanza. The formal criterion is strong, and it is possible that this stanza is Litla Skálda’s and Snorri’s Edda’s source for the kenning, as well as being one of the few stanzas that the author of Grettis saga drew on in order to construct its prosimetrical form. 282 Grettis saga, ed. Guðni Jónsson, pp. 170–2; Skj B 1, p. 288; A 1, pp. 310–11.

270

4 Prosimetrical Narrative

extended metaphor and complex mythological references are all combined in a dense web of obscure diction. The predilection for wordplay and extended metaphor is plain throughout the saga, except in the four stanzas or parts of stanzas that can plausibly be shown to be older, since they are quoted in other sources.283 The recurrent stylistic features indicate that we are dealing mainly with one and the same poet. He probably had a relatively small corpus of poetry to draw on: enough to construe Grettir as a poet and to give free rein to his own creativity. Most of the numerous examples of wordplay in the poetry of the saga are very complex.284 A famous instance is the kenning involving the god Bragi’s wife Iðunn: Braga kvónar fjón (the enmity of Bragi’s wife; see p. 45).285 This equals Iðunnar fjón (the enmity of Iðunn). The context, however, is that Grettir is swimming under a waterfall, and the reader or listener should thus understand not Iðunnar fjón, but iðunnar fjón (the enmity of the whirling stream). Here, mythological and referential complexity converge. Another representative example is the circumlocution for the name Þorsteinn drómundr found in Ævikviða I: stallgoðs bana ok Regins skáli Rauðahafs stórskip (cliffgod’s [giant’s] bane’s [Þórs-] and Regin’s (a dwarf’s) hall [stone, ON steinn > Þorsteinn] Red Sea’s large ship [the ship type drómundr, used on the Red Sea > Þorsteinn drómundr]).286 This circumlocution is followed by a mythological reference in the second half of the stanza: Býleists bróðurdóttir (Býleistr’s brother’s [Loki’s] daughter [Hel = death]). The poetry in the saga also exhibits a penchant for extended metaphor. The most elaborate instance in the saga and, at least with regard to its impact on the

283 Finnur Jónsson, Den oldnorske og oldislandske litteraturs historie, 1, p. 514; Grettis saga, ed. Guðni Jónsson, pp. xli, 92–93, 207–08, 234–35. 284 Three instances of ofljóst, apart from the one in Ævikviða II discussed above, designate Grettir himself: fjalla þinull (the mountains’ string [snake, grettir = Grettir]) (Grettis saga, ed. Guðni Jónsson, p. 98; Skj B 1, p. 288; A 1, p. 311), lautar áll (the land’s eel [snake, grettir = Grettir]) (Grettis saga, ed. Guðni Jónsson, p. 150; Skj B 2, p. 469; A 2, p. 438) and grundar hœngr (the ground’s salmon [snake, grettir = Grettir]) (Grettis saga, ed. Guðni Jónsson, p. 176; Skj B 2, p. 470; A 2, p. 439). Other wordplay: lítill steinn ok land hnefa (small stone [ON hallr] and fist’s land [hand, ON mund, i.e. Hallmundr]) (Grettis saga, ed. Guðni Jónsson, p. 176; Skj B 2, pp. 470; A 2, p. 439), Hafla fangvinr (Hafli’s (a giant’s) wrestling-friend [Þórr, also known as Bjǫrn; it seems that the two names are here to be combined into Þorbjǫrn] (Grettis saga, ed. Guðni Jónsson, p. 156; Skj B 1, p. 289; A 1, p. 312) and Arfs ok Gneista afl (Arfr’s (heiti for ox) and Gneisti’s (heiti for ox) force [oxen-force, ON øxnamegin, cognomen of Þorbjǫrn]) (Grettis saga, ed. Guðni Jónsson, p. 156; Skj B 1, p. 289; A 1, p. 312). Note that in the last case, the poet has taken pains to give two oxen to match the plural genitive øxna-. Another stanza has hlébarðr ( Bersi]) (Grettis saga, ed. Guðni Jónsson, p. 87; Skj B 1, p. 387; A 1, p. 310), and yet another has Jalfaðr ([heiti for Óðinn, here Auðunn, another heiti for Óðinn]) (Grettis saga, ed. Guðni Jónsson, p. 97; Skj B 1, p. 288; A 1, p. 311). 285 Grettis saga, ed. Guðni Jónsson, pp. 216–17; Skj B 2, p. 473; A 2, p. 442. 286 Grettis saga, ed. Guðni Jónsson, p. 86; Skj B 1, p. 287; A 1, p. 309.

4.7 Pseudo-Grettir

271

surrounding prose, in Old Icelandic literature at large, is found in stanza 11. There, an axe is referred to with the kenning hamartroll (hammer-troll [axe, the hammer being the hammer of the axe]; the interpretation ‘cliff-troll’ is, however, also possible), and this troll ran (hljóp) briskly to its victim. The troll is then specified as gunnar Gríðr (battle’s Gríðr [trollwoman; axe]). Gríðr in this kenning is then personified in the second half of the stanza: ‘sú gein harðmynnt ok sparði lítt vígtenn’ (she grinned with a hard mouth and spared but little her battle-teeth [edge]). In the background, one may hear the Old Norse name for an axe’s edge, muðr (mouth). The metaphor is so consistently sustained that the listeners in the saga take the stanza to mean that the victim was killed by an actual troll, analysing hamartroll as ‘cliff-troll’, rather than ‘hammer-troll’. Only the keen-witted Þorkell understands the true meaning of the stanza, namely that Grettir himself had directed the jaws of this particular ogre.287 The focus within the prose on this extended metaphor indicates that poet and author shared an interest in the same poetic features and are likely to have been one and the same person. Gunnar Gríðr is a conventional kenning, but its potential is here realized in new ways. The same holds true for a kenning in the next stanza: móins leiðar beiðendr (móinn’s [a snake’s] path’s [gold’s] cravers [men]).288 A kenning like this would normally be value neutral, but here, the imagery of riches and, potentially, greed is used to enhance the message of the stanza: these móins leiðar beiðendr are ‘heldr auðigir’ (quite rich), and yet have sent Grettir away ‘snauðr’ (poor.) Three more noteworthy examples of extended metaphor are found in the poetry of the saga. Two of these have been discussed above, namely stanza 47, based on Snorri’s example in Háttatal, and Ævikviða II. In Ævikviða I, already noted for its wordplay on Þorsteinn drómundr, the imagery of arson is used: Bersi wished to ‘brenna bragða borg hlífar eldi’ (burn the clever counsels’ citadel [breast or head] with the shield’s fire [sword]).289 Two of the poetic quotations just mentioned, namely stanza 47 and Ævikviða I, are also interesting because they are authenticating rather than situational quotations. These are the only examples of which I am aware where authenticating quotations of skaldic poetry can with a high degree of plausibility be shown to be spurious. The very rarity of such cases suggests that the distinction, for all its crudeness, provides a strong but not absolute dating criterion. Even if this book treats only a limited number of pseudonymous stanzas, I see no obvious reason why one should expect a completely different distribution in other cases. It is not entirely clear why the author of Grettis saga should have behaved differently from most authors in this regard, but he was certainly extremely creative, and it is possible that the relative waning of semi-secular centres of learning entailed changing norms in the treatment of secular poetry.

287 Grettis saga, ed. Guðni Jónsson, p. 47; Skj B 2, p. 464; A 1, p. 433; on this stanza, see also Stavnem, Stroferne i ‘Grettis saga’, pp. 20–22; O’Donoghue, Skaldic Verse, pp. 193–94. 288 Grettis saga, ed. Guðni Jónsson, p. 50; Skj B 2, p. 464; A 1, p. 433. 289 Grettis saga, ed. Guðni Jónsson, p. 185.

272

4 Prosimetrical Narrative

Pseudo-Grettir, then, appreciated the same features of diction as the learned poets of the twelfth and thirteenth centuries. His level of skaldic learning was extraordinary and is reminiscent of twelfth-century poets like Einarr Skúlason and Hallar-Steinn. This raises the question of who this remarkable poet may have been. The dating and geographical scope of the saga may offer some indications, provided that author and poet were the same person or at least somehow connected. I proceed now to a discussion of these matters. Several references to other sagas show that Grettis saga must belong to the late sagas of Icelanders (see below). For a more precise date, the most promising evidence is a description in the saga of when Grettir’s spear was found: ‘Spjótit þat, sem Grettir hafði týnt, fannsk eigi fyrr en í þeirra manna minnum, sem nú lifa; þat spjót fannsk á ofanverðum dǫgum Sturlu lǫgmanns Þórðarsonar’290 (The spear that Grettir had lost was not found any earlier than that people now living can remember it; that spear was found in the old age of Sturla Þórðarson the lawspeaker). Sturla was born in 1214 and died in 1284, and we may perhaps allow for his old age to occupy the years 1270–1284 or thereabouts. The reference to the memory of people living probably suggests that these were now old, but were young when the spear was found. It the finding was made as early as 1270 and ‘people’ were as old as fifteen or even twenty at the time, one might perhaps reckon with some 40–50 years for them to be referred to as ‘now [still] living’, giving a post quem at 1310–1320. If the spear was found around 1280 and the informants were ten at the time and around eighty when ‘now living’, we get an ante quem around 1350.291 The author is geographically best acquainted with the area of Húnavatnsþing in north-western Iceland, on such point of detail that he must have lived there for a long time.292 He explicitly refers to four sagas, namely Saga Eiríks jarls (now lost), Laxdœla saga, Bandamanna saga and Bjarnar saga Hítdœlakappa.293 It is also evident that he relies heavily on Landnámabók, and it is highly likely that he knew several other sagas as well.294 It is thus clear from the saga that he must have had access to a good library, and Húnavatnsþing could boast of one of the best in the country, namely that of Þingeyrarklaustr, Iceland’s oldest monastery.295 Since the author was an avid reader from the vicinity of Þingeyrar, it seems likely that he was in contact with the monastery, if he did not indeed belong to it. As for the poet, if his work is separated

290 Grettis saga, ed. Guðni Jónsson, p. 157. 291 This analysis is based on Grettis saga, ed. Guðni Jónsson, pp. lxix–lxx, 157. Guðni Jónsson posits a narrow time frame of between 1310 and 1320, seemingly to accommodate his candidate for authorship, Hafliði Steinsson, who died in 1319. 292 Grettis saga, ed. Guðni Jónsson, pp. lxxi–lxxii. 293 Grettis saga, ed. Guðni Jónsson, pp. xxv–xxvi, 25, 37, 62, 187. 294 Grettis saga, ed. Guðni Jónsson, pp. xvii–xxxi. 295 Grettis saga, ed. Guðni Jónsson, pp. lxxii–lxxv.

4.7 Pseudo-Grettir

273

from the prose, we can say little more about his intellectual background than that he was thoroughly acquainted with Snorri’s Edda. Although Snorri’s Edda was studied in different parts of the country, it is most strongly associated with Þingeyrar in the fourteenth century, and the debate about Snorri’s Edda at Þingeyrar will be the topic of the next chapter. The interests of Pseudo-Grettir thus fit with the overall intellectual milieu of Þingeyrar. He may in principle have been someone else than the saga author, but several factors make this unlikely or at least of mainly theoretical interest. The linguistic features of the poetry can hardly be accommodated before c. 1300, and the poet was thus probably not much earlier than the author. As seen above with regard to extended metaphor, the author and the poet were interested in the same poetic figures. Grettir, furthermore, was a local hero, and one would thus expect the impetus for pseudonymous composition to have been strongest within the geographical area associated with him. The degree to which pseudo-Grettir appears to have internalized Snorri’s Edda and the most inaccessible portions of skaldic poetry goes well beyond what one might expect of most poets, not to mention other members of society, indicating that he belonged to a highly literate milieu comparable to that of the author. In sum, author and poet were roughly contemporary, active in the same geographical area, interested in the same poetic features and probably both connected to Þingeyrar. The scale of pseudonymous composition indicates that it was undertaken for inclusion into the saga, and there is nothing in the medieval manuscript transmission to suggest later addition of stanzas.296 If the poet was not the author, he must at least have undertaken a full revision of the saga, which would in any event blur the border between author and poet. For the present purposes, then, the two may be treated as the same person. This person need not necessarily have been a monk; for instance, Hafliði Steinsson, who got his education at the monastery and became the priest of the richest living in the area – Breiðabólstaðr – has been suggested as the author of the saga.297 He was procurator of the monastery’s lands and would have had frequent dealings with the monks, as well as access to the library. He died in 1319, which may be somewhat early to make him a likely candidate for authorship, but pinpointing an individual is not my main concern here. More important is the connection to the intellectual context of Þingeyrar. With the identification of the influence of Snorri’s Edda on pseudo-Grettir, we have a new testimony to the strong focus on Snorri’s Edda in the learned milieux in or around Þingeyrar in the fourteenth century. Whatever the precise nature of the connections between the saga and the monastery, they suggest that monastic literary activities, pagan mythology and local saga heroes were all densely interwoven.

296 Grettis saga, ed. Guðni Jónsson, p. lxxvii. 297 Grettis saga, ed. Guðni Jónsson, pp. lxxiii–lxxv.

274

4 Prosimetrical Narrative

4.8 Summary: The Image of the Poet Here I shall outline some inferences, relevant for the Icelandic literary flowering of c. 1150–1350, which can be made based on the analyses in this chapter. First of all, the evidence of known works suggests that skaldic poetry and prose did not blend into a prosimetrical form until c. 1200 (although eddic poetry and some sort of prose probably did so much earlier). The classical saga form thus appears to have developed during the period under study. Notwithstanding the fact that the sagas are deeply indebted to local tradition and in many ways quite at odds with what Latin literature had to offer – the mode of poetic quotation being a case in point – they are nonetheless a product of the creative dynamics which emerged in the encounter between Old Norse and Latin learning. Their form is thus both traditional and new. Pseudonymous poetry highlights this creative tension. The breaking up of longer poems and the accommodation of these in a prose setting, as well as the inclusion of single, well-known stanzas into a grand narrative, were equally creative acts, but they are often invisible to us. In pseudonymous poetry, by contrast, we can come to grips with some of the methods involved and see just how dynamically poetry and prose were made to interact. The strong poetic tradition was all-important to this outcome, but pseudonymous poetry is a good reminder that tradition alone cannot account for the rise of saga literature. Also indebted to the poetic tradition, but evolving with the evolution of narrative texts about his character, was the poet himself. His or, rarely, her role was at the heart of Snorri’s Edda and the sagas alike, but he was treated somewhat differently in the two. I wish now briefly to dwell on the similarities and differences between Snorri’s Edda and the sagas, with a particular focus on pseudonymity. Snorri’s Edda, and Skáldskaparmál within it, constitutes a eulogy to Iceland’s poetic past, but as we have seen, it is highly unlikely that Snorri in that context composed pseudonymous poetry to glorify that past. This appears to hold true of Heimskringla as well. In composing Egils saga he appears to have done so, however, and a number of contemporary and later saga authors clearly did. This appears mainly to have happened when they wished to flesh out the historical image of the Icelandic poet, and when no important historical facts were at stake, such as a king’s battles. In these contexts, they also took the liberty to break the metrical rules that otherwise applied for poets of their own age. If Snorri’s Edda explains what skaldic poetry is, the sagas place the figure of the poet in a historical context. A clear example of this is Egils saga, but we also see it in Morkinskinna, where þættir about Icelandic poets provide colourful images of the interaction between poet and king, and the same can generally be said about sagas treating Óláfr Haraldsson. As we have seen, some pseudonymous poetry appears to have been composed for Morkinskinna and Styrmir’s saga of Óláfr. This suggests that the kings’ sagas, or at least the parts that are most reminiscent of sagas of Icelanders, are also indebted

4.8 Summary: The Image of the Poet

275

to an urge to situate the Icelandic poet historically that overrode the limitations imposed by earlier tradition. At this point, historiography and poetics had combined to produce a historical reflexion of the thirteenth-century poet in a comic but sympathetic light. If Snorri’s Edda had shown how poetry should be done, sagas of Icelanders and þættir in the kings’ sagas described who the poet was: dexterous in his poetics and clever and intrepid in his dealings with kings and magnates. In an age when Icelanders produced historical writings for a Norwegian market, the image of the historical poet showed how this had, in a way, always been the case, since the verbal skills of Icelanders had been highly prized in Norway since the establishment of the Icelandic nation.298 This self-perception need not have been entirely wrong, but the composition of pseudonymous poetry shows that it was actively cultivated in the thirteenth and fourteenth centuries and that historical reality was not at the heart of this endeavor. To the minds of the Icelandic historians, poetry as a national characteristic was as evident at home as abroad: whereas Morkinskinna, Styrmir’s book, and Egils saga bolster the image of the poet on an international scene, Grettis saga has its hero elaborate on Snorri’s Edda in a homely setting. On a more modest scale, this is true of Bandamanna saga as well. The body of pseudonymous compositions presented above is restricted so as to include only cases where a late date is highly plausible. Given the difficulties involved in separating authentic (or at least old) poetry from later compositions, this is not an insignificant corpus, and many sagas are involved: Morkinskinna, Styrmir’s book, Egils saga, Kormáks saga, Grettis saga, Stjörnu-Odda draumr, Bandamanna saga, Ragnars saga, Harðar saga and Víglundar saga. The pseudonymous poetry in these sagas indicates that composition in the name of the early poets was not uncommon and consequently that many authors were engaged in the project of solidifying and bolstering an impressive literary past at the same time. Snorri harked back to a poetic and poetological tradition that had been evolving for almost a century, and he provided it with its canonical shape in his Edda. With their literary output in general and their pseudonymous compositions in particular, he and other saga authors took this development one step further, tying poetics to vivid descriptions of the local past. The twelfth-century poets, the authors of kings’ sagas, Snorri and his nephews and many other saga authors were probably all active in closely related milieux. Many of them would have known each other personally, and through a number of different strategies,

298 The production of sagas of the Norwegian kings appears in several instances to have been intended for a Norwegian as well as an Icelandic audience (Sverris saga and Hákonar saga are the most obvious examples, but the great compilations at least partly served similar purposes). On manuscript export, see Stefán Karlsson, ‘Islandsk bogeksport til Norge i middelalderen’, in Stafkrókar, ed. Guðvarður Már Gunnlaugsson (Reykjavík: Stofnun Árna Magnússonar, 2000 [1979]), pp. 188–205.

276

4 Prosimetrical Narrative

they all contributed to providing Iceland with a history worthy of its poetically minded elite. Even though the plausibly pseudonymous corpus is relatively substantial, it should be remembered that it is very small in comparison to the corpus of skaldic poetry at large. Furthermore, the fact that it is possible to distinguish pseudonymous poetry from poetry that is probably authentic itself indicates that the bulk of the corpus is authentic or at least from more or less the time assigned to it. If, for instance, half or more of the poetry would have been pseudonymous, late features would probably have been so widely distributed throughout the corpus that it would have been nearly impossible to establish a plausible timeline for the evolution of individual traits. Since such a timeline is the primary tool for distinguishing pseudonymous from authentic poetry, a sifting could then probably only have been undertaken in exceptional cases. The same can be said about changes and updates in transmission generally; these no doubt occurred, but they have not been so profound as to upset the timeline, and the integrity of outdated forms generally appears to have been respected. Pseudonymous composition appears to have remained the exception, albeit not a very rare one. Its analytical importance does not lie in its representation of a typical method of the saga authors. Rather, it shows the creativity, enthusiasm and interest of the authors’ reception of tradition. Now, however, it is time to turn to a more restricted group of poets and grammarians connected to the monastery of Þingeyrar in the midfourteenth century. While they were fewer in number than the saga authors, they illustrate how the meeting of traditional poetics – now mediated by Snorri’s Edda – and Latin learning stimulated intellectual activity.

5 Poetry, Language and Snorri’s Edda in the Mid-Fourteenth Century At this point, I wish to draw on the discussion of Grettis saga in order to consider some remarkable intellectual developments in the monastic milieux of Northern Iceland. In the fourteenth century, these prominent centres of poetic studies played host to a remarkable discourse on language and poetics around Snorri’s Edda. We have already noted the exceptional output of pseudo-Grettir, who appears to have been active in or around Þingeyrar. Unlike most pseudonymous poets, he seems to have used not only Skáldskaparmál, but also Háttatal. Another important source to the interest in Snorri’s Edda is Codex Wormianus (W; AM 242 fol). This is by far the largest and most sophisticated of the Icelandic grammatical manuscripts, and it appears to have been composed at Þingeyrar around 1350.1 In support of this provenance, Jakob Benediktsson has discussed 11 manuscripts or fragments containing the hand of W and has argued convincingly that they belong in the North of Iceland.2 To these witnesses, Stefán Karlsson has added a fragment of a Latin Psalter.3 Finally, Guðbjörg Kristjánsdóttir has noted the relationship between the scribe’s, the rubricator’s and the illuminator’s hands in the Psalter fragment and in the Stjórn manuscript AM 227 fol, which suggests that we are dealing with a scriptorium of some prominence.4 Guðbjörg Kristjánsdóttir has also found an image from the same school on a page from a graduale from Höskuldsstaðir in the vicinity of Þingeyrar. It seems to have come there when it was new, since a church inventory for the year 1395 has been written on the originally blank verso page. The size of the scriptorium, the devotional content of several of the manuscripts in which the hand of W occurs and the geographical provenance of the graduale and other manuscripts, when known, suggest that the scriptorium in question was located at Þingeyrar.5

1 An extensive description of the contents and doctrines of W has been made in Cyril de Pins, ‘Hending ok kenning. Les théories linguistiques dans l’Islande médiévale (xiie–xive s.). Lecture du Codex Wormianus’ (PhD diss., Université Paris 7-Diderot, 2013). 2 ‘Catilina’ and ‘Jugurtha’ by Sallust and ‘Pharsalia’ by Lucan. In Old Norse: ‘Rómverja saga’, AM 595 a–b 4to, ed. Jakob Benediktsson. Early Icelandic Manuscripts in Facsimile, 13 (Copenhagen: Rosenkilde and Bagger, 1980), pp. 10–12. 3 ‘Saltarabrot í Svíþjóð með Stjórnarhendi’, Gripla, 5 (1982), pp. 320–22. 4 ‘Íslenskt saltarablað í Svíþjóð’, Skírnir, 157 (1983), pp. 64–73. 5 For a more detailed overview of the research on the provenance of W, see Johansson, Studier i Codex Wormianus, pp. 10–16. Notes: This chapter elaborates on my article ‘Wormianusredaktören: Språk, tro och sanning vid 1300-talets mitt’, Arkiv för nordisk filologi, 128 (2013), pp. 41–77. https://doi.org/10.1515/9783110643930-006

278

5 Poetry, Language and Snorri’s Edda in the Mid-Fourteenth Century

In W, the redactor not only includes Snorri’s Edda, but also displays much interest in it through interpolations and commentaries. W contains other indications of interest in poetry and Snorri’s Edda at Þingeyrar, as well. Most notable among these is 4GT, found only in this manuscript. It seems not to have been authored by the scribe, since there are some errors in the stanzas, but it probably cannot be older than 1319.6 This treatise is centred on poetic diction, and while most of the poetry in it seems to be of the author’s own composition, he also quotes stanzas 12 and 14 of Snorri’s Háttatal.7 Most likely, the author was a monk at Þingeyrar.8 One stanza that appears to be by the author himself has been composed in the metre alhent, that is, with four hendingar in each line – a considerable poetic feat.9 It shares this metre with the second to last stanza found in the W-redaction of ókennd heiti (the second part of Skáldskaparmál), which is said to have been composed by a bróðir Árni (brother Árni).10 Árni’s precise identity is not known, but his epithet suggests that he belonged to the monastery. These stanzas are the only known full stanzas apart from stanzas 43–44 in Snorri’s Háttatal to be composed in alhent.11 Three half stanzas in alhent have come down to us. One of these is by Snorri himself and quoted in Íslendinga saga.12 Of the two others, one is by Hofgarða-Refr, who was active in the first half of the eleventh century, and one is by Klœngr, who was bishop of Skálholt 1152–1176. These stanzas are cited in Skáldskaparmál and Háttatal, respectively.13 Apart from Snorri’s half stanza, all examples of this metre except the two stanzas in W are thus found in Snorri’s Edda, and mainly in Háttatal. Háttatal’s presence in W and the two quotations of it in 4GT attest to the prominent position of this work among the poetically inclined brethren, and it thus seems likely that the two stanzas in W are composed in emulation of Snorri’s metrical dexterity. The W-compilation as

6 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. lxiii; The Fourth Grammatical Treatise, ed. and trans. Clunies Ross and Wellendorf, pp. xi–xiv. 7 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, pp. 136–37, 272–73; The Fourth Grammatical Treatise, ed. and trans. Clunies Ross and Wellendorf, pp. 26, 108–10. 8 Apart from the general likelihood of this, his level of learning indicates that he must probably have received clerical training, and he exhibits a particular interest in allegorical interpretation (Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, pp. xliii–xliv; The Fourth Grammatical Treatise, ed. and trans. Margaret Clunies Ross and Jonas Wellendorf (London: Viking Society for Northern Research, 2014), pp. xii–xiv). 9 See Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, pp. lxxv–lxxvi, 124, 247–52; The Fourth Grammatical Treatise, ed. and trans. Clunies Ross and Wellendorf, pp. 6–8, 62–64. 10 See Edda Snorra Sturlusonar, Codex Wormianus AM 242, fol, ed. Finnur Jónsson (København and Kristiania: Gyldendal, 1924), p. 112; Skj B 2, p. 461; A 2, p. 430). 11 Snorri, Háttatal, ed. Faulkes, p. 21. 12 For Snorri’s half stanza and other occurrences of alhent, see Snorri, Háttatal, ed. Faulkes, p. 83 (Faulkes has not noted the stanza by brother Árni). 13 Snorri, Skáldskaparmál, ed. Faulkes, pp. 9–10 (17); Snorri, Háttatal, ed. Faulkes, p. 21.

5.1 The Wormianus Redactor

279

a whole, the two additional alhent stanzas quoted within it, and the two quotations from Háttatal in 4GT attest to a lively scholarly interest in poetry and Snorri’s Edda at Þingeyrar before and around middle of the fourteenth century, and Grettis saga corroborates this impression. In the following, I wish first to focus on the W-redactor’s reception of Snorri’s Edda, which is truly remarkable and innovative. Thereafter, I compare this to the reception of Snorri’s Edda by three other monastics: an abbot of Þingeyrar, an abbot of Þingeyrar’s sister monastery at Munka-Þverá and one monk of unknown provenance. As we shall see, reactions to and debates about Snorri’s Edda were as lively in the fourteenth century as they had been in the thirteenth, but they had now moved into the cloister.

5.1 The Wormianus Redactor W is completely centred on language and, above all, on poetry. Here is an overview of its contents, including some relevant material that is unique to W in the right column: . . . . . . . . . . . .

.

Prologue (SnE) Two long additions and one short addition, unique to W. Gylfaginning (SnE) Skáldskaparmál (SnE) Prologue to –GT Unique to W. GT GT One short addition in the beginning and one long in the end, unique to W. GT GT Maríukvæði Háttatal (SnE) Rígsþula Ókennd heiti (poetic synonyms = second part of Skáldskaparmál) Maríukvæði

1GT, 2GT and the first part of 3GT treat orthography and language. The remaining material either discusses poetry or consists thereof. The additional material in the right column betrays a particular interest in language and exhibits some recurrent

280

5 Poetry, Language and Snorri’s Edda in the Mid-Fourteenth Century

themes. Several factors combine to show that these texts all come from the pen of the redactor/scribe.14 To briefly review the main arguments, these are: 1. Linguistic. In the first addition to the prologue of Snorri’s Edda the words villa ((religious) error) and fjǫlgaz (multiply) are used three times each, whereas in the entire Edda, fjǫlgaz is otherwise used only once, and villa is not used at all. Both, however, are again used in the prologue to 1-4GT, which is clearly the work of the redactor, in reference to the prologue of Snorri’s Edda. 2. Ideological. Only in the first addition to the prologue of Snorri’s Edda and in the second addition to 2GT is there a focus on the division of tongues and its connection to truth.15 Furthermore, the idea of a universal phonology/script is found only in the prologue to 1–4GT and in the second addition to 2GT. 3. Graphic. External influence shows that this redactor was also the scribe. There are several descriptions of the events surrounding the Tower of Babel in Old Norse literature. Of these, one is found in the first addition to the prologue of Snorri’s Edda in W and another in the Old Norse adaptation of the Old Testament – Stjórn. These descriptions have many features in common, including, for instance, the description of Zoroaster, which is otherwise only found in one, somewhat later text.16 This portion of Stjórn also contains the embryo of the concept of a universal phonology as described in the second addition to 2GT (see below). The reason for these similarities is not long to be sought, since the hand of W is also found in two copies of Stjórn.17 The influence from Stjórn can thus be regarded as certain, and we can here discern a mind and its hand at work together. Since a number of features unique to W tie the additions together in various constellations, it is highly unlikely that the similarities are due to several stages of interpolation. There is no obvious reason why a later interpolator would favour material that had been interpolated at an earlier stage, in particular since he would probably be unable to distinguish it from the rest of the text. The prologue to 1–4GT is clearly the work of the redactor, referring both backwards and forwards in the compilation and tying its different parts together. This last observation, as well as

14 Males, ‘Wormianusredaktören’, pp. 41–53, 58. 15 The second addition in the prologue, furthermore, refers to the first. See Johansson, Studier i Codex Wormianus, p. 39. 16 AM 198 8vo, written in 1387. See Alfræði íslenzk. Islandsk encyklopædisk litteratur I. Cod. Mbr. AM. 198, 8vo, ed. Kristian Kålund (København: Samfund til udgivelse af gammel nordisk litteratur, 1908), p. 49. Apart from W and Stjórn, descriptions of the tower are found in Veraldar saga, Elucidarius, Hauksbók and AM 198 8vo (Veraldar saga, ed. Jakob Benediktsson (København: Samfund til udgivelse af gammel nordisk litteratur, 1944), pp. 14, 43; ‘Elucidarius’ in Old Norse Translation, ed. Evelyn Scherabon Firchow and Karen Grimstad (Reykjavík: Stofnun Árna Magnússonar, 1989), pp. 111–12; Hauksbók, ed. Finnur Jónsson, pp. 157–58; Alfræði íslenzk, ed. Kålund, pp. 8, 48–49). 17 Johansson, Studier i Codex Wormianus, p. 40.

5.1 The Wormianus Redactor

281

the internal similarities between the texts that are unique to W, indicate that they can all be ascribed to the redactor. This monastic redactor was passionately interested in the connection between language and truth, above all divine truth. Although he devotes some discussion to poetics in the prologue to 1–4GT, his heart seems to lie with his own remarkable blend of phonology and semiotics. His activities may thus appear somewhat tangential to a book on poetry, but his discourse is predicated not only on the phonological and orthographic analyses which he found in the grammatical treatises, but also on linguistic obscurity and paganism as represented by skaldic poetry and Snorri’s Edda. The redactor presents two visions of language – one of truth, one of falsehood – connected to 2GT and to Snorri’s Edda, respectively. We begin with 2GT and truth. The original 2GT featured two graphic figures containing letters. Both of these have been left out in W, but they are described in the text, and the redactor’s imagination more than makes up for their absence.18 One of the figures is modelled on the neck of a string instrument (the hurdy-gurdy), and in it, the author (not our redactor) has plotted Latin and Old Norse letters (see figure). This figure could in principle be extended to contain all letters/sounds in the world, and it would seem that this is what prompts the redactor to exclaim: Kallaz nú hvárir við aðra, stafróf ok gammi, ok taka nú hljóðstafir þar sín hljóð ok raddarstafir rǫdd, málstafir málit, ok safnaz til orðanna svá margra at ekki er þess mælt í heiminum at eigi sé þessir stafir til hafðir. Nú eru engi þau læti eða hljóð eða raddir at eigi muni þat allt finnaz í gammanum [. . .]19 Now they call out to each other, the alphabet and the musical scale, and the vowels (hljóðstafir) get their sound (hljóð) and the vowels (raddarstafir) get their voice (rǫdd), the consonants (málstafir) language (málit), and [the letters/sounds] are gathered into words, so many that nothing has been said in this world to which these letters cannot be used. There are no sounds or noises or voices that cannot all be found in the scale [. . .].20

The inspired redactor engages in a figura etymologica that makes little logical, but much associative sense: hljóðstafir and raddarstafir are synonyms (vowels), and the inclusion of both is probably intended to recall the progression from noise (hljóð) to inarticulate voice (rǫdd) to language (mál) that is given at the beginning of 2GT, as well as the discussion of sound (hljóð) of musical instruments immediately preceding this passage.21 He continues:

18 They are included in the other main witness to 2GT, Codex Upsaliensis, and are clearly original to the treatise (The So-Called Second Grammatical Treatise, ed. Raschellà, p. 17). 19 The So-Called Second Grammatical Treatise, ed. Raschellà, p. 45. 20 On the complications involved in this translation, see Appendix. 21 The So-Called Second Grammatical Treatise, ed. Raschellà, pp. 27–29. The opening of the treatise is replete with rhetorical triads, so that the redactor may also be picking up on the tenor of the treatise in a more general way.

282

5 Poetry, Language and Snorri’s Edda in the Mid-Fourteenth Century

This figure is found in the second manuscript containing 2GT (U), and it clearly belonged in the original treatise. Although it has been left out in W, its description was sufficient to stimulate the redactor’s outburst on phonological universalism.

Nú tekr svá fremi mikit um gjǫraz er orðin hefjaz upp ok hljómrinn vex ok raddirnar glymja á. Nefniz ok svá fremi sǫngr er þetta hefir allt til ok nú þyss sjá flokkr fram á leikvǫllinn ok ǫllum megin at stýrinu því er tungan heitir ok heita nú á hana til málsins ok orðanna ok sǫngsins at hon kveði þat allt upp, ok hon gjǫrir svá ok hneigir sik til stýrimannsins ok mælir svá: Osanna segir hon. Þat þýðiz á vára tungu svá: Græð þú oss. En þat er á ebresku mælt ok stakk hana náttúran til þess fyrir því at hon var fyrst ok gekk þá um allan heim þangat til er Guð skipti þeim. Nú segir þar til at henni þótti hann vera stýrimaðrinn er hann skapaði hana ok af Krists nafni er kristnin kǫlluð. Vér er kristnir erum kǫllum hann hǫfuð várt en vér hans limir ok liðir ok hans sonr er sá er hann sendi hingat í heim ok sá er vár faðir en vér hans bǫrn.22

22 The So-Called Second Grammatical Treatise, ed. Raschellà, pp. 45–47.

5.1 The Wormianus Redactor

283

The round figure is also found in U but not in W, but it is described in the text of the treatise and the redactor refers to it with the words ‘þessir .v. hríngar stafanna’ (these five rings of the letters; see Appendix).

Now much begins to happen, when the words are pronounced, the sound increases and the voices ring out over it all. That which contains all this is called song, and now this throng rushes forth onto the playground [that is, the mouth]23 and on all sides, towards the rudder which is called the tongue, and they [the letters/sounds] call upon it to receive language and words and song – that it should express all of this – and it does so and bows to the captain and says: Osanna! it says. That means, in our language: Heal us! And that is said in Hebrew and nature made it [the tongue] do so, because it [Hebrew] was first and was dispersed over

23 Compare earlier in 2GT: ‘muðrinn er leikvǫllr orðanna en tungan stýrit’ (the mouth is the playground of the words and the tongue is its rudder) (The So-Called Second Grammatical Treatise, ed. Raschellà, p. 29).

284

5 Poetry, Language and Snorri’s Edda in the Mid-Fourteenth Century

all of the world until God divided them [the languages]. Nu it should be said that it [the tongue] thought that he was captain, since he created it, and Christianity draws its name from the name of Christ. We who are Christian call him our head and we are his limbs and joints, and his son is the one whom he sent hither into the world, and he is our father and we his children.

After this, the redactor goes on to explain how we should place all our faith in the Son, the Father and the Passion. Then follows a description of the blessed in heaven after the Judgement: Þá skulum vér hefja upp: Alleluia! fyrir því at þat er eigi jarðneskr sǫngr. Syngja þetta þá allir saman tíu fylki guðsengla ok manna, þá er almáttigr Guð ferr meðr sína ferð heim í himinríkis dýrð [. . .] Amen.24 Then we shall proclaim: Alleluia! because that is no earthly song.25 Then all sing this together, the ten hordes of angels and men, when the almighty God travels home with his retinue to the glory of Heaven [. . .] Amen.

The redactor seems almost to be swept off his feet by his heavenly vision, but he retains a linguistic perspective throughout. The universal potential contained in the scale that can encompass all sounds leads his associations to Hebrew, the one true language. Indeed, the sum of all sounds storming out on the tongue prompts it, by nature, to speak Hebrew, and when speaking the true language, the tongue realizes the truth that God created it. The redactor’s line of thought is not easy to follow in point of detail, but perhaps that should be expected from such an outburst of theo-grammatical fervour. Nonetheless, the connection between a universal phonology, Hebrew, nature, truth and God is clear enough. The redactor probably developed these perceptions based on the description in 2GT itself and on a passage in Stjórn, with which he had been working. In the description of the division of tongues at the Tower of Babel we read in Stjórn: Petrus Comestor segir at Guð vann enga nýskapaða gjǫrð í þessu tungnaskipti, þvíat hinu sǫmu raddir ok orð eru af hinu sama efni meðr ǫllum þjóðum, en meðr ymissu kyni skildi hann ok greindi orðanna myndir ok málsháttu.26 Petrus Comestor says that God did not give rise to any new order in this division of tongues, since the same linguistic sounds and words are of the same substance among all peoples, but he divided and separated the forms and modes of expression of the words into different classes.

This is a somewhat elaborated translation of a passage in the Comestor’s Historia scholastica:

24 The So-Called Second Grammatical Treatise, ed. Raschellà, p. 47. 25 On the opposition between osanna and alleluia, see p. 319 n. 40. 26 ‘Stjórn’. Tekst etter håndskriftene, ed. Reidar Astås, 2 vols (Oslo: Riksarkivet, 2009), 1, p. 99.

5.1 The Wormianus Redactor

285

In hac diuisione nichil noui fecit Deus, quia uoces eedem sunt apud omnes gentes, sed dicendi modos et formas diuersis generibus diuisit.27 In this division [of tongues] God created nothing new, since linguistic sounds are the same among all peoples, but he divided modes and forms of expression into different kinds.

In the Historia scholastica and Stjórn, the point seems to be that the division of tongues did not constitute a second creation: that work was over with the six days, as we read in Genesis 2. 2 ‘complevit Deus die septimo opus suum quod fecerat’ (on the seventh day God ended the work which he had made). The finality of creation is stressed in, for instance, Augustine’s De Genesi ad litteram.28 In Historia scholastica and Stjórn, the retention of the original sounds led to no observations on the differing potential for existing languages to communicate truth. The W redactor takes this thought to the next step. When faced with the possibility of plotting all the sounds of the world into one system, he realizes that this will result in nothing less than a reassembly of the original language of truth, namely Hebrew.29 He sees unity and truth as two sides of the same coin: the one true language leads to truth itself, which is nothing other than the one true God. In the first addition to the prologue of the Edda, by contrast, he makes the opposing connection between multiplicity and falsehood.30 I have put his words about error and multiplicity in bold in order to graphically show their juxtaposition: [Zoroastres leads the building of the tower of Babel] [. . .] en forsmiðir váru ii ok lxx ok svá margar tungur hafa síðan dreifz um verǫldina eptir því sem risarnir skiptuz síðan til landa ok þjóðirnar fjǫlguðuz [. . .] sem tungna skiptit var orðit þá fjǫlguðuz svá nǫfnin mannanna ok annarra luta [. . .] Zoroastres [. . .] af hánum hófz skurðgoða villa ok sem hann var blótaðr var hann kallaðr Baal. Þann kǫllum vér Bel. Hann hafði ok mǫrg ǫnnur nǫfn en sem nǫfnin fiǫlguðuz, þá týndiz með því sannleikrinn. [. . .] þar til er þessi villa gekk um allan heim, ok

27 Petri Comestoris ‘Scholastica historia’. Liber Genesis, ed. Agneta Sylwan (Turnhout: Brepols, 2005), p. 76. 28 ‘Quapropter sic accepimus deum requieuisse ab omnibus operibus suis, quae fecit, ut iam nouam naturam ulterius nullam conderet, non ut ea, quae condiderat, continere et gubernare cessaret [. . .]’. (For this reason we accept that God rested from all his works that he had made, so that he did not create any new, further nature, not in order that he cease to restrain and rule what he had created [. . .]) (De Genesi ad litteram iv. 12; Augustine, Sancti Aureli Augustini de Genesi ad litteram libri duodecim, eiusdem libri capitula, de Genesi ad litteram imperfectus liber, locationum in Heptateuchum libri septem, ed. Iosephus Zycha. CSEL, 28 (3: 2) (Pragae: Bibliopola academiae litterarum Caesareae Vindobonensis, 1894), p. 110). 29 Opinions on whether the original language was Hebrew or something else varied, but most opted for Hebrew (see Arno Borst, Der Turmbau von Babel, 4, pp. 1945–47, and references there). 30 The most recent discussion of the additions to the prologue is Wellendorf, Gods and Humans in Medieval Scandinavia, pp. 100–08. In contrast to earlier research, Wellendorf stresses how the additions result in a more negative image of paganism than that contained in the original prologue. His views are supported by the present argument.

286

5 Poetry, Language and Snorri’s Edda in the Mid-Fourteenth Century

svá vandlega týndu þeir sannleiknum at engi vissi skapara sinn, útan þeir einir menn sem tǫluðu ebreska tungu þá sem gekk fyrir stǫpulsmíðina [. . .].31 [Zoroastres leads the building of the tower of Babel] [. . .] and the master builders were 72, and so many languages have since spread across the world, as the giants afterwards divided the lands between them and the peoples multiplied. [. . .] when the division of tongues had occurred, then the names of humans and other things multiplied [. . .] Zoroaster [. . .] was the origin of the error of idolatry, and when they sacrificed to him they called him Baal. We call him Bel. He had many other names besides, and as the names multiplied, so truth was lost [. . .] until that error travelled all over the world and they lost the truth so thoroughly that no one knew his Creator, apart from them alone who spoke the Hebrew language that had preceded the building of the tower [. . .].

Many languages, then, lead to many gods and thus to error, whereas the one true language leads to the one God and thus to truth. Further on in the manuscript, in the prologue to 1–4GT, the redactor refers back to the prologue of the Edda, and the juxtaposition of the words villa and fjǫlgaz betrays that he is referring specifically to his own discussion in the passage above: En eigi skulu menn þessum frásǫgnum trúa framar en skynsamligt er, eptir þí sem segir í fyrsta hlut bókarinnar,32 með hverju villurnar fjǫlguðuz, ok af þí hefir hvert skáld sett sér reglu, þat sem eigi trúði réttlega, af þí at þeir hugðu Óðin guð v[er]it33 hafa ok alla þá með guðmagni, sem hánum þjónuðu, sem heyra hefir mátt í mǫrgum frásǫgnum þessar bókar.34 But one should not believe these tales [of Snorri’s Edda] beyond what is prudent, according to what is said in the first part of the book of how the errors were multiplied, and every poet who did not believe rightly has constructed his own rule, because they thought that Óðinn was a god, and that all who served him had divine powers, as you have been able to hear in the many tales of this book.

The redactor’s view of the early poets is similar to that of Snorri in Háttatal, but expressed in a delightfully naive manner: he sees them as so many Donatuses, each setting up his rule as best he can. Since the practice of the poets was based on a fragmented and erroneous understanding of the world, they could not come up with a coherent and correct rule to guide their art. Again, multiplicity and error are synonymous, and the repercussions of this fundamental flaw play out on the level of language, specifically poetic language. These condemnations of the error of polytheism and fragmentation are obviously prompted by the mythological content of Snorri’s Edda. Even though the paganizing exposition of that work causes the redactor to explain the linguistic

31 Edda Snorra Sturlusonar, ed. Finnur Jónsson, p. 3. 32 The reference is to the prologue of Snorri’s Edda, but the wording here is a paraphrase of a statement in the beginning of Snorri’s text in Skáldskaparmál (Snorri, Skáldskaparmál, ed. Falkes, p. 5). 33 Abbreviature for er faded. 34 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 152–53.

5.1 The Wormianus Redactor

287

background of polytheism, however, this does not seem to affect his appreciation for Snorri’s Edda in other regards. In the prologue to 1–4GT, he enjoins his readers not to abandon the traditional kennings that are treated in Skáldskaparmál.35 Somewhat later he says that poets should feel free to take kennings from ‘new books’, by which he appears to mean 3GT and 4GT, which treat diction and follow later in the compilation. These kennings, however, should not be lengra reknar en Snorri lofar (expanded beyond what Snorri allows).36 The reference is to Háttatal 8, to follow later in the manuscript, where Snorri says that one may not expand a kenning beyond the fifth definer.37 The ‘new books’ in question are not any books. The relevant parts of 3GT and 4GT are translations of Donatus’s Barbarismus and Alexander de Villa Dei’s Doctrinale, and the redactor thus has Snorri win out against the two greatest authorities on diction within the Latin tradition (though not in all instances, since he also stresses that we should avoid barbarisms).38 This is an extraordinary homage under any circumstances, and it becomes more remarkable still since the redactor has earlier taken pains to point out the falsehood of the content of Snorri’s Edda. It would seem that traditional poetic diction as defined by Snorri had such a high standing to him that its connection to truth became irrelevant. This is probably also how his additions to the prologue of the Edda should be understood: they point out its falsehood, but only to scatter any possible doubts in the reader’s mind about the propriety of immersing oneself into secular poetics. His arguments thus do not indicate aversion but, on the contrary, a deep appreciation for Snorri’s legacy. As we shall see, others were not so easily won over. Before turning to Snorri’s critics, however, I present a few observations on the care with which the redactor compiled W, above all with regard to Snorri’s Edda. The relation between the prologue to 1–4GT and Háttatal is particularly telling. The prologue contains five echoes of Háttatal, of which two have been noted by earlier scholars.39 The first is verbal, although its meaning has also been coloured by the redactor’s reading of 3GT. He uses the phrase ‘einnar tíðar fall’ (one and the same accent), which is found in Háttatal 16.40 The relation between the different texts and meanings here is complex, and I refer the reader to the discussion in the Appendix. The next instance is where the redactor enjoins his reader not to retain 35 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 153. 36 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 155. 37 Edda Snorra Sturlusonar, ed. Finnur Jónsson, p. 89. The rule is given on 55r in W. 38 Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 153. The formulation is ‘þat sem klerklegar bœkr banna’ (that which is forbidden in clerical books), which can in this context probably only refer to barbarisms and solecisms (on the identification of the books and what is forbidden, see Males, ‘Wormianusredaktören’, pp. 54–55. 39 Number three and four above were noted by Finnur Jónsson (Edda Snorra Sturlusonar, ed. Finnur Jónsson, p. v). 40 Snorri, Háttatal, ed. Faulkes, p. 11.

288

5 Poetry, Language and Snorri’s Edda in the Mid-Fourteenth Century

‘hvert orðskrípi, þat sem fornskáldin nýttu’ (every verbal monstrosity, which the early poets used). The wording here seems to be based on Háttatal 8 – on kennings which extend beyond the fifth reference: ‘en þótt þat finnisk í fornskálda verka, þá látum vér þat nú ónýtt’ (and although this [too long kennings] may be found in the works of the early poets, we now avoid using it).41 The word orðskrípi (verbal monstrosity) also points to the discussion of extended metaphor in Háttatal 6. Snorri’s term is nykrat (mermanized), also invoking the image of a monster. The third echo is an explicit reference, urging the reader to avoid kennings that are ‘lengra reknar en Snorri lofar’ (expanded beyond what Snorri allows), referring to Háttatal 8. The fourth is a rephrasing of some words in the opening of Háttatal; when describing the phonological and orthographic content of the following grammatical treatises, the redactor says that one can there learn ‘hversu hverr stafr hljóðar med lǫngu hljóði eðr skǫmmu, hǫrðu eðr linu’ (how each letter sounds, with a long or a short, with a hard or a soft sound). This is a slight variation on the opening of Háttatal: ‘hljóð greinir þat at hafa samstǫfur langar eða skammar, harðar eða linar’ (sound is distinguished by having long or short, hard or soft syllables).42 The fifth echo of Háttatal comes in the redactor’s injunction to devote sufficient time and energy to the composition of poetry, since its aesthetic qualities are all-important. He says that the poets ‘leiti eptir sem vandligast [. . .] hversu fegrst er talat’ (should inquire most thoroughly [. . .] how they may speak with the greatest elegance), echoing Háttatal 16: ‘orðin munu finnask ef vandlega er leitat’ (the [right] words will be found if one inquires thoroughly).43 The reference to and paraphrases of Háttatal not only indicate the redactor’s opinion of the authority of this text, but they also show the degree of planning behind the compilation of W. The prologue to 1–4GT has several references both backwards and forwards, showing that the redactor must have had a clear idea of how the final product would look (the prologue cannot have been inserted last, since it is placed in the middle of a quire).44 The original place of the the three last items in the manuscript (Háttatal, Rígsþula and Ókennd heiti) before the book was bound in the seventeenth century is somewhat uncertain: Háttatal is contained in its own quire (missing one bifolio, and thus the beginning and end of the poem), whereas Rígsþula has its own folio and Ókennd heiti its own bifolium.45

41 Snorri, Háttatal, ed. Faulkes, p. 8. 42 Snorri, Háttatal, ed. Faulkes, p. 3. He here uses Snorri’s terminology, rather than Óláfr’s (snarpr and linr), although one may suspect that he is really referring to Óláfr’s discussion, which is fuller and less confused (see pp. 127–29; Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, pp. 39, 54). 43 Snorri, Háttatal, ed. Faulkes, p. 11. 44 See Johansson, Studier i Codex Wormianus, pp. 24–29. 45 Johansson, Studier i Codex Wormianus, pp. 59–60; Edda Snorra Sturlusonar, ed. Finnur Jónsson, pp. iii, 88–103.

5.1 The Wormianus Redactor

289

Karl G. Johansson has pointed out that Ókennd heiti contains a reference to Rígsþula and that the two in all likelihood belonged together in the original compilation.46 This also shows that Rígsþula, at least in this context, was intended as a list of heiti in poetic form.47 Furthermore, Johansson observes that the information given in the prose introduction to Rígsþula – that Rígr really was the god Heimdallr – is unique to W and appears to be an attempt at harmonizing the poem with Snorri’s Edda and thus to adapt it to the compilation as a whole.48 The reference to and paraphrases of Háttatal supports this observation; the final three items are intended for the book and not, for instance, to be used as loose quires for study. While these details do not tell us much about the placing of those items, they must have come at the end and have presumably retained their original place when the book was bound, since there is no room for them among the existing quires.49 Based on these observations, W can be divided into a theoretical part, containing the ‘old’ and ‘new’ grammatical treatises (the Edda and 1–4GT), and a practical part, containing lists of metres and poetic synonyms (Háttatal, Rígsþula, Ókennd heiti). Although in other manuscripts, Ókennd heiti is the second part of Skáldskaparmál, it seems that the lists of names contained within it prompted the redactor to place it in the end. W shares this structural division into treatises followed by lists and examples with other Icelandic grammatical manuscripts, as well as with Snorri’s Edda itself (see pp. 109–10). The references in the prologue to 1–4GT show that the redactor had conceived of this structure at an early stage (he refers to Snorri’s Edda, 2GT, 3GT and Háttatal). It thus becomes clear just how much preparation and thought has been put into the compilation beforehand in order to turn it into a full and user-friendly curriculum of Old Icelandic poetics.50 The redactor’s heart-felt concern for poetics and Snorri’s Edda appears to have provoked thoughts about linguistic obscurity and paganism, prompting him to seek ways of circumventing these errors. When Stjórn and 2GT offered such an opportunity, he eagerly grasped at it. The uniqueness of the theory that resulted, which conceived of a universal phonology of all languages that could be a route to orthodoxy, is predicated at least partly on the particular cultural setting in which Snorri’s poetics stood at the centre of learned attention. This is not to say that the idea of a universal language was unique – on the contrary, it was mainstream. What lacks clear parallels is the redactor’s description of the recovery of this language. The redactor’s thoughts are more directly indebted

46 See Karl G. Johansson, ‘Rígsþula och Codex Wormianus: Textens funktion ur ett kompilationsperspektiv’, Alvíssmál, 8 (1998), p. 71. 47 Johansson, ‘Rígsþula’, p. 75 48 Johansson, ‘Rígsþula’, p. 78–79. 49 See Johansson, Studier i Codex Wormianus, pp. 9, 24–29; Johansson, ‘Rígsþula’, p. 72. 50 Cf. Johansson, Studier i Codex Wormianus, pp. 61, 66.

290

5 Poetry, Language and Snorri’s Edda in the Mid-Fourteenth Century

to 2GT than to Snorri’s Edda, since the figure of a musical instrument as a matrix for phonological representation, as found in 2GT, is in itself a unique feature. An argument based on that figure was thus all but bound to deviate from commoner intellectual templates. The redactor’s thoughts on the connection between multiplicity and error, however, were deeply influenced by Snorri’s Edda. These thoughts stimulated him to seek out the way to singularity and truth, which he found in 2GT. W as a whole is incomparably the most exhaustive witness to monastic interest in Snorri’s Edda, poetics and language in the fourteenth century. There is, however, an inner tension in the redactor’s output which is left unresolved. In his additions to the prologue of Snorri’s Edda, he takes pains to describe the fallacies of language and their dire consequences, and in the addition to 2GT, his joy at the recovery of the one true language knows no bounds. Nonetheless, he enjoins his readers to follow the authority of Snorri and his poetic language based on lies and obscurity. The explanation of what appears to be a flat contradiction may be that he saw no interference between these two linguistic registers, since he did not consider traditional poetics to be a vector for divine truth. It is easy to see, however, how one could arrive at other conclusions and be tempted to jettison Snorri altogether within a monastic setting.

5.2 Monastic Poets and Snorri’s Edda Three poets of the mid-fourteenth century opted for this latter solution. Their reaction is at the same time both European and staunchly local, since their rhetoric is indebted to the long-standing tension between clarity and obscurity in the rhetorical and exegetical traditions, while their arguments and solutions emerge from a poetic cosmos where the centre of gravity was Snorri’s Edda.51 Two of these poets can be located in the Benedictine milieux of northern Iceland, but the one likely to have begun the debate cannot be located with certainty. To the degree that the late sources about him mention his affiliation, they say that he was a monk either in southern Iceland or in Niðaróss, Norway.52 Several sources agree that his name was Eysteinn, and his one preserved poem – the envy of all Icelandic poets – is called Lilja (lily, referring to the Virgin).53 Whatever Eysteinn’s precise affiliation may have been, his reference to Snorri’s Edda presupposes an Icelandic audience, and he seems to have had some contacts with Benedictine milieux in the North since, as we shall see, two of his

51 On obscurity in the rhetorical and exegetical traditions, see Jan M. Ziolkowski, ‘Theories of Obscurity’. 52 SkP 7, p. 554. 53 Cf. the statement of bishop Finnur Jónsson in 1774, since turned into an Icelandic saying: ‘Öll skáld vildu Lilju kveðið hafa’ (all poets wished that they had composed Lilja) (SkP 7, p. 561).

5.2 Monastic Poets and Snorri’s Edda

291

younger contemporaries there picked up on his rhetoric. The statement of his poetic programme is found in stanzas 97 and 98 of Lilja: Veri kátar nú virða sveitir; vætti þess, í kvæðis hætti várkynni, að verka þenna vanda eg minnr, en þætti standa. Varðar mest, að allra orða undirstaðan sie riettlig fundin, eigi glögg þó at Eddu regla undan hljóti at víkja stundum. Sá, er óðinn skal vandan velja, velr svá mörg í kvæði að selja hulin fornyrðin; trautt má telja; tel eg þenna svá skilning dvelja. Vel því að hier má skýr orð skilja, skili þjóðir minn ljósan vilja; tal óbreytiligt veitt að vilja; vil ek, að kvæðið heiti Lilja.54 May the throngs of men now rejoice. With regard to the form of the poem, I expect forgiveness for embellishing this work less then would seem appropriate. It is most important that the correct meaning of all words is found [by the reader/listener], even though the obscure rules of the Edda [that is, the obscure meaning that results from following them] must sometimes give way. He who will choose [words for] an artful poem chooses to convey so many hidden archaic words [that is, which hide the meaning] that they can hardly be counted. I think that such a choice [ þenna for þenna kost] inhibits understanding. Since it here easy to understand the clear words, may people understand my manifest wish – unchanging speech [that is, having only one meaning], willingly performed; I wish that the poem shall be called Lilja.

To one who has not struggled to find salvation under the poetic yoke of Snorri’s Edda, the novelty of these stanzas, and of Lilja generally, may not be immediately obvious. But to medieval Icelanders, no matter their degree of personal piety, it must have come as a shock: there are no kennings here. 55 No known poet within the entire skaldic tradition had ever composed like this, but from now on, the Lilja style would become a possibility within the pious register. Such simplicity can occasionally be found in an anonymous couplet in a saga or on a runic stick, but was inconceivable in a poem on the stylistic order of

54 SkP 7, pp. 672–75. 55 There are many metaphors in the poem (‘fountain of mercy’ etc.) which are, from a theoretical point of view, equivalent to kennings. They are not, however, contained within the traditional kenning system and they seem to have been perceived as something other than kennings.

292

5 Poetry, Language and Snorri’s Edda in the Mid-Fourteenth Century

Lilja, which is highly elaborate even for being a drápa.56 This unprecedented step was not a thing to be taken lightly. To present a viable alternative to Snorri’s Edda, Eysteinn had to show that there were other ways to reach the hights of poetic excellence, and he mustered the greatest authorities of Latinity and the monastic tradition to plead his case. The rhetorical edge of the second stanza is subtle: it is composed in a dunhent variety of hrynhent which, in addition to the ordinary phonetic features of dróttkvætt, has the second hending of odd lines forming an aðalhending to the first hending of even lines, and the stem of the words is identical (velja: velr, telja: tel, etc).57 Furthermore, the stanza has end-ryme in the scheme aaaa bbbb.58 This plea for simplicity is thus borne up by a degree of metrical complexity with few counterparts in Old Norse poetry. While Eysteinn here demonstrates how metrics, rather than kennings, could allow the skill of the poet to come to the fore without obfuscating the meaning, another feature is even more telling of the care with which he had to tread when questioning the authority of Snorri’s Edda. The words ‘at Eddu regla undan hljóti at víkja stundum’ (though the obscure rules of the Edda must sometimes give way) contain an allusion to some famous and often quoted words in the prologue to a widespread exegetical classic – Gregory’s Moralia in Iob: ‘[. . .] quia indignum vehementer existimo, ut verba caelestis oraculi restringam sub regulis Donati’59 ([. . .] for I consider it highly inappropriate if I should fetter the words of the heavenly oracle under the rules of Donatus).60

56 Even the number of stanzas – one hundred – has been carefully chosen (cf. The Divine Comedy and Decameron). 57 Lilja as a whole is composed in hrynhenda, a variety of dróttkvætt with two extra syllables per line. 58 Foote, ‘Latin Rhetoric’, p. 266 59 Gregory the Great, Gregorius Magnus, Moralia in Iob, ed. M. Adriaen. Corpus Christianorum. Series Latina, 143 (Turnhout: Brepols, 1979), p. 7. 60 Moralia, unlike parts of Homiliae and Dialogi, seem not to have been translated into Old Norse. Excerpts are found in AM 671 4to (fourteenth century) and Moralia were among the books at the episcopal see at Hólar (Kirsten Wolf, ‘Gregory’s Influence on Old Norse-Icelandic Religious Literature’, in Rome and the North. The Early Reception of Gregory the Great in Germanic Europe, ed. Rolf H. Bremmer Jr, Cornelis Dekker, David F. Johnson (Paris: Peeters, 2001), p. 269). The passage is quoted by, for instance, Gratian (Decretum 1, Distinctio 38) and twice by Abelard (Theologia christiana 3,126; 4,77; further examples can be collected through a search on sub regulis Donati in the databases Patrologia latina and Library of Latin Texts; se also discussion in Foote, ‘Latin Rhetoric’, p. 266). Moralia is mentioned in Gregorius saga: ‘Síðan gǫrði Gregorius bók þá, er menn kalla Moralia Iob, ok skýrði í hálva fjórða tigi bóka ǫll tákn, þau er hann þóttisk finna at fólginn væri í bók Iobs.’ (Heilagra manna søgur, ed. Unger, 1, p. 386. Other books by Gregory are mentioned on p. 390) (Then Gregory made the book that people call Moralia Iob, and in 35 books he elucidated all the signs that he thought were hidden in the Book of Job).

5.2 Monastic Poets and Snorri’s Edda

293

The signal word here is regla, which is not otherwise used in Old Icelandic poetics, and which points to the word regula in Gregory’s phrase.61 The paraphrase of an authority like Gregory shows us just how much was at stake here and, furthermore, the argument presupposes that Snorri’s Edda had a standing in vernacular poetics equal to that of Donatus’s Barbarismus in Latin schooling. This bears comparison to the W-redactor who, as we have seen, thought that the Icelandic translation of Barbarismus could be used to complement, but in no way to replace, Snorri’s Edda. In his innovative and skilful break with tradition, then, the Lilja poet ends up confirming just how important Snorri’s Edda was in his own time. It has been convincingly shown that the Lilja poet was inspired to his thoughts on clarity by a passage in Poetria nova (the new poetics) of Geoffrey of Vinsauf, an immensely popular work of the late twelfth century which partly complemented and partly replaced the Poetria vetus (the old poetics, that is, Horace’s Ars poetica).62 The Lilja poet’s dependence on Poetria nova is shown most clearly by stanzas 65–66, which are a translation of Poetria nova ll. 1139–50.63 The passage on clarity comes slightly earlier (ll. 1074–80), and the parallels to Lilja are found in the description of inaccessible diction versus ordinary speech: Poetria nova has ‘abdita verba [. . .] obscuris verbis [. . .] socialis eloquii’ (hidden words [. . .] obscure words [. . .] social speech), whereas Lilja has ‘eigi glögg regla [. . .] hulin fornyrðin [. . .] skili þjóðir tal óbreytiligt’ (obscure rules [. . .] hidden archaic words [. . .] may people understand my unchanging speech). In socialis and skili þjóðir a focus on intelligibility and social behavior converges. In light of the translation from Poetria nova in stanzas 65–66, these parallels must probably be due to direct influence.64 Eysteinn thus draws both on the role-model of European monasticism and on the main contemporary authority of Latin poetics to plead his case. There is, however, one crucial difference between the relevant passage in Poetria nova and its counterpart in Lilja: in Poetria nova, the injunction to simplicity and clarity is of a general nature, but the Lilja poet picks up on it as an asset in his revolt against the obscurity of Snorri’s Edda. This is not unlike the W-redactor’s treatment of his sources: he sought inspiration to his thoughts on the one true language in Stjórn and 2GT, but expressed these influences in relation to Snorri’s Edda and the image of traditional poetics presented there. For both of these monks, the study of Snorri’s Edda led to a focus on semantics that strongly influenced their reading of

61 The one instance that I have been able to find where regla is used in a linguistic, albeit not a poetic, setting is in the translation of Donatus’s Barbarismus (Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, p. 72). See Fritzner, Ordbog, 3, s.v.; LP, s.v.; ONP: http://dataonp.ad. sc.ku.dk/wordlist_d.html. 62 Foote, ‘Latin Rhetoric’, pp. 259–68; Friis-Jensen, ‘The Reception of Horace in the Middle Ages’, pp. 300–302. 63 Foote, ’Latin Rhetoric’, pp. 260–61. 64 Foote, ‘Latin Rhetoric’, pp. 265–68, though I have added some parallels to the ones he adduces.

294

5 Poetry, Language and Snorri’s Edda in the Mid-Fourteenth Century

literature in the Latin tradition, and which made them seek out arguments and theories that could be used when they positioned themselves in relation to Snorri’s Edda. In spite of the fact that one of them chose to embrace, the other to refute, the authority of Snorri’s Edda, their strategies for doing either were much the same. The next poem – Guðmundarkvæði by the abbot Arngrímr Brandsson of Þingeyrar – can be dated to 1345.65 It was composed in honour of Guðmundr, bishop of Hólar (d. 1237). It belongs to the same milieu and roughly the same time as W. It has clearly been influenced by Lilja, as may be seen from the stanza below and some other passages.66 Lilja, in turn, displays fourteenth-century linguistic forms and has several religious references that were new to that century, and may thus be tentatively dated to c. 1330–1345.67 In Guðmundarkvæði, the poet’s relation to Snorri’s Edda is expressed through a topos of humility (stanza 2): Rædda ek lítt við reglur Eddu ráðin mín, ok kvað ek sem bráðast vísur þær, er vil ek ei hrósa, verkinn erat sjá mjúkr í kverkum; stirða hefir ek ár til orða, ekki má af slíku þekkjaz, arnar leir hefik yðr at færa, emka ek fróðr hjá skáldum góðum.68

65 In stanza 49.1–4 the poet says that 108 years have passed since the death of Guðmundr (1237). On Arngrímr, see Karl G. Johansson, ‘Texter i rörelse. Översättning, original textproduktion och tradering på norra Island 1150–1400ʹ, in Übersetzen im skandinavischen Mittelalter, ed. Vera Johanterwage and Stephanie Würth (Wien: Fassbaender, 2007), p. 86. In spite of its name the poem is actually a drápa (stef in stanzas 22, 26, 30, 34, 38, 42). 66 See Hans Schottmann, Untersuchungen zur volksprachigen Mariendichtung des Mittelalters, (München: Wilhelm Fink, 1973), p. 248. Several common phrases show that one of the poems must have affected the other. Schottmann points out that some of these are suited to the description of God, but not of Guðmundr, who is the object of Guðmundarkvæði. Most striking is the phrase ‘senn verandi úti ok inni/uppi och niðri ok þar í miðju’ (who is, at the same time, outside and inside / above and below and in between [about God]) (Lilja 1.5–6). The second line occurs also in Guðmundarkvæði: ‘sóttir græðir allra átta / uppi ok niðri ok þar í miðju’ (you heal the diseases of all kin groups / above and below and in between [about Guðmundr]) (Guðmundarkvæði 59.1–2). When used of Guðmundr, the phrase uppi och niðri ok þar í miðju makes sense as a literary borrowing, but since one of these poems must be the giver, it seems likely that it is the one which stresses the omnipresence of God, rather than of the would-be saint. The same can be said about Guðmundarkvæði 56 and Lilja 51.7. In Guðmundarkvæði 56 nature and the entire creation are forced to bow to Guðmundr’s will; 56.7 skepnan ǫll verðr skyld at halda is a slight paraphrase of Lilja 51.7 skepnan ǫll er skyld at falla, where the entire creation kneels to God. Again, omnipotence is likely to first have been stressed with reference to God. Lilja is thus in all likelihood older than Guðmundarkvæði. 67 SkP 7, p. 555. 68 Skj B 2, p 372; A 2, pp. 348–49.

5.2 Monastic Poets and Snorri’s Edda

295

My deliberations were hardly formed in accordance with the rules of the Edda, and I hastily proclaimed these stanzas, which I will not praise: this poem is not soft on the throat. My oar of words [tongue] is stiff, such things merit no acclaim. I have only eagle’s clay [bad poetry] to offer. I am ignorant among skilful poets.

Arngrímr’s rhetoric is as elegant as the Lilja poet’s: in Guðmundarkvæði, unlike Lilja, some kennings are used, but they are all relatively simple. The one blatant exception is the kenning ‘eagle’s clay [bad poetry]’ in the stanza above, which requires specific mythological knowledge. That the necessary knowledge is found in Snorri’s Edda seems to be the rhetorical point of the kenning (see above pp. 160–61). Arngrímr knows his Edda and seems keen to make sure that his listeners are aware of this, so that no one can accuse him of abandoning its stylistics for lack of poetic skill. The underlying message of this stanza is thus of a less humble nature than it would first appear, and Arngrímr is not about to let his traditionalizing brethren, including perhaps the W-redactor, invoke their secular poetics to criticize him. The third protagonist in this poetic debate was Árni Jónsson, abbot of MunkaÞverá, a Benedictine monastery with close contacts to Þingeyrar. Since he died after 1379, his floruit is likely to have been after 1345, and his Guðmundardrápa is thus probably composed last of the three.69 This poem also treats the sanctity of Guðmundr and its message is reminiscent of that of Lilja (stanza 78): Yfirmeisturum mun Eddu listar allstirður sjá hróður virðaz þeim er vilja svá grafa ok geyma grein klókasta fræðibóka; lofi heilagra líz mér hæfa ljós ritninga sætra vitni, en kenningar auka mǫnnum engan styrk en70 fagnað myrkva.71 The foremost masters of the eddic art, who wish to bury and hide the wisdom of the most profound books of instruction, are likely to find this poem very stiff. I find the light of the writings of wonderful witnesses apt to the praise of the holy ones, while kennings lend us no support, but rather dims our joy.72

Here, kenning are identified as the problem with Snorri’s Edda, because they bar the way to true happiness. Neither Árngrímr nor Árni, however, follow their own programme as consistently as the Lilja poet, since their poems do contain

69 See Páll Eggert Ólason, Íslenzkar æviskrár frá landnámstímum til ársloka 1940, 1 (Reykjavík: Hið íslenzka bókmenntafélag, 1948), p. 52. 70 Emendation. Manuscripts have ‘nie’; ‘er’; ‘at’ (Skj A 2, p. 429). 71 Emendation. Manuscripts have ‘myrkra’; ‘myrknar’ (Skj A 2, p. 429). 72 The last clause is based on emended text. In any event, it is clear that darkness and kennings are connected here.

296

5 Poetry, Language and Snorri’s Edda in the Mid-Fourteenth Century

conventional kennings, albeit relatively few, simple ones. This indicates how deeply entrenched traditional diction was among the learned in Iceland, even to those who tried to liberate themselves from it. Presumably for this reason also, these poets’ new mode of clarity proposed did not generally win out, although many later religious poets exploited it.73 In general, even after the relative decline of skaldic poetry and the rise of the rímur, mythological and other kennings were retained, and Snorri’s Edda remained in use.74 This should not be construed as a failure on the side of the monastic poets, since they probably did not mean to question the authority of Snorri’s Edda for secular poetics, just as Gregory, one would assume, would not find fault with Donatus for the interpretation or composition of texts that were not inspired. They asked only for freedom to worship God without the mediation of Snorri’s Edda.

5.3 Summary To conclude, Grettis saga, the stanzas by bróðir Árni and the Fourth Grammarian, as well as the texts by the W-redactor, represent a strong pro-Edda position in or around the northern Benedictine milieux, to the point of ranking the authority of Snorri’s Edda above that of Donatus (prologue to 1–4GT). This extreme reverence for Snorri’s Edda within the ranks of their own order may explain why the three poets discussed above reacted so strongly against the need to follow the precepts of this secular text. Furthermore, all of these authors and poets collectively show that even though the literary landscape had changed by the fourteenth century and its centre of gravity had shifted from chieftains and their priests to the cloister, literary debates were still lively and remained focused on poetry and Snorri’s Edda. The force of the poetic tradition was far from spent and continued to serve as a catalyst, transforming European learning into forms that served to bolster the poetic past and which were helpful for meditating upon the properties of poetic language or even for revolting against that very poetic tradition.

73 Foote, ‘Latin Rhetoric’, pp. 268–69. 74 Björn K. Þórólfsson, Rímur fyrir 1600 (Kaupmannahöfn: Hið íslenzka fræðafélag, 1934), 88–90, 142; Sverrir Tómasson, ‘Nýsköpun eða endurtekning? Íslensk skáldmennt og Snorra Edda fram til 1609ʹ, in Guðamjöður og arnarleir. Safn ritgerða um eddulist, ed. Sverrir Tómasson (Reykjavík: Háskólaútgáfan, 1996), 1–64; Margrét Eggertsdóttir, ‘Eddulist og barokk í íslenskum kveðskap’.

6 Conclusions In this book, I have treated the portions of Old Icelandic literature which can, to borrow a phrase from Anthony Faulkes, be described as ‘totally Icelandic’.1 The question I have addressed may be expressed as: What were the crucial ingredients in the development of a characteristically Icelandic literature? In its shortest form the answer has turned out to be: Norse poetry and Latin methodology. Of these, the latter generally remains the least visible to the naked eye. This book’s focus on poetry offers an alternative to a long-standing debate within Old Norse scholarship about the influence of foreign models on native literature, and more precisely the influence of hagiography on the style and content of saga narrative.2 In Gabriel Turville-Petre’s famous words, ‘the learned literature did not teach the Icelanders what to think or what to say, but it taught them how to say it’.3 Subsequent scholarship has qualified all of these claims, suggesting that learned literature did sometimes teach Icelanders what to think and say, but they would often insist on saying it differently.4 While these studies may tell us something about the degree of European influence on Old Icelandic literature, they tend not to reveal the methods of the authors. Attention to the many formal features of poetry, by contrast, allows us to discern how the authors worked, and since poetry and the discourse around it were essential parts of the evolution of much of Old Icelandic literature, the findings are relevant far beyond the confines of language bound by metre. Chapters One and Two describe poetic developments preceding Snorri’s Edda and saga writing. In the twelfth century a grammatical mindset began to affect poetic composition. The new poets systematically analysed and imitated various elements of early metrics, as well as intricate features of early skaldic diction, such as specific mythological references, wordplay, extended metaphor and tmesis. The style that resulted differs somewhat from that of the early poets, since these features were now used systematically. They were no longer transmitted exclusively by practice, but were extracted and reproduced through normative and analytic

1 Faulkes, ‘The Sources of Skáldskaparmál’, p. 75. 2 The starting point for the modern debate was Nygaard, ‘Den lærde stil i den norrøne prosa’ (1896). 3 Turville-Petre, Origins of Icelandic Literature, p. 142. 4 See, for instance, Carl Phelpstead, Holy Vikings. Saints’ Lives in the Icelandic Kings’ Sagas (Tempe: ACMRS, 2007); Bagge, Sverre, ‘The Old Norse Kings’ Sagas and European Latin Historiography’; Grønlie, ‘Saint’s Life and Saga Narrative’; eadem, The Saint and the Saga Hero; Peter Foote, ‘Saints’ Lives and Sagas’, in Saints and sagas. A Symposium, ed. Hans Bekker-Nielsen and Birte Carlé (Odense: Odense Universitetsforlag, 1994), pp. 5–26; Jónas Kristjánsson, ‘Learned Style or Saga Style?’, in Speculum Norroenum: Norse Studies in Memory of Gabriel Turville-Petre, ed. Ursula Dronke et al. (Odense: Odense University Press, 1981), 260–92. https://doi.org/10.1515/9783110643930-007

298

6 Conclusions

study. The twelfth century also saw the first attempts at composing something like poetic treatises, namely Háttalykill and Litla Skálda (the latter text may possibly date as late as the beginning of the thirteenth century). In Chapter Three, I discuss the impact of these works and of the preceding poetic developments on Snorri’s Edda. Litla Skálda has not previously been recognized as a source for Snorri, and its identification contributes to our overall understanding of the evolution of Icelandic grammatica. In this new analysis, Snorri’s Edda does not come off as quite as unprecedented it is often thought to be. Earlier poetic trends shed some light on the background to the mythological focus of Snorri’s Edda and to the mythological revival of the thirteenth century in general – we are dealing not so much with antiquarianism in a general sense as with mythology as a corollary to grammatical studies. This mythologizing tendency gradually gained force. It was of no great concern to the author of Litla Skálda, but with Snorri’s Edda, it occupied the centre stage of poetological discourse. In Chapter Three I also discuss other branches of grammatical literature and their relation to Snorri’s Edda. The contents of Icelandic grammatical manuscripts show that eddic poetry, including the famous CR, should also be included in the grammatical corpus. I note that Gylfaginning is our most informative source on the reception of eddic poetry and that the first four poems in CR form the backbone of Gylfaginning. This overlap suggests that Snorri’s Edda may carry a considerable part of the responsibility for the collection of eddic poetry, at least in the form that it has come down to us. According to my analysis in Chapter Three, Snorri comes across as a more reliable transmitter of lore than he is sometimes credited with having been. In Skáldskaparmál, he appears to have added details to pre-existent motifs, but not much more. In Gylfaginning, he added a Christian dimension to the interpretation of pre-Christian poetry. This, however, seems mostly to have amounted to shifting the interpretive angle, and by far his most invasive act was that of discovering a flood in a stanza that mentions nothing of the sort. His treatment of unknown sources and traditions must remain unclear, and here as elsewhere in the book I have attempted to base my analyses on criteria that are as measurable and falsifiable as my topic will allow. This is also reflected in the Index where, for instance, dating criteria are classified as weak, strong and absolute. After discussing Snorri’s Edda and eddic poetry, I turn to the so-called grammatical treatises, which are more clearly indebted to the Latin tradition but still display considerable intellectual independence. Again, the causes for their relative autonomy are largely to be sought within the poetic tradition. In 1GT, the successful description of Icelandic phonology seems to be based on a highly developed sensitivity to rhyme, and the author’s final argument in support of his own orthography, if nothing else helps, is to call on the authority of the poets. 3GT is a translation of Priscian and Donatus, and as such one might expect the space for independent movement to be curtailed. This, however, turns out not to be the case. Its author

6 Conclusions

299

Óláfr observes that Icelandic poetry is the product of Trojan nobles, and whereas the Latins had to be content with translations, Icelanders were in possession of the originals. The implicit argument seems to be that adapting Donatus to skaldic poetry really amounted to restoring the work to the form it ought initially to have had, and as for Priscian, his Latin letters could easily be replaced with runes, which were in any event more pristine than anything the Romans had to offer. For Icelandic grammatica, the shift from compositions in a nativizing style without loan words or clearly Latin organizing principles in 1GT, Litla Skálda and Snorri’s Edda to translated compositions with retained Latin terminology in 3GT was drastic. Yet just as I argue that Snorri’s Edda had deeper historical roots than previous scholarship has shown, so 3GT was preceded by the hybrid 5GT, which treated concepts belonging to Latin grammatica but without using loan words or adhering to Latin organizing principles. Over some two or three generations, the nativizing way of doing grammatica had grown stronger, and turning away from it appears to have taken some intellectual effort. 5GT bears witness to this process, particularly since it can be shown to have been known to and perhaps composed by Óláfr, author of 3GT. 3GT and 4GT represent an anomaly among Icelandic translations. The material on poetics in both treatises is much longer and more complex than in their Latin originals, whereas other translations into Icelandic simplify and abbreviate, or never go beyond the literal. These are the only two translations of works on poetics into Icelandic, and they expand and develop the contents of their originals precisely with regard to poetic language. Their deviant mode of translation bears strong witness to the interference of the poetic tradition in the process of cultural and textual diffusion. In Chapter Four, I treat the prosimetrical form. Skaldic prosimetra are found in the two genres of Icelandic historical narrative that have attracted most admiration by posterity: kings’ sagas and sagas of Icelanders. Several factors indicate that this form was not taken over from oral tradition in a straightforward way. For one thing, a comprehensive study of datable texts suggests that in the twelfth century, prose and skaldic poetry were generally kept apart and were only gradually integrated into a prosimetrical form in the decades around 1200. For another, it is clear that in many cases, long poems have been broken up and inserted into the prose. Furthermore, substantial numbers of stanzas can be shown to have been composed for inclusion in the sagas, indicating that the form, in these cases at least, was not pre-existent. The functions of poetic quotation differ considerably from what one finds in Latin prosimetra. Only in Irish sagas does one find a similar distribution of functions. This indicates that the form had roots in a shared Irish-Norse mode of narration, dating back to at least the ninth century. Since the runes were not used for longer texts before the advent of book culture, this mode must have been exclusively oral in the Norse world before c. 1200. The gradual development of skaldic

300

6 Conclusions

prosimetra takes place only in the decades around 1200, and a number of sources indicate that the oral transmission of skaldic poetry could take the format of a minimal frame holding a group of stanzas together, but nothing like the later saga narrative. The oral precursors of the prosimetrical sagas are therefore highly unlikely to have been skaldic prosimetra. This leaves us with eddic prosimetra, that is, fornaldarsögur, and this hypothesis is corroborated by the fact that the only Latin text with a distribution of poetic functions similar to that of Irish and Icelandic sagas is Saxo, who drew on stories which we recognize as fornaldarsögur. These thus appear to have a long prehistory and to have conditioned the emergence of skaldic prosimetra in the decades around 1200. While the bulk of the poetry included in the sagas appears to be authentic or at least old, the limited pseudonymous corpus allows us to outline how later authors of prosimetra worked. In cases where stylistic demands outran what local tradition could supply, authors produced what was missing with the methodologies of the classroom. Pseudo-Egill (here identified as Snorri), Pseudo-Ragnarr and others made systematic observations on early metre. This activity was predicated on their acquaintance with the modes of analysis of the schoolroom, which provided them with a means for communication with the past, but which also predisposed them to posit a dichotomy between rules and their absence which in various ways resulted in a misrepresentation of early poetics. Others, including Pseudo-Grettir, made use of Snorri’s Edda – the canonical expression of native grammatica – to make their heroes compose with diction worthy of Icelandic warrior poets. The influence of Latin learning on native poetics was all-important for the development of the prosimetrical form of the Icelandic saga, but due to the tradition of oral eddic prosimetra and the tenacity of local metre and poetic language, its form remained distinctly un-Latin, indeed ‘totally Icelandic’. An important corollary to the study of pseudonymous poetry is that the very possibility of identifying such poetry suggests that most skaldic poetry is authentic or at least from more or less the time assigned to it. If that were not the case, late features would probably have been too common across the corpus for a plausible timeline to be established. A similar observation can be made regarding changes in transmission generally. While variant readings show us beyond a doubt that changes occurred in transmission, they have not upset the timeline, and it would thus appear that the transmission of skaldic poetry, including of such poetry that was composed with outdated metrical patterns, was remarkably faithful. The focus on pseudonymous poetry should thus not be taken to suggest that historical poetry was typically fabricated. Pseudo-Grettir appears to have been active in the first half of the fourteenth century, in or around the monastery at Þingeyrar in northern Iceland. In Chapter Five, I analyse the intellectual activities of those milieux, which remained closely tied to poetics in general and to Snorri’s Edda in particular. One attractive but little studied medieval Icelandic intellectual was the monk who redacted and transcribed

6 Conclusions

301

W – by far the richest of the grammatical compilations. Snorri’s mythological exposition prompted this monk to muse over the division of tongues at the Tower of Babel and the the rise of paganism out of the resulting incomprehension. Thus far, his thoughts belonged to the mainstream of European thinking on linguistic diversity and religious error, even if the connection of these to native poetics did not. When confronted with Snorri’s Edda in conjunction with 2GT, however, the redactor’s preoccupation with poetics and paganism incited him to present a truly original thought, namely that the construction of a universal phonology for all languages should somehow give us the truth of Hebrew back and lead the way to salvation. His argument is associative and incoherent but makes it clear that the interface of traditional poetics and Latin learning abounded in creativity. Snorri’s Edda, CR and the sagas are probably what most people would single out if asked to name Iceland’s most original contributions to Western literature. In this book, I have attempted to demonstrate that much of what distinguishes them is connected to the use of poetry, but I have also tried to show how the use of poetry changed over time. When Icelandic authors availed themselves of the resources of European learning, the poetic tradition proved a powerful tool for turning the various genres of Latin literature into something unmistakably their own. Once bolstered by grammatical methodology and analysis, the poetic tradition became just what these authors needed to construct their own counterpart to the classics of Latin literature. Like the Latins, Icelanders also had poets who laid the foundations for all literature to come, drawing on a heritage which was every bit as illustrious as that of Rome. And glory was not all that history had to offer – the lives and words of the poets were also riddles to be contemplated, discussed and possibly solved by their descendants. On this distant island, the wit and flare of the poetic past shot a ray of playful light through the endless winter gloom, and while the body fed on sourmilk and dried fish, the mind was invited to a feast of intellectual sensations.

Appendix: Texts by the Wormianus Redactor In the book, I have used normalized spelling of Old Icelandic. Below, I give the manuscript reading of relevant texts by the Wormianus redactor with translation and commentary.1 I have opted for a close translation in order to give an impression of the associative and somewhat incoherent style of the redactor.

1 Some material, notably the second addition in the prologue of Snorri’s Edda (Edda Snorra Sturlusonar, Codex Wormianus, ed. Finnur Jónsson, p. 4.18 (‘Leíkr þat’) – 6.38 (‘dæmí epter’)) and the beginning of the addition to 2GT (Rachellà p. 43 (until ‘Nu uerðr’)), has been left out, since it is of limited importance for the argument of this book. https://doi.org/10.1515/9783110643930-008

304

Appendix: Texts by the Wormianus Redactor

First addition to the Prologue of Snorri’s Edda. W pp. 3. 3 – 4. 9 (Þat uissu þeir æigi huar rriki hans uar. enn þui truðu þeir at hann reð ollum lutum a iorðu ok í loptí himins ok himintunglanna siouar ok ueðranna. Enn til þess at helldr mætti fra segía eða í minní festa þa gafu þeir nofn með sialfum ser ok hefer þessi atrunaðr a marga lund bræyst. sua sem þioðernar skiptuz ok tungurnar greinduz) uar efnat þat smíðí er mest hefer uerit i u[er]oldunní. I ellí sinni skipter noe heímínum með sonum sínum. ætlaði hann chám uestr haálfu Enn iapheth norðr haalfu. Enn sém suðr haalfu með þeiri parteran sem siðar greiner i þriðiunga skipti heimsins. I þann tíma sem syner þessara manna uoru í uerolldinni þa ǫflgaðiz þegar agirní til fiar ok mannmetnaðar af þui at þa kenduz þær margar lister sem aðr hofðu ekkí fundnar uerít. ok hof sik huerr epter sinní iþrott. ok sua langt færðu þeir framm sinn metnat at þeir Affricaní komner af chám heriuðu i þann luta ueralldar sem bygði afspringr sems frænda þeira. ok sem þeir hofðu þa yfer unnít þotti þeim ser æigi nægiaz heímrenn ok smiðuðu æinn stopul með tigl ok gríot þann er þeir ætluðu at uínna skylldi til hímíns a þeim uelli er kallaz sennáár ok þa er þetta smíði uar sua mikit uorðit at þat tok upp or ueðrum ok æigi hofðu þeir at minni fysi til at hallda framm smiðinní ok sem guð ser huersu þeira ofsi gæísar haatt þa sier hann at hann uerðr með æínshueríu niðr at sla. ok sa sami guð sem allz er uolldugr ok allt þeira uerk mattí a æínu auga bragði nídr sla en þa sialfa lata uerða at duptí. þa uilldi hann þo helldr með þi eyða þeira fyrer ætlan at þeir maætta uið kenaz sínn lítilleik með þí at æíngi þeira skildi huat annarr taladi ok þi uissi æingi þeira huat huerr bauð oðrum ok braut þat annarrr níðr sem annarrr uilldi uppreísa þar til er þeir stridduz a sialfer ok með þi æyddiz ǫll þeira fyrerætlan i upp tekinni stopul smið ok sa e[r] fremztr uar het zoroastres

2 The redactor shifts between the forms ‘þui’ and ‘þi’, the latter presumably corresponding to his pronunciation. 3 The note for ‘er’ seems to have been forgotten. 4 Hypercorrection from metnað to ‘metnat’, showing that –Vt# has become –Vð# in the redactor’s speech. 5 W displays diphtongization of previous /e:/. 6 The redactor occasionally uses d for /ð/. 7 Abbreviation mark seems to have been forgotten.

Appendix: Texts by the Wormianus Redactor

305

Translation First addition to the Prologue of Snorri’s Edda (And they did not know where his [God’s] realm lay. But they did believe that he governed all things on earth and in the air of heaven and of the celestial bodies, of the sea and of the clouds. In order to be able to more easily relate or remember these things they named them among themselves, and this superstition has since changed in many ways. When the peoples split apart and the languages were divided,) they began the building project that has been the greatest in the world. In his old age, Noah divided the world between his sons. He bequeathed the western part to Ham, and the northern to Japheth and the southern to Shem, with the partition that afterwards divides the world into three parts. In the time when the sons of these men were at large in the world, the lust for riches and honour increased, since then many arts were taught that had not existed before. And each man elevated himself according to his art, and so far did they proceed in their pride that the Africans, who came from Ham, ravaged the part of the world where the offspring of their relative Shem lived. And when they had defeated them, they found that the world was not enough, and on the field which is called Shinar they built a tower of tiles and stones that they intended to construct up to heaven, and when this building was so great that it protruded above the clouds and they still were no less eager to continue its construction, and as God saw how high their pride was surging, then he realized that he had to strike down in some way. And the very God who is omnipotent and who could strike down all their works in the blink of an eye and turn them into dust, he then preferred to shatter their ambitions in such a way that they would realize their own impotence in the manner that none of them understood what the other was saying, and therefore none of them knew what each ordered the other, and one broke down what the other was trying to raise, until they quarreled among themselves, and so their ambitions in the construction of the tower came to naught. And the one who led them was called Zoroaster.

8 The ‘air of the sea’ refers to the waters in heaven after the separation of the waters (Genesis 1: 6–7). 9 Without the interpolation, the text reads ‘[. . .] and this superstition has since changed in many ways as the peoples split apart and the languages were divided.’ 10 Having Zoroaster rather than Nimrod lead the building is very rare. See Jonas Wellendorf, ‘Zoroaster, Saturn, and Óðinn: The Loss of Language and the Rise of Idolatry’, in The Performance of Christian and Pagan Storyworlds, ed. Lars Boje Mortensen and Tuomas M. S. Lehtonen (Turnhout: Brepols, 2013), p. 157).

306

Appendix: Texts by the Wormianus Redactor

hann hlo fyrr en gret er hann kom i uerolldina. Enn forsmiðer voro ij. ok lxx. Ok sua margar tungur hafa siðan dreífz um uerolldína epter þi sem risarner skiptuz siðan til landa ok þioðernar fiolguðuz. I þessum sama stad uar gíorr æín hín agiætazta borg ok dregit af nafní stopulsins ok kollud babilon ok sem tungna skiptið uar orðít þa fiolguðuz sua nofnin mannanna ok annarra luta. ok sia samí zoroastres hafði morg nỏfn ok þo at hann understædí at hans ofsi uæri lægðr af sagðri smið þa færði hann sig þo framm til ueralldligs metnaðar ok let taka sik til konungs yfer mǫrgum þioðum assiríorum. Af honum hofz skurðgoða uilla ok sem hann uar blotaðr uar hann kallaðr baal. þann kollum uer bel. hann hafðí ok morg onnur nofn Enn sem nofnín fiǫlguðuz. þa týndiz með þui sanleíkrinn. Ok af fyrstu uillu þa blotaði huerr maðr epterkomandí sinn formeistara dyr eða fugla loptin ok himintunglín ok ymisliga dauðlega lutí þar til er þessi uilla gekk um allan heím ok sua uandlegha tyndu þeir sannleiknum at æingi uissi skapara sinn. utan þeir æinir menn sem toluðu ebreska tungu þa sem gekk fyrer stǫpul smíðina ok þo tyndu þeir æigi likamlegum giptum er þeim uoru ueíttar (ok þi skilðu þeir alla lutí iarðligri skilningu at þeim uar æigi gefen andleg spekten. Sua skilðu þeir at aller luter uæri smíðaðir  af nokkuru efní.)

11 Here, the redactor represents his own pronuciation (earlier –Vt# has become –Vð#) – cf. n. 4. 12 This form presumably represents the redactor’s pronunciation, whereas most of the time he uses the archaic sik. 13 The redactor seems to display some vowel harmony, at least to such an extent that if there is /i/ in the stressed syllable, he has it also in the ending (cf. ‘skirðir’, ‘liðir’, but ‘faðerinn’, ‘aller’ below). I have, however, opted for always resolving the abbreviation with er.

Appendix: Texts by the Wormianus Redactor

307

He laughed before he cried when he came into the world.14 And the master builders were 72, and so many languages have since spread across the world, as the giants afterwards divided the lands between them and the peoples multiplied. In this very place a splendid city was made, and it took its name from the tower and was called Babylon. And when the division of tongues had occurred, then the names of humans and other things multiplied. And this Zoroaster had many names and even though he understood that his pride was lowered by the aforementioned construction, he nonetheless promoted himself with temporal honours and had himself proclaimed king over the many peoples of the Assyrians. He was the origin of the error of idolatry, and when they sacrificed to him they called him Baal. We call him Bel. He had many other names besides, and as the names multiplied, so truth was lost. And in the first error, each descendant sacrificed to his founding master,15 to animals and birds, to the heavens and the celestial bodies and various mortal things, until that error travelled all over the world and they lost the truth so thoroughly that no one knew his Creator, apart from them alone who spoke the Hebrew language that had preceded the building of the tower. And even so, they [the gentiles] did not lose the physical gifts that they had been given, (and because they had not been given spiritual wisdom, they understood all things with an earthly understanding. Thus they understood that all things were made from some substance.)

14 Zoroaster’s laughter is also described in Stjórn and in AM 198 8vo (‘Stjórn’, ed. Astås, p. 154; Alfræði íslenzk, ed. Kålund, p. 49). The description can be traced back to Pliny’s Naturalis historia and seems to have come through Augustine to Vincent of Beauvais and thence into Stjórn (see Wellendorf, ‘Zoroaster, Saturn, and Óðinn’, p. 158–59). 15 The ideas of ancestor and leader worship seem to be conflated here.

308

Appendix: Texts by the Wormianus Redactor

Prologue to 1–4GT. W p. 83. 1–83.3216 Nu um hrið hefer sagt uerit huersu kenna skal þa luti sem frammi standa [i bok] þessarí megu þær kenningar a margan vegh bræytaz epter þi sem nu finna ny skaald ok taka til ok setia reglur epter ymisligum bokum. skal þo æigi at helldr laata þat unytt uera sem fornskaalldin hafa fundit er efní ok grundvǫllr er allz skaalldsk [a]par. enn æigi skulu menn þessum fra sǫgnum trua framarr en skynsamlígt er epter þi sem seger i fyrsta lut bokarennar með hveriu [u]illu[r]nar fiolguðuz ok af þi hefer hvert skalld sett ser reglu þat sem æigi truði [re]ttlegha af þi at þeir hugðu oðín guð uerit hafa ok alla þa með guð magní sem hanum þionuðu sem hæyra hefer maatt imorgum frasǫgnum þessar bokar. Enn nu sk[al] lysa huersu ny skalld ok fræði menn ok æinkannlega klerkarner uilia lofaz lááta huersu kveða skal ok onyta æigi at helldr þat sem forner menn hafa framit utan þat sem klerklegar bækr banna. þiat þat er natturuligt at menn se nu smasmuglarí sem fræði bækrnar dreifaz nu uiðara. laata froðer klerkar hueriar bækr sem þeir finna snara til þeirar þioðar tungu sem i þi landi talaz sem þa eru þeir. æigi at æins huersu tala skal helldr ok Iamuel huersu huerr stafr hlioðar með lǫngu hlioði ęðr skǫmmu hǫrðu ęðr linu. ok huat hvergí þeira hefer af ser fall eðr tima eðr þeim sem fyrer hanum stendr ok epter sem yðr mun synt uerða i þeim greinum sem siðarr eru skrifaðar

16 In this section and the next, commas seem to have been added by the same eighteenth-century hand as the one who has written ‘proemium Grammaticum’ above the text, namely that of Jón Ólafsson úr Grunnavík (Kålund, Katalog, 1, p. 214). I have not represented these in the text. 17 Lacuna. 18 Lacuna. 19 Text damaged. 20 Text damaged. 21 Only the bottom of ‘sk’ visible.

Appendix: Texts by the Wormianus Redactor

309

Prologue to 1–4GT Now some time has been spent explaining how one shall define the things that are found in the beginning of this book. These kennings can be varied in many ways according to how the new poets invent, adopt, and establish rules according to various books. Even so, one should not leave that behind which the early poets have invented and which is the substance and foundation of the entire art of poetry. But one should not believe these tales [of Snorri’s Edda] beyond what is prudent, according to what is said in the first part of the book of how the errors were multiplied, and every poet who did not believe rightly has constructed his own rule because they thought that Óðinn was a god, and that all who served him had divine powers, as you have been able to hear in the many tales of this book. But now we shall learn how the new poets and scholars, and in particular the clerics, allow that one should compose, but we should not forget to avail ourselves of that which men of old have achieved, except what is forbidden in the books of the clerics, since it is natural that people are more perceptive, now that scholarly books are spreading more widely. Wise clerics see to it that the books that they find are translated into the language which is spoken in the country where they are at the time – not only how one should speak, but also how each letter sounds, with a long sound or a short, hard or soft, and what accident and duration belongs to each of them or to the letters that stand before or after it, as will be demonstrated to you in the treatises that are written

22 The reference is to the prologue of Snorri’s Edda, in particular to the redactor’s own interpolation given above here (see pp. 285–86). The wording, however, is a paraphrase of a statement in the beginning of Snorri’s text in Skáldskaparmál (Snorri, Skáldskaparmál, ed. Falkes, p. 5). 23 It is clear from the context that these ‘rules’ must concern kenning formation, which Snorri has classified into many categories which are placed immediately before the prologue in W (Skáldskaparmál). These categories are presented with phrases like ‘how shall one define Óðinn?’ and could thus be perceived as ‘rules’, and they are often based on mythological motifs or narrative. The redactor perceives of the activity of the early poets as that of the grammarian, setting up individual rules that may not be broken. The multiplicity of gods has, however, given rise to a corresponding multiplicity of rules. He seems to see the new treatises presented in his book as a means to attain a unified poetic counterpart to the unification of all languages into a reconstruction of Hebrew (and, thus, of truth). If the pagan rules are upgraded in accordance with the books that are approved by the Church, all rules can be combined into a unified poetics which does more justice to the single, omnipotent Maker. 24 That which is ‘forbidden’ are barbarisms and solecisms, as found in 3GT, which follows later in W. 25 I.e. aspirated or unaspirated. 26 The redactor is mixing different taxonomical levels here. Aspiration and duration belong to the accidents of a letter or syllable. 27 The references in this sentence seem be to both 2GT and 3GT. The accidents of the letter and syllable are described in 3GT, but the length of consonants in relation to their position before or after a vowel in 2GT.

310

Appendix: Texts by the Wormianus Redactor

epter þeira manna upp tekinni stafasetningar reglu sem uer hyggium uel hafa kunnat ordograffiam28 ok þo at sína figuru hafi huerr þeirra til sinnar sagnar þa syniz mǫnnum aller þeir fagrliga skipat hafa. hefer huerr sett stafina epter þeiri tungu sem þeir hafa talað. ok þo at þeira verk se saman borín þa bregðr ekki þeira annars reglu. skal yðr syna hinn fyrsta letrs haatt sua ritinn epter sextan stafa stafrofí i danskri tungu epter þi sem þoroddr runa meistarí ok ari prestr hinn froði hafa sett i motí latinu manna stafrófi29 er meistarí priscianus hefer sett. hafa þeir þi fleiri hlioðs greíner með huerium raddar staf sem þessi er tungan fa talaðrí sua at þat ma understanða með hlioði umbæygiligu huossu ok slíofu

28 = orthographiam. 29 This is a rare case where the redactor marks vowel length with a superscript stroke, as is found in some early manuscripts.

Appendix: Texts by the Wormianus Redactor

311

later [in the book], in accordance with the adoption of the spelling rule of those people whom we believe to have been knowledgeable in orthography. And even though each of them has chosen to shape the letters in accordance with his own idiom,30 people still find that they have all arranged them in a pleasing manner. Each has established the letters according to the language that they spoke, and even if their works are compared, none of them breaks the rule of another.31 You will be shown the first way of writing, written down according to the sixteen-letter alphabet of the Danish tongue [=Old Norse], following [the alphabet] that Þóroddr Master of Runes and the priest Ari the Wise have defined against the alphabet of the Latins, which master Priscian has defined. They [the runes] have more accents to every vowel in proportion to the fewer sounds of this language [Old Norse],32 so that it may be understood with circumflex or acute or grave accent, in order that

30 The meaning of sǫgn here is somewhat obscure. It normally means ‘story’ or ‘word’, and the latter is the meaning generally found in the grammatical texts. Given the context, though, I have preferred to translate ‘idiom’. 31 The reference is to the redactor’s own interpolation at the end of 2GT (though at this point probably only planned). 32 Björn M. Ólsen emended fátalaðri (with fewer sounds) to fástafaðri (with fewer letters), based on the fact that Old Norse had more, not fewer, vowels than Latin (Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, pp. 154, 306; see the discussion in Fritzner, Ordbog, 1, p. 395; but accepting the emendation in 4, p. 93). The basis for this passage, however, is 3GT, and the different number of vocalic phonemes in the two languages is on the whole suppressed in that text. Furthermore, since the text is contemporaneous to the manuscript, the emendation, resulting in a hapax, is unwarranted (Males, ‘Wormianusredaktören’, pp. 41–53, 58). What the redactor means by ‘fewer [vowel] sounds’, then, is that runes have no accents (sound and sign are here conflated).

312

Appendix: Texts by the Wormianus Redactor

sua at eínnar tiðar fall værí i huaru tveggia stafrofí til þess at skalldín mætti þa miukara kveða epter ny fundinni letr list. enn hafa æigi huert orðzskrípí þat sem fornskalldin nyttu en haalfu siðr auka i enn uerrum orðum en aðr hafa fundín uerít. þiat at uandara var þeim að tala sem ekki hofðu fyrer ser enn þeim sem nu hafa ymisligar fręði bękr. enn uel ma nyta at hafa epter þeim heití ok kenningar æigi lengra reknar enn Snorri lofar.

Appendix: Texts by the Wormianus Redactor

313

there be one and the same accent in both alphabets,33 so that the poets can compose more softly according to the newly received art of letters and not retain every verbal monstrosity which the early poets used,34 but much less add even worse words than those that have been invented before. For it was more difficult for them to speak, who had nothing in front of them, than for those who now have various scholarly books. But it is good to take from them heiti and kennings that are not extended beyond what Snorri allows.35 Those who now wish to follow the new

33 The phrase einnar tíðar fall literally means ‘the case of one time’, but has here been translated as ‘one and the same accent’. The background to this phrase is complex, since its meaning and its form do not appear to derive from the same text. I begin with the meaning: this seems to be drawn from the formally similar, but not identical tíðar tilfelli (the accident of time), which is found in the section on accent in 3GT, but which is there mentioned as an aside about quantity (Den tredje og fjærde grammatiske avhandling, ed. Björn M. Ólsen, pp. 55–56). Its location in the chapter on accent seems to have suggested to the redactor the use of tíðar fall to denote accent, but there is probably more to the story than sloppy reading. Óláfr’s translation of tenor (accent) is hljóðsgrein (phonetic distinction), which, in the context of the two writing systems described in 3GT (runes and Latin letters), might convey a sense of division between the two languages – precisely what the redactor wants to avoid. With the use of fall, by contrast, he can hint that both belong to one and the same case (though precisely of what remains unclear) and thus retain an image of unity. Similarly, another instance of apparent confusion in this passage may be due not so much to a mistake as to adaptation: the redactor says that Old Norse has fewer sounds than Latin, which in practice means that there are no accents above the runes, but that one should understand the runes as if they had them, so as to arrive at the same system (einnar tiðar fall) in both cases. The confusion between sound and sign is here seemingly complete, since understanding the runes as if they had accent can hardly change the Old Norse language as such. Without this confusion, however, the redactor could not have projected the image of unity so dear to him. The difficulties in this passage, then, are partly due to his source text and partly due to his wish to convey a unified phonographemic structure which covers both languages involved. On the verbal level, the phrase einnar tíðar fall is derived from Háttatal 16: ‘Í þeima hætti skal velja saman þau orðtǫk er ólíkust sé at greina ok hafi þó einnar tíðar fall bæði orð ef vel skal vera’ (in this metre [refhvǫrf], one should juxtapose those expressions that differ most in meaning, but that, if it is performed well, both have the same quantity [i.e. the same length, occupying the same number of metrical positions]) (Snorri, Háttatal, ed. Faulkes, p. 11). This is one of several examples that show that the redactor had thoroughly internalized Háttatal, but that he would only use that text as he saw fit. As for the translation of Snorri’s text, I take orð to refer back to orðtak, and that Snorri means that the two expressions should occupy half the line each. He strives to achieve this in the stanza, but he is not successful in each line. In the following sentence, he explicitly admits this: ‘En til þessa háttar er vant at finna ǫll orð gagnstæðlig, ok er hér fyrir því sum orð dregin til hœginda’ (But in this metre, it is difficult to find all expressions that balance each other out, and some expressions are therefore here chosen to facilitate the process). This interpretation differs somewhat from that of Faulkes (Snorri, Háttatal, ed. Faulkes, pp. 11, 53–54). 34 Based on Snorri’s words about metre: ‘en þótt þat finnisk í fornskálda verka, þá látum vér þat nú ónýtt’ (and though this [too long kennings] may be found in the works of the early poets, we now avoid using it), but here seemingly referring to nykrat (see p. 287). 35 The reference is to Háttatal 8 (see p. 287).

314

Appendix: Texts by the Wormianus Redactor

Leiti epter sem uandligast þeir sem nu uilía fara at nyium hattum skalldskapar huersu fegrst er talat enn æigi huersu skiott er ort. þiat at þui uerðr spurt huerr kvað þa er fra liðr enn æigi huersu lengi lengi uar at uerit ok þeir sem nu uilia með nyiu kveða hafi smasmugul ok huoss ok skygn hugsunar augun að sia huað yðr er nu synt i þessum frasǫgnum.

Appendix: Texts by the Wormianus Redactor

315

modes of poetic composition should consider with utmost care how one may speak with the greatest elegance, since with the passing of time, people will ask who the poet is, and not how much time he spent in composition,36 and the mind’s eyes of those who now wish to compose in new ways should be perceptive and sharp and piercing, so as to see what pleases you in these expositions.

36 Echoing Snorri’s ‘orðin munu finnask ef vandlega er leitat’ (the [right] words will be found if one inquires thoroughly) (see p. 287).

316

Appendix: Texts by the Wormianus Redactor

Addition to 2GT. W p. 93. 10–94. 7 Nu uerðr þetta allt saman stafrof kallat. þesser stafer giora allt maal ok hender maalit ymsa sua til at iafna sem horpu strenger giora hlioð eða eru læyster luklar i simphoníe eða þa er organ gengr upp ok níðr aptr ok framm um allan gamma þann er með ser hefer nítian lukla ok aatta radder. ok nu koma til motz þeser .v. hríngar stafanna er aaðr uar um rætt. kallaz nu huarer uið aðra stafrof ok gammí ok taka nu hlioðstafer þar sin hlioð ok raddar stafer rǫdd. maalstafer malít ok samnaz til orðanna sua margra at ekki er þess mælt i heiminum at eigi se þesser stafer til hafðer. Nu eru æingi þau lætí eða hlioð eða radder at æigi muni þat allt finnaz i gamanum. Nu tekr sua fremí mikit um gioraz er orðin hefiaz upp ok hliomrenn uex ok raddernar glymia a. Nefniz ok sua fremí sǫngr er þetta hefer allt til ok nu þyss sia flokkr framm á leikvǫllinn ok ǫllum megín at styrínu þui. er tungan heiter ok heita nu a hana til maalsins ok orðanna ok sǫngsins at hon kveðí þat allt upp ok hon giorer sua ok hneiger sik til styrimanzins ok maler sua Osanna seger hon. þat þyðiz a uaara tungu sua. græð þu oss. Enn þat er a ebresku mælt ok stakk hana natturan til þess fyrer þui at hon var fyrst ok gekk þa um allan heim þangat til er guð skiptí þeim. Nu seger þar til at henní þotti hann uera styri maðrínn er hann skapaði hana ok af kristz nafní er kristnin kolluð. uer er kristner erum kollum hann hofuð uaart enn uer hans limer ok liðir ok hans sonr er sa er hann sendi hingat i heím ok sa er váár faðer en uer hans born. uar ok faðerinn uęnligr til at stiorna sinum bornum sua sem bezt gegndi. uar þi orðit or messunní til tekit at hann uissi huerr lofsongr honum þottí mestr framm fluttr þessa heims uið sik sialfan. er þar ok uaar hialp oll i folgín er um hans pisl er rætt ok saar er hann þolði a krossinum helga er or rann bæði bloð ok vatn ok iþi erum uer skirðir er rett truum a almattkan guð. ok þat hans holld ok bloð er i messunni er framm flutt er vart farnest þa er uer forum af þessum heímí. Nu skal þat uaan uaar at uætta þess at sua fremí fariz oss vel er sua uerðr

37 One would have expected mæler.

Appendix: Texts by the Wormianus Redactor

317

Addition to 2GT

Now this is all called the alphabet. These letters produce all language and the language treats the various letters as when the strings of a harp produces sounds or when the keys of a hurdy-gurdy are released or when the organ goes up and down, back and forth across the entire scale which has nineteen keys and eight notes. And now these five rings of letters which were previously discussed come to meet [the scale]. Now they call out to each other, the alphabet and the musical scale, and the vowels (hljóðstafir) get their sound (hljóð) and the vowels (raddarstafir) get their voice (rǫdd), the consonants (málstafir) language (málit), and [the letters/sounds] are gathered into words, so many that nothing has been said in this world for which these letters cannot be used. There are no sounds or noises or voices that cannot all be found in the scale. Now much begins to happen, when the words are pronounced, the sound increases and the voices ring out over it all. That which contains all this is called song, and now this throng rushes forth onto the playground [the mouth] and on all sides, towards the rudder which is called the tongue, and they [the letters/sounds] call upon it to receive language and words and song – that it should express all of this – and it does so and bows to the captain and says: Osanna! it says. That means, in our language: Heal us! And that is said in Hebrew and nature made it [the tongue] do so, because it [Hebrew] was first and was dispersed over all of the world until God divided them [the languages]. Now it should be said that it [the tongue] thought that he was captain, since he created it, and Christianity draws its name from the name of Christ. We who are Christians call him our head and we are his limbs and joints, and his son is the one whom he sent hither into the world, and he is our father and we his children, and the father could be expected to govern his children in the best way. That word [osanna] was extracted from the mass because he knew which praise of this world which, when offered to him, pleased him the most. And all our help is contained in when we speak of his passion and the wound he received on the holy Cross, out of which ran both blood and water, and in it we who believe correctly in the almighty God are baptized. And his flesh and blood which is presented in the mass is our travel provisions when we depart from this world. Now it shall be our hope to bear witness to that it will turn out well for us insofar as things will become

38 Compare earlier in 2GT: ‘muðrinn er leikvǫllr orðanna en tungan stýrit’ (the mouth is the playground of the words and the tongue is its rudder) (The So-Called Second Grammatical Treatise, ed. Raschellà, p. 29).

318

Appendix: Texts by the Wormianus Redactor

sem hann hefer fyrer sieð at bæði se at hann er i fỏr með oss ok uer með honum þa er uer forum hæim til foður leifðar uaarrar. ok þa er hann hefer skipt sinu liði sier til hægrí handar epter doms dag. þa skulum uer hefia upp alleluia fyrer þuí at þat er æigi iarðneskr sỏngr. syngia þetta þa aller saman tíu fylkí guðsengla ok manna þa er almattigr guð ferr meðr39 sina ferð hæim i himinrikis dyrð ok skulum þa una i sifellu sua at alldrí skal epter verða með guði almatkum þar sem hann er æ ok æ með feðr ok syní ok helgum anda sa er lifer ok riker æinn guð of allar allder ueralda[r] amen.

39 Here, the redactor has the form ‘meðr’, whereas he has just previously used the form ‘með’, both unabbreviated. Similarly, he uses both ‘eða’ and ‘eðr’.

Appendix: Texts by the Wormianus Redactor

319

as he has foreseen, namely that he is our travelling companion and we his when we journey home to our paternal inheritance. And when he has divided the throngs on his right side after doomsday, then we shall proclaim: Alleluia! because that is no earthly song.40 Then all sing this together, the ten hords of angels and men, when the almighty God travels home with his retinue to the glory of Heaven, and we shall then be joyous forever with almighty God, so that there will never be an afterwards, for he is forever and ever with the Father and Son and Holy Spirit, the one God who lives and reigns through all the ages of the world. Amen.

40 The background to the opposition between osanna and the unearthly (i.e. heavenly) alleluia is not entirely clear. Jerome devoted a letter to pope Damasus to the correct translation of Osanna (which, as he says, is our way of saying Osianna, and he translates ‘osi (salvum fac), anna (obsecro)’ and thus ‘obsecro, salvum [me] fac’ (I pray, make me healthy) (PL 22, cols 375–79), and in his dictionary of Hebrew words, he translates salvifica (PL 23, cols 842, 1156). For alleluia, he gives laudate Dominum (praise the Lord) (PL 23, cols 857, 1150). In Old Norse expositions of the mass (one of them a translation of Honorius Augustodunensis, Gemma Animae), alleluia is described as the song of angels or the song of men after doomsday (Messuskýringar. Liturgisk symbolik frå den norsk-islandske kyrkja i millomalderen, ed. Oluf Kolsrud (Oslo: Kjeldeskriftfondet, 1952), pp. 17, 45–46). Compare, for instance, Gregory’s Homiliae in Ezechielem I. 9: ‘Alleluia […] Hoc nobis carmen coelestis patriae nuntiare cives eius venerant, qui concorditer clamabant: Gloria in excelsis Deo […]’ (Alleluia […] this song of the heavenly fatherland its citizens had come to proclaim to us, when they harmoniously shouted: Glory to God on high […]) (PL 76, col 885). In the Old Norse expositions of the Mass, osanna is said to be how people greeted Christ on his last trip to Jerusalem, that is, while he was still here in the flesh (Messuskýringar, ed. Kolsrud, p. 50). This may perhaps explain why it belongs in the Mass in this world, in contradistinction to the heavenly alleluia.

Bibliography Manuscripts Ip: various paper manuscripts of Sturlunga saga (Sturlunga saga, ed. Kålund, pp. xlii–lv, lxxvi) A: Reykjavík, AM 748 Ib 4to, c. 1300–1325 (ONP Registre, p. 464) B: Reykjavík, AM 757 a 4to, c. 1400 (ONP Registre, p. 464) Br: Brit. Mus. Addit. 11, 127 fol. (Sturlunga saga, ed. Kålund, pp. lvi–lviii, lxxvi) C: Reykjavík, AM 748 II 4to, c. 1400 (ONP Registre, p. 464) Copenhagen, AM 45 fol (Fríssbók), c. 1300–1325 (ONP Registre, p. 432) Copenhagen, AM 685d 4to, c. 1400–1500 (ONP Registre, p. 462) Copenhagen, AM 732b 4to, c. 1300–1325 (ONP Registre, p. 463) CR: Copenhagen, GKS 2365 4to (Codex Regius of eddic poetry), c. 1270 H: Holm. Papp. 8 4to (Sturlunga saga, ed. Kålund, pp. liii–liv, lxxvi) R: Copenhagen, GKS 2367 4to (Codex Regius), c. 1300–1350 (ONP Registre, p. 472) Reykjavík, AM 132 fol (Möðruvallabók), c. 1330–1370 (ONP Registre, p. 433) Reykjavík, AM 162 A γ fol. (c. 1275–1300) (ONP Registre, p. 434) Reykjavík, AM 162 A δ fol (c. 1300) (ONP Registre, p. 434) Reykjavík, AM 162 A ζ fol. (c. 1250–1300) (ONP Registre, p. 434) Reykjavík, AM 162 A θ fol. (c. 1250) (ONP Registre, p. 434) Reykjavík, AM 748 Ia 4to (c. 1300–1325) (ONP Registre, p. 464) Reykjavík, Lbs fragm 82 (c. 1258–1264) (Stefán Karlsson, ‘Kringum kringlu’) Stockholm, Kungliga biblioteket, Holm. Papp. 25 8vo (c. 1650–1700) (Háttalykill enn forni, ed. Jón Helgason and Anne Holtsmark, pp. 7–21) Stockholm, Holm. Perg. 7 4to (c. 1300–1325) (ONP Registre, p. 474) Stockholm, Holm. Perg. 15 4to (c. 1200) (ONP Registre, p. 474) T: Utrecht University Library MS 1374 (Codex Trajectinus), c. 1600 U: Uppsala, Universitetsbiblioteket, DG 11 4to (Codex Upsaliensis), c. 1300–1325 (ONP Registre, p. 469) Uppsala, Universitetsbiblioteket, R: 683, Salanska samlingen 28 fol. (c. 1650–1700) (Gödel, Katalog öfver Upsala universitets biblioteks fornisländska och fornnorska handskrifter, p. 27) V: Advoc. Library 21–3–17 (Sturlunga saga, ed. Kålund, pp. lx–liv, lxiii) W: Copenhagen, AM 242 fol. (Codex Wormianus) c. 1340–70 (ONP Registre, p. 438; Karl G. Johansson, Studier i Codex Wormianus, pp. 17–18) Wolfenbüttel, WolfAug 9 10 4to (c. 1330–1370) (ONP Registre, p. 493)

Primary Sources Accessus ad auctores. Bernard d’Utrecht. Conrad d’ Hirsau: ‘Dialogus super auctores’, ed. R. B. C. Huygens (Leiden: Brill, 1970) Ágrip af Nóregskonunga sǫgum. Fagrskinna – Nóregs konunga tal, ed. Bjarni Einarsson. Íslenzk fornrit, 29 (Reykjavík: Hið íslenzka fornritafélag, 1985) Alfræði íslenzk. Islandsk encyklopædisk litteratur, 1: Cod. Mbr. AM. 198, 8vo, ed. Kristian Kålund (København: Samfund til udgivelse af gammel nordisk litteratur, 1908) The Anglo-Saxon Chronicle. MS C, ed. Katherine O’Brien O’Keeffe. The Anglo-Saxon Chronicle. A Collaborative Edition, 5 (Cambridge: D.S. Brewer, 2001) The Arnamagnæan Manuscript 674A, 4to. ‘Elucidarius’, ed. facs. Jón Helgason (Copenhagen: Munksgaard, 1957) https://doi.org/10.1515/9783110643930-009

322

Bibliography

Árni Magnússons levned og skrifter, ed. Finnur Jónsson, 2 vols (Kommissionen for det arnamagnæanske legat: København, 1930) Laxdœla saga, ed. Einar Ól. Sveinsson. Íslenzk fornrit, 5 (Reykjavík: Hið íslenzka fornritafélag, 1934) Les arts poétiques du xiiᵉ et du xiiiᵉ siècle: Recherches et documents sur la technique littéraire du moyen age, ed. Edmond Faral. Bibliothèque de l’Ecole des hautes études, Sciences historiques et philologiques, fasc. 238 (Paris, 1924) Augustine, Sancti Aureli Augustini de Genesi ad litteram libri duodecim, eiusdem libri capitula, de Genesi ad litteram imperfectus liber, locationum in Heptateuchum libri septem, ed. Iosephus Zycha. CSEL, 28 (3: 2) (Pragae: Bibliopola academiae litterarum Caesareae Vindobonensis, 1894) Augustine, Confessions, ed. and trans. William Watts, 2 vols. Loeb Classical Library, 26–27 (Cambridge: Harvard University Press, 1912) Augustine, De doctrina Christiana, ed. and trans. R. P. H. Green (Oxford: Clarendon Press, 1995) Auraicept na-néces. The Scholars’ Primer, ed. George Calder (Edinburgh: John Grant, 1917) Austfirðinga sǫgur, ed. Jón Jóhannesson. Íslenzk fornrit, 11 (Reykjavík: Hið íslenzka fornritafélag, 1950) Bandamanna saga, ed. Hallvard Magerøy (Oslo: Dreyers forlag, 1981) Benedict of Nursia, Benedicti Regula, ed. Rudolph Hanslik. Corpus Scriptorum Ecclesiasticorum Latinorum, 75 (Vindobonae: Hoelder/Pichler/Tempsky, 1960) Biskupa sögur, 1, ed. Sigurgeir Steingrímsson, Ólafur Halldórsson and Peter Foote, 2 vols. Íslenzk fornrit, 15 (Reykjavík: Hið íslenzka fornritafélag, 2003) Biskupa sögur, 2, ed. Ásdís Egilsdóttir. Íslenzk fornrit, 16 (Reykjavík: Hið íslenzka fornritafélag, 2002) Biskupa sögur, 3, ed. Guðrún Ása Grímsdóttir. Íslenzk fornrit, 17 (Reykjavík: Hið íslenzka fornritafélag, 1998) ‘Bréf Alexandri Magni. Den norrøne oversættelse af Epistola Alexandri Magni ad Aristotelem udgivet sammen med forlægget’, ed. Povl Skårup, in Opuscula, 9. Bibliotheca Arnamagnæana, 39 (Hafniæ: Reitzel, 1991), pp. 19–99 Brennu-Njáls saga, ed. Einar Ól. Sveinsson. Íslenzk fornrit, 12 (Reykjavík: Hið íslenzka fornritafélag, 1954) ‘Catilina’ and ‘Jugurtha’ by Sallust and ‘Pharsalia’ by Lucan. In Old Norse: ‘Rómverja saga’, AM 595 a–b 4to, ed. Jakob Benediktsson. Early Icelandic Manuscripts in Facsimile, 13 (Copenhagen: Rosenkilde and Bagger, 1980), pp. 10–12) Codex Frisianus. En samling af norske konge-sagaer, ed. C. R. Unger (Christiania: Mallings, 1871) Codex Regius of the Younger ‘Edda’. MS No. 2367 4ᵗᵒ in the Old Royal Collection in the Royal Library of Copenhagen, ed. Elias Wessén (Copenhagen: Munksgaard, 1940) Codex Trajectinus. The Utrecht Manuscript of the ‘Prose Edda’, ed. Anthony Faulkes. Early Icelandic Manuscripts in Facsimile, 15 (Copenhagen: Rosenkilde and Bagger, 1985), pp. 19–20 Danakonunga sǫgur, ed. Bjarni Guðnason. Íslenzk fornrit, 35 (Reykjavík: Hið íslenzka fornritafélag, 1982) Dante Alighieri, Opere, 1: Rime, Vita nova, De vulgari eloquentia, ed. and trans. Claudio Giunta, Guglielmo Gorni, Mirko Tavoni (Milano: Mondadori, 2011) Den norsk-isländska skaldediktningen, ed. Ernst A. Kock, 2 vols (Lund: Gleerup, 1946–49) Den tredje og fjærde grammatiske avhandling i Snorres Edda tilligemed de grammatiske avhandlingers prolog og to andre tillæg, ed. Björn M. Ólsen. Skrifter, 12 (København: Samfund til udgivelse av gammel nordisk litteratur, 1884)

Bibliography

323

Donat et la tradition de l’enseignement grammatical. Étude sur l’‘Ars Donati’ et sa diffusion (ivᵉ–ixᵉ siècle) et édition critique, ed. Louis Holtz (Paris: CNRS éditions, 1981) ‘Echtrae Chonnlai’ and the Beginnings of Vernacular Writing in Ireland: A Critical Edition with Introduction, Notes, Bibliography and Vocabulary, ed. Kim McCone. Maynooth Medieval Irish Texts, 1 (Maynooth: Department of Old and Middle Irish, National University of Ireland Maynooth, 2000) Edda Snorra Sturlusonar, Codex Wormianus AM 242, fol, ed. Finnur Jónsson (København and Kristiania: Gyldendal, 1924) Edda Snorra Sturlusonar. Edda Snorronis Sturlæi, ed. Jón Sigurðsson et al., 3 vols (Hafniæ: Legatum Arnamagnæanum, 1848–87) Eddukvæði, ed. Jónas Kristjánsson and Vésteinn Ólason, 2 vols. Íslenzk Fornrit (Reykjavík: Hið íslenzka fornritafélag, 2014) Egils saga Skallagrímssonar tilligemed Egils större kvad, ed. Finnur Jónsson. Samfund til udgivelse av gammel nordisk litteratur, 17 (København: Møller, 1886–88) Egils saga Skalla-Grímssonar, ed. Sigurður Nordal. Íslenzk fornrit, 2 (Reykjavík: Hið íslenzka fornritafélag, 1933) Einarr Skúlason, Geisli, ed. Martin Chase (Toronto: University of Toronto Press, 2005) ‘Elucidarius’ in Old Norse Translation, ed. Evelyn Scherabon Firchow and Karen Grimstad (Reykjavík: Stofnun Árna Magnússonar, 1989) Encomium Emmae reginae, ed. Alistair Campbell. Camden Third Series 72 (London: Offices of the Royal Historical Society, 1949) Eyrbyggja saga, ed. Einar Ól. Sveinsson and Matthías Þórðarson. Íslenzk fornrit, 4 (Reykjavík: Hið íslenzka fornritafélag, 1935) ‘Eyrbyggja saga’. The Vellum Tradition, ed. Forrest S. Scott. Editiones Arnamagnæanæ A, 18 (Copenhagen: Reitzel, 2003) Eyfirðinga sögur, ed. Jónas Kristjánsson. Íslenzk fornrit 9 (Rekjavík: Hið íslenzka fornritafélag, 1956) The First Grammatical Treatise. Introduction, Text, Notes, Translation, Vocabulary, Facsimiles, ed. Hreinn Benediktsson. University of Iceland Publications in Linguistics, 1 (Reykjavík: Institute of Nordic Linguistics, 1972) The First Grammatical Treatise. The Earliest Germanic Phonology. An Edition, Translation and Commentary, ed. Einar Haugen, 2nd ed. (London: Longman, 1972) Flateyjarbók. En samling af norske konge-sagaer med inskudte mindre fortellinger om begivenheder i og udenfor Norge samt annaler, ed. Guðbrandr Vigfusson and C. R. Unger, 3 vols (Christiania: Mallings, 1860–68) Fornaldar sögur norðurlanda, ed. Guðni Jónsson, 4 vols (Reykjavik: Íslendingasagnaútgáfan, 1950) Fornmanna sögur, 12 vols (Kaupmannahøfn: Popp, 1825–37) The Fourth Grammatical Treatise, ed. and trans. Margaret Clunies Ross and Jonas Wellendorf (London: Viking Society for Northern Research, 2014) Færeyinga saga. Óláfs saga Tryggvasonar eftir Odd munk Snorrason, ed. Jónas Kristjánsson and Þórður Ingi Guðjónsson. Íslenzk fornrit, 25 (Reykjavík: Hið íslenzka fornritafélag, 2006) Geoffrey of Monmouth, The History of the Kings of Britain. An Edition and Translation of De Gestis Britonum [Historia Regum Britanniae], ed. Michael D. Reeve and trans. Neil Wright. Arthurian Studies 69 (Woodbridge: Boydell, 2007) ‘Grágás’. Islændernes Lovbog i Fristatens Tid udgivet efter det Kongelige Bibliotheks Haandskrift og oversat, ed. and trans. Vilhjálmur Finsen, 2 vols (Kjøbenhavn: Berling, 1852–70) ‘Grágás’ efter det Arnamagnæanske Haandskrift, nr. 334 fol., Staðarhólsbók (Kjøbenhavn: Gyldendal, 1879) Grammatici latini, 2, ed. Heinrich Keil (Leipzig: Teubner, 1855)

324

Bibliography

Gregory the Great, Grégoire le Grand, Dialogues, 2, ed. Adalbert de Vogüé and trans. Paul Antin. Sources Chrétiennes, 206 (Paris: Éditions du Cerf, 1979) Gregory the Great, Gregorius Magnus, Moralia in Iob, ed. M. Adriaen. Corpus Christianorum. Series Latina, 143 (Turnhout: Brepols, 1979) Grettis saga, ed. Guðni Jónsson. Íslenzk fornrit, 7 (Rekjavík: Hið íslenzka fornritafélag, 1936) Harðar saga, ed. Þórhallur Vilmundarson and Bjarni Vilhjálmsson. Íslenzk fornrit, 13 (Reykjavík: Hið íslenzka fornritafélag, 1991) Háttalykill enn forni, ed. Jón Helgason and Anne Holtsmark. Bibliotheca Arnamagnæana, 1 (København: Munksgaard, 1941) ‘Hauksbók’ udgiven efter de arnamagnæanske håndskrifter no. 371, 544 og 675, 4°, ed. Finnur Jónsson (København: Det kongelige nordiske oldskrift-selskab, 1892–96) Heilagra manna søgur. Fortællinger og legender om hellige mænd og kvinder efter gamle haandskrifter, ed. Christian R. Unger, 2 vols (Christiania: Bentzen, 1877) Heimskringla, ed. Bjarni Aðalbjarnarson, 3 vols. Íslenzk fornrit, 26–28 (Rekjavík: Hið íslenzka fornritafélag, 1941–51) Henry Archdeacon of Huntingdon, Historia Anglorum. The History of the English People, ed. and trans. Diana Greenway (Oxford: Clarendon Press, 1996) Homiliu-bók: Isländska homilier efter en handskrift från tolfte århundradet, ed. Theodor Wisén (Lund: Gleerup, 1872) Horace, Satires. Epistles. Ars Poetica, ed. and trans. H. R. Fairclough (Cambridge: Harvard University press, 1929) ‘Hrólfs rímur Gautrekssonar’, ed. Jóhanna Katrín Friðriksdóttir and Haukur Þorgeirsson. Gripla, 26 (2015), pp. 81–137 Isidori hispalensis episcopi etymologiarum sive originum libri XX, ed. W. M. Lindsay, 2 vols (Oxford: Clarendon, 1911) Islands runeindskrifter, ed. Ander Bæksted. Bibliotheca Arnamagnæana, 2 (Hafniæ: Munksgaard, 1942) Íslendingabók. Landnámabók, ed. Jakob Benediktsson. Íslenzk fornrit, 1 (Reykjavík: Hið íslenzka fornritafélag, 1986) Íslenzk miðaldakvæði, ed. Jón Helgason, 2 vols (København: Munksgaard, 1936–38) Íslenzkar þjóðsögur og ævintýri, ed. Jón Árnason, 2nd ed., ed. Árni Böðvarsson and Bjarni Vilhjálmsson, 6 vols (Rekjavík: Bókaútgáfan þjóðsaga, 1954–61) ‘Jómsvíkinga saga’ efter arnamagnæanska handskriften N: o291. 4: toi diplomatariskt aftryck, ed. Carl af Petersens (København: Samfund til udgivelse af gammel nordisk litteratur, 1882) ‘Jómsvíkinga saga’ efter skinnboken N: o7, 4: toå kungl. Biblioteket i Stockholm, ed. Gustaf Cederschiöld (Lund: Berlings, 1875) ‘Jómsvíkinga saga’ i latinsk oversættelse af Arngrim Jonsson, ed. A. Gjessing (Kristiansand: Steens bogtryckeri, 1877) Kjalnesinga saga, ed. Jóhannes Halldórsson. Íslenzk fornrit, 14 (Rekjavík: Hið íslenzka fornritafélag, 1956) Kommentar zu den Liedern der Edda, 3: Götterlieder, ed. Klaus von See et al. (Heidelberg: Carl Winter, 2000) Kommentar zu den Liedern der Edda, 5: Heldenlieder, ed. Klaus von See et al. (Heidelberg: Carl Winter, 2006) Laurentius von Durham, Consolatio de morte amici, (ed. Udo Kindermann PhD diss.FriedrichAlexander-Universität,, 1969) Legendarisk Olafssaga, ed. facs. Anne Holtsmark. Corpus Codicum Norvegiorum medii aevi. Quarto serie, 2 (Oslo: Selskapet til utgivelse av gamle norske håndskrifter, 1956)

Bibliography

325

The Life of St. Gregory and his Dialogues. Fragments of an Icelandic Manuscript from the 13th Century, ed. Hreinn Benedikttson. Editiones Arnamagnæanæ B, 4 (Copenhagen: Munksgaard, 1963) Lives of Saints. Perg. Fol. Nr. 2, Royal Library, Stockholm, ed. Peter Foote. Early Icelandic Manuscripts in Facsimile, 4 (Copenhagen: Rosenkilde and Bagger, 1962) Monumenta historica Norvegiæ. Latinske kildeskrifter til Norges historie i middelalderen, ed. Gustav Storm (Kristiania: A. W. Brøgger, 1880) Morkinskinna. The Earliest Icelandic Chronicle of the Norwegian Kings (1030–1157), trans. Theodore Andersson and Kari Ellen Gade. Islandica, 51 (Ithaca: Cornell University Press, 2000) Morkinskinna, ed. Ármann Jakobsson and Þórður Ingi Guðjónsson, 2 vols. Íslenzk fornrit, 23–24 (Reykjavík: Hið íslenzka fornritafélag, 2011) Morkinskinna, ed. Finnur Jónsson. Samfund til Udgivelse af Gammel Nordisk Litteratur, 53 (København: Samfund til Udgivelse af Gammel Nordisk Litteratur, 1932) Messuskýringar. Liturgisk symbolik frå den norsk-islandske kyrkja i millomalderen, ed. Oluf Kolsrud (Oslo: Kjeldeskriftfondet, 1952) Norges innskrifter med de yngre runer, 2, ed. Magnus Olsen (Oslo: Jacob Dybwad, 1951) Norrœn fornkvæði. Islandsk samling af folkelige oldtidsdigte om Nordens guder og heroer, almindelig kaldet Sæmundar Edda hins fróða, ed. Sophus Bugge (Christiania: Mallings, 1867) Notices et extraits de divers manuscrits Latins pour servir à l’histoire des doctrines grammaticales au moyen âge, ed. Charles Thurot (Paris: Imprimerie impériale, 1868) ‘Olafs saga hins helga’ efter pergamenthaandskrift i Uppsala Universitetsbibliotek, Delagardieske samling nr. 8ᴵᴵ, ed. Oscar Albert Johnsen (Kristiania: Jacob Dybwad, 1922) Orkneyinga saga, ed. Finnbogi Guðmundsson. Íslenzk fornrit, 34 (Reykjavík: Hið íslenzka fornritafélag, 1965) Otte bruddstykker av den ældste saga om Olav den Hellige, ed. Gustav Storm (Christiania: Det norske historiske kildeskriftsfond, 1893) Peter Comestor, Petri Comestoris ‘Scholastica historia’. Liber Genesis, ed. Agneta Sylwan. Corpus Christianorum. Series Latina, 191 (Turnhout: Brepols, 2005) Philippe de Thaon, ‘Le livre de Sibile’ by Philippe de Thaon, ed. Hugh Shields (London: Anglo-Norman Text Society, 1979) Plácidus saga. With an Edition of ‘Plácitus drápa’ by Jonna Louis-Jensen, ed. John Tucker. Editiones Arnamgnæanæ B, 31 (Copenhagen: Reitzel, 1998) The Poetic Edda, ed. and trans. Ursula Dronke, 3 vols (Oxford: Oxford University Press, 1969–2011) The Poetry of Arnórr jarlaskáld. An Edition and Study, ed. Diana Whaley. Westfield Publications in Medieval Studies, 8 (Turnhout: Brepols, 1998) Rímnasafn: Samling af de ældste islandske rimer, ed. Finnur Jónsson, 2 vols (København: Møller, 1905–22) ‘Saga Óláfs konungs hins helga’. Den store saga om Olav den hellige efter pergamenthåndskrift i Kungliga Biblioteket i Stockholm nr. 2 4to med varianter fra andre håndskrifter, ed. Oscar Albert Johnsen and Jón Helgason, 2 vols (Oslo: Kjeldeskriftfondet, 1941) Saxo Grammaticus, The History of the Danes. Books I–IX, ed. Hilda Ellis Davidson and trans. Peter Fisher, 2 vols (Cambridge: D. S. Brewer, 1979–80) Saxo Grammaticus, Gesta Danorum: Danmarkshistorien, ed. Karsten Friis-Jensen and trans. Peter Zeeberg, 2 vols (København: Gads forlag, 2005) Scriptores minores historiæ Danicæ, 1, ed. M. Cl. Gertz (København: Gad, 1917) Scriptores rerum mythicarum latini Romae nuper reperti, ed. Georgius Henricus Bode (Cellis: 1834) Skj = Den norsk-islandske skjaldedigtning, 4 vols (A 1–2; B 1–2), ed. Finnur Jónsson (København: Gyldendal, 1912–15)

326

Bibliography

SkP = Skaldic Poetry of the Scandinavian Middle Ages, ed. Margaret Clunies Ross, Kari Ellen Gade, Diana Whaley et al. (Turnhout: Brepols, 2007–) Småstykker. Samfund til Udgivelse af Gammel Nordisk Litteratur, 13 (København: Møller, 1884–91) Snorri Sturluson, Edda, trans. Anthony Faulkes (London: Everyman, 1987) Snorri Sturluson, Edda. Prologue and ‘Gylfaginning’, ed. Anthony Faulkes, 2nd ed. (London: Viking Society for Northern Research, 2005) Snorri Sturluson, Edda. Skáldskaparmál, ed. Anthony Faulkes (London: Viking Society for Northern Research, 1998) Snorri Sturluson, Edda. Háttatal, ed. Anthony Faulkes, 2nd ed. (London: Viking Society for Northern Research, 2007) Snorri Sturluson, Edda Snorra Sturlusonar, ed. Finnur Jónsson (København: Komissionen for det arnamagnæanske legat, 1931) Snorri Sturluson, Háttatal Snorra Sturlusonar, ed. Th. Möbius, 2 vols (Halle: Verlag der Buchhandlung des Waisenhauses, 1879–1881) Snorri Sturluson, Snorre Sturlasons ‘Edda’. Uppsala-handskriften DG 11, ed. Anders Grape, Gottfrid Kallstenius and Olof Thorell (Uppsala: Almqvist & Wiksell, 1962–1977) Snorri Sturluson, The Uppsala ‘Edda’, ed. Heimir Pálsson and trans. Anthony Faulkes (London: Viking Society for Northern Research, 2012) The So-Called Second Grammatical Treatise. An Orthographic Pattern of Late Thirteenth-Century Icelandic. Edition, Translation and Commentary, ed. Fabrizio Raschellà (Firenze: Felice Le Monnier, 1982) ‘Stjórn’. Tekst etter håndskriftene, ed. Reidar Astås, 2 vols (Oslo: Riksarkivet, 2009) ‘Sturlunga Saga’ Including the ‘Islendinga Saga’ of Lawman Sturla Thordsson and Other Works, ed. Gudbrand Vigfusson, 2 vols (Oxford: Clarendon Press, 1878) ‘Sturlunga saga’ efter membranen Króksfjarðarbók udfyldt efter Reykjarfjarðarbók, ed. Kristian Kålund, 2 vols (København: Gyldendal, 1906–11) Sturlunga saga, ed. Jón Jóhannesson, Magnús Finnbogason, and Kristján Eldjárn, 2 vols (Reykjavík: Sturlunguútgáfan, 1946) Sverris saga, ed. Þorleifur Hauksson. Íslenzk fornrit, 30 (Reykjavík: Hið íslenzka fornritafélag, 2007) The Vatican Mythographers, trans. Ronald E. Pepin (New York: Fordham University Press, 2008) Vatnsdœla saga, ed. Einar Ól. Sveinsson. Íslenzk fornrit, 8 (Reykjavík: Hið íslenzka fornritafélag, 1939) Veraldar saga, ed. Jakob Benediktsson. Samfund til udgivelse af gammel nordisk litteratur, 61 (København: Samfund til udgivelse af gammel nordisk litteratur, 1944) Two Versions of ‘Snorra Edda’, ed. Anthony Faulkes, 2 vols (Reykjavík: Stofnun Árna Magnússonar, 1977–79) Vestfirðinga sǫgur, ed. Björn K. Þórólfsson and Guðni Jónsson. Íslenzk fornrit, 6 (Reykjavík: Hið íslenzka fornritafélag, 1963) ‘Völuspá’ gefin út med skýringum, ed. Sigurður Nordal (Reykjavík: Prentsmiðjan Gutenberg, 1923)

Secondary Studies Abram, Christopher, Myths of the Pagan North. The Gods of the Norsemen (London: Continuum, 2011) Abram, Christopher, ‘The Post-Pagan Mythological Kenning’, in Eddic, Skaldic, and Beyond. Poetic Variety in Medieval Iceland and Norway, ed. Martin Chase (New York: Fordham University Press, 2014), pp. 44–61

Bibliography

327

Abram, Christopher, ‘Modeling Religious Experience in Old Norse Conversion Narratives: the Case of Óláfr Tryggvason and Hallfreðr vandræðaskáld’, Speculum, 90 (2015), 114–57 Acker, Paul, Revising Oral Theory. Formulaic Composition in Old English and Old Icelandic Verse (New York: Garland Publishing, 1998) Adams Leeming, David, The World of Myth. An Anthology (Oxford: Oxford University Press, 1990) Andersen, Merete Geert, Katalog over AM Accessoria 7. De latinske fragmenter. Bibliotheca Arnamagnæana, 46 (Hafniæ: Reitzel, 2008) Andersson, Theodore, ‘Skalds and Troubadours’, Mediaeval Scandinavia, 2 (1969), pp. 7–41 Andersson, Theodore, ‘Kings’ Sagas (Konungasögur)’, in Old Norse-Icelandic Literature. A Critical Guide, ed. Carol J. Clover and John Lindow. Islandica, 45 (Ithaca: Cornell University Press, 1985) Andersson, Theodore, The Growth of the Medieval Icelandic Sagas (1180–1280) (Ithaca: Cornell University Press, 2006) Ármann Jakobsson, ‘The Life and Death of the Icelandic Short Story’, Journal of English and Germanic Philology, 112 (2013), pp. 257–91 Arndís Hulda Auðunsdóttir and Aðalheiður Guðmundsdóttir, ‘“Berðu mér ei blandað vín”’. Um rímnakveðskap Sigurðar Breiðfjörð, skáldskaparmjöðinn og áfengið’, Són, 13 (2015), pp. 11–35 Ascoli, Albert Russell, Dante and the Making of a Modern Author (Cambridge: Cambridge University Press, 2008) Ásgeir Blöndal Magnússon, Íslensk orðsifjabók (Reykjavík: Mál og menning, 1989) Ashman Rowe, Elizabeth, The Development of Flateyjarbók. Iceland and the Norwegian Dynastic Crisis of 1389 (Odense: The University Press of Southern Denmark, 2005) Ashman Rowe, Elizabeth, ‘Quid Sigvardus cum Christo? Moral Interpretations of Sigurðr Fáfnisbani in Old Norse Literature’, Viking and Medieval Scandinavia, 2 (2006), pp. 167–200 Astås, Reidar, An Old Norse Biblical Compilation. Studies in’ Stjórn’ (New York: Peter Lang, 1991) Bagge, Sverre, ‘The Old Norse Kings’ Sagas and European Latin Historiography’, Journal of English and Germanic Philology, 115 (2016), pp. 1–38 Baldur Hafstað, Die ‘Egils saga’ und ihr Verhältnis zu anderen Werken des nordischen Mittelalters (Reykjavík: Rannsóknarstofnun kennaraháskóla Íslands, 1995) Beck, Heinrich, ‘Snorri Sturlusons Mythologie: Euhemerismus oder Analogie?’, in Snorri Sturluson – Historiker, Dichter, Politiker, ed. Heinrich Beck, Wilhelm Heizmann, and Jan Alexander van Nahl (Berlin: de Gruyter, 2013), pp. 1–21 Benedikt Eyþórsson, ‘Reykholt and Church Centres’, in Church Centres. Church Centres in Iceland from the 11th to the 13th Century and their Parallels in other Countries, ed. Helgi Þorláksson (Reykholt: Snorrastofa, 2005), pp. 105–16 Bergsveinn Birgisson, ‘Inn i skaldens sinn. Kognitive, estetiske og historiske skatter i den norrøne skaldediktningen’ (PhD diss., University of Bergen, 2007) Beyschlag, Siegfried, ‘Möglichkeiten mündlicher Überlieferung in der Königssaga’, Arkiv för nordisk filologi, 68 (1953), pp. 109–39 Bishoff, Bernard, ‘Das griechische Element in der abendländischen Bildung des Mittelalters’, Byzantinische Zeitschrift, 44 (1951), pp. 27–55 Bjarni Aðalbjarnarson, Om de norske kongers sagaer. Skrifter utgitt av det Norske VidenskapsAkademi i Oslo, II. Hist.-filos. klasse, 1936, No. 4 (Oslo: Jacob Dybwad, 1937) Bjarni Einarsson, Skáldasögur. Um uppruna og eðli ástaskáldasagnanna fornu (Reykjavík: Bókaútgáfa menningarsjóðs, 1961) Bjarni Einarsson, ‘The Lovesick Skald: A Reply to Theodore Andersson’, Mediaeval Scandinavia, 4 (1971), pp. 21–41

328

Bibliography

Bjarni Einarsson, ‘On the Rôle of Verse in Saga-Literature’, Mediaeval Scandinavia, 7 (1974), pp. 118–25 Bjarni Einarsson, Litterære forudsætninger for ‘Egils saga’ (Reykjavík: Stofnun Árna Magnússonar, 1975) Bjarni Einarsson, ‘The Last Hour of Hallfreðr vandræðaskáld as Described in Hallfreðarsaga’, in Proceedings of the Eighth Viking Congress, Århus 24–31 August 1977, ed. Hans Bekker-Nielsen, Peter Foote and Olaf Olsen (Odense: Odense University Press, 1981), pp. 217–21 Bjarni Einarsson, ‘Íslendingadrápa’, Tímarit Háskóla Íslands, 4 (1989), pp. 127–31 Björn K. Þórólfsson, Rímur fyrir 1600 (Kaupmannahöfn: Hið íslenzka fræðafélag, 1934) Björn M. Ólsen, Runerne i den oldislandske literatur (København: Gyldendal, 1883) Björn M. Ólsen, ‘Landnáma og Egils saga’, Aarbøger for nordisk oldkyndihed oh historie, 2: 19 (1904), pp. 167 –247 Björn M. Ólsen, ‘Landnáma og Eyrbyggja saga’, Aarbøger for nordisk oldkyndighed og historie, 2: 20 (1905), pp. 81–117 Björn M. Ólsen, Om Gunnlaugs saga ormstungu. En kritisk undersøgelse (København: Andr. Fred. Høst & sønn, 1911) Bjorvand, Harald and Fredrik Otto Lindeman, Våre arveord. Etymologisk ordbok, 2nd ed (Oslo: Novus forlag, 2007) Boer, R. C., ‘Om kommentaren til Háttatal’, Arkiv för nordisk filologi, 43 (1927), pp. 262–309 Borst, Arno, Der Turmbau von Babel. Geschichte der Meinungen über Ursprung und Vielfalt der Sprachen und Völker, 4 vols (Stuttgart: Anton Hiersemann, 1957–63) Boulhosa, Patricia Pires, Icelanders and the Kings of Norway. Mediaeval Sagas and Legal Texts (Leiden: Brill, 2005) Breatnach, Liam, ‘Zur Frage der roscada im Irischen’, in Metrik und Medienwechsel. Metrics and Media, ed. Hildegard L. C. Tristram(Tübingen: Gunter Narr, 1991), pp. 197–205 Breatnach, Liam, ‘An Mheán-Ghaeilge’ in Stair na Gaeilge in ómós do Phádraig Ó Fiannachta, ed. Kim McCone et al. (Maigh Nuad: Roinn na Sean-Ghaeilge, Coláiste Phádraig, 1994), pp. 221–333 Brøndsted, K., ‘En kirkelig allegori og en nordisk myte’, Historisk Tidskrift (norsk), 2:3 (1882), pp. 21–43 Brown, Ursula, ‘The Saga of Hrómund Gripsson and Þorgilssaga’, Saga-Book of the Viking Society, 13 (1947–48), pp. 51–77 Burnett, Charles S. F., ‘A Note on the Origins of the Third Vatican Mythographer’, Journal of the Warburg and Courtauld Institutes, 44 (1981), pp. 160–66 Cannon, Christopher, ‘The Middle English Writer’s Schoolroom: Fourteenth-Century English Schoolbooks and Their Contents’, New Medieval Literatures, 11 (2009), pp. 19–38 Carey, John, ‘On the Interrelationships of Some Cín Dromma Snechtai Texts’, Ériu, 47 (1995), pp. 71–92 Carruthers, Mary, The Experience of Beauty in the Middle Ages (Oxford: Oxford University Press, 2013) Charles-Edwards, T. M., ‘The Context and Uses of Literacy in Early Christian Ireland’, in Literacy in Medieval Celtic Societies, ed. Huw Pryce (Cambridge: Cambridge University Press, 1998), pp. 62–82 Chase, Martin, ‘Introduction’, in Eddic, Skaldic, and Beyond. Poetic Variety in Medieval Iceland and Norway, ed. Martin Chase (New York: Fordham University Press, 2014), pp. 1–15 Chase, Martin, ‘Devotional Poetry at the End of the Middle Ages in Iceland’, in Eddic, Skaldic, and Beyond. Poetic Variety in Medieval Iceland and Norway, ed. Martin Chase (New York: Fordham University Press, 2014), pp. 136–49 Chase, Martin, ‘Framir kynnask vátta mál: The Christian Background of Einarr Skúlason’s Geisli’, in Til heiðurs og hugbótar. Greinar um trúarkveðskap fyrri alda, ed. Svanhildur Óskarsdóttir and Anna Guðmundsdóttir (Reykholt: Snorrastofa, 2003), pp. 11–32 Ciklamini, Marlene, ‘Divine Will and the Guises of Truth in Geirmundar þáttr heljarskinns’, Skandinavistik, 11 (1981), pp. 81–88 Clover, Carol, ‘The Long Prose Form’, Arkiv för nordisk filologi 101 (1986), pp. 10–39

Bibliography

329

Clunies Ross, Margaret, ‘Skáldskaparmál’. Snorri Sturluson’s ‘Ars Poetica’ and Medieval Theories of Language (Odense: Odense University Press, 1987) Clunies Ross, Margaret, ‘Snorri’s Edda as Medieval Religionsgeschichte’, in Germanische Religionsgeschichte. Quellen und Quellenprobleme, ed. Heinrich Beck, Detlev Ellmers, and Kurt Schier (Berlin: De Gruyter, 1992), pp. 633–55 Clunies Ross, Margaret, A history of Old Norse Poetry and Poetics (Cambridge: D. S. Brewer, 2005) Clunies Ross, Margaret, ‘Stylistic and Generic Identifiers of the Old Norse Skaldic Ekphrasis’, Viking and Medieval Scandinavia, 3 (2007), pp. 159–92 Clunies Ross, Margaret, ‘Verse and Prose in Egils saga Skallagrímssonar’, in Creating the Medieval Saga. Versions, Variability and Editorial Interpretations of Old Norse Saga Literature, ed. Judy Quinn and Emily Lethbridge (Copenhagen: University Press of Southern Denmark, 2010), pp. 191–211 Clunies Ross, Margaret, ‘Reginnaglar’, in News from Other Worlds. ‘Tíðendi ór ǫðrum heimum’. Studies in Folklore, Mythology and Culture in Honor of John F. Lindow, ed. Merrill Kaplan and Timothy R. Tangherlini (Berkeley: North Pinehurst Press, 2012), pp. 3–21 Clunies Ross, Margaret, ‘Snorri Sturluson and the Construction of Norse Mythography’, in Writing Down the Myths, ed. Joseph Falaky Nagy. Cursor Mundi, 17 (Turnhout: Brepols, 2013), pp. 201–21 Clunies Ross, Margaret and Kari Ellen Gade, ‘Cosmology and Skaldic Poetry’, Journal of English and Germanic Philology, 111:2 (2012), pp. 199–207 Cole, Richard, The Jew Who Wasn’t There: Studies on Jews and their Absence in Old Norse Literature (PhD Diss., Harvard University, 2015) Copeland, Rita, Rhetoric, Hermeneutics, and Translation in the Middle Ages. Academic Traditions and Vernacular Texts (Cambridge: Cambridge University Press, 1991) Copeland, Rita, ‘Naming, Knowing and the Object of Language in Alexander Neckham’s Grammar Curriculum’, The Journal of Medieval Latin, 20 (2010), pp. 38–57 Copeland, Rita, ‘Grammar, Rhetoric, and Figurative Language: Learned Innovations and Vernacular Receptions’, in Intellectual Culture in Medieval Scandinavia c. 1100–1350, ed. Stefka G. Eriksen (Turnhout: Brepols, 2016), pp. 213–42 Cormack, Margaret, ‘Egils saga, Heimskringla, and the Daughter of Eiríkr blóðøx’, Alvíssmál, 10 (2001), pp. 61–68 Coulson, Frank T., ‘Hitherto Unedited Medieval and Renaissance Lives of Ovid’, Mediaeval Studies, 49 (1987), pp. 152–207 Coulson, Frank T., ‘Ovid’s Metamophoses in the School Tradition of France, 1180–1400: Texts, Manuscript Traditions, Manuscript Settings’, in Ovid in the Middle Ages, ed. James G. Clark, Frank T. Coulson, and Kathrym L. McKinley (Cambridge: Cambrdige University Press, 2011), pp. 48–82 Crook, Eugene J., ‘Gregory’s Dialogi and the Old Norse Sagas: Njáls Saga’, in Rome and the North. The Early Reception of Gregory the Great in Germanic Europe, ed. Rolf H. Bremmer Jr, Kees Dekker, and David F. Johnson (Paris: Peeters, 2001), pp. 275 –85 Dating the Sagas. Reviews and Revisions, ed. Else Mundal (Copenhagen: Museum Tusculanum Press, 2013) Derolez, René, Runica Manuscripta. The English Tradition (Brugge: Rijkuniversiteit te Gent, 1954) Dronke, Peter, Fabula. Explorations into the Uses of Myth in Medieval Platonism (Leiden: Brill, 1974) Dronke, Ursula and Peter, ‘The Prologue of the Prose Edda: Explorations of a Latin Background’, in Sjötíu ritgerðir helgaðar Jakobi Benediktssyni, ed. Einar G. Pétursson and Jónas Kristjánsson (Reykjavík: Stofnun Árna Magnússonar, 1977), pp. 153–76 Dronke, Ursula, ‘Sem jarlar forðum. The Influence of Rígsþula on two saga-episodes’, in Speculum Norroenum: Norse Studies in Memory of Gabriel Turville-Petre, ed. Ursula Dronke et al. (Odense: Odense University Press, 1981), pp. 56–72

330

Bibliography

Duncan, Thomas G., ‘”Quid Hinieldus cum Christo”. The Secular Expression of the Sacred in Old and Middle English Lyrics’, in Sacred and Secular in Medieval and Early Modern Cultures. New Essays, ed. Lawrence Besserman (Palgrave Macmillan: New York, 2006), pp. 29–46) Egeler, Matthias, ‘Eikþyrnir and the Rivers of Paradise. Cosmological Perspectives on Dating Grímnismál 26–28’, Arkiv för nordisk filologi, 128 (2013), pp. 17–39 Einar Ól. Sveinsson, ‘Dróttkvæða þáttur’, Skírnir, 121 (1947), pp. 5–32 Einar Ól., Dating the Icelandic Sagas. An Essay in Method (London: Viking Society for Northern Research, 1958) Einar Ól., ‘Kormákr the Poet and his Verses’, Saga-Book, 17 (1966), pp. 18–60 Einar Ól., ‘An old Irish Verse form wandering in the North’, in Proceedings of the Seventh Viking Congress, Dublin 15–21 August 1973, ed. Bo Almqvist and David Greene (Dublin: Royal Irish Academy, 1976), pp. 141–52 Einar Sigmarsson, ‘Hamskipti eða endaskipti? Um nykur og nykrað, finngálkn og finngálknað’, Gripla, 16 (2005), pp. 287–98 Eisner, Martin, Boccaccio and the Invention of Italian Literature. Dante, Petrach, Cavalcanti, and the Authority of the Vernacular (Cambridge: Cambridge University Press, 2013) Elliott, Kathleen O. and J. P. Elder, ‘A Critical Edition of the Vatican Mythographers’, Transactions and Proceeding of the American Philological Association, 78 (1947), pp. 189–207 Faulkes, Anthony, ‘The Sources of Skáldskaparmál: Snorri’s Intellectual background’, in Snorri Sturluson. Kolloquium anläßlich der 750. Wiederkehr seines Todestages, ed. Alois Wolf (Tübingen: Gunter Narr Verlag, 1993), pp. 59–76 Faulkes, Anthony, Poetical Inspiration in Old Norse and Old English Poetry (London: Viking Society for Northern Research, 1997) Fidjestøl, Bjarne, ‘The Tale of Haraldr harðráði and Þorgils the Fisherman’, in Bjarne Fidjestøl, Selected Papers, ed. Odd Einar Haugen and Else Mundal (Odense: Odense University Press, 1997 [1971]), pp. 277–93 Fidjestøl, Bjarne, Det norrøne fyrstediktet (Øvre Ervik: Alvheim & Eide, 1982) Fidjestøl, Bjarne, ‘Sogekvæde’, in Deutsch-nordische Begegnungen. 9. Arbeitstagung der Skandinavisten des deutschen Sprachgebiets 1989 in Svendborg, ed. Kurt Braunmüller and Mogens Brøndsted (Odense: Odense University Press, 1991), pp. 57–76 Fidjestøl, Bjarne, ‘Pagan Beliefs and Christian Impact: The Contribution of Skaldic Studies’, in Viking Revaluations: Viking Society Centenary Symposium 14–15 May 1992, ed. Anthony Faulkes ochRichard Perkins (London: Viking Society for Northern research, 1993), pp. 100–20 Fidjestøl, Bjarne, The Dating of Eddic Poetry, ed. Odd Einar Haugen (Hafniæ: Reitzel, 1999) Finlay, Alison, ‘Skalds, Troubadours and Sagas’, Saga-Book, 24 (1995), pp. 105–53 Finlay, Alison, ‘Skald Sagas in their Literary Context, 2: Possible European Contexts’, in Skaldsagas. Text, Vocation, and Desire in the Icelandic Sagas of Poets, ed. Russell Poole (Berlin: De Gruyter, 2001) pp. 232–71 Finnur Jónsson, Kritiske studier over en del af de ældste norske og islandske skjaldekvad (København: Gyldendal, 1884) Finnur Jónsson, ‘Edda Snorra Sturlusonar – dens oprindelige form og sammensætning’, Aarbøger for nordisk oldkyndighed og historie, 2nd series, 13 (1898), pp. 283–357 Finnur Jónsson, Det norsk-islandske skjaldesprog omtr. 800–1300 (København: Møllers bogtrykkeri, 1901) Finnur Jónsson, ‘Versene i Hallfredssaga’, Arkiv för nordisk filologi, 18 (1902), pp. 305–30 Finnur Jónsson, ‘Om Njála’, Aarbøger for nordisk oldkyndighed og historie, 2: 19 (1904), 89–166 Finnur Jónsson, ‘Tilnavne i den islandske oldlitteratur’, Aarbøger for nordisk oldkyndighet og historie, 2: 22 (1907), pp. 161–381.

Bibliography

331

Finnur Jónsson, ‘Sagaernes lausavísur’, Aarbøger for nordisk oldkyndighed og historie, 3: 2 (1912), pp. 1–57 Finnur Jónsson, Den oldnorske og oldislandske litteraturs historie, 2nd ed., 3 vols (København: G. E. C. Gads forlag, 1920–24) Finnur Jónsson, ‘Snorre Sturlusons Háttatal’, Arkiv för nordisk filologi, 45 (1929), pp. 229–69 Finnur Jónsson, Lexicon Poeticum antiquae linguae septentrionalis. Ordbog over det norsk-islandske skjaldesprog, oprindelig forfattet av Sveinbjörn Egilsson, 2nd ed. (København: Møllers, 1931) Finnur Jónsson, ‘Kenningers led-omstilling og tmesis’, Arkiv för Nordisk filologi, 49 (1933), pp. 1–23 Fjellhammer Seim, Karin, ‘Er “et upåaktet gammelnorsk ord: hausi” belagt på Bryggen i Bergen?’, Maal og Minne (1984), pp. 85–89 Foote, Peter, ‘Notes on the Prepositions of and um(b) in Old Icelandic and Old Norwegian’, Studia Islandica, 14 (1955), pp. 41–83 Foote, Peter,, ‘An Essay on the Saga of Gisli’, in The Saga of Gisli, trans. George Johnston (Toronto: University of Toronto Press, 1963), pp. 93–134 Foote, Peter,, ‘Beginnings and Endings: Some Notes on the Study of Skaldic Poetry’, in Peter Foote, Aurvandilstá, ed. Michael Barnes, Hans Bekker-Nielsen, and Gerd Wolfgang Weber (Odense: Odense University Press, 1984 [1976]), pp. 236–48 Foote, Peter,, ‘Latin Rhetoric and Icelandic Poetry. Some Contacts’, in Aurvandilstá. Norse Studies, ed. Michael Barnes, Hans Bekker-Nielsen, Gerd Wolfgang Weber (Odense: Odense University Press, 1984), pp. 249–70 Foote, Peter,, ‘Sagnaskemtan: Reykjahólar 1119’, in Peter Foote, Aurvandilstá, ed. Michael Barnes, Hans Bekker-Nielsen, and Gerd Wolfgang Weber (Odense: Odense University Press, 1984), pp. 65–83 Foote, Peter,, ‘Saints’ Lives and Sagas’, in Saints and sagas. A Symposium, ed. Hans Bekker-Nielsen and Birte Carlé (Odense: Odense Universitetsforlag, 1994), pp. 5–26 Förster, Max, ‘Beiträge zur mittelalterlichen Volkskunde, 4’, Archiv für das Studium der neueren Sprachen und Literaturen, 125 (1910), pp. 39–70 Frank, Roberta, ‘Onomastic Play in Kormákr’s Verse: The Name Steingerðr’, Mediaeval Scandinavia, 3 (1970), 7–34 Frank, Roberta, Old Norse Court poetry. The ‘Dróttkvætt’ Stanza (Ithaca, N.Y: Cornell University Press, 1978) Frank, Roberta, ‘Snorri and the Mead of Poetry’, in Speculum Norroenum: Norse Studies in Memory of Gabriel Turville-Petre, ed. Ursula Dronke et al. (Odense: Odense University Press, 1981), pp. 155–70 Frank, Roberta, ‘Skaldic Poetry’, in Old Norse-Icelandic Literature. A Critical Guide, ed. Carol J. Clover and John Lindow. Islandica, 45 (Ithaca: Cornell University Press, 1985), pp. 157–96 Frans, Victor, ‘Sub regulis Donati’. ‘Ars Brugensis’ och Den Tredje grammatiska avhandlingen (MA thesis, University of Oslo, 2019) Friis-Jensen, Karsten, Saxo Grammaticus as a Latin Poet. Analecta Romana Instituti Danici. Supplementum, 14 (Roma: L’erma di Bretschneider, 1987) Frank, Roberta, ‘The Reception of Horace in the Middle Ages’, in The Cambridge Companion to Horace, ed. Stephen Harrison (Cambridge: Cambridge University Press, 2007), pp. 291–304 Frank, Roberta, ‘Saxo Grammaticus og fornaldarsagaerne’, in Fornaldarsagaerne. Myter og virkelighed, ed. Agneta Ney, Ármann Jakobsson, and Annette Lassen (København: Museum Tusculanum, 2009), pp. 67–77 Fritzner, Johan, Ordbog over det gamle norske Sprog, 3 vols (Kristiania: Den norske forlagsforening, 1883–96) Gade, Kari Ellen, The Structure of Old Norse ‘Dróttkvætt’ Poetry. Islandica, 49 (Ithaca: Cornell, 1995) Gade, Kari Ellen, ‘Penile Puns: Personal Names and Phallic Symbols in Skaldic Poetry’, Essays in Medieval Studies, 6 (1989), pp. 57–67

332

Bibliography

Gade, Kari Ellen, ‘The Dating and Attribution of Verses in the Skald Sagas’, in Skaldsagas. Text, Vocation, and Desire in the Icelandic Sagas of Poets, ed. Russell Poole (Berlin: De Gruyter, 2001), pp. 50–74 Gade, Kari Ellen, ‘Ælfric in Iceland’, in Learning and Understanding in the Old Norse World. Essays in Honour of Margaret Clunies Ross, ed. Judy Quinn, Kate Heslop, and Tarrin Wills (Turnhout: Brepols, 2007), pp. 321–39 Ghisalberti, Fausto, ‘Medieval Biographies of Ovid’, Journal of the Warburg and Courtauld Institutes, 9 (1946), pp. 10–59 Ghosh, Shami, Kings‘ Sagas and Norwegian History. Problems and Perspectives (Leiden: Brill, 2011) Gjessing, Gustav Antonio, ‘Egils-saga’s Forhold til Kongesagaen’, Arkiv för nordisk filologi, 2 (1885), pp. 289–318 Gneuss, Helmut, Hymnar und Hymnen in englischen Mittelalter. Studien zur Überlieferung, Glossierung und Übersetzung lateinischen Hymnen in England. Mit einer Textausgabe der lateinisch-altenglischen Expositio Hymnorum (Tübingen: Max Niemeyer, 1968) Gödel, Vilhelm, Katalog öfver Upsala universitets biblioteks fornisländska och fornnorska handskrifter. Skrifter utgifna af Humanistiska Vetenskapssamfundet i Upsala, 2: 1 (Uppsala: Almqvist & Wiksell, 1892), pp. 27–29) Grønlie, Siân, ‘Saint’s Life and Saga Narrative’, Saga-Book, 36 (2012), pp. 5–26 Grønlie, Siân, The Saint and the Saga Hero. Hagiography and Early Icelandic Literature (Woodbridge: Boydell and Brewer, 2017) Grotans, Anna, Reading in Medieval St. Gall (Cambridge: Cambridge University Press, 2006) Grove, Jonathan, ‘Skaldic Verse-Making in Thirteenth-Century Iceland: The Case of the Sauðafellsferðarvísur’, Viking and Medieval Scandinavia, 4 (2008), pp. 85–131 Gschwantler, Otto, ‘Christus, Thor und die Midgardschlange’, in Festschrift für Otto Höfler zum 65. Geburtstag, ed. Helmut Birkhan and Otto Gschwantler, 2 vols (Wien: Notring, 1968), 1, pp. 145–68 Guðbjörg Kristjánsdóttir, ‘Íslenskt saltarablað í Svíþjóð’, Skírnir, 157 (1983), pp. 64 –73 Guðmundur Finnbogason, ‘Nogle bemærkninger om skjaldedigtningens kenningar’, Acta philologica Scandinavica, 9 (1934–35), pp. 69–75 Guðrún Nordal, Tools of Literacy. The Role of Skaldic Verse in Icelandic Textual Culture of the Twelfth and Thirteenth Centuries (Toronto: University of Toronto Press, 2001) Guðrún Nordal, ‘Samhengið í íslenskum fornbókmenntum’, in Sagnaheimur. Studies in Honour of Hermann Pálsson on his 80th Birthday, 26th May 2001, ed. Ásdís Egilsdóttir and Rudolf Simek (Wien: Fassbaender, 2001), pp. 91–106 Guðrún Nordal, Skaldic Versifying and Social Discrimination in Medieval Iceland. The Dorothea Coke Memorial Lecture in Northern Studies delivered at the University College London 15 March 2001 (London: Viking Society for Northern Research, 2003) Guðrún Nordal, ‘Ars Metrica and the Composition of Egils saga’, in Scandinavia and Christian Europe in the Middle Ages. Papers of the 12th International Saga Conference, Bonn, Germany, 28th July – 2nd August 2003, ed. Rudolf Simek and Judith Meurer (Bonn: Hausdruckerei der Universität Bonn, 2003), pp. 179–86 Guðrún Nordal, ‘The Art of Poetry and the Sagas of Icelanders’, in Learning and Understanding in the Old Norse World. Essays in Honour of Margaret Clunies Ross, ed. Judy Quinn, Kate Heslop, and Tarrin Wills (Turnhout: Brepols, 2007), pp. 219–37 Guðrún Nordal, ‘The Dialogue between Audience and Text: The Variants in Verse Citations in Njáls saga’s Manuscripts’, in Oral Art Forms and their Passage into Writing, ed. Else Mundal and Jonas Wellendorf (Copenhagen: Museum Tusculanum Press, 2008), pp. 185–202 Guðrún Nordal, ‘Skaldic Citations and Settlement Stories as Parametres for Saga Dating’, in Dating the Sagas. Reviews and Revisions, ed. Else Mundal (Copenhagen: Museum Tusculanum Press, 2013), pp. 195–212

Bibliography

333

Guðrún Nordal, ‘Skaldic Poetics and the Making of the Sagas of Icelanders’, in New Norse Studies. Essays on the Literature and Culture of Medieval Scandinavia, ed. Jeffrey Turco. Islandica, 58 (Ithaca, NY: Cornell University Library, 2015) Guðrún Nordal, Sverrir Tómasson and Vésteinn Ólason, Íslensk bókmenntasaga, 1 (Reykjavík: Mál og Menning, 1992) Gunnar Harðarson, ‘“Alls vér erum einnar tungu”. Um skyldleika ensku og íslensku í Fyrstu málfræðiritgerðinni’, Íslenskt mál, 21 (1999), pp. 11–30 Gurevich, Elena A., ‘The System of Kennings’, Nordica Bergensia, 3 (1994), pp. 139–56 Hagland, Jan Ragnar, ‘Eit skriftleg førelegg for Øyvind Skaldespillars dikt i Fagrskinna?’, Maal og Minne (1975), pp. 12–19 Hægstad, Marius, ‘Um maalet i Ólafs saga hins helga’, in Olafs saga hins helga efter pargamenthaandskrift i Uppsala Universitetsbibliotek, Delagardieske samling nr. 8ᴵᴵ, ed. Oscar Albert Johnsen (Kristiania: Jacob Dybwad, 1922), pp. xxviii–lvii Hallberg, Peter, Snorri Sturluson och ‘Egils saga Skallagrímssonar’. Ett försök till språklig författarbestämning. Studia Islandica, 20 (Reykjavík: Bókaútgáfa menningarsjóðs, 1962) Hallberg, Peter, Stilsignalament och författarskap i norrön sagalitteratur. Synpunkter och exempel (Göteborg: Acta Universitatis Gothoburgensis, 1968) Halvorsen, Eyvind Fjeld, ‘Fagrskinna’, in Kulturhistoriskt lexikon för nordisk medeltid, 22 vols (Malmö: Allhems förlag, 1956–78), 4, cols 139–40 Haraldur Bernharðsson, ‘Göróttur er drykkurinn. Fornmálsorð í nútímabúningi’, Gripla, 17 (2006), pp. 37–73 Harris, Joseph, ‘The Prosimetrum of Icelandic Saga and Some Relatives’, in Prosimetrum. Crosscultural Perspectives on Narrative in Prose and Verse, ed. Joseph Harris and Karl Reichl (Cambridge: D. S. Brewer, 1997), pp. 131–63 Harris, Joseph and Tom D. Hill, ‘Gestr’s “Prime Sign”: Source and Signification in Norna-Gests þáttr’, Arkiv för nordisk filologi, 104 (1989), pp. 103–22 Haug, Walter, Vernacular Literary Theory in the Middle Ages. The German Tradition, 800–1300, in its European Context, trans. Joanna M. Catling (Cambridge: Cambridge University Press, 1997 [1985]) Haukur Þorgeirsson, ‘Late Placement of the Finite Verb in Old Norse Fornyrðislag Meter’, Journal of Germanic Linguistics, 24 (2012), pp. 233–69 Haukur Þorgeirsson, ‘Snorri versus the Copyist. An Investigation of a Stylistic Trait in the Manuscript Traditions of Egils Saga, Heimskringla and the Prose Edda’, Saga-Book, 38 (2014), pp. 61–74 Haukur Þorgeirsson, book review: Lasse Mårtensson, Skrivaren och förlagan. Amsterdamer Beiträge der älteren Germanistik, 76 (2016), pp. 152–53 Haukur Þorgeirsson, ‘The Dating of Eddic Poetry – Evidence from Alliteration’, in Approaches to Nordic and Germanic Poetry, ed. Kristján Árnason et al. (Reykjavík: Háskólaútgáfan, 2017), pp. 33–61 Haukur Þorgeirsson, ‘A Stemmatic Analysis of the Prose Edda’, Saga-Book, 41 (2017), 49–70 Haukur Þorgeirsson, ‘How Similar are Heimskringla and Egils saga? An Application of Burrow’s Delta to Icelandic Texts’, European Journal of Scandinavian Studies, 48 (2018), pp. 1–18 Haukur Þorgeirsson, ‘Reflections on the Creation of Snorri Sturluson’s Prose Edda’, Scripta Islandica, 68 (2017), pp. 189–232 Heimir Pálsson, ‘Að sætta heimildir. Lítið eitt um starfsaðferðir Snorra Sturlusonar’, in Heiðin minni. Greinar um fornar bókmenntir, ed. Haraldur Bessason and Baldur Hafstað (Reykjavík: Heimskringla, 1999), pp. 159–79 Helgi Þorláksson, ‘Snorri í Odda. Um menntun Snorra Sturlusonar, uppeldi og mótun’, Skírnir (2014), pp. 353–80 Heller-Roazen, Daniel, Dark Tongues. The Art of Rogues and Riddlers (New York: Zone Books, 2013) Hermann Pálsson, Sagnaskemmtun íslendinga (Reykjavík: Mál og Menning, 1962)

334

Bibliography

Heusler, Andreas, Die gelehrte Urgeschichte im altisländischen Schrifttum. Abhandlungen der Königlichen Preussischen Akademie der Wissenschaften philosophisch-historische Classe 3 (Berlin: Verlag der Königlichen Preussischen Akademie der Wissenschaften, 1908) Hill, Thomas D., ‘Beer, Vomit, Blood, and Poetry. Egils saga, Chapters 44–45’, in New Norse Studies. Essays on the Literature and Culture of Medieval Scandinavia, ed. Jeffrey Turco. Islandica, 58 (Ithaca, NY: Cornell University Library, 2015), pp. 243–54 Hofmann, Dietrich, ‘Das Reimwort giǫr in Egill Skallagrímssons Hǫfuðlausn’, Mediaeval Scandinavia, 6 (1973), pp. 93–101 Holtsmark, Anne, En islandsk scholasticus fra det 12. århundre (Oslo: Jacob Dybwad, 1936) Holtsmark, Anne, ‘Bjarne Kolbeinsson of hans forfatterskap’, Edda (1937), pp. 1–17 Holtsmark, Anne, ‘Det norrøne ord lúðr’, Maal og Minne (1946), pp. 49–65 Holtsmark, Anne, book review: Odd Nordland, Hǫfuðlausn i Egils saga. Ein traditionskritisk studie (Oslo: Samlaget, 1956), Maal og Minne, 1956, pp. 130–38 Holtsmark, Anne, ‘Kjærlighetsdiktning’, in Kulturhistoriskt lexikon för nordisk medeltid, 22 vols (Malmö: Allhems förlag, 1956–78), 8, cols 438–44 Holtsmark, Anne, Studier i Snorres mytologi (Oslo: Universitetsforlaget, 1964) Holtsmark, Anne, ‘Heroic Poetry and Legendary Sagas’, Bibliography of Old Norse-Icelandic Studies (1965), pp. 9–21 Hreinn Benediktsson, ‘Phonemic Neutralization and Inaccurate Rhymes’, Acta philologica Scandinavica, 26 (1964), pp. 1–18 Hreinn Benediktsson, Early Icelandic Script as Illustrated in Vernacular Texts from the Twelfth and Thirteenth Centuries. Íslenzk handrit. Series in Folio, 2 (Reykjavík: The Manuscript Institute of Iceland, 1965) Hreinn Benediktsson, ‘Relational Sound Change: vá > vo in Icelandic’, in Lingustic Studies, Historical and Comparative, ed. Guðrún Þórhallsdóttir et al. (Reykjavík: Institute of Linguistics, 2002 [1979]), pp. 227–42 Hume, Kathryn, ‘Thematic Design of Grettis saga’, Journal of English and Germanic Philology, 73 (1974), pp. 469–86 Hunt, Tony, Teaching and Learning Latin in 13th-Century England, 3 vols (Cambridge: D. S. Brewer, 1991) Indrebø, Gustav, Fagrskinna (Kristiania: Grøndahl, 1917) Irvine, Martin, The Making of Textual Culture. ‘Grammatica’ and Literary Theory 350–1100 (Cambridge: Cambridge University Press, 1994) Jakob Benediktsson, ‘Hvar var Snorri nefndur höfundur Heimskringlu?’, Skírnir, 129 (1955), pp. 118–27 Jakob Benediktsson, ‘Sturlunga saga’, in Kulturhistoriskt lexikon för nordisk medeltid, 22 vols (Malmö: Allhems förlag, 1956–78), 17, pp. 355–59 Jakob Benediktsson, ‘Hafgerðingadrápa’, in Speculum Norroenum: Norse Studies in Memory of Gabriel Turville-Petre, ed. Ursula Dronke et al. (Odense: Odense University Press, 1981), pp. 27–32 Janson, Henrik, ‘Edda and “Oral Christianity”: Apocryphal Leaves of the Early Medieval Storyworld in the North’, in The Performance of Christian and Pagan Storyworlds. Non-Canonical Chapters of Nordic Medieval Literature, eds. Lars Boje Mortensen and Tuomas M. S. Lehtonen with Alexandra Bergholm (Turnhout: Brepols, 2013), pp. 171–97 Janus Jónsson, ‘Um vísurnar í Harðar sögu Grímkelssonar’, Tímarit hins íslenzka bókmenntafjelags, 13 (1892), pp. 259–75 Jesch, Judith, ‘In Praise of Ástríðr Óláfsdóttir’, Saga-Book, 24 (1994–97), pp. 1–18 Jesch, Judith, ‘Skaldic Verse in Scandinavian England’, in Vikings and the Danelaw. Select papers from the proceedings of the Thirteenth Viking Congress, Nottingham and York, 21–30 August 1997, ed. James Graham-Campbell et al. (Oxford: Oxbow Books, 2001), pp. 313–25

Bibliography

335

Jesch, Judith, ‘The Orcadian Links of Snorra Edda’, in Snorres ‘Edda’ i europeisk og islandsk kultur, ed. Jon Gunnar Jørgensen (Reykholt: Snorrastofa, 2009), pp. 145–72 Jochens, Jenny, ‘Mansǫngr and the Jómsvíkingar’, in Le secret d’Odin. Mélanges offerts à Régis Boyer, ed. Marc Auchet (Nancy: Presses universitaires de Nancy, 2001), pp. 17–26 Johansson, Karl G., Studier i Codex Wormianus. Skrifttradition och avskriftsverksamhet vid ett isländskt skriptorium under 1300-talet (Göteborg: Acta universitatis gothoburgensis, 1997) Johansson, Karl G., ‘Rígsþula och Codex Wormianus: Textens funktion ur ett kompilationsperspektiv’, Alvíssmál, 8 (1998), pp. 67–84 Johansson, Karl G., ‘Översättning och originalspråkstext i handskriftstraderingens våld: Merlínuspá och Vǫluspá i Hauksbók’, in Neue Ansätze in der Mittelalterphilologie, ed. Susanne Kramarz-Bein. Texte und Untersuchungen zur Germanistik und Skandinavistik, 55 (Frankfurt a.M.: Lang, 2005), pp. 97–113 Johansson, Karl G., ‘Texter i rörelse. Översättning, original textproduktion och tradering på norra Island 1150–1400’, in Übersetzen im skandinavischen Mittelalter, ed. Vera Johanterwage and Stephanie Würth (Wien: Fassbaender, 2007), pp. 83–106 Johansson, Karl G., ‘Vǫluspá, the Tiburtine Sibyl, and the Apocalypse in the North’, in The Nordic Apocalypse. Approaches to Vǫluspá and Nordic Days of Judgement, ed. Terry Gunnell and Annette Lassen (Turnhout: Brepols, 2013), pp. 161–84 Jón Helgason, ‘Ek bar sauð’, Acta philologica Scandinavica, 23 (1957), pp. 94–96 Jón Helgason, ‘Höfuðlausnarhjal’, in Einarsbók. Afmæliskveðja til Einars Ól. Sveinssonar 12. desember 1969, ed. Bjarni Guðnason et al. (Reykjavík, 1969), pp. 156–76 Jón Jóhannesson, Gerðir Landnámabókar (Reykjavík: Hið íslenzka bókmenntafjelag, 1941) Jón Þorkelsson, Om digtningen på Island i det 15. og 16. århundrede (København: Høst, 1888) Jón Karl Helgason, Sigurður Ingibergur Björnsson and Steingrímur Páll Kárason, ‘Fingraför fornsagnahöfunda. Fráleiðsla í anda Holmes og stílmæling í anda Burrows’, Skírnir, 191 (2017), pp. 273–309 Jónas Kristjánsson, ‘Íslendingadrápa and Oral Tradition’, Gripla, 1 (1975), pp. 76–91 Jónas Kristjánsson, ‘Var Snorri Sturluson upphafsmaður íslendingasagna?’, in Sagnalíf. Sextán greinar um fornar bókmenntir, ed. Þórður Ingi Guðjónsson (Reykjavík: Stofnun Árna Magnússonar í íslenskum fræðum, 2015 [1990]), pp. 250–74 Jónas Kristjánsson, ‘Kveðskapur Egils Skallagrímssonar’, Gripla, 17 (2006), pp. 7–35 Jónas Kristjánsson, ‘Learned Style or Saga Style?’, in Speculum Norroenum: Norse Studies in Memory of Gabriel Turville-Petre, ed. Ursula Dronke et al. (Odense: Odense University Press, 1981), 260–92 Jørgensen, Jon Gunnar, ‘“Snorre Sturlesøns fortale paa sin chrønicke”. Om kildene til opplysingen om Heimskringlas forfatter’, Gripla, 9 (1995), pp. 43–62 Jørgensen, Jon Gunnar, ‘Om verdien av sagaavskrifter fra 1600-tallet’, Collegium medievale, 11 (1998): 87–103 Jørgensen, Jon Gunnar, ‘Óláfr Haraldsson, King, Warrior and Saint: Presentations of King Óláfr Haraldsson the Saint in Medieval Poetry and Prose’, in King and Warrior in Early North-West Europe, ed. Jan Erik Rekdal and Charles Doherty (Dublin: Four Courts Press, 2016), pp. 345–98 Jørgensen, Jon Gunnar, The Lost Vellum Kringla (Copenhagen: Reitzel, 2007) Kålund, Kristian, Katalog over den arnamagnæanske håndskriftsamling, 2 vols (København: Kommissionen for det arnamagnæanske legat, 1889–94) Karlsen, Espen, ‘Fragments of Patristic and Other Ecclesiastical Literature in Norway from c. 1100 until the Fifteenth Century’, in Latin Manuscripts in Medieval Norway. Studies in Memory of Lilli Gjerløw, ed. Espen Karlsen (Oslo: Novus press, 2013), pp. 215–69 Kay, Sarah, ‘Occitan Grammar as a Science of Endings’, New Medieval Literatures, 11 (2009), pp. 39–61

336

Bibliography

Kendrick, Laura, The Game of Love. Troubadour Wordplay (Berkeley: University of California Press, 1988) Klingenberg, Heinz, ‘Edda’: Sammlung und Dichtung. Beiträge zur nordischen Philologie, 3 (Basel: Helbing und Lichtenhahn, 1974) Klingenberg, Heinz, ‘Gylfaginning. Tres vidit unum adoravit’, in Germanic Dialects. Linguistic and Philological Investigations, ed. Bela Brogyani and Thomas Krömmelbein (Amsterdam: John Benjamin, 1986), pp. 627–93 Knirk, James E., ‘Runes from Trondheim and a Stanza by Egill Skalla-Grímsson’, in Studien zum Altgermanischen, ed. Heiko Uecker. Ergänzungsbände zum Reallexikon der Germanischen Altertumskunde, 11 (Berlin: de Gruyter, 1994), pp. 411–20 Kock, Ernst Albin, Notationes Norroenae. Anteckningar till edda- och skaldediktning (Lund: Gleerup, 1923–44) Kock, Ernst Albin, ‘Ett kapitel nordisk metrik och textkritik’, Arkiv för nordisk filologi, 49 (1933), pp. 279–94 Konráð Gíslason, Efterladte skrifter, 1: Forelæsninger over oldnordiske Skjaldekvad (København: Gyldendal, 1895) Kraus, Manfred, ‘Grammatical and Rhetorical Exercises in the Medieval Classroom’, New Medieval Literatures, 11 (2009), pp. 63–89 Kristján Árnason, ‘Um Háttatal Snorra Sturlusonar. Bragform og braglýsing’, Gripla, 17 (2006), pp. 75–124 Kristján Árnason, ‘On the Principles of Nordic Rhyme and Alliteration’, Arkiv för nordisk filologi, 122 (2007), pp. 79–114 Kristján Árnason, ‘Vernacular and Classical Strands in Icelandic Poetics and Grammar in the Middle Ages’, North-Western European Language Evolution, 69: 2 (2016), pp. 191–235 Krömmelbein, Thomas, ‘Die Spitzenstillung des Geirmundar þáttr heljarskinns innerhalb der Kompilation Sturlunga saga’, Alvíssmál, 4 (1994), pp. 33–50) Kruken, Kristoffer and Ola Stemshaug, Norsk personnamnleksikon, 3rd ed. (Oslo: Samlaget, 2013) Kuhn, Hans, ‘Das nordgermanische Heidentum in den ersten christlichen Jahrhunderten’, Zeitschrift für deutsches Altertum und deutsche Literatur, 79 (1942), pp. 133–66 Kuhn, Hans, ‘Vor Tausend Jahren. Zur Geschichte des skaldischen Innenreims‘, in Speculum Norroenum: Norse Studies in Memory of Gabriel Turville-Petre, ed. Ursula Dronke et al. (Odense: Odense University Press, 1981), pp. 293–309 Kuhn, Hans, Das Dróttkvætt (Heidelberg: Carl Winter, 1983) La Farge, Beatrice and John Tucker, Glossary to the Poetic Edda. Based on Hans Kuhn's Kurzes Wörterbuch. Skandinavistische Arbeiten, 15 (Heidelberg: Winter, 1992) Larsson, Patrik, Yrrunan: Använding och ljudvärde i nordiska runinskrifter. Runrön, 17 (Uppsala: Institutionen för nordiska språk vid Uppsala universitet, 2002) Lassen, Annette, Odin på kristent pergament. En teksthistorisk studie (København: Museum Tusculanum, 2011) Lavender, Philip, ‘Merlin and the Vǫlva’, Viking and Medieval Scandinavia, 2 (2006), pp. 111–39 Law, Vivien, ‘Anglo-Saxon England: Aelfric’s “Excerptiones de arte grammatica anglice”’, Histoire Épistémologie Langage, 9 (1987), pp. 47–71 Law, Vivien, The History of Linguistics in Europe. From Plato to 1600 (Cambridge: Cambridge University Press, 2003) Leclercq, Jean, The Love of Learning and the Desire for God: A Study of Monastic Culture, trans. Catharine Mishrahi (New York: Fordham University Press, 1961 [1957]) Leclercq, Jean, Monks and Love in Twelfth-century France. Psycho-Historical Essays (Oxford : Oxford University Press, 1979)

Bibliography

337

LeFèvre, Yves, L’‘Elucidarium’ et les Lucidaires. Contribution, par l’histoire d’un texte, à l’histoire des croyances religieuses en France au moyen âge. Bibliothèque des Écoles Françaises d’Athènes et de Rome, 180 (Paris: Boccard, 1954) Lethbridge, Emily, ‘Dating the Sagas and Gísla saga Súrssonar’, in Dating the Sagas. Reviews and Revisions, ed. Else Mundal (Copenhagen: Museum Tusculanum Press, 2012), pp. 77–113 Lie, Hallvard, ‘Skaldestil-studier’, Maal og Minne (1952), pp. 1–92 Lie, Hallvard, Natur og unatur i skaldekunsten (Oslo: Aschehoug, 1957) Liestøl, Aslak, Wolfgang Krause, and Jón Helgason, ‘Drottkvætt-vers fra Bryggen i Bergen’, Maal og Minne (1962), pp. 98–108 Lind, E. H., Norsk-isländska personbinamn från medeltiden. Samlade och utgivna med förklaringar (Uppsala: Lundequistska bokhandeln, 1920–21) Lindblad, Gustaf, Studier i Codex Regius av Äldre Eddan (Lund: Gleerup, 1965) Lindow, John, ‘Riddles, Kennings, and the Complexity of Skaldic Poetry’, Scandinavian Studies, 47 (1975), pp. 311–27 Lindow, John, ‘Two Skaldic Stanzas in Gylfaginning: Notes on Sources and Text History’, Arkiv för nordisk filologi, 92 (1977), pp. 106–24 Lönnroth, Lars, ‘Hjalmar’s Death-Song and the Delivery of Eddic Poetry’, Speculum, 46 (1971), pp. 1–20 Lönnroth, Lars, review of Klaus von See, Mythos und Theologie, Skandinavistik, 20 (1990), pp. 43–47 Louis-Jensen, Jonna, Kongesagastudier. Kompilationen Hulda-Hrokkinskinna. Bibliotheca arnamagnæana, 32 (Hafniæ: Reitzel, 1977) Louis-Jensen, Jonna, ‘Heimskringla – Et verk af Snorri Sturluson?’, Nordica Bergensia, 14 (1997), pp. 230–45 Louis-Jensen, Jonna, ‘Heimskringla og Egils saga – samme forfatter?’, in Studier i Nordisk. Selskab for Nordisk Filologi. Foredrag og årsberetning, ed. Dorthe Duncker (København: Selskab for Nordisk Filologi, 2009), pp. 103–11 Louis-Jensen, Jonna, ‘Dating the Archetype. Eyrbyggja saga and Egils saga Skallagrímssonar’, in Dating the Sagas. Reviews and Revisions, ed. Else Mundal (Copenhagen: Museum Tusculanum Press, 2013), pp. 133–47 LP = Finnur Jónsson, Lexicon Poeticum antiquae linguae septentrionalis. Ordbog over det norsk-islandske skjaldesprog, oprindelig forfattet av Sveinbjörn Egilsson, 2nd ed. (København: Møllers, 1931) Lummus, David, ‘Bocaccio’s Poetic Anthropology: Allegories of History in Genealogie deorum gentilium libri’, Speculum, 87 (2012), pp. 724–65 Mac Cana, Proinsias, ‘Mongán Mac Fiachna and Immram Brain’, Ériu, 23 (1972), pp. 102–42 Mac Cana, ‘Prosimetrum in Insular Celtic Literature’, in Prosimetrum. Crosscultural Perspectives on Narrative in Prose and Verse, ed. Joseph Harris and Karl Reichl (Cambridge: D. S. Brewer, 1997), pp. 99–130 Magerøy, Hallvard, Studiar i Bandamanna saga: Kring gjerd-problemet. Bibliotheca Arnamagnæana, 18 (København: Munksgaard, 1957) Magnús Stefansson, ‘De islandske stadenes egenart og eldste historie’, in Church Centres. Church Centres in Iceland from the 11th to the 13th Century and their Parallels in other Countries, ed. Helgi Þorláksson (Reykholt: Snorrastofa, 2005), pp. 117–25 Males, Mikael, ‘Egill och Kormákr – tradering och nydiktning’, Maal og Minne (2011), pp. 115–46 Males, Mikael,, ‘Wormianusredaktören: Språk, tro och sanning vid 1300-talets mitt’, Arkiv för nordisk filologi, 128 (2013), pp. 41–77 Males, Mikael,, ‘Allegory in Old Norse Secular Literature: Theoretical and Methodological Challenges’, Viking and Medieval Scandinavia, 9 (2013), pp. 99–132

338

Bibliography

Males, Mikael,, ‘Er Ólafur Þórðarson höfundur Eglu?’, Són, 13 (2015), pp. 173–79 Males, Mikael,, ‘1100-talets pseudonyma skaldediktning: En kritisk granskning’, Maal og Minne (2017), pp. 1–24 Males, Mikael,, ‘The Last Pagan’, JEGP, 116 (2017), pp. 491–514 Males, Mikael,, ‘Character, Provenance, and Use of the Icelandic Fifth Grammatical Treatise’, Arkiv för nordisk filologi, 132 (2017), pp. 121–38 Males, Mikael,, ‘Pseudo-Remigius and the Old Icelandic Barbarismus: A Pilot Study’, in Reading Slowly. A Festschrift for Jens E. Braarvig, ed. Lutz Edzard, Jens Wilhelm Borgland and Ute Hüsken (Wiesbaden: Harrassowitz Verlag, 2018), pp. 321–331 Marchand, James, ‘Leviathan and the Mousetrap in the Niðrstigningasaga’, Scandinavian Studies, 47 (1975), pp. 328–38 Margrét Eggertsdóttir, ‘Eddulist og barokk í íslenskum kveðskap á 17. öld’, in Guðamjöður og arnarleir. Safn ritgerða um eddulist, ed. Sverrir Tómasson (Reykjavík: Háskólaútgáfan, 1996), pp. 91–116 Marold, Edith, ‘Nýgerving und nykrat’, North-Western European Language Evolution, 21/22 (1993), pp. 283–302 Marold, Edith, ‘Der Dialog in Snorris Gylfaginning’, in Snorri Sturluson. Beiträge zu Werk und Rezeption, ed. Hans Fix (Berlin: De Gruyter, 1998), pp. 131–80 Marold, Edith, ‘Mansǫngr – a Phantom Genre’, in Learning and Understanding in the Old Norse World. Essays in Honour of Margaret Clunies Ross, ed. Judy Quinn, Kate Heslop, and Tarrin Wills (Turnhout: Brepols, 2007), pp. 239–62 Mårtensson, Lasse, ‘Översikten över Háttatal i DG 11 4to – dess funktion och ursprung’, Gripla, 20 (2009), pp. 105–45 Mårtensson, Lasse, Skrivaren och förlagan. Norm och normbrott i Codex Upsaliensis av ‘Snorra Edda’ (Oslo: Novus forlag, 2013) Maurer, Konrad, Ueber die Ausdrücke. Altnorwegische, altschwedische und isländische Sprache (München: Straub, 1867) McKinnell, John, ‘The Making of Hávamál’, Viking and Medieval Scandinavia, 3 (2007), pp. 75–115. McKinnell, John, ‘Vǫluspá and the Feast of Easter’, Alvíssmál, 12 (2008), pp. 3–28 McTurk, Rory, Studies in Ragnars saga loðbrókar and its Major Scandinavian Analogues (Oxford: Society for the Study of Medieval Languages and Literatures, 1991) Meissner, Rudolf, Die Kenningar der Skalden. Ein Beitrag zur skaldischen Poetik (Bonn: Kurt Schroeder, 1921) Meldahl, Per. Norges keisere. Kongesagaen i klassisk-retorisk perspektiv. Om den Svetonius-imiterende strukturen i Snorres ‘Heimskringla’ og andre kongesagaer (Bergen: University of Stavanger, 2007) Meulengracht Sørensen, Preben, The Unmanly Man. Concepts of Sexual Defamation in Early Northern Society (Odense: Odense University Press, 1983) Meulengracht Sørensen, Preben, ‘Om eddadigtenes alder’, in At fortælle Historien – Telling History. Studier i den gamle nordiske litteratur – Studies in Norse Literature, ed. Preben and Sofie Meulengracht Sørensen (Trieste: Edizioni Parnaso, 2001 [1991]), pp. 143–50 Meulengracht Sørensen, Preben, ‘The Prosimetrum Form, 1: Verses as the Voices of the Past’, in Skaldsagas: Text, Vocation, and Desire in the Icelandic Sagas of Poets, ed. Russell Poole. Ergänzungsbände zum Reallexikon der Germanischen Altertumskunde, 27 (New York: de Gruyter, 2001), pp. 172–90 Micillo, Valeria, ‘Classical Tradition and Norse Tradition in the “Third Grammatical Treatise”’, Arkiv för nordisk filologi, 108 (1993), pp. 68–79

Bibliography

339

Micillo, Valeria, ‘Die grammatische Tradition des insularen Mittelalters in Island. Spuren insularer Einflüsse im Dritten grammatischen Traktat’, in Übersetzung, Adaptation und Akkulturation im Insularen Mittelalter, ed. Erich Poppe and Hildegard L. C. Tristram (Münster: Nodus Publikationen, 1999), pp. 215–29 Möbius, Theodor, ‘Vom Stef’, Germania, 18 (1873), pp. 129–47 Möbius, Theodor, ‘Vom isländischen Mansöngr’, Zeitschrift für deutsche Philologie. Ergänzungsband (1874), pp. 42–61 Mogk, Eugen, ‘Zur Bewertung der Snorra-Edda als religionsgeschichtliche und mythologische Quelle des nordgermanischen Heidentums’, Berichte über die Verhandlungen der Sächsischen Akademie der Wissenschaften zu Leipzig. Philologisch-historische Klasse, 84 (1932), pp. 3–18 Mundal, Else, ‘Heilagmann som sa sex. Lausavise nr. 7 etter Óláfr Haraldsson’, in Fjǫld veit hon frœða. Utvalde arbeid av Else Mundal, ed. Odd Einar Haugen, Bernt Øyvind Thorvaldsen, and Jonas Wellendorf (Oslo: Novus, 2012 [1984]), pp. 97–116 Müllenhoff, Karl, Deutsche Altertumskunde, 5 vols (Berlin: Weidmann, 1887–1900) Myrvoll, Klaus Johan and Tryggve Skomedal, ‘Tonelagsskilnad i islendsk i Tridje grammatiske avhandling’, Maal og Minne (2010), pp. 68–97 Myrvoll, Klaus Johan, ‘Kronologi i skaldekvæde. Distribusjon av metriske og språklege drag i høve til tradisjonell datering og attribuering’ (PhD diss., University of Oslo, 2014) Myrvoll, Klaus Johan, ‘The Authenticity of Gísli’s Verse’, Journal of English and Germanic Philology (forthcoming) van Nahl, Jan Alexander, ‘Snorri Sturlusons Konstruktion eines Deus omnipotens’, in Snorri Sturluson – Historiker, Dichter, Politiker, ed. Heinrich Beck, Wilhelm Heizmann, and Jan Alexander van Nahl (Berlin: de Gruyter, 2013), pp. 23–48 Noreen, Adolf, Altisländische Grammatik, 4th ed. (Halle: Niemeyer, 1923) Noreen, Erik, ‘Studier i fornvästnordisk diktning’, 3 vols. Uppsala universitets årsskrift. Filosofi, språkvetenskap och historiska vetenskaper, 3–5 (Uppsala: Akademiska bokhandeln, 1921–23) Nygaard, Marius, ‘Den lærde stil i den norrøne prosa’, in Sproglig-historiske studier tillegnede professor C. R. Unger (Kristiania: Aschehoug, 1896), pp. 143–70 Nygaard, Marius, Norrøn syntax (Kristiania: Aschehoug, 1905) O’Connor, Ralph, ‘History or Fiction? Truth-Claims and Defensive Narrators in Icelandic RomanceSagas’, Mediaeval Scandinavia, 15 (2005), pp. 101–69 O’Donoghue, Heather, The Genesis of a Saga Narrative. Verse and Prose in ‘Kormáks saga’ (Oxford: Clarendon Press, 1991) O’Donoghue, Heather, Skaldic Verse and the Poetics of Saga Narrative (Oxford: Oxford University Press, 2005) Ólafur Halldórsson, ‘Sagnaritun Snorra Sturlusonar’, in Snorri. Átta alda minning, ed. Gunnar Karlsson (Reykjavik: Sögufélag, 1979), pp. 113–38 Olmer, Emil, Boksamlingar på Island 1179–1490 enligt diplom. (Göteborg: Zachrissons boktryckeri, 1902) Olsen, Magnus, ‘Eldste forekomst av navnet Hlaðir’, Maal og Minne (1942), pp. 154–56 Olszowy-Schlanger, Judith, ‘The Knowledge and Practice of Hebrew Grammar among Christian Scholars in Pre-Expulsion England: The Evidence of “Bilingual” Hebrew-Latin Manuscripts’, in Hebrew Scholarship and the Medieval World, ed. Nicholas de Lange (Cambridge: Cambridge University Press, 2001), pp. 107–28 ONP = Ordbog over det norrøne prosasprog: http://www.onp.hum.ku.dk/ONP Registre = Ordbog over det norrøne prosasprog. Registre (København: Den arnamagnæanske kommission, 1989) Óskar Guðmundsson, Snorri. Ævisaga Snorra Sturlusonar 1179–1241 (Rekjavík: Forlagið, 2009) Paasche, Fredrik, Snorre Sturlason (Kristiania: Cappelen, 1922)

340

Bibliography

Paasche, Fredrik‘Esras aabenbaring og pseudo-Cyprianus i norrön litteratur’, in Festskrift til Finnur Jónsson, 29. Maj 1928, ed. Johannes Brøndum-Nielsen et al. (København: Levin & Munksgaard, 1928), pp. 199–205 Páll Eggert Ólason, Íslenzkar æviskrár frá landnámstímum til ársloka 1940, 1 (Reykjavík: Hið íslenzka bókmenntafélag, 1948) Pétur Pétursson, ‘Manifest and Latent Biblical Themes in Vǫluspá’, in The Nordic Apocalypse. Approaches to Vǫluspá and Nordic Days of Judgement, ed. Terry Gunnell and Annette Lassen (Turnhout: Brepols, 2013), pp. 185–201 Phelpstead, Carl, Holy Vikings. Saints’ Lives in the Icelandic Kings’ Sagas (Tempe: ACMRS, 2007) de Pins, Cyril, ‘Hending ok kenning. Les théories linguistiques dans l’Islande médiévale (xiie–xive s.). Lecture du Codex Wormianus’ (PhD diss., Université Paris 7-Diderot, 2013) Poole, Russell, ‘Compositional Technique in some Verses from Gunnlaugs saga’, Journal of English and Germanic Philology, 80 (1981), pp. 469–85 Poole, Russell, ‘The Origins of the Máhlíðingavísur’, Scandinavian Studies, 57 (1985), pp. 244–85 Poole, Russell, ‘The “Conversion Verses” of Hallfreðr vandræðaskáld’, Maal og Minne (2002), pp. 15–37 Poole, Russell, ‘Some Examples of the Adverbial Genitive in Skaldic Poetry’, Mediaeval Scandinavia, 14 (2004), pp. 115–31 Poole, Russell, ‘Scholars and Skalds. The Northward Diffusion of Carolingian Poetic Fashions’, Gripla, 24 (2013), pp. 7–44 Poole, Russell, ‘Identity Poetics among the Icelandic Skalds’, in New Norse Studies. Essays on the Literature and Culture of Medieval Scandinavia, ed. Jeffrey Turco. Islandica, 58 (Ithaca, NY: Cornell University Library, 2015) Priest, Paul, ‘Allegory and Reality in the Commedia’, Dante Studies, 96 (1978), pp. 127–44 Rathbone, Eleanor, ‘Master Alberic of London, “Mythographus tertius vaticanus”’, Medieval and Renaissance Studies, 1 (1943), pp. 35–38 Reynolds, Suzanne, Medieval reading. Grammar, Rhetoric and the Classical Texts (Cambridge: Cambridge University Press, 1996) Sävborg, Daniel, ‘Blockbildningen i Codex Upsaliensis. En ny metod att lösa frågan om Snorra Eddas ursprungsversion’, Maal og Minne (2012), pp. 12–53 Schottmann, Hans, Untersuchungen zur volksprachigen Mariendichtung des Mittelalters (München: Wilhelm Fink, 1973) von See, Klaus, ‘Der Skald Torf-Einar’, Beiträge zur Geschichte der deutschen Sprache und Literatur, 82 (1960), pp. 31–43 von See, Klaus, ‘Polemische Zitate in der Skaldendichtung. Hallfrøðr vandræðaskáld und Haldórr ókristni’, Skandinavistik, 7 (1977), pp. 115–19 von See, Klaus, ‘Skaldenstrophe und Sagaprosa. Ein Beitrag zum Problem der mündlichen Überlieferung in der Altnordischen Literatur‘, Mediaeval Scandinavia, 10 (1977), pp. 58–82 von See, Klaus, ‘Mündliche Prosa und Skaldendichtung. Mit einem Exkurs über Skaldensagas und Trobadorbiographien’, Mediaeval Scandinavia, 11 (1981), pp. 82–91 von See, Klaus, ‘Das Problem der mündlichen Erzählprosa im Altnordischen. Der Prolog der Þiðriks saga und der Bericht von der Hochzeit in Reykjahólar’, Skandinavistik, 11 (1981), pp. 91–95 von See, Klaus, Mythos und Theologie im skandinavischen Hochmittelalter (Heidelberg: Carl Winter, 1988) von See, Klaus, ‘Zum Prolog der Snorra Edda’, Skandinavistik, 20 (1990), pp. 111–26 Seznec, Jean, The Survival of the Pagan Gods. The Mythological Tradition and its Place in Renaissance Humanism and Art (New York: Bollingen, 1953 [1940]) Sigurður Nordal, Om Olaf den helliges saga. En kritisk undersøgelse (København: Gad, 1914) Sigurður Nordal, Snorri Sturluson (Reykjavík: Þór B. Þorláksson, 1920)

Bibliography

341

Sigurður Nordal, ‘Sagalitteraturen’, in Litteraturhistorie, B: Norge og Island, ed. Sigurður Nordal. Nordisk kultur, 8: B (Stockholm: Bonniers, 1953), pp. 180 –273 Sigurjón Páll Ísaksson, ‘Ólafs saga helga’ og ‘Heimskringla’ (Reykjavík: Skúmhöttur, 2014) Simms, Katherine, ‘Literacy and the Irish Bards’, in Literacy in Medieval Celtic Societies, ed. Huw Pryce (Cambridge: Cambridge University Press, 1998), pp. 238–58 Sims-Williams, Patrick, and Erich Poppe, ‘Medieval Irish Literary Theory and Criticism’, in The Cambridge History of Literary Criticism, 2: The Middle Ages, ed. Alastair Minnis and Ian Johnson (Cambridge: Cambridge University Press, 2005), pp. 291–309 Sjöström, Henning, ‘Magister Albericus lundoniensis, mythographus tertius vaticanus. A Twelfth Century Student of Classical Mythology’, Classica et Mediaevalia, 29 (1968), pp. 249–64 Solvin, Inger Helene, ‘Litla Skálda’ – Islands første poetiske avhandling? Et forsøk på å etablere en relative kronologi mellom ‘Skáldskaparmál’ og ‘Litla Skálda’ (MA thesis, University of Oslo, 2015) Stattin, Jochum, Näcken (Malmö: Carlssons, 1992) Stavnem, Rolf, Stroferne i ‘Grettis saga’. Deres funktion og betydning (København: Det armnamagnæanske Institut, 2000) Steinsland, Gro, ‘Vǫluspá and the Sibylline Oracles with a Focus on the “Myth of the Future”’, in The Nordic Apocalypse. Approaches to Vǫluspá and Nordic Days of Judgement, ed. Terry Gunnell and Annette Lassen (Turnhout: Brepols, 2013), pp. 147–60 Stefán Karlsson, ‘Islandsk bogeksport til Norge i middelalderen’, in Stafkrókar, ed. Guðvarður Már Gunnlaugsson (Reykjavík: Stofnun Árna Magnússonar, 2000 [1979]), pp. 188–205 Stefán Karlsson, ‘Kringum Kringlu’, in Stafkrókar, ed. Guðvarður Már Gunnlaugsson (Reykjavik: Stofnun Árna Magnússonar, 2000 [1977]), 251–73 Stefán Karlsson, ‘Saltarabrot í Svíþjóð með Stjórnarhendi’, Gripla, 5 (1982), pp. 320–22 Stifter, David, ‘Metrical Systems of Celtic Traditions’, North-Western European Language Evolution, 69: 1 (2016), pp. 38–94 Ström, Folke, ‘Poetry as an Instrument of propaganda. Jarl Hákon and his Poets’, in Speculum Norroenum. Norse Studies in Memory of Gabriel Turville-Petre, ed. Ursula Dronke et al. (Odense: Odense University Press, 1981), pp. 440–58 Strömbäck, Dag, ‘Näcken’, in Kulturhistoriskt lexikon för nordisk medeltid, 22 vols (Malmö: Allhems, 1955–78), 12, cols 432–38 Sveinbjörn Rafnsson, ‘Merlínuspá og Völuspá í sögulegu samhengi’, Skírnir, 173 (1999), pp. 377–419 Sveinbjörn Rafnsson, Ólafs sögur Tryggvasonar. Um gerðir þeirra, heimildir og höfunda (Reykjavík: Háskólaútgáfan, 2005) Sveinbjörn Rafnsson, Um ‘Snorra Eddu’ og ‘Munkagaman’. Drög til menningarsögu íslenskra miðalda. Ritsafn Sagnfræðistofnunar, 43 (Reykjavík: Háskólaútgáfan, 2016) Sverrir Tómasson, Formálar íslenskra sagnaritara á miðöldum. Rannsókn bókmenntahefðar (Reykjavík: Stofnun Árna Magnússonar, 1988) Sverrir Tómasson, ‘Nýsköpun eða endurtekning? Íslensk skáldmennt og Snorra Edda fram til 1609’, in Guðamjöður og arnarleir. Safn ritgerða um eddulist, ed. Sverrir Tómasson (Reykjavík: Háskólaútgáfan, 1996), 1–64 Sverrir Tómasson, ‘“Upp í garð til Sæmundar”. Lærdómssetrið í Odda og Snorri Sturluson’, Goðasteinn, 33 (1997), pp. 188–202 Sverrir Tómasson, ‘Skáldskapur og fræði fyrir stokk innan’, in Freyjas psalter. En psalter i 40 afdelinger til brug for Jonna Louis-Jensen, ed. Bergljót S. Kristjánsdóttir and Peter Springborg, 2nd ed. (København: Det arnamagnæanske Institut, 1997), pp. 190–92

342

Bibliography

Teichert, Matthias, ‘Das Goldzeitalder: Latinität und Fragmente eines griechisch-römischen Mythos in der Gylfaginning’, in Snorri Sturluson – Historiker, Dichter, Politiker, ed. Heinrich Beck, Wilhelm Heizmann, and Jan Alexander van Nahl (Berlin: de Gruyter, 2013), pp. 129–50 Texts and Transmission. A Survey of the Latin Classics, ed. L. D. Reynolds (Oxford: Clarendon, 1983) Thurneysen, Rudolf, ‘Zur irishen Accent- und Verslehre’, Revue celtique, 6 (1883–85), pp. 309–47 Thurneysen, Rudolf, A Grammar of Old Irish, trans. and rev. D. A. Binchy and Osborn Bergin (Dublin: Dublin Institute for Advanced Studies, 1946) Torfi H. Tulinius, The Matter of the North. The Rise of Literary Fiction in Thirteenth-Century Iceland, trans. Randi C. Eldevik (Odense: Odense University Press, 2002) Torfi H. Tulinius, Skáldið í skriftinni. Snorri Sturluson og ‘Egils saga’ (Reykjavík: Hið íslenska bókmenntafélag, 2004) Townend, Matthew, ‘Pre-Cnut Praise-Poetry in Viking Age England’, The Review of English Studies. New Series, 51 (2000), pp. 349–70 Tranter, Stephen N, ‘Clavis Metrica’. ‘Háttatal’, ‘Háttalykill’ and the Irish Metrical Tracts. Beiträge zur nordischen Philologie, 25 (Frankfurt: Helbing & Lichtenhahn, 1997) Turville-Petre, Gabriel, Origins of Icelandic Literature (London: Clarendon Press, 1953) Turville-Petre, Gabriel, ‘Gísli Súrsson and his Poetry: Traditions and Influences’, in Nine Norse Studies (London: Viking Society for Northern Research, 1972 [1944]), pp. 118–153 Turville-Petre, Gabriel, Scaldic Poetry (Oxford: Clarendon, 1976) Vésteinn Ólason, ‘Er Snorri höfundur Egils sögu?’, Skírnir, 142 (1968), pp. 48–67 de Vries, Jan, De skaldenkenningen met mytologischen inhoud (Haarlem: Tjeenk Willink, 1934) de Vries, Jan, Altnordisches Etymologisches Wörterbuch, 2nd ed. (Leiden: Brill, 1962) de Vries, Jan, Altnordische Literaturgeschichte, 2 vols. Grundriss der germanischen Philologie 15–16 (Berlin: De Gruyter, 1964–67) Walter, Ernst, ‘Argumente zur Bestimmung des Alters der Íslendingadrápa Hauks Valdísarsonar’, in Deutsch-nordische Begegnungen. 9. Arbeitstagung der Skandinavisten des deutschen Sprachgebiets 1989 in Svendborg, ed. Kurt Braunmüller and Mogens Brøndsted (Odense: Odense University Press, 1991), pp. 96–103 Wanner, Kevin, Snorri Sturluson and the Edda. The Conversion of Cultural Capital in Medieval Scandinavia (Toronto: University of Toronto Press, 2008) Weber, Gerd Wolfgang, ‘Siðaskipti. Das religionsgeschichtliche Modell Snorri Sturlusons in Edda und Heimskringla’, in Sagnaskemmtun. Studies in Honour of Hermann Pálsson, ed. Rudolph Simek, Jónas Kristjánsson, and Hans Bekker Nielsen (Wien: Böhlau, 1986), pp. 309–29 Weber, Gerd Wolfgang, ‘Intellegere historiam. Typological perspectives of Nordic prehistory (in Snorri, Saxo, Widukind and others)’, in Tradition og historieskrivning. Kilderne til Nordens ældste historie, ed. Kirsten Hastrup and Preben Meulengracht Sørensen (Århus: Aarhus universitetsforlag, 1987), pp 95–141 Wellendorf, Jonas, ‘Zoroaster, Saturn, and Óðinn: The Loss of Language and the Rise of Idolatry’, in The Performance of Christian and Pagan Storyworlds, ed. Lars Boje Mortensen and Tuomas M. S. Lehtonen (Turnhout: Brepols, 2013), pp. 143–70 Wellendorf, Jonas, ‘No Need for Mead. Bjarni Kolbeinsson’s Jómsvíkingadrápa and the Skaldic Tradition’, North-Western European Language Evolution, 69: 2 (2016), pp. 130–54 Wellendorf, Jonas, Gods and Humans in Medieval Scandinavia. Retying the Bonds (Cambridge: Cambridge University Press, 2018) Whaley, Diana, ‘Skalds and Situational Verses in Heimskringla’, in Snorri Sturluson. Kolloquium anläßlich der 750. Wiederkehr seines Todestages, ed. Alois Wolf (Tübingen: Gunter Narr Verlag, 1993), pp. 245–66

Bibliography

343

Whaley, Diana, ‘The “Conversion Verses” in Hallfreðar saga: Authentic Voice of a Reluctant Christian?’, in Old Norse Myths, Literature and Society, ed. Margaret Clunies Ross (Odense: University Press of Southern Denmark, 2003), pp. 234–57 Wills, Tarrin, The Foundation of Grammar. An Edition of the First Section of Óláfr Þórðarson’s Grammatical Treatise (PhD diss., University of Sydney, 2001) Wills, Tarrin, ‘The Development of Skaldic Language’, in Á austrvega. Saga and East Scandinavia. Preprint Papers of the 14th International Saga Conference. Uppsala 9th – 15th August 2009, ed. Agneta Ney, Henrik Williams, Fredrik Charpentier Ljungqvist (Gävle: Gävle University Press, 2009), pp. 1032–38 Wills, Tarrin, ‘The Thirteenth-century Runic Revival in Denmark and Iceland’, North-Western European Language Evolution, 69: 2 (2016), pp. 114–29 Wolf, Alois, ‘Zur Rolle der Vísur in der altnordischen Prosa’, in Festschrift Leonhard C. Frans zum 70. Geburtstag, ed. Osmund Menghin and Hermann M. Ölberg (Innsbruck: Innsbrucker Gesellschaft zur Pflege der Geisteswissenschaften, 1965), pp. 459–84 Wolf, Kirsten, ‘Gregory’s Influence on Old Norse-Icelandic Religious Literature’, in Rome and the North. The Early Reception of Gregory the Great in Germanic Europe, ed. Rolf H. Bremmer Jr, Cornelis Dekker, David F. Johnson (Paris: Peeters, 2001), pp. 255–74 Wood, Cecil, ‘The Skald’s Bid for a Hearing’, Journal of English and Germanic Philology, 59 (1960), pp. 240–54 Würth, Stephanie, Der ‘Antikenroman’ in der isländischen Literatur des Mittelalters. Eine Untersuchung zur Übersetzung und Rezeption lateinischer Literatur im Norden (Basel: Helbing und Lichtenhahn, 1998) Þorgeir Sigurðsson, ‘Rímstuðlar. Um tengsl ríms, atkvæðaskiptingar og stuðla í íslenskum skáldskap’, Íslenskt mál og almenn málfræði, 23 (2001), pp. 215–27 Þorgeir Sigurðsson, ‘Þróun dróttkvæða og vísuorðhlutar að hætti Hans Kuhn’, Són, 8 (2010), pp. 9 –21 Þorgeir Sigurðsson, ‘Nýjar skjálfhendur á 12. öld’, Són, 12 (2014), pp. 55–67 Þórunn Sigurðardóttir, ‘Tvær ritgerðir um skáldskap í kvæðabók úr Vigur (AM 148 8vo)’, Gripla, 19 (2008), pp. 193–209 Zeeman, Nicolette, ‘In the Schoolroom with the ‘Vulgate’ Commentary on ‘Metamorphoses’ I’, New Medieval Literatures, 11 (2009), pp. 1–18 Zieman, Katherine, Singing the New Song. Literacy and Liturgy in Late Medieval England (Philadelphia: University of Pennsylvania Press, 2008), pp. 1–39 Ziolkowski, Jan M., ‘Theories of Obscurity in the Latin Tradition’, Mediaevalia, 19 (1996), pp. 101–70 Ziolkowski, Jan M., ‘The Prosimetrum in the Classical Tradition’, in Prosimetrum. Crosscultural Perspectives on Narrative in Prose and Verse, ed. Joseph Harris and Karl Reichl (Cambridge: D. S. Brewer, 1997), pp. 45–65 Ziolkowski, Jan M., ‘Medieval Latin Mythography as Death and Resurrection of Myth’, in Writing Down the Myths, ed. Joseph Falaky Nagy. Cursor Mundi, 17 (Turnhout: Brepols, 2013), pp. 87–106

Index The Index is selective, avoiding excessive overlap with the Table of Contents and listing only pages that provide information about the entry, rather than mere mention of it. The Index is intended as a mind map, and longer entries may show ancillary topics of the book. 1GT 31, 99, 165 reference to story about Þórr, 175–77 minimal pairs modelled on differentiae, but aided by skaldic competence; poetic precedence explains the independence and quality of 1GT, 182 mentions Hebrew letters 2GT 99, 281–85 additions to 2GT 3GT 78, 88, 91–93, 115, 123–24, 126, 178 n. 267 authorship, 178–80 equality of Latin and Norse poetry, 180–81 adaptation of Latin concepts to Norse hendingar and alliteration, 181 relevance for Norse poetry prioritized also in discussion of letters and syllables, 182–83 Latin letters versus Hebrew, Greek and runic letter names, 183 matres lectionis and runes (not part of Hebrew alphabet lore n. 291), 183–84 argument about ↕ based on alphabet lists, 185 order of runes, 186 rational names; theoretical premise for native learning, matching the saga authors’ practical endeavours, 186–87 elaborated, rather than simplified and abbreviated, translations due to poetic topic; translation of grammatical texts late due to the organic development of vernacular poetics, 189 link to 5GT, 287 4GT 115, 186–87 longer than exemplar, 287 5GT 187–88 colophon and text, 189 vices are a Latinate concept; link to 3GT; calque from Latin, 190 native terminology; 5GT hybrid between Latinizing and nativizing approaches, 190–91 Óláfr Þórðarson likely author, 191 the hybridity of 5GT attests to the efforts of arriving at translation as an approach within grammatical studies, and it fleshes out the intellectual development from Snorri’s Edda to 3GT A: AM 748 Ib 4to 107, 110, 130–33, 134 n. 120, 139 nn. 133–34 aðalhending 22 def.

https://doi.org/10.1515/9783110643930-011

Ágrip 194 Albericus lundoniensis see Third Vatican Mythographer Alexander of Villedieu 99, 187, 287 alhent 29, 31, 278 alliteration 180 Óláfr’s adaptation of parhomoeon to Norse alliteration. See also metre and dating criteria (alliteration in vr-) anacrusis see onset Anglo-Saxon Chronicle 208 archaization 34–35 regularization of archaic features in Jómsvíkingadrápa; archaic action rather than date accounts for the archaization (similarly Pseudo-Ásbjǫrn 249–50), 35–37 regularization of archaic features in Háttatal, 215 pseudonymous poets could archaize metre and style, but not phonology and morphology. The latter is at odds with widespread scholarly assertions (also 249), 225 Bárøðr for Bárðr – modern, not medieval archaization. Archaic name forms were perceived as synchronic variants in the Middle Ages, 244–45 Kormáks saga, exaggeration of archaic features, 246 the archaizing cluster – metrical fillers and irregular hendingar – found in Egils saga, Kormáks saga, Ragnars saga and Stjǫrnu-Odda draumr, 247–48 PseudoRagnarr’s failed attempt at archaization; 248–49 Krákumál model for PseudoRagnarr’s archaization, 252 Skáldskaparmál as a tool for archaization Arngrímr Brandsson 294 Árni (bróðir) 278 Árni Jónsson 295 Arnórr jarlaskáld 24, 42, 59, 65–67, 70–71 Augustine of Hippo 37, 97, 285 authority 118–19, 287

346

Index

B: AM 757 a 4to 107–08, 110, 130–33, 134 n. 120, 139 nn. 133–34 Bandamanna saga 253–54 Bárðar saga 251 Begleitprosa 212 rejected Bergbúa þáttr 251 Bible, Biblical material 170–71 Flood, 171–72 Babel motif (also 280, 284–86, 304–09), 172–73 Trinity, 174 creatio ex nihilo Bjarnar saga Hítdœlakappa 67 Bjarni (in Skáldatal) 69 Bjarni Kolbeinsson 34, 38, 107, 194 Bjǫrn Breiðvíkingakappi 74 Bjǫrn Hítdœlakappi 27, 42, 65, 68 Boccaccio 96, 118–19 Boethius 207 Bragi (god) 45 Bragi (poet) 22, 53 Britain 30 innovation in skaldic poetry related to Britain and Ireland Búadrápa 194 catalexis 20, 24–26 (def. p. 24), 30 Christ 86 Conrad of Hirsau 96–98 CR 3, 71 n. 128, 108–09 why CR rather than Poetic Edda, 110, 157 date, 165–66 collections of eddic poems probably indebted to Snorri’s Edda Dante 11–12, 96, 118–19 Dares Phrygius 187 Dating criteria (weak, strong, absolute when applicable) 60 metrical, 61 a/ǫ (absolute), 71–72 vr- (absolute), 72–75 stylistic criteria – normally weak, but here supported by circumstances and transmission, 75 -t/-ð (strong), 157 vr-, 166–67, vr-; correlation of V2 breaks and of/um (strong), 168 V2 breaks and of/um; the mythological slump c. 1015–1120, 214 authenticating quotation strong dating criterion for the post-settlement period, but not before (see also 271 – only two spurious stanzas in authenticating quotations), 221–222 clusters of dating criteria (individually mostly weak, but collectively strong to absolute, since every new criterion amounts to a testing of

previous ones), 230 compensatory hendingar (strong–absolute), 251–52 on the need for clearly articulated dating criteria, 253 circumstances allowing for intertextual dating, 252 dependence on Skáldskaparmál’s deviations from earlier tradition (weak on its own, but strong if supported by additional criteria), 263 few formal criteria for dating within the period c. 1000–c. 1250, 264–65 linguistic dating criteria in Grettis saga, Harðar saga and Víglundar saga; rarity of such criteria indicates limited pseudonymous composition in the fourteenth century, 265 little comprehensive discussion of measurable dating criteria from the early twentieth century until 2014, 266–67 intertextual dating of ‘saying’ in Ævikviða, 272–73 the dating of Grettis saga De vulgari eloquentia see Dante Denmark 56 Devil see Satan diction (advanced) 39, 57–62 reasons for slump, 75–76 social setting in early versus late period, 93–94 interconnectedness of mythological references, extended metaphor and wordplay; partial independence of tmesis, 93–94 methodological innovation and formal conservatism, 100 preconditions for revival Donatus 9, 37, 88, 99, 125, 178–81 in 3GT, 287, 292–93 drápa 28 def. Draumr Þorsteins Síðu-Hallssonar 251 dróttkvætt 21–22 def. of classical/regular and archaic/irregular dróttkvætt Dublin 30 dunhent 31 n. 48, 114 n. 47, 292 dwarf 30, 64–66, 68–69, 72, 81, 90–91, 153–54, 161–64 ‘dwarves’ ship’, 253 eddic poetry 5, 20 def., 208–10 oral transmission (also 212), 214 typical of presettlement prosimetra, 217 Egill Skallagrímsson 26–27, 38, 45, 49, 51, 56, 90 Egils saga 219 pseudonymous poetry before c. 1250, 220 irregular hendingar most

Index

promising feature, 221–222 clusters of dating criteria, 225–57 influence of of Gregory’s Dialogues on Egils saga, 228 all references to runes traceable to the author, not to Egill, 228–31 authentic model stanzas for the author’s irregular ones, 232 four hending styles in Egils saga – three attributable to Egill, one to the author, 234 Óláfr not a better candidate for authorship than Snorri, 235–36 circumstantial evidence for Snorri’s authorship – if so, composed before Heimskringla, 236–39 stylistic comparison to Heimskringla and Snorri’s Edda – the three form a very distinct group within the Old Icelandic corpus, 241–44 Snorri composed no pseudonymous poetry in his Edda or Heimskringla, but did so in Egils saga Eilífr Goðrúnarson 50, 53–54, 56, 149, 256–58 Einarr skálaglamm 53–54, 64, 148–59 Einarr Skúlason 24, 27, 53, 76–89, 134, 145, 211 relevance of paraphrase of the cluster of stanzas in Hallfreðar saga to oral skaldic prosimetra Eiríkr jarl 54–55, 63–64, 69, 75, 262 Eiríksmál 168 Elucidarius 10, 187 emulation 99–100 end-rhyme 26–27, 38 Epistola Alexandri 123–24 Eyjolfr dáðaskáld 28, 42, 63–64, 258–59, 259, 262 Eyrbyggja saga 74, 216 Eysteinn (composer of Lilja) 290–96 Eyvindr skáldaspillir 44, 56, 66, 85–86, 168 extended metaphor 39, 43, 46 equivalent to both nýgerving and nykrat, 46 n. 25 Old Icelandic meaning of nýgerving, 46 n. 27 syntactic scope, 46–47 def., 52–53 early period; Snorri’s misrepresentation of its use, 57–58 intermediate period, 74–75 Pseudo-Óláfr, 76 Haukr Valdísarson, 80 Einarr; extended metaphor + oxymoron, 92 Sturla, 93 connection to other features of diction, 121–24 Snorri’s use of Horace to name and condemn inconsistent use of extended metaphor (nykrat)

347

Fagrskinna 198, 213 Fenrir 44, 86–87, 129, 138, 141–44 Fifth Grammatical Treatise see 5GT First Grammatical Treatise see 1GT fornyrðislag 21 fornaldarsögur 35, 205, 208–10 oral transmission (also 212), 246, 250 Fóstbrœðra saga 216 Fourth Grammatical Treatise see 4GT Freyja 41, 50, 72, 79–82, 172 Freyr 43, 79 Gamli kanóki 24, 195 Geoffrey of Monmouth 187 Geoffrey of Vinsauf 293 giant 68, 72, 153–54, 253–54 giantess/troll-woman 70–71, 271 Gísla saga 251 Gizurr gullbrárskáld 30 Gospel of Nicodemus 86 Grágás 12–14 grammatica 32–33, 77, 93, 95 n. 219 def.; Greek and Latin, 99, 103–06 contexts of study; social setting, 107–10 manuscripts; contents of Icelandic grammatica, 109–10 grammatical manuscripts structured according to Snorri’s Edda, 120 rules and regularity, 125 Latin, basic, 175 nativizing and Latinizing grammatica (also 124, 187, 189–91); poetry cause for adaptation of Latin grammatica (see also 128, 180–81), 187 translation of grammatical texts late due to the organic development of vernacular poetics, 248–49 grammatical training as precondition for misconstruing early poetry Greek letters 182 Greenland 5 Gregory the Great 225–27 influence on Egils saga, 292 Grettis saga 45, 264–73 (272–73 dating of the saga; connection to Þingeyrar; author and poet appear to be the same person) Grímnismál 164, 168, 171–72 Gunnlagur Leifsson 168 Gunnlaugr ormstunga 27, 42 Gunnlaugs saga 27, 77 n. 154, 251 Gunnr (Valkyrie) 43 Guthormr sindri 56

348

Index

Gylfaginning 7, 85–87, 164 Gylfaginning probably indebted to Litla Skálda for mythological prose narrative based on poetry, but vastly more elaborate; Gylfaginning our best source to reception of eddic poetry, 165–66 Gylfaginning is the first datable collection of eddic poetry; Hávamál as a framing device; same four poems in Gylfaginning and the beginning of CR, probably indicating that CR was composed to complement Gylfaginning; A also related to Snorri’s Edda, suggesting a comparable background, 167–69 Snorri’s perception of the poetry quoted in Gylfaginning: he may well have thought that it was composed by the gods (=the Trojans), but his quotation of the twelfthcentury Vǫluspá in skamma (=Hyndluljóð) complicates the matter, 170–71 no Flood before Snorri, 171–72 Snorri explains Óðinn’s names in a Biblical as well as a commonsensical way, 172–73 heterodox Trinity extracted from Grímnismál, 173–74 merely historia, no spiritual meaning, 174 Snorri sometimes has biased interpretations, but probably did not alter or produce any of the poetry in Gylfaginning or Skáldskaparmál haðarlag 24 Hafgerðingadrápa 24 Hákon Hákonarson (king †1263) 110 Hákon jarl 50, 54–56, 63, 66, 69, 75 hálfhneppt 24 Hallar-Steinn 33–34, 76, 89–91, 194 Halldórr ókristni 64 Hallfreðar saga 54, 60–61, 83–85 the saga’s cluster of stanzas was a cluster already to Einarr Skúlason (first half of twelfth century), 211 conceptual frame of the cluster of stanzas; relevance for oral skaldic prosimetra Hallfreðr vandræðaskáld 42, 50 n. 41, 54, 60–61, 64, 66, 68–69, 71, 74–75, 83–85 Hallr Þórarinsson 31, 111 Hallvarðr háreksblesi 25, 30, 42 Haraldr harðráði 27, 42, 57, 65, 70–71 remarkably many mythological references

connected to Haraldr (also 74–75, 261–62), 255, 259 Harðar saga 250, 265 Háttalykill 28, 31, 111–12 Háttatal 29, 35–37 regularizing the early poets, 38, 112–14 expansion and regularization of its model Háttalykill, 114–18 authorship of commentary (116–18 U-scribe’s perception of Háttatal), 118–19 how commentary conveys authority to the core text, 119–20 regularizing the early poets, 121–24 paraphrase of Horace (124 Snorri’s nativizing versus Óláfr’s Latinizing paraphrase), 125–29 Háttatal and Latin grammar (125–26 basic grammar, 127–29 the problem of h, 129 grammatical beginning of the commentary for rhetorical effect – the body of the commentary more pragmatic than 3GT; methodological innovation and formal conservatism), 268–69 use by Pseudo-Grettir háttlausa 31 n. 48 háttr see metre Haukr Valdísarson 107 Hávamál 164 Hebrew 182–86, 282–86 Heimdallr 289 Heimskringla 23 reliability of poetry, 69, 198, 203 prologue, 213, 233–34, 235–39 comparison to Egils saga, 239–41 composed by Snorri, 241–43 Snorri composed no pseudonymous poetry in Heimskringla heiti see synonyms Helinand 6 hending 21 def., 22 irregular hendingar, 34–37 irregular hendingar and Bjarni’s and Snorri’s reception of them; 180 Óláfr’s adaptation of paronomasia to hendingar and of hendingar to end-rhyme, 220 most promising dating criterion in Egils saga, 228–31 compensatory hendingar serve as model for Pseudo-Egill’s irregular ones, 232 four hending styles in Egils saga – three attributable to Egill, one to the author Hervarar saga 212 hjástælt 29 Hlaðir 51

Index

Hnoss 79 Hofgarða-Refr 29, 31, 42, 63, 65, 75, 278 homonymy 44, 75, 88, 91, 94 Honorius Augustodunensis see Elucidarius Horace 121–24 hrynhenda (hrynjandi, liljulag) 24, 292 Hungrvaka 195 Hyndluljóð 168–69 iðrmælt 31 n. 48, 114 n. 47 Iðunn 45 Illugi Bryndœlaskáld 71–72 Ireland 30 innovation in skaldic poetry related to Britain and Ireland, 95, 105, 208–10 Irish prosimetra are similar to Icelandic ones and differ from Latin (except Saxo) in the same ways. This suggests an oral background to the prosimetrical fornaldarsögur tradition Isidore 124, 190, 247 Íslendinga saga 278 Íslendingabók 194 Jerome 182 Jómsvíkinga saga 35 Jóns saga helga 195 Jóns saga postula 195 Jǫrð 55 Karlevi stone 21 kenning 39–40 def., 43 def., 48 term inverted kenning rejected, 62–63 uniqueness of individual kennings, 153–56 redundancy (overdetermined kennings), 163 n. 221 kennings as base-words klifat 114 n. 47 klofastef see refrain Klœngr (bishop) 97, 278 Knútr (of England and Denmark) 30, 63, 75 Knýtlinga saga 233–34 composed by Óláfr Þórðarson, 241 Kolbeinn Tumason 196 Kormákr Ǫgmundarson 28–29, 50, 71, 149 n. 165 Krákumál 195, 248–49 Kristni saga 195, 210 relevance for oral skaldic prosimetra kviðuháttr 21

349

Latinizing approach 124 Legendary sagas see fornaldarsögur Leviathan 86 liðhendur 114 n. 47 Litla Skálda 15, 92, 129–30 not previously recognized as model for Snorri’s Edda; functional similarity to Gylfaginning and Skáldskaparmál, 130–33 manuscripts A and B; Skáldskaparmál not redacted to avoid repetition, but to remove narrative; B reintroduces missing material on kennings from W branch and sees Edda as the narrative part of Snorri’s Edda, 133–40 comparison of Litla Skálda and Skáldskaparmál (134–35 Litla Skálda, Háttalykill and Háttatal breaking the rules of Skáldskaparmál; 134 Einarr inspired Snorri to distinction between silver and gold; 136 Litla Skálda could not have extracted the sea-king rule from Skáldskaparmál, but has based it on independent observation; 137 kennings for gods; 137–40 associative versus preconceived structure); 140–44 intertextual connections between Litla Skálda and Snorri’s Edda (142 poetry in Litla Skálda reworked into prose in Gylfaginning; 143 variants in the Fenrir text used by Snorri; 143–44 influence of Litla Skálda on Gylfaginning and Skáldskaparmál, respectively), 144–45 Snorri’s authorship unlikely due to his traditionalizing approach, 145–47 Litla Skálda’s relevance for understanding the rationale behind the composition of Snorri’s Edda ljóðaháttr 21 Loki 50 Lucan 187 máláháttr 21 Markús Skeggjason 24, 57–58 Martianus Capella 207 metre 20 def., 93–94 methodological innovation and formal conservatism, 34–37 early metre misconstrued (Bjarni Kolbeinsson, Háttatal) (see also 119–20 [Háttatal], 228–31 [Egils saga], 243–46 [Egils saga and Kormáks saga], 246–49

350

Index

[Ragnars saga]), 34–35 metre as marker for fantastic narrative (also 250) Miðgarðsormr 70, 86–87 Morkinskinna 77, 198, 213, 216 high incidence of situational quotations, 255–59 SnegluHalla þáttr, 259–62 the following þáttr and its implications for oral skaldic prosimetra Munka-Þverá 295 munnvǫrp 31 n. 48, 34 myth 40 def., 95 lack of medieval term, 96 disbelief as precondition for transmission mythography 7, 95–100 Europe and Iceland mythological references 28, 39, 40–41 def. of generic and specific mythological references, 41–43 statistics, 49 why generic references are left out, 53–55 early period, 61–62 why generic references were harmless, 62–63 why generic references were retained, 63–68 specific references connected to others than kings during the intermediate period, 68–75 specific references connected to kings, intermediate period, 76 Haukr Valdísarson, 92 Snorri, Óláfr and Sturla, 93 connection to other features of diction, 100 preconditions for revival Niðaróss 77, 290 Nikulás Bergsson 195 Njáls saga 210 relevance for oral skaldic prosimetra, 232 n. 136 the younger stratum of stanzas, 251 Northumbria 30 Norway 33, 54–55, 66, 71, 104 nýgerving 46 n. 25; see also extended metaphor nykrat 46 n. 25; see also extended metaphor nykrinn 121–24 obscurity 98–100, 189–295 Occitan 11–12, 106 Oddaverja annáll 235 Óðinn 30, 41, 53, 55, 65, 68, 79, 83, 86, 90, 94, 154, 158–59, 171–72 ofljóst see also wordplay; 44 (n. 18)–45 (def.), 49–51 early period, 79 Einarr, 88–89 Einarr, 89 Hallar-Steinn, 91 Hallar-Steinn Óláfr Haraldsson 56–60, 64–65, 69–70, 72, 82–83, 85–87

Óláfr Leggsson svartaskáld 190 Óláfr Tryggvason 39, 56, 59–60, 63, 68, 70–71, 75 Óláfr Þórðarson hvítaskáld 9, 32, 88, 92, 105 bio, 178 bio Óláfs drápa Tryggvasonar 194 Óláfs saga hins helga 57 Legendary Saga, 72–75 Greater Saga, 196–97 *Oldest Saga and Legendary Saga bear witness to Icelandic interests, 203 Separate Saga (prologue), 216 Heimskringla and Separate Saga: high incidence of situational quotations, 217 *Oldest saga Óláfs saga Tryggvasonar 69 Heimskringla, 196 by Oddr munkr, 217 by Oddr munkr, 240–41 Greater Saga – relevance for Snorri’s authorship of Heimskringla Ónarr 55 onset 22, 36 oral transmission see eddic poetry and skaldic poetry Orkney 3, 30–31, 33–34, 66, 104 Orkneyinga saga 31, 198, 217, 240 Orms þáttr Stórólfssonar 249–50 Otfrid of Weissenburg 97 Óttarr svarti 24, 30, 42 Ovid 34, 96–97, 109, 169 Páls saga 195 paganism 61, 66, 83 cult, 97 Manicheism, 258–59 agency attributed to the gods in the eleventh century (also 262), 284–87 rise of paganism through linguistic confusion paraphrase 61 Einarr Skúlason of Hallfreðr (twice), 64 Eyjolfr of Einarr skálaglamm, 66–67 Arnórr of poetry to jarls and of Vǫluspá, 70 n. 122 Rǫgnvaldr of Haraldr harðráði, Haraldr of Glúmr Geirason, 71 Illugi of Kormákr and Fáfnismál, 72 Arnórr, Illugi and paraphrase in the intermediate period, 74 Pseudo-Óláfr of Bjǫrn Breiðvíkingakappi, 80–82 Einarr’s paraphrase of Hallfreðr’s Hákonardrápa and Hnoss stanza, 83–84 Einarr’s paraphrase of Hallfreðr’s ‘Fyrr vas hit es harra’, 87 Snorri of Einarr, 90 Þórðr’s of Egill, Hallar-Steinn’s of both, 93 general observations, 179 n. 274 Óláfr of prologue

Index

to Snorri’s Edda, 286 the W redactor of himself, 287–88 W redactor of Háttatal, 292 Eysteinn’s of Gregory, 294 Arngrímr of Eysteinn and Snorri’s Edda, 306 n. 24 the W redactor of Skáldskaparmál paronymy 44 def. personification 55 Hallfreðr, 79 Einarr, 271 Pseudo-Grettir Petrus Comestor 284–85 Plácitúsdrápa 194 Poetic Edda see CR Priscian 99, 126, 178, 181–86 in 3GT prologue to 1–4GT 115, 287–88, 308–15 pseudonymous poetry 72–75 Sighvatr and Óláfr Haraldsson (also 262–63), 174 Snorri composed no pseudonymous poetry in his Edda, 218 main productive period of pseudonymous poetry that of saga writing, in particular the thirteenth century (also 231 n. 135, 232 n. 136, 252 n. 216, 265), 241–44 Snorri composed no pseudonymous poetry in his Edda or Heimskringla, but did so in Egils saga, 244–46 Pseudo-Kormákr, 246 PseudoStjǫrnu-Oddi, 246–49 Pseudo-Ragnarr, 249–50 Pseudo-Ásbjǫrn, 250 Harðar saga and Víglundar saga, 253–54 PseudoÓfeigr, 255–59 Pseudo-Þjóðolfr, 262–63 may be more common in þættir than we can tell, since mainly stylistic and few formal criteria are helpful, 264–73 PseudoGrettir, 265 composing the bulk of the poetry of a saga very rare in the thirteenth century, but possibly more common in the fourteenth, 275 list of texts containing pseudonymous poetry, 276 pseudonymous poetry small portion of the poetic corpus R: AM 2367 4to 107, 110, 267 Ragnars saga loðbrókar 246–49 refrain 28–30 klofastef, (28) stál; 29 stefjabálkr; 29–30 alhent (also 38) Rígsþula 288–89 runes 16, 33, 104, 182–87, 222–23, 225, 228 all references to runes in Egils saga traceable to its author, not to Egill, 234–35 Rǫgnvaldr jarl 32–33, 111

351

sagas of Icelanders 199–200 prose versus prosimetrum; prosimetrical in the first half of the thirteenth century, with the exception of the courtly Laxdœla saga, 250–51 why many sagas are left out of the study Saint Magnus Cathedral 32 Sallust 187 Satan 86 Saxo Grammaticus 208 functions of poetry like in Old Icelandic saga narrative Second Grammatical Treatise see 2GT Servius 118 commentary on the Æneid, 247 De centum metris Sighvatr Þórðarson 24, 30, 57–58, 72–75 Sigurðr Fáfnisbani 71–72, 86–87 Sigurðr Hlaðajarl 66 Sigurður Breiðfjörð 94 Sigurðr slembidjákn 24 Skáldatal 54, 64, 68, 77 skaldic poetry 5, 20 def., 22–23 def. of authenticating and situational quotation, 39 def., 210–12 oral transmission (also 80–85, 259–62), 212–17 authenticating and situational quotation (214 strong dating criterion for the post-settlement period, but not before; 216–17 type of quotation not governed by genre or date of the saga, but by character of action and elevation of subject), 271 only two instances of spurious poetry in authenticating quotations, Skáldskaparmál 23 reliability of poetry, 29, 53, 55, 59, 65, 78, 84, 147–49 Snorri’s use of his sources – presentation of method and sources, 150–55 Boðn, 156–57 Óðrerir (the conception of Óðrerir as a vat did not originate with Snorri), 158–59 Kvasir (lack of definer for ‘mead’ means that Kvasir must be animate), 159–64 narrative formed from kennings that are not quoted (159–60 illumination at Óðinn’s and Ægir’s feast, 160–61 arnar leir(r) older than Snorri, 161–64 ‘dwarves’ ship’ – only one poetic occurrence, probably transformed into narrative by Snorri, 164 Based on poetic precedence, Snorri added details to pre-existent motifs, but did not invent motifs), 252 tool for stylistic archaization

352

Index

(also 253–54 use by Pseudo-Ófeigr, 255–59 use by Pseudo-Þjóðolfr, 266–68 use by Pseudo-Grettir), 288–89 Ókennd heiti in W skjálfhenda 25, 34 skothending 21 def. Skúli jarl (†1240) 30, 110 Skúli Þorsteinsson 63–64 Sneglu-Halla þáttr 255–59 Snorri (†23 Sept. 1241) 91–92, 111 bio, 174 composed no poetry himself in his Edda or in Heimskringla (also 241–42), but did so in Egils saga (also 243) Snorri’s Edda 6–9 compared to European counterparts, 9–11 rationale, 102 apex of poetic and scholarly developments; reason for analysing it from end to beginning, 109–10 grammatical manuscripts structured according to Snorri’s Edda, 118 the U-scribe’s intention of reproducing Snorri’s Edda as Snorri wrote it, 145–46 the rationale for the composition of Snorri’s Edda – traditional explanation untenable, 147 coherence of Snorri’s Edda, 174 Snorri composed no pseudonymous poetry in his Edda, 174–75 use of older poetry for corroboration in both Gylfaginning and Skáldskaparmál: religious interpretation in Gylfaginning and narrative flavour in Skáldskaparmál, 204 the rubric in U is the only discussion of the form, rather than the content, of a prose or prosimetrical text, 235 authorship of Snorri’s Edda (also 116–18), 237–39 comparison of prose style in Snorri’s Edda and Egils saga, 278–96 reception among monastics spurious poetry see pseudonymous poetry stál see refrain Steinunn (mother of Hofgarða-Refr) 210 stef see refrain stefjabálkr see refrain Stjórn 280, 284–85 Stjǫrnu-Odda draumr 246 Sturla Þórðarson 13, 47, 92 Styrmir’s book 73–75 Suetonius 207 Svarfdœla saga 251

Sverris saga 195 synonyms (poetic = heiti) 43 def. of poetic usage, 45–46 use in wordplay, 61 in generic mythological references, 78, 88, 138–40 T (Codex Trajectinus) 166 Theodoricus 4, 206 Third Grammatical Treatise see 3GT Third Vatican Mythographer 6, 96 tmesis 39, 47–49 def., 56 early period, 58 intermediate period, Sighvatr, 76 Haukr Valdísarson, 92 Snorri; lack of tmesis in Sturlung poetry, 93 connection to other features of diction, yet partial independence translation 2, 186–87, 227 n. 114, 293 Troy 179 tvískelft 34 tøglag (tøgdrápulag) 24 Þangbrandr 25, 210 Þingeyrar (monastery) 272–73, 277–78, 294 Þjalfi 50 Þjóðolfr Arnórsson 27, 42, 57, 255–62 Þjóðolfr ór Hvini 26, 49–50, 53, 56 Þórarinn stuttfeldr 24, 30 Þórðar saga hreðu 251 Þórðr Kolbeinsson 90 Þórðr Særeksson 28, 30, 57–58, 64 Þorfinnr jarl 65 Þorgils saga ok Hafliða 200–01 description of the wedding at Reykjahólar in 1119, 201–02 form of *Orms saga, 203–05 author not interested in saga form, but in content; poetry discussed as corroboration of content Þorláks saga 195 Þorleifs þáttr jarlsskálds 251 Þormóðr Trefilsson 24 Þórr 86–87, 255–58 Þorvaldr veili 24–26 þríhent 31 n. 48 þættir 263 U (Codex Upsaliensis, DG 11 4to) 107, 109–10, 115–18, 132–33, 160, 166, 204, 282–83 Ulfr Uggason 53 Vafþrúðnismál 164, 170

Index

Vergil 33, 37, 95, 109, 118–19, 169, 178–79 Víglundar saga 250, 265 Vǫluspá 86, 164, 168 Vǫluspá in skamma see Hyndljuljóð W (Codex Wormianus, AM 242 fol) 107, 110, 118, 166, 184–85, 267, 276–90, 303–19 Walter of Châtillon 187 William of Sabina 199 wordplay see also ofljóst; 39, 43, 44–46 def., 49–51 early period, 57 intermediate period, 76 Haukr Valdísarson, 88–89

353

Einarr; ofljóst, 89 Hallar-Steinn; ofljóst, 91 Hallar-Steinn; ofljóst, 92 Snorri, 93 connection to other features of diction Ymir 41, 65–66, 170–71, 174 York 38 Zoroaster 280, 286, 304–09 Ægir 159