In Search of the Unknown in Medieval Architecture 190459736X, 9781904597360

Citation preview

In Search of the Unknown in Medieval Architecture

In Search of the Unknown in Medieval Architecture

John James

The Pindar Press London 2007

Published by The Pindar Press 40 Narcissus Road London NW6 1TH · UK

British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library

ISBN 978-1-904597-36-0 (hb) ISBN 978-1-904597-53-7 (pb)

Printed by Estudios Gráficos ZURE 48950 Erandio Spain This book is printed on acid-free paper

Contents Introduction

i

HISTORY I

A Revised Chronology for French High Gothic: Consequences of The Ark of God, Part A

II

An Investigation into the Uneven Distribution of Early Gothic Churches in the Paris Basin before 1240

25

III

Memes and Assumptions 

56

IV

Chartres was Lucky the Parisians were Busy

64

1

TOICHOLOGY V

The Discipline of Toichology

89

VI

Were the Chartres transept porches added?

91

VII Anomalies in the Ascension and Incarnation portals of Chartres cathedral

100

VIII Multiple contracting in the Saint-Denis chevet

117

IX

Could Suger have built the choir of Saint-Denis in four years?

149

X

La construction de la façade occidentale de la cathédrale de Senlis

153

XI

A northern master for Clermont-Ferrand?

167

XII The end of discontinuous contracting

172

XIII Building the narthex at Chartres

178

VAULTING XIV The evolution of structure between 1080 and 1150

199

XV The peaked arch, and the earliest domical rib vaults in the Paris Basin

202

XVI The rib vaults of Durham Cathedral

222

XVII Rib vaults in Italy

241

XVIII Flying Buttresses before 1180

250

TECHNIQUES XIX Medieval mortar and the Constraints of Formwork

287

XX

295

The master masons as contractors

SCULPTURE AND DECORATION XXI Frenchmen in Peterborough

315

XXII The Rustic Master — A Chartrain Sculptor in Italy

324

XXIII Chapiteaux a feuilles d’acanthe de 1138 à 1140

334

XXIV Describing Chevron Patterns at Durham

337

GEOMETRY XXV The significance of geometry

345

XXVI Measuring and extracting geometry

356

XXVII The foot of the master

366

XXVIII The tools of Hues Libergier, master mason

371

XXIX The Creation Figure of Chartres cathedral

382

XXX Four ‘identical’ windows

394

contents

XXXI The western rose at Chartres

399

XXXII Discrepancies in medieval architecture, careless or deliberate?

413

XXXIII Gothic pinnacles 

434

XXXIV The geometric layout for the Sainte-Chapelle

445

MEANING XXXV The canopy of paradise

455

XXXVI The mystery of the great labyrinth at Chartres

467

XXXVII The zodiac on the Royal Portal, Chartres

487

ECONOMICS XXXVIII What price the cathedrals?

503

XXXIX

How many built All the churches?

520

XL

Funding the Early Gothic churches of the Paris Basin

524

Bibliography

557

Index

581

Introduction



T

The texts are not everything. The monument itself will reveal some of its secrets to those who know how to seek them. Yves Delaporte

HESE forty articles have accumulated over the past 35 years, often in little-known or remote journals. I am personally excited to have been given this opportunity by Pindar Press to collect and reprint the more important. Though some editing has been necessary I have tried to maintain the style and feeling tone of the originals. They range from my first on medieval architecture published in 1972 to today. As time and understanding do not stand still, all have been edited to improve clarity, to remove irrelevant or repetitive material or to correct those occasional statements that no longer accord with more recent research. My passion for medieval architecture is rooted in the remarkable fact that such a revolutionary manner of building could be created by a society that outwardly maintained that tradition was the basis of all law and all relationships. I have set myself what must be in the end an improbable goal, to understand how this revolution in stone was able to arise, both socially and technically, out of the conservative and fragmented societies of the twelfth century. I do not know of any other culture before our own, be it Chinese or Classical or Indian, that was able to throw over its established building forms and totally reinvent itself. It could even be said that our own culture is deeply indebted to the Gothic masons, for the concept of the lightweight skeletal frame with infilling walls that became the hallmark of modern architecture was first displayed in the buildings of 1200.

ii

Thus I became fascinated by the questions that have been explored in these articles: Where were the elements of Gothic invented? And who were the powerful individuals who did it? How did they evolve the techniques to enable these great structures to stand up? What were their means of documentation and communication on large building sites? Where did the enormous sums of money come from? And not least, what impelled these people to so exert themselves in the first place and why did it end? I am still far from these ultimate goals, but certain aspects have become clearer, especially with the gradual completion of the series of studies I have called The Ark of God.

History

I A Revised Chronology for French High Gothic: Consequences of The Ark of God, Part A*

I

N Part A of The Creation of Gothic Architecture — an Illustrated Thesaurus, also called The Ark of God, I assembled a large archive of foliate capitals from the Paris Basin. It had been remarked by Viollet-le-Duc, Jalabert and others, that the style of Gothic foliate carving between 1170 and 1250 changed gradually from printemps à l’été, from simpler more spring-like forms to highly complex summer-like images.1 In The Ark I have confirmed this observation, mainly from foliage that is clearly dated by the documents. They enabled me to define the carving manner of each of these eight decades Fig. I–1.2 The definitions were then applied to all those churches and parts of churches for which there was little or no documentation. Thus the one single item of foliage, common to every one of these buildings, could be used to lay the foundation for a chronology for over 400 churches, if in most cases only to the decade. These included all those that form the substance of historical analysis. With Chris Henige’s help, 300 documents pertinent to construction have been included. Each has been verified, the wording checked with the originals, and translated. They show that during the eighty years covered by these two volumes only parts of seventeen buildings can be dated with any certainty. Forty-five other buildings have texts with varying degrees of relevance and uncertainty as to time or place. Everything written on the creation of Gothic

* Adapted from a talk given at Leeds Medieval Conference, July 2003. 1 Jalabert, Flore sculptée, ch 1; Lambin, La flore gothique; Viollet-le-Duc, Dictionnaire raisonné. 2 James, Ark of God, Part A.

2

Fig. I–1. Samples of foliage from each decade between 1170 and 1240.

Fig. I–2. Typical formal foliage, from Senlis nave aisle 1155±.

Fig. I–3. Typical natural foliage from Chartres ambulatory 1203.

A chronology for french high gothic

3

architecture depends on them. Such a minuscule foundation has resulted in many unfortunate assumptions and memes.3 Scholars have become used to this arrangement, and though some have seriously questioned details — Caviness, Blum, Bruzelius and myself, to mention a few4 — it would seem that a global reassessment is now needed. My purpose here is to stimulate such a reassessment by setting out some consequences of the dating presented in The Ark for the period after 1170. Any reexamination of the earlier, and often more important buildings, will have to wait on the next volumes which are now in preparation. Not all conclusions will be unexpected and many dates are close to those used by historians but, as Bony wrote, “the margin of even a few years makes a great difference in terms of historical meaning”.5 The first conclusion was that during only one decade, the 1170s, the design of foliate carving in the Greater Seine valley was transformed from the formal style that Denise Jalabert called généralisée to a more natural style of carving, Fig. I–2 and I–3.6 The transformation was a localized phenomenon, and seldom appeared elsewhere for another twenty years. This conclusion is particularly vital for historians, for it provides a reliable watershed date whereby every building in the Paris Basin with only formal capitals should be dated before 1170±, and all those with only natural capitals should be dated after 1180±. For example, the west portal at Senlis and the Valois portal at Saint-Denis should both be dated before 1170 as all the foliage is formal. The second conclusion was that Gothic was not created solely in the great cathedral workshops from which ideas spread outwards to the minor sites, but that invention was an on-going continuous process, often found in the smaller buildings long before being used in the cathedrals. This contradicts earlier beliefs that, for example, the ideas generated in Reims were “passed on to a large number of new shops that it formed and colonized” and “once the work at Saint-Denis was finished . . . the workshop gathered together by Suger was moved to Chartres”.7

James, Template-makers, ch 1; “III: Memes and assumptions”. Caviness, Braine: primary sources, 524–48; Blum, Lateral Portals, 190–227; Bruzelius, NotreDame, 540–69. 5 Bony, French gothic, 273. 6 Jalabert, Flore gothique, 181–246. 7 Branner, Saint Louis, 380.

3 4

4

Fig. I–4. Saint-Germer-de-Fly interior 1160±.

In the grand histories that dwell mainly on the larger buildings and on the flow of styles and ideas, it is usual to credit the High Gothic tall clerestory to Chartres, the Rayonnant integration of triforium with clerestory to SaintDenis, to locate the invention of tracery in Reims and of flying buttresses in Paris — all major centers. The analysis in The Ark shows that none of this is true, and that the picture is more complex and more interesting. The essential ingredients of the High Gothic style created to the north and east of Paris were tall aisles and spaciousness, enormous windows stretching full-width between thin vaulting shafts and elongated below the vault springing into the triforium, and the whole articulated with delicate stone tracery in one plane and flying buttresses in the other. Whatever the decorative niceties applied to this format over the centuries, these were the core inventions from which all others stemmed. The third conclusion was that after 1160 major sources of inspiration for this format lay with the great Benedictine and Cistercian abbeys to the north and east of Paris, rather than the cathedral sites: with Saint-Germerde-Fly, Preuilly, Saint-Remi, Braine, Longpont, Chaalis, Orbais, Essômes and, finally, Saint-Martin-aux-Bois. The possible logic is that the abbeys may have been more inclined to investigate deeper religious issues, and may have more naturally discussed their ideas in a way that stimulated the masons to redesign their spaces, than the cathedrals that were more concerned with political issues.

A chronology for french high gothic

5

To illustrate these conclusions this article will address four of the major issues: The loneliness of Notre-Dame that inspired very little; Chartres was the outcome, not the origin, or invention was continuous and ubiquitous; The first High Gothic format was already present in the 1160s, and The concepts that made the Rayonnant of Saint-Germain-en-Laye began twenty years earlier. The loneliness of Notre-Dame It is usual to consider Notre-Dame the greatest work of the ‘Paris school’ rather than an exception.8 Apart from a couple of local buildings — Santeuil around 1176 and Chars started just a few years later — the tall-and-thin format of Notre-Dame seems to have had almost no immediate effect in the Ile-de-France. The elevational arrangement had been formulated in the later 1150s at the Benedictine abbey of Saint-Germer-de-Fly. Full of innovation, Saint-Germer presents the first tall-and-thin proportion with a four-story elevation, galleries that were intended to be vaulted and with supports for flying buttresses, Fig. I–4.9 The internal clerestory passage created a skeletal appearance as there was no walling adjacent to the ribs — a system that was to have a major influence in the Laonnais. Indeed, Chars, Fig. I–5, with its internal clerestory passage, should be seen as a more sophisticated version of Saint-Germer, a local facsimile rather than “an imitation of the original elevation of Notre-Dame”.10 From Aubert onwards scholars have dated the first flyers from those in the Paris nave, and Bony’s comparison of just one pair of capitals in NotreDame and Canterbury sufficed for scholars to place the nave in the 80s, Fig. I–6. However, the style of foliage of the first six bays shows they were carved in the early 1190s, while the more westerly bays were carved in the next decade. Not that it is now so important to keep the origin of flyers in Paris, as it is understood they were first planned and built thirty years earlier during the later 1150s along the Seine and Oise rivers.11 Bony, French gothic, 137; James, Template-makers, 112–17. James, Flying buttresses, 261–87. 10 Bony, French gothic, 141. 11 James, Flying buttresses, 284; James, Saint-Denis, 42–62. 8 9

6

Fig. I–5. Chars choir interior 1170s.

Fig. I–6. Paris, Notre-Dame nave to east, begun 1190s.

Fig. I–7. Champigny-sur-Marne nave interior 1220s.

A chronology for french high gothic

7

The height of the Notre-Dame aisles was determined when the external walls were built in the 1140s, and suggests an interior akin to Saint-Germaindes-Prés. This height could still have been intended in the gallery level of the late 60s if there was to have been no triforium, as at Mantes or Santeuil. The decision to raise the height had been made before Torigny’s letter of 1177.12 Having created this showpiece, great height hardly ever appeared in later years in the Parisis, but it was taken up with enthusiasm in the bishoprics of the north-east. Squat is far more usual. Aisles with about three-meter piers were normal for the area, and remained so until the 1250s, Fig. I–7. In the majority of Ile-de-France churches the vaults spring from below the triforium or gallery capitals, and the clerestory is pulled downwards and often reduced to a small oculus.13 In some cases it has been eliminated altogether. This is seldom discussed in the histories, which are generally more interested in following the tall-and-thin story being pursued in other regions. Squat has been unkindly condemned as having “no proper understanding of the major problems of the day”14 — yet who decides what is a “major problem” if not the locals? While the rest of the Basin was pushing upwards, the Parisis was determinedly hunkering even lower.15 Some scholars would wish to date the Notre-Dame western portals before the Chartres transept porches.16 The foliage indicates that the capitals over the northern embrasure figures were carved just before 1205 integrating some earlier work from the 80s, those alongside the central door around 1208 and the south just after 1210 incorporating earlier work from the 40s.17 At Chartres the south porch embrasure figures were in place by 1202 and all the northern embrasure figures were completed in 1208.18 If placement reflected execution, then this may settle the argument on whether the Chartres figures were earlier or later than Paris — they were either carved at the same time or, in some cases, Chartres may have been a year or two earlier.

Delisle, Robert de Torigny, 68. James, Template-makers, 115–16. 14 Bony, Resistance to Chartres, 40–1. 15 James, Template-makers, 112–6. 16 Lefèvre-Pontalis, Architectes, 113; Sauerländer, Notre-Dame in Paris, 1–56. 17 James, Ark of God, 940–44. 18 James, Contractors, 382–93. 12 13

8

Chartres was the outcome, not the origin For over a century Chartres has been seen as the source of any building that carried its genes. Consequently Braine, Soissons, Longpont and Orbais were all dated after it. Yet the foliage in the capitals is unambiguous in showing that the essential parts of all of these buildings were completed or under construction many years, even decades, before Chartres. They are the source, not the progeny. The vocabulary of forms at Chartres had been established over the previous two decades, the only difference being that at Chartres “power is the key: in constructional engineering, in the carving of space”.19 Saint-Yved in Braine: Madeline Caviness has argued from the documents that “cumulative though largely circumstantial evidence suggests that the Premonstratensian abbey of Braine was begun about 1176”.20 As there are only a couple of formal capitals in the aisles of the choir and first bays of the nave their foliage would have been carved within a year of 1180, with the footings laid down a little earlier, as Caviness suggests. This is almost twenty years before Chartres. There are two elements at Braine that were to have a profound impact on the future: the continuous triforium and the layered clerestory, Fig. I–8. The Mans nave from the 1140s and Juzières ten years on were earlier, as were those in the Laon and Noyon cathedrals from the 60s where the triforium was less important than the vaulted gallery.21 But in Braine it formed the center of an austere and unadorned three-story elevation that was to become extremely popular. By extending the triforium full-width between the vaulting shafts any remaining wall was squeezed out. The triforium arcade appears to continue behind the shafts before reappearing on the other side. Woven in front of this, string courses and the edges of walkways continue across the outside of the shafts. Weaving horizontal and vertical elements reduced mass to line and pattern, the first step in what was to be applied to every element of the building in order to reduce heavy masonry to a skeleton. The second element found at Braine is what Bony called recession, a concept of ‘diminishing bulk’ with stepped-back planes and external Bony, French gothic, 233. Caviness, Braine: primary sources, 524. 21 James, Template-makers, 116; Bony, French gothic, 105 et seq. 22 Bony, French gothic, 187–9. 19 20

A chronology for french high gothic

9

Fig. I–8. Braine interior to north-east begun 1176±.

walkways in the clerestory, Fig. I–9.22 It was also used at Saint-Michel-enThierace, which is contemporary with Braine in the lower story but a decade later in the upper. The earliest example of recession may be the clerestories of Guignicourt from the later 60s and Mons-en-Laonnais from the early 70s. In any case, it comes from the Laonnais and seldom strayed beyond until the next century. As the external walkways in the Chartres ambulatory of 1202/3 were designed by a builder from the north-east this is not an exception. Soissons cathedral: Our understanding of documents depends on interpretation, and there have been as many missed opportunities at Soissons as there were at Chartres, where the misreading of the Chartres moneychangers document by Lefèvre-Pontalis led to seventy years of error that was finally corrected by Robert Branner.23 During and after the 80s money was given and chaplaincies founded for the choir, and there were annual processions around the ambulatory.24 These one could not imagine if the choir aisles were not in place. These documents together with the 23 24

Branner, Chartres. James, Ark of God, 1597–98.

10

Fig. I–9. Braine apse with recession in clerestory 1180s.

Fig. I–10. Soissons cathedral tall clerestory 1192±.

A chronology for french high gothic

11

interconnected coursing between the choir and the south transept,25 suggest a commencement date of about 1182. The gift for the small roses of the clerestory in 1202 should have alerted historians to a more realistic completion date for the clerestory walls, Fig. I– 10.26 We may guess that, as with most fund-raising, donors were encouraged to pay for specific items to provide a sense of pride and personal achievement, especially effective if the donor could see an immediate outcome for his money. It is therefore likely that the upper parts of the clerestory walls were being placed in 1202. Both these dates are confirmed in the foliage of the capitals. In Chartres it needed thirteen years from the first stones of the nave triforium to the underside of the clerestory roses. Construction at Soissons may have been a little faster as only the choir was being erected, but even so, the decision for what Carl Barnes has called the ‘tall clerestory’ in which the high vault capitals were raised above the sill, would have been made at least ten years before, around 1192. Therefore Soissons definitely predates Chartres. Orbais-l’Abbaye: Two chapels were dedicated in 1180, and the capitals over the drum piers confirm that the aisle vaults could have been completed by then, though the walls had been built almost 30 years earlier, Fig. I–11.27 Another document states that the work was completed by Count Thibaut during his reign, 1197–1203. The capitals in the triforium and clerestory will not construe a date later than this, in spite of arguments by some historians.28 They indicate the mid-90s. Thus the tall clerestory of Orbais would seem to be virtually contemporary with Soissons. There is also the interesting choir at Moret-sur-Loing where the clerestory capitals were raised just above the sills in the last years of the century, see Fig. I–12. Thus by 1200 three buildings had begun to raise the springing of the high vaults, over a decade before the decision was taken to do the same at Chartres. The key inventions that went into the making of Chartres are all from the north-east: the stretched aisle, the full-width window, the giant rose, the tall clerestory and the tas-de-charge.

James, Template-makers, 128–29. James, Template-makers; I have not found any errors in the toichological observations. 27 James, Template-makers, 52–53. 28 Villes, dismisses their value, St-Pierre, 549–551; Héliot, Deux églises, 87–112. 25 26

12

Fig. I–11. Orbais clerestory complete before 1203. Fig. I–12. Moret-sur-Loing choir 1190s.

A chronology for french high gothic

13

The stretched aisle: Piers over 8 meters tall became almost standard in the great buildings of the 1230s and 40s. Up to the 20s tall aisles were found almost exclusively in the north and east. These aisles are usually supported on thin drum columns. The earliest would be the early 80s — in the choirs of the two Cistercian abbeys of Longpont and, to a lesser extend, Chaalis,29 and in the ambulatories of Soissons and Saint-Quentin, both around 1182.30 The Longpont drum piers are just under 9 meters high, which is almost the same as Chartres, Fig. I–13. The trend to stretch the aisle upwards culminated in the enormously tall aisles of Reims, Beauvais, Amiens and Saint-Martin-aux-Bois. The simplicity of these stretched buildings was in marked contrast to the complexity of the four-story format in Laon and Noyon of the 60s, with heavy piers and shafts, “with multiplicity of divisions . . . and a tight grid of small-scale units”.31 Where Bony refers to it as “an ideal gridwork for the development of height”, the very mediocre heights achieved with a four-story elevation puts this in doubt.32 The Soissons south transept of the 70s was a last attempt to thin the members while keeping to four stories, Fig. I–14, an attempt that was superceded almost immediately in the same building by the stretched aisle and elongated drum piers of the choir. In the Chartres nave the bases for the first pilier cantonné were laid in 1196.33 The porch at Corroy in the Marne may have been earlier, though the capitals are very worn and hard to decipher. This was followed in the 1200s by the naves of Montataire, Pacy-sur-Eure, Saint-Leu-d’Esserent and NotreDame in Paris. It was never as popular as the drum, by one to five.34 The full-width window: In polygonal apses it was common to set the window shafts alongside those of the ribs, at least from the time of SaintMartin-des-Champs in the 1130s. This was not applied to the wider walls

For dating “XXXV: Canopy of Paradise”. Tall thin piers, often built as en délit monoliths, had been used sixty years earlier south of the Loire, as in Fontevrault and Paray-laMonial. 30 The date for Saint-Quentin was determined by the height of the chapel walls and Soissons by the level of the window arches that were in place before the gallery in the adjacent transept was vaulted, around 1182; James, Template-makers, 129; James, Ark of God, 485 and 495. 31 Bony, French gothic, 227. 32 Bony, French gothic, 107. 33 By Rose in campaign ‘C’, James, Contractors, 117. 34 James, Template-makers, 105–6. 29

14

Fig. I–13. Longpont nave, choir begun 1180s.

Fig. I–14. Soissons cathedral south transept begun 1178±.

A chronology for french high gothic

15

in straight bays until Saint-Remi. In its axial chapel in the 60s the framing shafts of the straight bays merge into the rib shaft, Fig. I–14. All trace of the wall had disappeared. This was followed in the triplet-windows on the eastern apse wall at Fossoy and in the Saint-Remi clerestory a few years later, Fig. I–15. By the 80s the idea had entered the mainstream. Eliminating the walling alongside the windows advanced the weaving process that had been begun in the triforium into the clerestory, and further added to the skeletalization of the interior. However, the shafts used to frame the triplet windows tended to diminish this process by drawing attention to the frames rather than to the glass. The Orbais clerestory avoided this by eliminating the framing shafts. The next move was to replace the triple window with double lights, and to set a rose above that. As with nearly every item discussed here, their first use was among churches of the north-east rather than in the Ile-de-France. Of all the double-windows-with-rose in the Basin almost fifty percent were built in the north-east before 1200 compared to 12 percent elsewhere, while after that only one-fifth was erected in the north-east. The great rose windows: Most roses from the 1130s onwards were, like other windows, narrower than the bay and were surrounded by large areas of walling. Spreading the window across the entire bay with only minimal walling on each side — just enough to stabilize the masonry in the corners — was first used in the 60s in the Cistercian abbey of Preuilly, now in ruins, Fig. I–16. This was followed by Guignicourt in the late 70s and many other Laonnais buildings in the 80s and, further afield, the Donnemarie choir to the south later in that decade. The extension to the Laon cathedral choir and the westernmost bay of the nave were both started within a year of 1180, and completed by the mid90s. The foliage on the capitals quite clearly dates these parts, Fig. I–17.35 As well, the profiles in both end walls are almost identical, and are recognizable in two masters who later worked at Chartres: Bronze in the lower level and Scarlet in the upper.36 The tall clerestory: As long as the vaults sprang from below the clerestory window sills they were still being visibly supported by the wall. 35 36

James, Template-makers, 84. Besides their foot units in the templates, the major items for Bronze are porch details, eastern window frames, and corbels; and for Scarlet the corbels, string courses, roses and stair windows.

16

Fig. I–15. Saint-Remi axial chapel begun 1165±.

Fig. I–16. Ruins of Preuilly choir 1160s.

A chronology for french high gothic

17

Fig. I–17. Laon cathedral dado capital inside western porches early 1180s.

The masonry between the arches of the triforium and the clerestory sills was the last remaining functional masonry visible from the inside. To bring the whole building into consistency with the illusional concept of skeletal weaving, the supportive function of the wall had to be removed. Bony calls this “the point of departure for the achievement of gigantic height”.37 It was normal in all single-story churches for the vaulting capitals to be placed higher than the window sills. It was only a matter of time before this concept was applied to the clerestory of a large multi-story building. The first lowering of the sill below the capitals of the vault spring was taken either at Soissons or Orbais just before 1200 — a decade before the same scheme was begun at Chartres, Fig. I–10 and I–11. It could be argued that the tall clerestory had been planned in these buildings when the footings were being laid. At Soissons this is belied by the date the crossing capitals next to the south transept were raised, and at Chartres there is evidence for a change to the design in 1210.38 Work paused just below the clerestory walkway at the most advanced western end. As there was no technical reason for a pause at this level, one wonders whether

37 38

Bony, French gothic, 225–26. James, Contractors, 438. Grant Hildebrand and I reported this process in James, Pioneers, 8. For Soissons James, Template-makers, 132-39.

18

work was deliberately held back while a more advanced design was being considered. If this were the case, it coincided neatly with the completion of the choir clerestory and vaults at Soissons, just before the clergy occupied it in 1212. The tall clerestory would not have been possible without the tas-decharge. This constructional device was first used just before 1150,39 for in thinning the shafts the width of the impost that supported the close-packed vaulting arches was greatly reduced, and a neat way had to be found to merge the arches with the ribs. This was done by cutting all the lower voussoirs of all three arches from a single stone. In the narrow piers of the tall clerestory this technique was adapted and improved. The fact that the horizontal bed joints in the tas-de-charge raised the effective height of the springing opens the question of whether the masters realized that in a number of earlier buildings they had already, to all intents and purposes, raised the vault springing above the sills. If so, the structural modeling of, say, Saint-Martin at Etampes, with the top of the tas-de-charge set two courses above the sills, would have been perceived as acting in the manner of a tall clerestory, the only difference being that visually the capital still lay below the sill.40 All the evidence points to the fact that the bishoprics of Laon, Soissons and Reims, with a little help from the Champagne, created the vocabulary for High Gothic before 1200. It was only gradually that the south and west of the Basin took these ideas on until, in an extraordinary moment, they transformed them all into the intricacies of the Rayonnant. The first High Gothic format The axial chapel in the abbey of Saint-Remi has received occasional credit for the paradigm it created, though not enough. Contemporary with the aisle piers of Notre-Dame, yet totally unlike it, Saint-Remi contained most elements of what was to become the High Gothic style, though with no great height in the aisles or gallery, Fig. I–15.

39 40

James, Template-makers, 108-12. Aubert, Plus anciennes, 36, refers to the corbelling of the lower courses of the ribs in the Durham nave (1128/33) as a tas-de-charge, for the bed joints are set perpendicular to the ribs and not to the doubleau.

A chronology for french high gothic

19

The foliage on the capitals bear out Anne Prache’s analysis that dates the chapel foundations to 1165± and the gallery to around 1172.41 Though the upper flying buttress arches do not seem to have been erected until the high vaults were completed in the 90s, the foliage on the bosses, the gallery vaults and the clerestory walls and timbers of the main roof show that all were in place by 1175/7. The foliate evidence for this is that three-quarters of the capitals in the upper stories are formals, which indicates the early 70s.42 Jean Bony wrote that “in the last quarter of the twelfth century architects ran a fine surface network over the inner face of the building to give the impression of a coherent armature which, in spite of the thinness, appears to support the whole fabric: an illusory effect but not an irrational one”.43 In Saint-Remi this occurred earlier than that. The chapel is the first example of the light-weight skeletal double-wall system, and the clerestory the first example of the triplet window linked to the triforium that created a ‘surface network’ across the upper two stories. The chapel fits between huge supports for later flyers, full-width windows and minimal-sized shaft members that are as thin as any built over the next fifty years. There is full dissolution of the wall face behind the passage screen and the dado arcades. The lightness of the dado arcade feels like a lower story, with a strong division under the window sill emphasized by the walkway string course and tall windows. The vaulting capitals are placed above the sills producing a miniaturized version of the tall clerestory. Other oft-quoted examples of the double-wall window and recessed upper story are later: the gallery chapels of the Laon transepts and SaintJulien-le-Pauvre from the late 70s, and the twin double-wall clerestories of the Noyon transepts that were not erected until well into the next decade. Saint-Remi’s first clone, the choir of Notre-Dame-en-Vaux in Châlonsen-Champagne, was begun as the roof was being laid over Saint-Remi, and took as long to build. Its semi-clone in the Saint-Quentin ambulatory chapels was begun as Notre-Dame was being finished. The concepts that made Rayonnant Where the axial chapel of Saint-Remi was a model for the High Gothic, so the royal chapel of Saint-Germain-en-Laye, completed in 1238, displays all Prache, Saint-Remi. James, Ark of God, 11–42. 43 Bony, French gothic, 158 et seq. 41 42

20

the major qualities of the new paradigm that came to fulfillment over the next ten years or so — tracery, dado arcades with tri-lobed arches, recession with passageways under the windows and comprehensive undercutting to profiles and the arches of the dado, Fig. I–18. It was part of the resistance in the Ile-de-France to what we call Gothic that the region was affected by neither the tall clerestory nor tracery until the 1230s, when the ‘Court School’ materialized. It is surprising that this happened at all, considering that the nearby pilgrimage center at Chartres had so little influence on the Parisis. Chartres denied all those values which the ordinary Parisians held most dear. When Rayonnant emerged it collected almost no ideas from the Demesne, but imported its major vocabulary from the north and east where they had been invented over the previous thirty years. These elements were first assembled in Saint-Germain.44 There are no plain surfaces — all are covered with traceried patterns, even to the spandrels between the windows and the roof cornice. Only the stretched aisles, and the linked and glazed triforium had to wait on a larger building. In all these things Saint-Germain is, at one level, just a highly decorated version of the Saint-Remi axial chapel built seventy years before. Comparing the two, the basic forms are no different. It is the surface patterning of tracery that turned Saint-Germain into a new archetype. Tracery: In small oculi, solid plates of stone were used to divide the circle into trefoils and quatrefoils. The earliest are in the north and east in the 1140s and 50s: Oulchy-la-Château, Notre-Dame-en-Vaux transepts and nearby Francheville. This is essentially plate tracery. A larger format developed during the late 70s in the same area in which a central opening was surrounded by four, six or eight smaller openings, such as the Laon north rose, the Couvrelles west wall and the Boult-sur-Suippe clerestory, all very close to 1180. Sometimes the small oculi over paired windows or doorways would be treated in the same way, as in the north arm of Saint-Eugène in the early 90s, the Orbais clerestory and the Grisy and Champagne-sur-Oise west doors a decade later. Bar tracery first appeared where the plates in oculi were replaced with thin bands of curved stone that left cusp-shaped openings that were filled with glass, probably first in the narthex at Glennes in the 80s. More complex

44

Champceuil in the south from the 20s is the only exception.

A chronology for french high gothic

21

arrangements may have appeared early in decorative work, as in the panels under the stair windows of the Chartres transept porches where small basreliefs show an imaginative collation of windows and oculi.45 This introduced a significant new concept into medieval architecture: before, the geometric form of every element (like a window was designed symmetrically around an axis, and the centers of any arcs lay within the element. Buildings were an assemblage of such forms. But these cusps are not centralized forms. They are geometrically accidental, being the by-product of other actions. They allowed the indeterminate to have an equal presence with the determined. The great tracery windows would have been inconceivable if the concept of the ‘accidental form’ had not taken on. It was at the Augustinian abbey of Essômes-sur-Marne that an accidental form was first allowed to penetrate the walling between the window openings, in or just before 1210, Figs. IV–4 to 6.46 This set the stage for the marvelous traceried aisle windows of Reims that were installed some ten years later. However, apart from a related group of churches centered on the Marne, bar tracery did not become a wide-spread concept for another generation.47 In the Ile-de-France even the extremely tall Villers-Saint-Paul windows of the 40s had no use for anything but plate tracery. The trefoil arch has a not dissimilar effect, being decidedly horizontal and non-structural. The earliest seems to be the dado of Avenay-Val-d’Or from the 1210s, followed by Mont-Notre-Dame and the Notre-Dame Gallery of Kings shortly afterwards. The extension of tracery patterns beyond the frame of the window, be it lancet or rose, seems to have first appeared in the lower spandrels under the north rose at Chartres, around 1233.48 After Saint-Germain-en-Laye patterns of tracery were applied to any bare wall surface and to the cusp-like spandrels over aisle and triforium arches, playing such revolutionary games as had never been seen in the north-east where the concept had come from. Saint-Martin-aux-Bois is a splendid example of this. The linked and glazed triforium: The first linkage was in the choir of Saint-Remi just before 1175,49 with six more before 1230: Orbais, James, Ark of God, 522 and 535. James, Template-makers, 156–59. 47 James, Template-makers, ch. 8 and map 177. 48 Probably financed, like Saint-Germain, by the Court. James, Contractors, 495–97. 49 The vaults were not completed until the 90s. Prache, Saint-Remi, 95–7 and Notre-Dame-enVaux, 29–92; James, Template-makers, 46–49. 45 46

22

Fig. I–18. Saint-Germain-en-Laye interior completed 1238.

Fig. I–19. Vaudoy apse 1215±.

A chronology for french high gothic

23

Notre-Dame-en-Vaux, Essômes, Vaudoy, and in Reims the cathedral apse and Saint-Jacques. Though clearly not exceptionally popular, it was taken over by Rayonnant builders to enable them to absorb the triforium into the clerestory. The first triforium with back-lighting was in the western wall of Couvrelles around 1180, the Saint-Leu-d’Esserent choir later in the 80s, and Mello twenty years later.50 The first with windows the same size as the triforium arcade was Vaudoy in the later 1210s, Fig. I–19. From foliage and profiles I cannot accept Branner’s view that Vaudoy was a copy of Beauvais.51 It stands on its own as an independent creation. The effect was, as with linkage, to merge the clerestory with the triforium. This usually meant that the roof over the aisles had to be kept away from the triforium wall instead of leaning up against it, involving box gutters which have never been renowned for being fully watertight. From a roofing point of view it was an uncomfortable solution which stretched the plumber’s skills to the utmost. Additional small shafts were added around the panel containing the triforium that pushed it back from the plane of the wall. Tracery was applied over the junction between clerestory and triforium until at Saint-Denis the pointed gables of the triforium look like the skyline of the Heavenly City.52 Conclusions In short, we can say that the evolution of Gothic was not a process that exploded out of a few powerful city workshops to seed ideas elsewhere. Creativity was a continuous process, and the great cathedrals were the culmination of decades of prior experimentation. During the years before 1150 invention and originality came from the whole of the Paris Basin, and in particular from the Oise, but after that all the inventions that created the vocabulary of High Gothic came from outside the Ile-de-France, being most importantly, interconnected and fluid spaces, thin walls with slender members and flying buttresses. Thus for the years after 1160, and for the key elements of thirteenth century architecture, we need to reassess von Simson’s “Gothic is the style of the Ile-de-France” and Bony’s “Gothic surged suddenly For Couvrelles and Mello, James, Template-makers, 58–61, 165, n. 51; Bruzelius, SaintDenis, 150. 51 Branner, Saint Louis, 22. 52 James, Traveller’s key, ch 3. 50

24

in the Ile-de-France and developed its conquests until the final years of the twelfth century”.53 In fact, the bulk of innovative architecture came from the abbeys of the northeast — the Soissonais, the Reimois and the Laonnais — and not from the cathedrals. That these were some of the major Benedictine and Cistercian houses in the area suggests that the concepts may have been consciously evolved in the intellectual atmosphere found in many of the abbeys. The lists notes where these concepts first appeared. The abbeys, marked in bold, predominate. Dates are within a couple of years, and are not for the commencement of the building, but for when the item was constructed. Designs were changed so often it would not be wise to assume that, for example, the Orbais tall clerestory was intended when the plan was prepared over forty years earlier. 1160 Saint-Germer-de-Fly: tall-and-thin, internal recession 1165 Preuilly: full-width rose window 1165 Laon: continuous triforium 1167 Saint-Remi chapel: full-width windows with passage 1170 Notre-Dame Paris: extreme height 1172 Saint-Remi choir: linked triforium 1180 Longpont or Chaalis choirs: stretched aisle 1185 Braine clerestory: external recession 1192 Orbais or Soissons: tall clerestory 1196 Chartres: pilier cantonné 1210 Essômes: bar tracery These concepts played decisive roles over the following decades:54 together they enabled the upper stories to become a replica of the Heavenly Jerusalem floating above the congregation.55 Based on the chronology evolved in The Ark of God, the 1170s transformed more than foliate carving from formal to natural: it simultaneously redefined the nature of architecture. Some very deep psychic adjustment had occurred in the community to bring about such a transformation in aesthetics. The recession in building that can be traced at this time may have been the cause, the trigger or even the result.56 Whatever the reason, there had been a crucial change in mood. Von Simson, Gothic cathedral, 64; Bony, Genesis of gothic, 18–20. James, Template-makers, 179 et seq. 55 James, Template-makers, 115–20; “XXXV: Canopy of Paradise”. 56 “XL: Funding”. 53 54

II An investigation into the uneven distribution of early gothic churches in the Paris Basin before 1240 *

G

OTHIC architecture evolved in the Paris Basin during the century that separated Saint-Denis from the Sainte-Chapelle. There are more than one thousand five hundred churches containing something from this intensely creative period. During the summer of 1980, I visited all the buildings noted in the literature and in the archives as being of this time.1 By the end of the year, the list had grown to 315, over thirty of which I had found by accident.2 It was easy to suspect that there would be more buildings, and so dur­ing the summer of 1981 I made a complete survey. In over seven months I visited every town in the Basin, and saw some 3,500 churches. This increased the list from 315 to 1284 — to four where there had been one. In the twentyfive years since then that number has been increased to over 1400.3 French scholars were not surprised that the original list was neither accurate nor complete. Map II–1 shows the zone within the Paris Basin that contains most of these churches. There are relatively few beyond this, and though the list contains cathedrals such as Amiens, le Mans and Bourges, they are peripheral to the most energetic centres of production. The limits were established

* Adapted from The Art Bulletin, lxvi 1984, 13–46. 1 Principally from the Inventaire Général des Richesses de France; Bibliothèque du Patrimoine; Bulletin Monumental; Congès archéolo­gique; and Brosse, Dictionnaire des églises. 2 James, Pioneers. 3 The full list, larger that that which appeared in the Art Bulletin, may be found on my web site, www.johnjames.com.au.

26

Map II–1 . The distribution of churches in the Paris Basin. Those from category 1–3 are marked with a solid black dot, and category four with an open circle. The dashed line notes the limit of the total search.

early gothic churches in the paris basin

27

when I might drive for a day and not find any churches, or only a few of the simpler sort. Besides the royal censuses taken around 1300, there is little data avail­able from earlier medieval France that is not piecemeal. This survey seems to provide the most comprehensive sta­tistical map of any medieval activity for the area. Few have had the opportunity to visit all the buildings from a particular time and place. There is a great assurance in such a total survey, for it provides a solid idea of the period’s diversity and conventions. Style was necessarily the basis for selection, as there is little assured dating.4 The dates for these churches lie from the oldest in the eleventh century to the Sainte-Chapelle, as the survey was extended over the years to include every pre-1250 church in the Paris Basin. It is a bewilderingly diverse period, and a simple clear definition of its stylistic characteristics remains elusive. However, eliminating those elements which varied from place to place and from one contractor to another, there seem to be a number of common ideas that can be defined. The earlier buildings can be distinguished from Roman­esque by the thinning of the shafts and ribs and their gradual emancipation from the wall. Space and movement come to dominate the inertness of mass. This can be seen in the shafts of Antony, the ribs of Saint-Denis, and the spaces of Nouvion and Saint-Remi, Reims. Throughout the century and half, the origins of each element re­main reasonably clear, as does the expression of its func­tion. But towards the end of the period, these distinctions blur as the hitherto separate mouldings begin to fuse and their identities are lost. The projection of the impost is re­duced, the scotia in the torus disappears, and window shafts not only move closer to the wall shafts, as in Avernay, but the junctions between them become fluid so they tend to merge into a bundle of functionless roll‑moulds, as in the Sainte-Chapelle. This merging becomes more universal as the importance of the wall‑as‑structure diminishes. In some areas, windows become the dominant element, but this is not universal. Similarly, tallness is not to be found everywhere. Though the skeletization of structure is perhaps the most important contribution for the future, during this century it is not particularly common. The dif­ferent ways of handling the forms, in small buildings as in large, all share the same

4

Documents discussed in James, The Ark of God, ch. 2.

28

approach to details. Thus the Sen­lis choir lies within this definition, while the Morienval nave does not, and the Reims cathedral aisle windows lie within it, while the choir windows of Damery do not. The survey included every building containing any work from the period, on the principle that even remains, however small, would indicate where something had been built and so contribute to the significance of the distribution. It was assumed that if a building had been totally recon­ structed without leaving any trace, it would either have suffered a major tragedy like those destroyed in the recent wars, or would have been poorly built. I will describe later how I have compensated for the former, while instances of the latter will still indicate something about the funds and materials available, for insubstantial construction may be as indicative of the conditions around 1200 as more positive evidence. To the extent that this may be true, the map de­limits those areas which had both the cash to build and the will to dedicate that cash to quality building and not to some other purpose. As these matters of cash and quality were considered important, the churches were separated into four cate­gories. The first includes the most complex, being the ca­thedrals and the larger churches like Orbais. The second contains those with a number of parts like Lhuys, or of more than two stories like Jouy‑le‑Moutier and Crècy‑la­-Chapelle. The third category comprises those with only one major part from this period, like Hautefontaine, SaintFram­bourg in Senlis, or Mennency. These divisions were based on the measure of complexity and thus the number of “projects” and, for all three categories, the care taken in detailing, carving, and construc­tion. These are the more significant buildings for the historian. Into a fourth category, I placed the crudely built, simple, or rustic churches, and included those which had been restored beyond recognition, like Saint-Vincent in Senlis, Bondy, and Brenelle. Also included were earlier buildings into which vaulting or ribs were inserted, like Juziers, or where the building campaigns had been so numerous, and the resulting confusion so great, that I felt they would con­tribute little to our understanding of the period, like Cra­moisy and Villers‑Hélon. This category also included all buildings where only the tower was of this period, like Var­infroy and Cauffry. This group numbered 616, or 48 percent of the total. It should not be thought that the buildings in the third or fourth category were all constructed by local or rela­tively unsophisticated

early gothic churches in the paris basin

29

builders, nor, conversely, that these local men never worked on the better buildings.5 Similarly, it should not be thought that the quality of design and workmanship in the second and third categories was necessarily any worse than the first. They were at times built more economically, with consequent problems from damp, poor foundations, or inadequate buttressing. Yet their careful detailing and high‑quality finish, the exact geometry and skilled carving in the capitals, show they were no less cunningly built and designed than the best‑known works, and often by the same men. In Map 1 the two major groups are distinguished. This distinction has been useful, for the number of less important churches can distort our analysis of the maps. One would think that there had been a great deal of activity in the Vexin Français, the Gâtinais, and the Sénonais, unless one knew that more than two‑thirds of the buildings were in the fourth group. By comparison, only nine percent of those around Braine are unimportant. Much has been destroyed, though the amount of violent destruction from all earlier centuries may be less than the devastation of commercial pillage since 1800 and of the co­lossal firepower of the last two wars, which was more likely to obliterate a building than any number of marauding fourteenth‑century armies — the witnesses to whose dam­age may be anything but trustworthy.6 To offset this bias, I have included in the map any build­ings from this period destroyed since 1900.7 We still have photographs and sketches of One of the masters at Chartres, Green, seems to have been a local man judging from the roughness of his work and the absence of details attributable to him in buildings to the north — James, Contractors, 484. In other buildings, crude and simple work that appears to be by local teams lies alongside masonry by the best: as in Dourdan where the eastern triforium and the lowest courses of the clerestory were erected by Ruby to the same plan and details used at Chartres, and in part of the triforium in Saint-Léger of Soissons and Saint-Jacques of Reims. 6 Medieval descriptions of damage, particularly during the Hundred Years’ War, are possibly exaggerated. Many churches reputedly destroyed, like Chalmaison, remain today much as they must have been in 1250. The Wars of Religion and the Protestant League may have wrought more dam­age than the Burgundians and the English, as their actions were specifically directed against the Church, and they used artillery. Denifle, Guerre de Cent Ans; Réau, Monuments. 7 Inadequate French records suggest that 71 of the churches destroyed since 1900 contained work from this period. Of these, half would seem to have been category four. Also the 46 ruins or totally rebuilt churches are included in their appropriate categories, including those turned into granaries or houses like Montrou and Montmort. 5

30

some of them,8 and though they do make a difference to the map in the area around the Chemin‑des‑Dames and Reims, most of the war zones lay in areas where there were few Early Gothic churches. I know of no way to determine whether the density found today is still in proportion to what it was in the thirteenth century, but having placed these destroyed churches on the map, I discussed the matter with three French scholars — Franoçois Salet, Anne Prache, and Léon Pressouyre — who felt the losses may have been fairly evenly spread, and that the map should be a reasonable representation of the orig­inal densities. To check these matters, I made a detailed examination of the area around Paris. Within some fifty kilometres of the centre, and excluding the churches built within Phi­lippe‑Auguste’s walls (as their numbers may have distorted comparisons between the Paris region and the rest of the Basin), the more detailed maps of the sev­enteenth century show 238 towns with churches, spaced roughly three kilometres apart.9 By basing the search on these old maps, the endless churches built in the metropolis since then were eliminated. The population density of the area around Paris varied from six hearths/ hectare in the south to eighteen in the north. At 4–4½ people per hearth, the population of this area would have been some 120,000, excluding the city of Paris, or more than 400 per church.10 Seventy‑one of these 238 churches had been rebuilt since the maps were prepared; seventy‑six were Renaissance and fourteen were medieval from after 1240. Of the remaining seventy‑seven with something in them from around 1200, fully thirty‑five were category 4.

Bibliothèque du Patrimoine; Archives du Département de l’Aisne; Den­ifle, Guerre de Cent Ans; Lefèvre‑Pontalis, Architecture religieuse, Lefèvre‑Pontalis, Architecture dans la Champagne; Moreau‑Nelaton, Fère-en-Tardenois; Moreau-Nelaton, Chateau­-Thierry; Moreau-Nelaton, Arondissement de Soissons; Moreau-Nelaton, Soissons; Moreau-Nelaton, Chez‑Nous; Trouvelot, Monuments historiques, 329–30; Cameron, Eglises détruites. 9 The Beaurain Collection, Atlas Géographique, vin, 1749; Cassini, Carte; Mariette, Diocése; Mariette, Ev­esché; de Vaugonoy, Atlas; Christ, Eglises Parisiennes; Petit, Val‑de‑Marne; and Inventaire Général des oeuvres d’art, 9 vols., Paris, 1886 on. 10 Dupaquier, Paroisses et communes; Dupaquier, Population. This is confirmed by the 1328 census which disclosed that there were on average 101 hearths per parish — Fawtier, Capetin kings, 96; Fourquin, Population, 63–64; Baldwin in Masters, Princes, 132; estimated the Paris population at 20–50,000; Fourquin, Population, 83, at 61,000; Lot, L’etat des paroisses, 297, at 200,000; and Russell, Medieval Regions, at 80,000. 8

early gothic churches in the paris basin

31

Paris, with all of its metropolitan attractions, its growing population, and its power to draw wealth towards itself, is surprisingly almost average for the region. First, 32 percent of the churches are Early Gothic, compared to 33 percent for the whole Basin. Secondly, the proportion of more significant to fourth category buildings was a little less than average — being 12:10 compared to 10.8:10 overall. This propor­tion may have been similar in 1240, as the records tell us that most of the churches rebuilt in the Paris area after 1800 were the smaller and simpler ones. In the Val‑de‑Marne, we know of forty‑two churches founded or mentioned before 1250.11 Twenty of these, or a little less than half, still remain in whole or in part. There is no way of knowing whether this ratio may be true for the region, nor whether the depredations were greater in the Paris area as the suburbs advanced and as the fashion­able views of the centre took their toll. Nor is there any certain way of knowing whether those churches which dis­appeared were equal in quality to those which remain. Yet the figures for the proportion of extant Early Gothic churches compared to those from other periods is fairly equal in the Paris area to elsewhere, and I am therefore tempted to suggest that, excluding both the cathedrals and the smallest chapels, we may have lost not much more than half of those built around 1200, and that most of those lost may have been in the larger towns. While the absolute numbers are interesting, it is the dis­tribution that is important, and from which major conclusions can be drawn. The most important group lies in the small valleys around Soissons and Laon, and tends to peter out towards Reims and the Marne. From a sample of 135 contiguous churches from this area almost 91 percent were worked on during this century. The next most concentrated group lies along the riverine areas of the Oise and its tributaries, which loses itself in the Vexin and towards Meaux. In a similar sample taken around the Oise, 58 percent contain Early Gothic remains. A third group follows the Seine and its tributaries, fanning out south of Paris to dissipate over the gentle flat lands of the Brie and Gâtinais. Another sample taken within sixteen kilometres of the walls of Paris shows, perhaps surpris­ingly, that around the city only 34 percent of the churches come from this period — again only the average for the whole Basin and a lot less than in these three groups.

11

Petit, Val‑de‑Marne.

32

Besides these three centres, there are ribbons around Châlons, along the valley of the Essonne and its surround­ing plains to the south, and, joining them, scattered groups through Moret and Provins. Between them lie notably va­cant areas around Meaux, Epernay, and the Hurepoix where the numbers never exceed 6 percent of the total. These densities are paralleled by the proportion of cat­egory 1–3 churches to those in category 4. In the environs of Paris, the proportion is 12:10, which comes a poor third to 28:10 near the Oise and a colossal 52:10 in the Sois­sonais.12 When we conjoin these two sets of figures the numerical importance of the north-eastern region in the con­struction of Early Gothic churches is established over all the others. We will now isolate those factors which might explain these anomalies, especially in the four areas of the Parisis, the Oise. Meaux. and the Soissonais. Societal Factors Map II–2 shows there is no perceivable relationship be­tween the distribution of the churches and the ancient counties,13 though their names are still useful. Viollet‑le‑Duc’s dictum that “the French cathedral was born with the monarchical power” suggests there were political reasons for the construction of religious buildings.14 Yet a glance at the political boundaries on Map II–3 shows there is no corre­lation between the Royal Domain and these churches.15 There are many in the north of the king’s territory and few to the south. Similarly, to the east the densest concentration of all straddles the frontiers between the Champagne, Sois­sons, and Reims without extending across the entire ter­ritory of any. Clear injunctions to build would have resulted in visible concentrations of churches within the political boundaries concerned, and this is not the case. More new towns may have been founded in the king’s territories than elsewhere, for there are ten Villeneuves with As most of the rebuilt churches in the Paris area seem to have been of the simpler type, any disproportionate amount of rebuilding would only have accentuated this contrast — Petit, Val-de-Marne. 13 These boundaries have been drawn from maps by Beaurain, Atlas Géographique and Cassini, Carte; and Delisle, Provinces. 14 Viollet‑le‑Duc, Dictionnaire raisonné, 11and 284. 15 Lognon, Atlas histoire; Newman, domaine royal; Schraden, Atlas; and the author’s discussions with Bernard Ancien, Robert Fossier, and Jean Hubert. 12

early gothic churches in the paris basin

33

Map II–2. The ancient counties of France.

Map II–3. The political boundaries between the Royal Domain and the County of Champagne.

34

churches from this period of which nine lie within the Royal Domain, while the tenth, Villeneuve­-la‑Lionne in the Marne, was of Roman origin.16 The Artois, the Poitou, and even Normandy show little sign today of Early Gothic. This was not from lack of wealth, for the riches of Eleanor’s possessions were leg­endary and Normandy brought the Capetians as much in­come as all their other possessions combined.17 If churches were the outward expression of royal authority, why were so few erected in the new territories? The comment by Fawtier,18 that Gothic may have reflected some­ thing of the attitude engendered by the growing royal power, but cannot be said to have been pursued by royal initiative, is therefore more correct than Simson’s that “we must assume that Gothic was considered the expression of ideas with which the crown wished to be associated”.19 The strength of the great seigneurs over their vassals grew as fast in Provins and Meaux, where there are few churches, as in Oulchy and Soissons where there are many. There were powerful seigneurs at Fère‑en‑Tardenois, Braine, Coulchy, and Pierrefonds. Many churches were built around the first two, and almost none in the territories of the second. So firm feudal authority was not in itself a factor. Two of the more important families, well‑favoured by the king, and substantial in their revenues and prospects, were the powerful lords of Montmorency and Mello. Both were in a strong position to reflect the wishes of their master and to fund major building campaigns, and their best can be seen at Taverny and Mello. Yet there are only a few interesting churches in the Montmorency domain, and over forty in the other. Thus, although there may have been local political factors, depending on individual generosity, there are few signs of any long‑term royal or seigneural policy towards the construction of churches. The vacant land around Meaux lies along the boundary between the Capetians and the Counts of Champagne. Yet the boundary itself was more Higounet, Paysages. Though deliberate encouragement of building may have been absent, the many small churches on the plains south of Paris may have had something to do with Louis VII’s positive policy towards rural improvements. 17 In preparing this section, I have consulted Baldwin, Masters, princes; Bournazel, Government; Bur, Conté de Champagne; Evergates, Feudal society; Fawtier, Capetin kings; Hallam, Capetian France; Hubert, Monde antique; Longnon, Documents relatifs; Longon, Vassaux du roi. 18 Fawtier, Capetin kings, 165 19 Simson, Gothic world, 64. 16

early gothic churches in the paris basin

35

a “zone of influence” than a frontier. The Count’s writ was theoretically valid in the Meaux, but was in practice far from exercisable. This va­cant land has been termed a “no man’s land”20 between the two great contestants for power, a remnant of the huge wooded barriers that at one time divided most of the major political powers from one another.21 The many Templar commanderies along the border suggest a peace‑keeping force, if only because there are so few of those other settlers, the Cistercians. Hubert lists eighteen commanderies be­tween Montereau and Meaux (of which six remain) compared to two Cis­tercian. Yet there is no such barrier between the Champagne and the bishops of Soissons and Reims, nor between Picardy and the Ile‑de‑France. Skirmishes and attempts to grab ter­ritory went on all the time, and feelings would have been just as fragile between these lords and bishops as between them and the crown. So we must rule out military con­siderations. except perhaps in the Vexin. The few churches and many castles in the border areas of both Vexins bear witness to the long struggle between the Norman kings and the French. For twenty years the taxes imposed by Phi­lippe‑Auguste to pursue his wars with the English may also have inhibited building in his domain, as a significantly large proportion of lle‑de‑France churches date from after the conquest, and this may be one reason for the Braine area being so much better stocked than the Parisis. There is a closer correspondence between the episcopal boundaries and the vacancies around Meaux, as can be seen in Map II–4,22and between the densest collection of churches and the diocese of Soissons. Yet the group of churches to the east of Meaux belies any consistent policy by its bish­ops, as does the spread of churches beyond the borders of the Soissonais. Similarly, the dense zone around Mello straddles three bishoprics, two of which show little build­ing activity elsewhere. Finally, there are very few category 1–3 buildings around Noyon, nor in the archbishopric of Sens, nor, in spite of the greatest cathedral of the age at Chartres, in the Beauce and the Perche. So we must rule out ecclesiastical policy. Jean Hubert in conversation. Hubert, Frontière occidentale, 14–30. Higounet, Paysages; Higounet, Forêts de I’Europe, 343–98; and Higounet, Forestiéres de l’Europe, 213–18. 22 The boundaries are those of the 17th century, but do not seem to have changed substantially since 1100. They are drawn from maps of Mar­iette, Diocése; Duval‑Arnould, Diocèse de Soissons, 159–266; Vaugonoy, Atlas; Kaiser, Diozese Soissons; no author, Plans des forests, bois, et buissons . . . de l’Isle de France, Brie, Perche, Picardie et Pays Reconquis, Paris, 1668. 20 21

36

Map II–4 . The Episcopal boundaries.

We cannot even attribute these churches to the growing seigneural power of the bishops. For in some bishoprics, like Soissons where the counts had been largely overshad­owed by the bishops, there was lots of building work, while in other bishoprics which may have been more, if not equally independent, like Reims, Laon or Noyon, there was only scattered activity.23 There was work on every cathedral site in the Basin. One third were instigated by fire. Six were being worked on in the 1170s, 13 around 1200, and at least ten were still being built at the end of the period. Though all were under the jurisdiction of bishops associ­ated with the king,24 they were Yet as the church extended its ownership, prosperity may have in­creased. Disputes were reduced, titles were better recognized, and administration was longstanding and well‑regulated. Unlike lands belonging to lay lords which could be transferred for marriage or debts to new owners with new policies, church tenants could look forward to a continuity of effort and to stable taxes which would only encourage prosperity. Further research may permit a more accurate assessment. 24 Pacaut, Louis VII. 23

early gothic churches in the paris basin

37

equally associated with trading cities whose burghers sought the protection of the Crown, and whose trading profits helped to finance these works,25 as well the many important urban churches, abbeys, and collegials of the period. Though the cathedrals had the inestimable advantage of being able to impound funds from the whole diocese, the monies spent on them seem to represent only the tip of the iceberg. As can be seen in Fourquin’s map of the territories owned by the Church in the Paris area,26 there were many more churches built in the domains owned by Saint-Denis and Saint-Germain‑des‑Près than in those owned by the Chapter of Notre-Dame. This suggests that the cathedral may have discouraged building by usurping regional funds for the benefit of its own program. The same may have happened in Reims, where nearly all the churches in the diocese seem to date from either before or after the building of the ca­thedral. The uprising of 1233 may have reflected the harsh way funds were raised for the building works,27 there being only four churches in the rest of the diocese being worked on while the cathedral was building — Villers‑aux‑Noeuds around 1215, and Crugny, Pevy, and Prouilly in the thirties. In contrast, vast building programs were being carried out in the diocese of Soissons and to the south of Laon at the same time as their cathedrals, that further emphasize the richness of this area. Forests and Marsh It is thought that during the Merovingian period the for­ests cleared in Roman times reclaimed most of the country,28 especially the uplands. As the population increased, the forests were cleared once more, starting naturally where the soil was best, and prompting Higounet’s phrase that the geology of the soil determined the geography of the forests.29 By 1000, forests

Kraus, Mortar, sug­gests that the cathedrals expressed the independence of the merchants, and sometimes of the bishops, from local seigneurs, and that the cathedrals were mostly a manifestation of pride and wealth. 26 Fourquin, Compagnes, map IV, is from the end of the Middle Ages. My sen­tence applies after deleting later acquisitions. Kraus, Mortar, 25–26. 27 Reinhardt, Reims, 72–74, and Branner, Historical Aspects, 23–37. 28 Higounet, Forêts de l’Europe, 53. In a more conservative study, it was considered that only one third was forested in the 10th century. Devèze, Forêt française. 29 Higounet, Paysages, 45 and Roblin, Terroir, 263. 25

38

probably covered twice as much land as today, and since then the largest have re­mained substantially untouched, while the smaller have been heavily reduced.30 It must not be thought that every­one wanted to turn forest over to food production, for the woods were an important source of revenue, the king raising one quarter of his income from forests.31 All the ar­able and well‑drained land had been cleared by 1250, though the plateaux were not fully cleared until 1700.32 Unforested areas do not, however, always coincide with quantities of buildings, as to the east of Reims and south of Chartres, nor can the dense forests to the northwest of Soissons be alone responsible for the paucity of churches there. Similarly, the hunt did little to affect construc­tion, since areas like Senlis are surrounded by churches, while others like the Hurepoix and Fontainebleau have few. Marshland, on the other hand, had a direct effect. Cal­culations and records suggest that almost twice as much river frontage was marsh as is found today, examples being the Seine and Yonne upstream from Montereau as far as Sens and Nogent, the Marne around Meaux, and much of the upper Oise and the Vesle west of Reims.33 As Map 5 shows, these are just those riverine areas least serviced by churches, no matter how rich the adjoining farmlands may have been. This shows how essential fertile valleys were to the successful villages of 1200. Population and Trade Dense populations, in spite of the strong internal markets they engendered, were not necessarily in areas of high con­struction. Normandy was 40 percent more densely inhabited than the Ile-de‑France, or the Soissonais.34 The Exceptions can be found in Brunet, Restauration, 65–99; Devèze, Forêt française and Moreau­-Nèlaton, Fère-en-Tardenois, 145. A study of the oldest maps shows that the great hunting parks are substantially as they were in the 1650’s — Ainville collection, Forets de France; Carte archèologique de la France, Orléans, 1948, and other maps cited earlier. Devèze, Histoire, 268, gives today’s forest cover as 18%. 31 Devèze, Histoire, 64, and Fourquin, Part de forêt. 32 Robert Fossier in conversation. Largely unnoted are scrub‑lands which would support neither forest nor agriculture. 33 Prepared from Fourquin, Compagnes, 60; Roblin, Terroir, 254, and an exami­nation of those river flats which today rely on drainage for their useful­ness, based on maps at the Institut Géographique National, Saint-Mande, Paris. 34 Dupaquier, Cartographie , 981. 30

early gothic churches in the paris basin

39

Map II–5. The geology of the Paris Basin with the grossier substratum shaded. The limits of the Lutétian seas are dashed. The outcrops of lucastre are dotted. The boundary between surface limestone and chalk is delineated with a line of crosses.

Plaine de France, north of Paris, was one of the most closely settled regions in the 1328 census, where some of the best wheat was grown. Yet it has few churches.35 On the other hand, sparse populations and lack of churches do go together, as in the Hurepoix, and around Bourges and Troyes — although poor soil may also have been responsible.36 Total population is hard to estimate, though we can count communes or hearths to gain some idea of relative densities. The hearth seems to have represented both the num­bers of people and wealth. It was a sort of fiscal 35

36

Fourquin, Population, 63–64, notes the number of hearths per hectare at nine­teen in La France, ten‑twelve in the Vexin, sixteen along the Seine south of Paris and only six‑nine in the Hurepoix. Also Brunet, Peuplement de la Brie, 161–69. Fourquin, Compagnes; Fossier, Médiévale, 144–­165; and Fossier, Peuplement de la France, 59–99.

40

coefficient.37 Using a factor of 4 to 4.5 people per hearth, the population of France — the smaller France of medieval times — may not have been far from 12 million in 1328, and maybe 9 million in 1200.38 The Paris Basin may have supported a quarter of these people. Whatever the actual population of the Basin may have been, the absolute figure is less important than its growth, for between 1100 and 1348 it is estimated that the popu­lation increased some threefold.39 Yet there is no direct correlation between population growth and prosperity. Even the growth of Paris may not indicate a general prosperity, for the very presence of this huge agglomeration seems to have made nearby towns anaemic.40 As trade was increasingly concentrated at Paris after 1200, smaller towns would have lost some of their earlier importance, as may be seen in the interrupted building campaigns at Lagny and Chennèvieres. There were a few areas of manufacture in the Paris Basin, especially weaving. Along the Thierace, weaving had been practiced since ancient times in an area that extended north through Amiens to Flanders. This area still has many of the most ancient churches, while those nearer the Oise are generally later.41 Save perhaps in the major towns, industry was not therefore a primary influence on construction,42 though suc­cessful manufacture may have encouraged building once it was initiated.43 The fairs held in the four Champagne trading towns of Troyes, Bar‑sur‑Aube, Provins, and Lagny seem to have encouraged some buildings in their immediate vicinity. This is particularly apparent around the last two towns where a few satellite buildings are surrounded by emptiness. It is surprising that these trading areas, like Paris, lie to the south of the most important church‑building areas. Lot, L’etat des paroisses, 289. Fourquin, Population, 1956, 63–67. 39 Duby and Wallon, France rurale, 403. 40 Dupaquier, Cartographie, 991: in the 18th century, towns like Melun and Etampes had a quarter of the number of hearths of Chartres or Beauvais, which were themselves a lot smaller than cities farther away from Paris like Rouen and Orleans. The increasingly visible tendency of Paris to monopolize all secondary and tertiary activity in the Ile‑de‑France would have been present in the 13th century, though impossible to quantify. 41 Feucheis, Somme et Seine, 1–37. 42 Kraus, Mortar; Lopez; Commercial Revolution and White, Medieval Technology. 43 Scale is important. Troyes is called a weaving city, yet its 2,000 lots annually cannot be compared to 50,000 from Ghent. The whole of the Troyes production could have come from 40 looms. 37 38

early gothic churches in the paris basin

41

Roads, Rivers, and Bridges The ballasted Roman roads appear to have had little maintenance before the eleventh century, and over long stretches they may have become unusable and unable to play any part in medieval trade. Dead‑ended paths may have been frequent, but total isolation would have been rare indeed. Bridle tracks linked every settlement, no mat­ter how small,44 for the collection of taxes and administra­tion, and well‑worn wagon tracks joined every hamlet or farm house to the outlying fields.45 I find that the maximum distance from home to field was two kilometres, both for safety and lunch. Most roads would have been dirt,46 and as in the country today the greater the traffic the worse the wear. Since travellers would have tended to select the better paths, the roads were being constantly modified,47 often enough to bypass one town in favour of another.48 Brigandage may also have been a factor. Local prosperity could have depended on the va­garies of these roads. Churches may have been begun dur­ing the days of prosperity, never to be completed once the road had been washed out or shifted. A walking pace was probably the mean for the transport of goods, though the distance covered by horse varied from forty‑five kilometres a day between Paris and Lille to eighty between Paris and Tours. The average was some fifty‑six kilometres a day.49 It would have taken no more than three days to cross from one extremity of the Paris Basin to the other. Masons and carpenters could have moved rapidly from their homes to any site. But the movement of ma­terials would have been more difficult, though improved breeding enabled a high proportion of trade to be carried by pack animals, especially mules, which may have reduced dependence on the cart and the necessity for better roads.50 Smaller churches with limited resources seem to have ob­tained the bulk of their stone from local quarries, but stone for the bigger jobs often came

This can be seen in maps such as Cassini, Carte. Also Es­tienne, Guide; Hubert, Routes de France, 25–56; Mesqui, Champagne. 45 Roblin, Terroir, 247. 46 Forbes, Land, and Roblin, Terroir. 47 Bautier, Recherches, 99–143; Imbardis, Médiévales, 95–98. 48 Imbardis, Médiévales, 97; Lopez, Commercial Revolution. 49 Bautier, Recherches, 112. 50 Lopez, Commercial Revolution 80. 44

42

great distances, and Gimpel suggested that 20 kilometres by road would have doubled the cost of the stone.51 Saint-Quentin is reputedly built from the quarries of Saint-Leu‑d’Esserent on the Oise.52 Some of the stone for the Royal Portal sculpture at Chartres was shipped some three hundred kilometres from the Saint-Maximum quarries;53 the stone for Bourges was trans­ported from quarries eight, seventeen and thirty‑eight kilo­metres distant.54 Though the cost of land transport was high, work on these giant buildings does not seem to have been noticeably discouraged by this financial burden. And if a road was needed, the most skilled men were on hand to construct it. So we can only say that the roads them­selves were no limit to building. The larger rivers formed the major transport links, es­pecially for stone and wine. Traffic by water was so cheap that people preferred to use the best water‑borne stone from the Oise and the Aisne than material from land‑locked quarries a tenth of the distance away. 55 The importance of these rivers can be seen in the numbers of interesting churches along the navigable stretches of the Seine, and considering the relative paucity of churches in the Parisis (only 32 percent) one wonders how many of these would have been built if the Seine could not have been used for the cheap transport of stone. In practice a river was just a different sort of road, and economically, a road may have merely been the means of getting from one river system to another. Few of the smaller rivers were navigable, and were too easily restricted with low country bridges and, though windmills were being installed after the mid‑12th century, most mills were water‑driven, making transport up the smaller rivers difficult if not impossible.56 Major rivers may not have limited the supply of materials, for there were at least eighty bridges in the Paris Basin before 1300, and where there were

Gimpel, Medieval Machine, 61. Bondeau and Viré in conversation. 53 James, Contractors, 83–84. 54 Blanc, Debrand‑Passard, Lebouteux, Matériaux. 55 Ellenberger, Viré, Marvy, Anciennes; French researchers suggest that the stone for Notre Dame in Paris came from the open‑cut quarries of Saint-Leu‑d’Esserent rather than from the underground workings of Arcueil and Gentilly which produced a similar stone only ten kilometres away. Gim­pel, Medieval machine, 60, contradicts this. 56 Bautier, Moulins hydrauliques, 567–­626; Roblin, Terroir, 246; White, Medieval technology, 84. However, the benefits from the mills in metalworking, fulling, and sawing (of both wood and stone) may have offset the higher cost of transport. Duby, “Révolution”, 359– 65; White, Medieval technology. 51 52

early gothic churches in the paris basin

43

none there were often ferries and fords.57 In spite of the rich valleys and agriculture around Paris, the wealth of the capital, it is said, came from its rivers rather than from its land.58 The distribution of the better churches in Map 1 shows that for the west and south (though not in the Sois­sonais), nearly every interesting building lies a short dis­tance from one of the major navigable rivers. Agriculture There is one factor that applies to nearly all the buildings of our period: they lie along the river valleys and not on the plateaux. This is especially true in the north where the valley walls are steeper than in the more gently rolling south. The advantages of the valleys are: • the lighter soil, largely of fertile limon eroded from the uplands, mixed with sands and constantly washed with leachings from above, while the limon on the plateaux is less porous and heavier to work,59 and is also liable to desiccation when dry;60 • the excellent natural drainage, unlike the uplands;61 • the noticeably warmer temperatures, and protection from the harsh winds; • the constantly flowing water for mills and fishponds;62 • the people seem to have been healthier and the popu­lation was greater;63 • rheumatism may have been common on the damper higher ground; • the small capital required for agriculture as the lighter soils could be worked by wooden scratch ploughs drawn by a single animal.64 The Boyer, French Bridges; Forbes, “Land Transport”. Fourquin, Compagnes, 62. 59 Dupuis, Explicative; and discussions at INRA. 60 Roblin, Terroir, 260. 61 There was little investment in drainage before the 16th century, though the techniques were known from Roman times. Fourquin, Compagnes, 75. 62 Roblin, Terroir, 246. Jean Jacquart and François Bucher in discussion. 63 Dupaquier, Cartographie, 987. 64 Horses were in common use in northern France by the 13th century — White, Medieval technology, 64–65. They greatly improved the yield and the amount of work a man could achieve in a day — Forbes, Technology, 83; White, Medieval technology, 68, suggests that 57 58

44

heavier plateau soils required more expensive equipment and more animals to draw them.65 Nevertheless, the fertile limons were extensively culti­vated and brought great riches to their proprietors,66 but during this century the plateau farm­ers seldom built churches that have lasted. Even on the best soils to the north of Paris, there are few churches. Because these plateau soils were heavy, farmers had to use the expensive wheeled mouldboard plough, or charrue, and large teams of horses, for which they reor­ganized the field boundaries. Aerial photography has shown that valley fields were one‑two acres in extent, while those on the plateau were expanded to five acres, with some up to 40 acres.67 Only large land owners could do this, so men were employed for a wage. By contrast, in the valleys the smallholders and vignerons could own their own tools and keep some of the profits from their labour.68 As the plateau yields were so much higher than those of the valley,69 one looks for some social force to provide an explanation for the higher incidence of churches in the valleys. It may be that where farmers could avoid having to work for a wage and were able to keep a part of their produce, enough funds were generated locally to pay for new churches. The best limons lie to the north and west of the Marne, the worst to the south and east,70 and it is significant that the area without churches lies the horse allowed people to live farther from their work, permitting towns to be enlarged. Costs for horses have been variously estimated as being as much as 4 times that of oxen — Duby and Wal­lon, France rural, 413, and Lopez, Commercial Revolution, 46. 65 Brunet, Structure agraire; Duby and Wallon, France rurale; Duby, “Révolution” and Me­dieval Agriculture; Fourquin, Population, 69 et seq.; White, Medieval technology. 66 Bourde, Agronomie; Fourquin, Compagnes, 8. 67 Duby and Wallon, France rural, 409; Fourquin, Compagnes 1964, 73–75; Higounet, Paysages, 75; White, Medieval technology, 47. 68 Duby and Wallon, France rural, 410; Higounet, Paysages, 75. 69 By 1300 yields had increased from two and a half grains for one in Carolingian times to three and even four, to be exceeded in the Plaine de France by as much as seven for one. Duby and Wallon, France rural, 462. 70 No two maps agree on the extent of the limon but, excluding a crescent through Troyes and Reims, it covers most of the higher ground over both the chalk and the limestone. South of the A4 and east of Fontainebleau it is impermeable, and it is more or less acidic between the Marne and the Morin. The maps all agree on the poorer soils of the Oise forests, the Noyon area, the Hurepoix, and the Brie. Ministère de l’Industrie, Cartes Pédologiques de la France, Paris, 1967, and the Centre National de Recherches Agronomiques, INRA, Versailles.

early gothic churches in the paris basin

45

astride both types of soil.71 Similarly, the better limons extend across the densest area of churches in the Soissonais and across the thinnest around Beauvais and Chartres.72 So here soil is not the major factor. The churches along the Essonne may owe their existence to the profits from agriculture supplemented by weaving and good quarries.73 Conversely, the poor dry lands of the Aube and Epernay supported few people and even fewer churches. Much of southern Picardy is similar, with a syl­van landscape with many small forests and little intensive agriculture.74 In the Hurepoix and around Fontainebleau the soils are poor, sandy, damp, and lacking in limon. Their populations have always been small. They remained cov­ered in forest until the building of royal palaces and the expansion of the metropolis. If the people in the Brie and to the north and east of Paris built churches, they must have been very simple for most were rebuilt after 1600 as the land was drained.75 There are almost no fourth category churches around Meaux today, suggesting that any built around 1200 were of wood, plaster, and rubble. No wooden ones remain, though there are over 20 in the forests east of Reims. This suggests that if of wood, then they were insubstantial. To the west and north of Soissons the soil is good, and there are both forests and churches, albeit simpler ones. However, the valleys are narrow and changes in climate could have had a profound effect on the viability of the region.76 The same may apply to the region southwest of the Hurepoix in the Perche where there are few churches after 1200, but many built before. Thus, one can say that the soil in itself appears to be only a contributory factor, for it cannot explain the dearth of churches in the areas where the limon is extremely fertile, nor the absence of churches around Meaux where there is both good and bad soil. Nor does it explain the general density within the Paris Basin, and the small numbers to be found beyond that. Certainly, we can conclude that within the Basin the churches are the product of the valleys, and of valley agriculture rather than of large‑scale plateau farming. Fourquin, Compagnes, 60; Roblin, Terroir, 270. Chèdeville, Campagnes. 73 In conversation Cambeillare suggested that these farmers combined grazing with cropping, using the excrement of the one and the straw from the other to produce a compost that greatly enriched the soil. Were these techniques used only here? 74 Roblin, Terroir, 275. 75 Roblin, Gallo‑Romaines, 234; Duby and Wallon, France rural, 400; and conversation with Jean Jacquart. 76 Roblin, Terroir, 273. 71 72

46

We can also see that poor soil and church building do not go together. If on one map we set out all the deleterious influences — marsh, forest, acidic soils, damp, and so on — we will still not have explained those positive forces which produced the dense concentrations of churches in the Oise and the Soissonais. However, there is one factor that does coincide precisely with these churches and that is stone. Limestone and Quarries An island of tertiary limestone called calcaire covers much of the Paris Basin, lying within a sea of secondary chalk stretching from Dunkirk to Nevers and Le Mans to Verdun — some 450 kilometres in each direction.77 Most of the chalk is soft, white, and fine‑textured. Like all sedimentary stones, it begins to harden after a few days exposure to the air, until it becomes more or less imperme­able. On the whole, it is a softer stone than calcaire, easily worked and sometimes weathering badly. The cathedrals of Beauvais, Amiens, and Le Mans are all built of it, as well as most of the churches in the chalk country. The calcaire is harder than chalk, of varying shades of yellow, with an even perceptible grain, and usually containing fossils. Masons consider the calcaire, be it grossier or vergelé, the finest for building and sculpture. It is an exquisite stone to work, capable of taking a razor‑like edge and holding it under the chisel, while hardening evenly so that the best will retain its sharpness and detail after cen­turies of weathering. The boundary of the Early Gothic churches coincides al­most exactly with that of the calcaire, as shown in Map II–5,78 while the densest concentration of buildings along the Oise and the Soissonais lies over the grossier. Grossier is the stone of Gothic. One is not conceivable without the other. It is the conjunction of grossier with the valleys which provides the best correlation so far. The best layers of grossier were laid down while a tongue of the Lutétien or Nordic sea covered part of northwest France.79 Its southern boundary coincides with the escarpments of the Marne, and continues under the Blondeau, Lutétien; Morellet, Bartonien; Pomerol, Guides. Major source Blondeau, Lutétien, 398. 79 The grossier was the product of the lower Lutétien. The various layers of limestone of Provins, Saint-Ouen, Champigny, Brie, and Beauce were laid down in later seas. 77 78

early gothic churches in the paris basin

47

sedimentary rocks of Paris, being exposed in an outcrop to the south of the City. Above and to the south of the grossier, the geologically later beds vary considerably, and though none come close to the quality of the vergelé, some make good building stone, if not the best capitals or details. The beds of lucastre around Provins, which reappear near Bourges and south of Chartres, also continue westwards with usable outcrops near Château‑Landon and Champigny. Building stone was quarried from all of these beds during the Middle Ages, although it is harder to work than grossier and not as capable of holding details. The beds of Château‑Landon and Champigny are exceptional portions of a large mass of travertin or bartonien superieur. This mass lies between the Marne and the Seine, and between Epernay and the eastern outskirts of Paris. It is exposed and accessible from nearly every valley, but in varied quality, and degraded where near the surface. Above this, and covering most of the Valois plateau in­cluding Meaux, lies the Saint-Ouen stone, which is totally unusable. Above the Champigny lies the sandier calcaire of the Brie, and this merges into the harder layers of the Beauce, which are also too obdurate to use, save for paving. Except for an important outcrop of grossier to the south of Paris, and of Champigny travertin to the east, the Seine­-Marne valley between Poissy and Meaux consists of sed­imentary sands and clays. Along this line the bedrock has sunk to form a trough, burying the calcaire under some thirty metres of later material. The absence of accessible building stone may explain why there are relatively few churches to the northeast and southwest of the metropolis. A band of gypsum ludien lies to the north of the Seine which makes a waterproof plaster. It was seldom used where buildings could be built of grossier. But where there was little good stone, plaster was the best finish available. The surviving buildings show that no templates were required and little design skill. Its presence may, paradoxically, have retarded the development of those ideas and skills which produced Gothic: What the potato was to Irish culture, plas­ter may have been to the medieval architect. The stone was sometimes quarried in open cut, but as the most accessible material was worked out, the quarry­men began tunnelling into the hillsides to avoid removing tons of overburden.80 These subterranean workings are 80

Ellenberger, Viré, Marvy, Anciennes; Roblin, Terroir, 248, and in discussion with Anne Blanc, Tourtebatte, Benoit, Chapelot, and Viré. Sometimes stone was accessed through vertical shafts, as at Chamigny, Laversines, and Saint-Remi.

48

Fig. II–1. Section through a typical valley, showing the steep escarpment between it and the plateau, the beds of grossier, and the usual relationship between the quarry and the nearby village church.

extraordinarily extensive, and were dug into the mountains as far as moisture and poor circulation of air would permit.81 The level ceilings over these tunnels were first supported on pillars of unexcavated rock left between the workings, but this was wasteful and later centuries learned to extract all the rock and to build pillars from the inferior stone as it was extracted.82 These caverns are still used for storage. Across much of the Paris Basin the rivers have been able to work their way through the strata to form deep and irregular valleys, eroded to a level de­termined more by the estuary level than any intrinsic bar­rier of nature. Along the sides of the escarpments flanking the valleys, the bands of calcaire have been exposed to view. Throughout this country nearly every church lies just beneath the level of the exposed grossier, and within a hundred metres or so from where it could have been mined on the escarpments, Fig. II–1. The same stone used in the church is to be found in the adjoining villages, suggesting that the location of the villages may have been influenced by the easy availability of stone. Deliv­ery would often have been downhill by trolley or chute, showing that the lack of roads was not a serious barrier to construction. We do not know whether these buildings were built from the nearest quarry, but in the grossier region almost any bank could have been used, even if small quantities of some better material had to be imported for hard‑wearing sur­faces, capitals, or sculpture. 81 82

Gimpel, Medieval Machine, ch .2; and White, Medieval technology. Noel, Département; Noel, Pierre; Pomerol, Trance; Tourtebatte, Réflexions.

early gothic churches in the paris basin

49

Map II–6. Ashlar churches shown as black dots, and rubble as circles. No category 4 churches have been included.

This phenomenon is universal, and any distance from stone yard to site dampened the urge to build. It seems to be as clear and fundamental as that. The exceptions are usually only the larger churches and most of the cathedrals. Their size shows they could afford the transport. The large central plateau around Meaux without churches lies in the middle of the grossier, but is here overlaid with more than ten meters of hard St. Ouen and Brie, strata that are so dense that the rivers were unable to erode their way through the surface. They have left a softly undulating countryside under which the grossier has remained unseen and inaccessible. Its inaccessibility, in conjunction with the lack of valleys and peasant agriculture, seems to have stopped nearly all building. South of the Marne, the bartonien is visible along many of the escarpments, but a large proportion of the ribs, caps, and the like, seem to have been cut from grossier which often had to be carried great distances. The consequences can be most clearly seen in the distribution of churches built of ashlar and rubble, Map II–6.

50

A rubble building, coursed or random, is less expensive than one in ashlar, no matter how close the quarry, for the stone only has to be taken out of the ground, placed as it comes, and the gaps filled with mortar. Ashlar on the other hand has to be selected, squared, the face cut back until it is even and true, and then trimmed in size to suit coursing heights and details. Everything has to be integrated so that coursing in the walling matches that in the jambs and buttresses. Further, one has to cope with the temptation to embroider by carving decorative niceties around openings and along projections and terminations. I would estimate that the cost of labour alone in ashlar could be a third more than rubble. There are three concentrations of ashlar: along the valleys south of Laon, along those of the Oise and its tributaries, and the riverine region of the Parisis. Elsewhere there are almost no ashlar churches at all. There is also some correspondence between these three areas and the proportion of villages with churches from this period discussed in the first part of this article: 91 percent around Braine and 58 percent near Senlis, compared to 34 percent near Paris, 24 percent around Provins, and 60 percent in the Meaux area. If we take the number of churches per village, and mul­tiply the category 1 to 3 buildings by an appropriate factor for the higher cost of ashlar over rubble, we get very crude figures for relative expenditure. The Parisis villages would have spent six times as much as those from Meaux, while those around Senlis and Soissons spent thirteen and twenty­-two times respectively. It shows that the Ile‑de‑France comes in a poor third, with the villages around Braine spending nearly four times as much as those around Paris.83 Approximate as these figures are, the trend is clear: there is something about the areas with the best stone, and the quarries they supported, that encouraged people in the en­thusiasm so essential to any prolonged building boom, and gave them the wherewithal to accomplish their dreams. Also, there may have been something about the stone itself that appealed to people of 1200, as exposed concrete appeals to us and cast iron did to the nineteenth century, for a large proportion of later buildings built over the grossier do not use it, but are built of rubble and plaster instead. Perhaps they were excited by the appearance and feel of smooth firm ashlar, with its flush, fortress‑like im­penetrability. If so, then the grossier was the only available material that could have satisfied that enthusiasm.

83

“XL: Funding”.

early gothic churches in the paris basin

51

Yet we have not explained why the best stone should be associated with such an enormous preponderance of high­ quality churches, especially in the Soissonais where, as already noted, the proportion of better churches to those in category four is 12:10 for the Basin as a whole, but 52:10 around Soissons. Nor have we explained why the boundary around these churches should coincide so clearly with that between the calcaire heartlands and the surrounding chalk. It is not as if this boundary has always marked the di­vision between church building and its absence. These two entities, geology and geography, have existed since time immemorial, and the chalk lands are full of churches from other periods, both earlier and later, including great cathedrals. In some regions, the chalk may have been less depend­able, but the existence of churches in every town you pass through makes it unlikely that any properties intrinsic to the stone itself may have discouraged building. Yet the fact remains that around 1200 there is nothing anywhere in France comparable to the passion for building in grossiere limestone. I have a feeling that the answer, when we find it, will have more to do with feelings than economics. Possible Answers Though the number of prospective answers to these questions are many, there are four that may have combined to produce the economic surplus necessary for construction. The first is the much‑admired stone that may have been exported. Profits from trans­port may also have been a factor. The numerous gifts of­fering free access to quarries suggest that the income from them was far from negligible. A closer study of medieval mining techniques may dis­close something of the relative costs of open cut, tunnelling, and transport, for the scale of exploitation may have made the Oise quarries so efficient that in spite of cartage costs the grossier could have been marketed in competition to inferior stone produced elsewhere. Secondly, it is a little difficult to visualize how the gros­sier could have been exported across such a large region without drawing with it many of the skilled men trained in working it. I have prepared maps of the jobs built by various contractors that show de­cided concentrations, which suggest

52

where they had their headquarters.84 Most of those who worked beyond their locality came from the region of the grossier,85 in­cluding eight of the nine masters at Chartres.86 Analysis may indicate whether they were attached to specific quarries, which were the larger contractors, and whether their teams were much larger than those of the small builders. This may in turn help to show whether such wide‑ranging work honed their skills and expertise, and whether one master could control a number of scattered workshops at one time while still working efficiently in his own locality. There is also the vexed question of the impact church building had on the rest of the economy.87 The income from the sale of stone and a skilled work force may to some extent explain the density of the riverine churches whose navigable waterways provided cheap transport, but it will not explain why the densest con­centration of churches to the northeast is scattered almost uniformly across a countryside without navigable rivers. For that, one suspects some third source of cash, and wine springs immediately to mind. Wine‑making was particularly im­portant in the valleys around Soissons and Laon,88 and being labour‑intensive, it supported a large popula­tion. The wine‑growing areas of the Soisson­ais, the Oise and southeast of Paris had twice the population density as non‑wine areas.89 Later figures show that the density was 25 hearths/hectare where vineyards occupied more than ten percent of the terrain, compared to ten hearths elsewhere.90 This part of France was famous for the high quality of its wines. Dion quotes poems celebrating the competitions held between the various areas to As with master M1, James, Template-makers, ch. 8. The major contractors can be found working as far away from their Valois‑Soissonais home bases as Blois, Bonneval, Pacy, Le Mans, Amiens, St. Quentin, Mouzon, St. Michel‑en‑Thierace, and Troyes — distances of 150–200 kilometres. The well‑known masters of the 15th century did the same — Parler, Ensingen and Stethairner, Wynford, Yevele, etc. 86 Maps in “IV: Chartres was lucky”. In the Contractors the master I called ‘Green’ seems to have been local, ‘Scarlet’ and ‘Bronze’ did most of their work in the triangle between Soissons, Laon, and Cohan, while ‘Olive’, ‘Jade,’ ‘Red,’ and perhaps ‘Cobalt’ concentrated along the Marne and near Reims, and ‘Ruby’ and ‘Rose’ along the Oise. 87 Johnson, Cathedral building, 121‑210. The discussion was continued in the 1969 issues of this journal by B. Alford and M. Smith with a rejoinder by Johnson. “39 How many built the churches”. 88 Dion, Histoire; Doehaerd, Laon, 145‑165; Petit‑Du­taillis, Communes. 89 Dupaquier, Cartographie, 985. 90 Fourquin, Compagnes, 63‑70. 84 85

early gothic churches in the paris basin

53

Fig. II–2 . The changes in temperature, shown dashed, and the rainfall over the centuries during the summer months and for the rest of the year. Adapted from Lamb, 1965

determine which was the best, finally adjudicating in favour of those from the Seine as being better than those from the Aisne that flowed through the north-east.91 Yet little attempt has been made to quantify the export and the possible incomes from wine, which may have been prodigious. Later figures suggest that the quantity of wine being transported through Paris was double the per capita production today, between ten and twenty million hectolitres passing through Paris each year. This is about twice as much as is produced, drunk, and exported per capita in France today. Three quarters of all fluvial trade was wine.92 These areas grow little wine today, however, posing a fourth line of investigation, that of climate. It is now firmly established that around 1200 the average summer temperature was markedly higher than today, and that a mini‑ice age began near the middle of the thirteenth cen­tury, Fig. II–2.93 Dion, Histoire, 83–92. Favier, Commerce; Pirenne, Vins de France, 225‑43. Early 14th‑century exports from Bordeaux to England had declined to only 0.7 million hectolitres. Duby, Medieval Agriculture, 45. 93 Lamb, Epoch, 13‑15; Lamb, Climate; Ladurie, L’an mille; Parry, Climatic change; Utterstrom, Climatic Fluctuations, 3‑47. Fig. II–2 is adapted from Lamb. James, Contractors, 11 and 554 n. 53; Alexandre, Regions voisines. 91 92

54

These changes reduced frost and encouraged growth in the Basin, and lengthened the grow­ing season. This is suggested in sculpture at Chartres where Labours of the Months place sowing in Scorpio, a little earlier than today, and the corn harvest which we commence in August is shown in July.94 While the temperature steadily rose during the twelfth century, so did expenditure on churches. With it came a corresponding decline in summer precipitation, which may have suited grape production in some areas more than others. But these favourable conditions did not continue beyond the end of the century. After some very wet years during the 1190s there was a prolonged dry spell lasting almost two generations, partly in drought, which could have affected all agriculture.95 The decline in construction after 124096 coincided with the increase in cold weather to the 1260s, a steadily worsening situation that was followed by deteriorating economic conditions “when Europe was swept by record storms, blizzards and cold weather, far exceeding anything which has occurred since . . . In the Alps, glaciers closed permanently passes which during the previous mild centuries had been routes of commerce”.97 The affect on agriculture, the source of people’s wealth, can be imagined. Recent evidence has attempted to link the profound climate changes over the twelfth to fourteenth centuries to ecological changes. Thomas van Hoof and his colleagues have studied pollen grains and leaf remains collected from lake-bed sediments in the southeast Netherlands, monitoring the quantity of cereal pollen (like buckwheat) and tree pollen (like birch and oak).98 From Fourquin, Compagnes, 80; Web­ster, Medieval art. In a letter Dr. Justin Schove suggests that French summers were par­ticularly favourable to wine production after the mid‑1150’s. In detail he suggests this table: 1113–1125 Often poor conditions 1129–1137 Dry, often warm 1138–1145 Often wet, cold 1151–1156 Often wet, cold 1157–1189 Often very warm and dry 1192–1201 Very wet 1205–1218 Very dry 1232–38 Dry except for 1233 1241–1253 Very dry 1254–1262 Wet 96 “XL: Funding”. 97 Willett, Meteorology, 49. 98 Van Hoof, “Forest re-growth”. 94 95

early gothic churches in the paris basin

55

this they have estimated changes in land-use between 1000 and 1500, and suggested that the increase in cereal pollen reflected farmland expansion. This would have been achieved through clearing large areas of forest and bushland, which released enormous quantities of carbon dioxide. As today, we know that this would have an upward impact on temperature. They then found that after 1350 there was a sharp decrease in pollen along with a lowering of temperature. This occurred after the decimation of the population with the plague, and with it a serious decline in agriculture and concomitant increase in bushland. They conjectured that this led to the mini ice-age of the next century. Another study correlated these dates with a ten percent deceleration of the Gulf Stream.99 Though many are sceptical about the interpretations of these results, research is confirming the intimate relationship that exists between human activity and agricultural ecosystems. More than any other factor, climate change may have caused the serious decline in wine production, and thus income and the construction industry that depended on it.100 When in earlier decades vineyards had been a major investor’s choice in which merchants and others with excess funds could acquire valuable assets, there is a distinct decline in the number of purchases of vineyards after 1220.101 The pronounced reduction in construction in the north-east after the 90s may have been directly attributable to the drought, probably influenced by global warming. The impact of drought may have been mitigated in Paris and along the Seine as royal conquests opened riverine traffic for Parisian goods through Rouen after 1205. By 1240 the decline in church building was evident everywhere, with climate changes being a primary factor. However, money from wine production will not be the only factor in the booming times of the 80s and 90s, for it does not explain the number of churches along the Essonne, around Oulchy and in the valleys of the Marne and nearby Surmelin where vineyards were not particularly plentiful, nor those scattered but important building areas to the south of Paris. Their funding may have had something to do with other aspects of agriculture or idiosyncrasies of patronage.

Lund, “Gulf Stream”. Lopez, Commercial Revolution, 163. 101 Duval­-Arnould, Vignoble, 207–236. 99

100

III Memes and Assumptions*

I

N art-historical studies there are two levels of analysis: what we may call an ‘upper’ level concerned with those broad issues of style and built form that have interested most art historians in this century, and a ‘lower’ level concerned with a large quantity of detailed and minuscule data about the day by day process of construction and the full range of decision-making that went into it. It forms one basis for upper-level analysis. Lower-level analysis is like an examination of the stresses in the frame of a violin. The knowledge we gain in no way affects our appreciation of the music played on it, but it does add to our understanding of both the instrument and the maker. Each level is a semiautonomous subsystem augmenting the way we know a violin — or a church. To proceed with an analysis at one level we actually put aside concepts inherited from any other level, just as theories about harmony will not be relevant to tests of material stress. Generally, the researcher believes he proceeds inexorably from one well-established fact to the next, never being influenced by any unproven conjectures. Nevertheless, the truth is that much of medieval art history is based on unverified assumptions that are quite tenaciously held. They have been taught and accepted for so long they have become the belief systems of the profession, so that even eminent historians have foundered. A useful way to conceptualise the nature of a priori assumptions is through Richard Dawkins’ concept of the meme, which he defines as a “gene-of-memory.” 1 A meme is like a catchy tune, a new fashion in clothes

* Extract from The Template-makers of the Paris Basin, Leura, 1989 1 Dawkins, Selfish gene, 143.

memes and assumptions

57

or a way of building an arch. When an architect hears about a good idea, he passes it on to his colleagues and students, and if it catches on it will, like a gene, propagate itself by spreading from brain to brain. This process can, in the broad sense, be called imitation. Yet memes are living structures, for “when you plant a fertile meme in my mind, you literally parasitise my brain, turning it into a vehicle for the meme’s propagation in just the way that a virus may parasitise the genetic mechanism of a host cell.” The consequences may be as physical as the impact of a gene on a living organism, for the ideas in the brain realize themselves physically. They form structures in the nervous system from where they are manifested in the world around us. Selection may favour those memes which exploit the cultural environment to their own advantage and, like a gene, the meme is the only thing we leave behind us when we die. In short, the meme is any idea which has ceased to be open to reflection and has been accepted to the point that it unconsciously directs our thoughts and actions. It is therefore invasive and concealed, parasitic and often, limiting. Some art-historical processes are based on sets of conventions that we might more properly call assumptions and memes. True, they have evolved over the past hundred years to provide an intellectual framework to order the often confusing, if not conflicting, evidence it deals with. These assumptions have been built into a logical, coherent and self-validating system, but like all mental constructs they tend to limit observation. Some of its proponents, like Lefèvre-Pontalis, hold such an exalted reputation that their conclusions have been questioned only on a piecemeal basis. No doubt we need coherent mental sets in order to observe in the first place, for the eye is not a camera to see everything at equal value, but a selective instrument. If programming is essential to learning, then deprogramming is fundamental to any growth in understanding. Art-historical assumptions stem in part from the way that stylistic analysis seldom concerns itself with the construction process. The first is the long-held belief that the inventions which made the Gothic revolution came from the great workshops around the larger cathedrals, whence they spread to other places. This might be called the “from-the-greater-to-the-smaller” assumption, and is a meme because it has formed part of the belief-system of more than a generation of scholars and for which there is insufficient evidence. For example, it was presumed that the abbey of Braine could not have been designed until the scheme for Chartres had been worked out: Longpont

58

had to come after the nearby cathedral in Soissons, and Soissons cathedral could not have been even begun until the footings at Chartres had been laid.2 The belief-system is summed up in Branner’s claim for Reims that the ideas generated there were “passed on to a large number of new shops that it formed and colonized,”3 or Mâle’s “once the work at Saint-Denis was finished . . . the workshop gathered together by Suger was moved to Chartres.” 4 Three further assumptions have flowed from this. First, that the masters in charge of the major workshops remained in one location for decades and that only the apprentices or minor craftsmen moved. This implied, second, that ideas were carried to the smaller workshops by lesser masters who had been influenced by one of the greater from a major shop, rather than being carried by the genius himself working on buildings of all sizes and taking his ideas with him. And finally, the major fallacy, that because Parisian workshops “established the mode that others followed” in the late thirteenth century, they had done so from the time of Abbot Suger.5 It may be hard to believe that these assumptions, when stated so baldly, could play any significant role in our scientific age, but close reading will show that nearly every writer has paid court to one or more of these concepts at some time. A second venerable meme is that the whole building as we see it today was conceived at the footings, as if a complete set of construction drawings were being followed during all the many decades needed for the work. Thus at Soissons it has been assumed that the present elevations of the south transept and the choir were determined when their respective footings were laid, from which it was argued that as the choir was eight meters higher than the south transept, the latter must have been completely finished before the choir was begun.6 A third is that chronology within one building, or even between buildings, is often established by noting that one thing is more stylistically advanced or retarded than another, as in most of the disagreements on whether the

Consistently promulgated by Lefèvre-Pontalis in his many publications, well-illustrated in his Les façades. His dates are now being picked off one by one, but his methodology has not yet been questioned, save obliquely: Branner, Saint-Louis, 15. 3 Branner, Saint-Louis, 28. 4 Mâle, Religious art, 380. 5 Branner, Saint-Louis, 2. 6 James, Template-makers, ch. 7. 2

memes and assumptions

59

choir or the nave of Chartres Cathedral had been built first.7 Related to this lies the ideal of progress, in which every stylistic concept is placed in an ordered sequence with a steady and, one fears, almost implacable order, as if the medieval artist was guided by destiny itself. This is the “logical bond which links together all individual trends . . . the forward march of a singularly progressive architecture, which steadily advances from generation to generation.”8 From this, the entire range of creative endeavour which does not conform to such ‘progress’ may be dubbed ‘regional’ or ‘archaic’ and relegated to some back chapter of history. Regionalism is not synonymous with backwardness; it is just different. With hindsight, we relegate these buildings to limbo which do not fit into our definitions of historic development. These definitions follow the threads of artistic evolution and are needed as teaching tools, for without them much human endeavour would be awfully hard to pass on. But our view of their part in historical progress need not match any contemporary assessment. Just because the pointed arch seldom appears in eleventh century buildings but is found in most fourteenth century churches, does not mean that around 1200 people felt the pointed arch was more advanced or progressive than the round. Indeed, more vaulting ribs were set to a pointed form around 1150 than around 1200.9 It says nothing about the inherent qualities of, say, African or Japanese art that neither was greatly appreciated in Europe before 1850, but that moment in history when they came to be loved says much about the Europeans. A fourth, if disappearing, meme is that bishops or other members of the clergy were responsible for innovation, a concept stimulated both by the medieval habit of stating that some bishop ‘constructed’ or ‘caused to be built’ something, and by Abbot Suger’s sophisticated involvement in the construction of his own abbey. From this, the travels of the clergy and patterns of ownership have received more prominence in the search for the source of ideas than the builders. One might feel that the repetition of these matters is no longer necessary, and that serious art historians are aware of these traps and have been avoiding them for some time. Unhappily, this is not the case, not when this was first

James, Contractors, 23–24. Bony, Hurlimann and Meyer, French cathedrals, 8. 9 James, Template-makers, ch 8. 7 8

60

written in 1989 nor today. It is still instructive to recognise the presence of assumptions, and to see how continue to influence professional opinion. Putting these assumptions aside threatens to unravel the hundred-year foundation of Gothic history. The analysis of Stoddard, Focillon, Branner, Mâle etc may have to lose some of the orderly simplicity which has made teaching so easy. The emerging picture is more complex, but also more interesting. No appreciation need be lost. The first step in this project is to complete the process already begun to revise the dating. Many scholars have been working on this since the War, including Madeline Caviness, William Clark, Anne Prache, Dominique Vermand, Malcolm Thurlby and Danielle Johnson. Below I will discuss some of the consequences that will emerge through the articles in this book. The most important technical innovations — structural rib vaults, flying buttresses, tracery, massive lifting gear and so on — have until recently been dated to coincide with their first use in a Gothic-like setting. For example, the first flyers were located in the Paris Notre-Dame nave in the 1180s just in time to support the tallest interior spaces yet built. However, a few years ago I showed that the first flyers were created forty years earlier than this, from which it follows that tall buildings were created only after the ability to support them had been developed. The area that was most creative in the structural sense lay along the Oise river. Mostly in the 1120s they created the peaked arch (a variation of the pointed arch), flyers, thin-wall construction, the structural buttress and the structural rib. The latter were most important for the evolution of Gothic. This may be from the presence of good local quarries and skilled carvers with some of the earliest building funds in the region. I have recently studied the relative flow of money into architecture. The graph over the page shows the different amounts being spent in the northeast and the Parisis. Though neither I nor my colleagues have any idea what caused the 1150s to 70s decline, it had a profound effect. In foliage naturalism replaced abstraction, and in architecture the northeast developed height, spaciousness, great glass walls and theatricality, while the Parisis tended to hold onto the opposite qualities until well after 1200. The decline in the 90s in the northeast coincided with the start of a fifty-year drought. As funding collapsed inventiveness disappeared with it, and designs reverted to both the ideas and the techniques of a generation earlier. Nowhere did style evolve continuously in the one direction, but there

memes and assumptions

61

was a coming and going in ideas and inventions. This brings up another assumption: that artistic ideas moved progressively towards one high goal. In fact, ideas that play a major role at a later date often seem to disappear for a while between their invention and their flowering. This is a curious paradox, and may shed light on the inner processes of creation which may be very illuminating. One other matter has arisen that may have a considerable bearing on any updated history - the origin of the concept itself. It has been written that Gothic was first manifested in the choir Abbot Suger built at Saint-Denis. Every book eulogizes its openness, wide windows and exquisite vaults. Rightly so. But excellent as it is, was it the first conscious step towards the new? It seems likely that, like every other church of the period, the choir was not the creation of one genius, but of many. In the ambulatory the walls with large windows were constructed by one master to support domical and groin vaults over the chapels. He was followed by another who installed drum piers with rib vaults. Though revolutionary and beautiful, no later building by either master show any influence from what they created at Saint-Denis. Coupled with the fact that Saint-Denis remained uncopied for over twenty years, it would seem that the choir was not a deliberate attempt to create a new spacial form, but the result of a fluke of appointments and opinions. This raises the possibility that the key aspects of Gothic were the natural product of another time. The ramifications of this possibility are enormous. Indeed, it would seem that the first building with the key elements of Gothic (extremely thin members, hollowed-out walls, flying buttresses, no wall between windows and shafts, totally integrated shafting and vaults) may be the axial chapel at Saint-Remi from the mid-60s. From here there is an unbroken line of development through the Soissons south transept to Braine and Chartres. These four key buildings are all by Soissonaise and Reimois builders, where most of the stylistic ideas that created Gothic architecture were made The role of the abbeys has seldom been mentioned. The redating of the austere Cistercian abbey at Longpont to 1180 suggests the monks may have inspired the extreme simplicity that developed in local carving after a long period of very ornamental work. The resulting simplicity may have concentrated architect’s attention on structural rather than decorative form, which is the heart of the Gothic revolution. There are personal connections between Soissons and Longpont that make this more likely. The eastern chapel in the Benedictine abbey of Saint-Remi of the 1160s may have made the first Gothic statement. Later, the abbeys of Orbais and

62

Essômes found how to create the impression that the main vaults could be designed to appear weightless, so they seemed to hover over the congregation, like a canopy. Tracery and the linked triforium were also invented in these Benedictine abbeys. One consequence is be that the driving creative force behind Gothic in the northeast may have been a popular enthusiasm kindled by the ruling bishops, funded by wine and guided by the monastic orders, in contrast to the oft-quoted version that Gothic was an expression of the emerging authority of the Capetians. Gothic — the quintessential architecture of the spirit — would then have been the creation of a spiritual/agricultural rather than an imperial/mercantile society. The size and location of the buildings in the Ile-de-France between 1130 and 1150 suggest that funds came from the royal treasury and those of Louis’ closest allies, and that after 1180 they came from the merchant and urban communities. In the east, on the other hand, the same factors suggest that money came from the farmers and vignerons. The possibility would radically alter our views of the political and social setting in which Gothic evolved. Concerning sculpture, some of the consequences are no less momentous. The fourteen decorated portals carved before 1170 are usually spread evenly over the five decades between 1130 and 1180. However, the evidence for dating is minimal. Rigorous studies of foliated capitals and the hands that carved them is providing an alternative dating procedure. This is showing that all may have been built between 1134 to 1145. It raises many important art historical issues that could fundamentally alter the way we understand the period. It also leaves a most intriguing hiatus between 1150 and 1190. To what extent were the portals inspired by Eleanor of Aquitaine? Her homeland has some of the most magnificent portals in the country. The construction of nearly all these portals continued only while she remained in Paris and — can we say? — in love with Louis, which was between 1137 and the mid-1140s.10 The majority of later work in the Ile-de-France before 1230 could not be called Gothic in the way we would define it. With a few exceptions, churches are squat, dark and cave-like. They do not soar to great heights with complex internal spaces and huge windows. They are simple rectangles with clerestories that may have no windows at all or only small oculi. The

10

James, Ark of God, projected in vol. 6.

memes and assumptions

63

role of the Paris region in these years would seem to have been overvalued. It was not until the late 1230s that a new and adventurous mode emerged that Branner called “The Court Style” of la Sainte-Chapelle. It was this, based on ideas imbibed from the northeast, that the High Gothic of Amiens and Beauvais was to develop. Ten years ago I attempted a trial run at reconsidering the history of Gothic in A Traveller’s Key. I had less understanding than I have now, and recent research has redefined some of the main issues. I have now reached the moment in my life when I can begin to gather all these parts together into an updated history of Gothic architecture, which I hope to publish in a later volume of The Ark of God.

IV Chartres was lucky the Parisians were busy *

I

N 1969 I came to Chartres with my wife Hilary and our three children, and began to probe into the cathedral’s history. Through all the years since then I have had the most generous support ­from the people of Chartres and the wonderful friends we have made here.­ Whatever I have discovered was made possible by their interest and encouragement. For this I am extremely grateful. Though I have ranged far from Chartres in my search for the origins of the masters who worked there, it is to Chartres I have returned again and again for sustenance and inspiration. With information derived from this wider search I want to show four things: 1) The major contractors who created the cathedral came from the region to the north-east of Paris, from the Soissonaise, the Laonnais and the Reimois. These great men were not trained in the workshops of the Ile-deFrance, but in those beyond the Royal Domain. 2) The cathedral Chapter may have chosen the builders for aesthetic reasons. The style of architecture then being created in the Paris region did not contain the ­elements that we are used to calling Gothic. With a few notable exceptions such as Notre-Dame, they were building squat churches with very little glass. 3) The Chapter may have chosen the builders for purely practical reasons. During 1190s there ­was a construction boom in the Ile-de-France, while the north-east was in economic decline. Parisian builders would have been busy in their own locality, while ­those in the north-east would have been actively seeking work elsewhere. * Translated from a talk given at the Colloque international sur l’occasion de 8o centenaire de la cathédrale de Chartres en 1994, et publié dans Monde médiéval et société chartraine, Paris 1997, 39–62.

chartres was lucky

65

4) Most of the inventions that created the Gothic style came from the­ north-east. Without the creativity of that region of prosperity and peacefulness Gothic as we know it may not have evolved. It was the role of the Ile-de-France to reassemble the inventions that came from the north-east into the ­next phase of Gothic that came after Chartres, the phase we call Rayonnant, but it is to the north-east that we must look for the origin of the powerful creative impulse that brought the Gothic style into being. In the five years I spent at Chartres after 1969 I was able to detail the construction sequence of the cathedral. To do this I had to create my own techniques I called Toichology.1 Some 10,000­ changes to the design of profiles and elements showed that the cathedral had been built by many teams of masons each working for short campaigns of construction some 32 separate campaigns from the footings to the completion of the choir vaults. The documents which help to date the work suggest each campaign lasted for about a year. I illustrated all of these campaigns in isometric drawings like this, Fig. IV–1.

Fig. IV–1. Isometric drawing showing where work had finished at the end of Bronze’s campaign in 1205.

1

James, Contractors; James, Master masons. Methodology described in James, Template-makers, especially in the appendix.

66

Each campaign was led by a different master. The masters were independent, and were guided by little more­than general principles laid down by the Chapter, and perhaps with a large-scale model. I suggest a model from the similarities and the differences between Saint-Remi, Notre-Dame-enVaux and Mouzon that suggest that the moulding details were too small to be included.2 Within the constraints of the model each master used the technical methods and profiles they liked best. The means for identifying the masters, both at Chartres and elsewhere, lie in these technical details, the elements and profiles, and the way each master used a personal geometry and foot measure.3 Eight masters shared the work to the completion of the choir vaults. On most occasions they brought their own carvers and masons with them.4 The chronology for the layers of construction was set between the documented dates. From this it was clear that the portal sculpture was earlier than thought, and led to the enormously important conclusion that the apogee of Gothic sculpture had occurred during the reign of Philipe Auguste, and not of Saint Louis. Nothing I have read or researched over the past thirty years has altered my earlier conclusion that every Early Gothic building in the Paris Basin was constructed in­the same way — in many short campaigns under the control of different masters. At this time there were no chantier under the direction of one permanently employed master. This applies to the whole period from the 1130s to the 1240s and includes works under royal patronage and with presumably adequate funding, like Saint-Denis5 and la Sainte-Chapelle.6 The Chartres Royal Portal was no exception: the irregularities in the carving being due to changes made by successive masters, not by the doorways being moved.7 Having identified the masters who constructed Chartres I searched for them in other buildings in the Paris Basin. Olive, for example, worked at Chartres five times, starting in 1197. I have found he also worked at 44 other churches between 1190 and 1230.8 At Chartres his design for the five great lancets under the south rose shows his interest in breaking windows James, Template-makers, 196–200. James, Contractors, ch. 5. 4 James, Contractors, ch. 16. 5 “VIII: Saint-Denis chevet”. 6 James, Template-makers, ch. 4. 7 “VII: Ascension and Incarnation portals”. 8 James, Template-makers, ch. 8 and 9. 2 3

chartres was lucky

Fig. IV–2. Chartres cathedral, external walkway under the lancet windows of the south transept, laid out by Olive in 1210.

Fig. IV–3. Abbey of Essômes, transepts looking east.

67

68

Fig. IV–4. Essômes, eastern window in the northern transept chapel.

Fig. IV–5. Essômes, eastern window in the southern transept chapel.

chartres was lucky

69

into their elements, Fig. IV–2. This was in 1210. Just afterwards he began the ­momentous process that was to lead to the invention of tracery. You can watch this creative process unfolding in the windows to the transept chapels at the Abbey of Essômes on the Marne, Fig. IV–3. In the northern two chapels the oculus presses down onto the arches over the lancets. There is a small triangular piece of stone between them, Fig. IV–4. In the two ­southern chapels he has become intrigued with this triangle and has hollowed it out, Fig. IV–5. Not all the way through, but far enough to create a void on the inside, where there was a solid. In the fifth window, the one facing to the south, he cut this triangular hole completely through the ­wall and filled it with glass, Fig. IV–6. This is the first such window in medieval ­architecture. Before then, every window (and, indeed, every form and element,­ be it the plan of a chapel, a porch or a door) had been generated from axes and­ centres that lay within the element itself. Windows were designed symmetrically around a central axis and the centres of their arcs lay within the window opening. This was the first time any medieval master formed an opening as the consequence of the forms around it. Its shape was, as it were, accidental. This was the first step in a transformation that was to have momentous consequences in later centuries – though there is one rider that is discussed in the addendum to this chapter. This led directly to Olive’s design for the chapel windows in the cathedral of Reims, Fig. IV–7, where he constructed a large segment of the aisle chapels from the walkway to the apex of the windows and the vaults over them. At Reims he built the first example of true tracery around 1216.9 When I identified the other buildings by Olive and plotted them on a map nearly every one of them are bunched together along the Marne, Map IV–1. This area would seem to have been ­his home base. Indeed, nine of his twelve earliest works lie near the junction between the Marne and its tributary, the Surmelin. This suggests that Olive may have lived in this area, and that from here he and his men travelled the 150 kilometres south-west to work at Chartres. I have accumulated this evidence by visiting all 3,500 churches in the Paris Basin to isolate those in which some part was constructed at this time,

9

James, Template-makers, 158–9.

70

Fig. IV–6. Essômes, south window in the southern transept chapel.

Fig. IV–7. Reims cathedral, apse chapel window.

71

chartres was lucky

Map IV–1. Sites where Olive worked.

Map IV–2. Sites where Bronze worked.

Map II-1.10 Among these, 650 have details and carved capitals that can help to identify the masters in charge and date their campaigns. It was my dream that by identifying the major innovators of the period I would be able to bring a greater sense of ­life and individuality to this most creative era. It is thrilling to discover these men ­and to discern each step in their inventive process, as in the invention of ­tracery. This opens a prospect that I find most exciting: to clothe a previously anonymous period with an individual variety, and to catch a­ glimpse, however brief, into the actual process of creation. Bronze, for example, worked at Chartres more than any other — eleven times including the lowest stones of the porches in 1198. I have evidence for him in over­fifty other churches between 1170 and 1240, Map IV–2. There 10

“II: An investigation”. page 26

72

is a definite ­concentration of work by Bronze in the Soissonaise, including extensive work on the cathedral of Soissons itself. What we now need to pursue these ideas further is a team of people to study all the ­fifty or so major contractors of the Paris Basin. Then we ­will be able to begin writing a history of the extraordinary achievements of ­the Early Gothic architects from a personal rather than from a stylistic point ­of view. It is significant before 1190 Bronze always worked within this core area, while after 1200­over half his work lay down the Oise valley and on into the south-west, Map IV–3. At first he was a local man, but he quickly learned to be an inter-regional contractor moving great distances between jobs. Over the many decades in which this was happening, the travelling would have had a profound effect on the way he was engaged, the way he worked and the way he employed his men. Regional teams based on local quarries and training workshops may have turned into looser organizations relying more and more on part-time subcontractors than on full-time labour. When I say ‘him’ for Bronze I refer to the team rather than to the man. Minor changes to geometry and profiles at Chartres have shown that the leader of­ this team was changed between 1195 and 1198, and again in the mid 1220s.11 Similarly with the Scarlet team where the inspired master who set out the first plan for the cathedral was­ replaced around 1210.12 He also set out the Abbey of Longpont, Fig. IV–8, and the choir chapels at Lagny. Map IV–4, shows that Scarlet’s earlier work, like that of Bronze, lay in the Soissonaise. After 1210 he worked almost all the time outside his home base. The same concentration of work and travel patterns occurred with Rose, Ruby and Cobalt, Maps IV–5, 6 and 7. The concentration of work in the north-east shows that the major contractors for Chartres came from the Soissonaise, the Laonnais and the Reimois. They were not, and could not have been, trained in the workshops of the Royal Domain. The reason for this is obvious once we look at what was being produced in the Paris region .­The predominant style of architecture did not contain the elements that we are used to calling Gothic. With few exceptions, they were building squat churches with little glass, Fig. IV–9, such as Bagneux,­ Ferrières and Villeneuve-le-Comte.

11 12

James, Contractors, ch. 22, n. 49. “XXXV: Canopy of Paradise”.

chartres was lucky

73

Map IV–3. Sites where Bronze worked before 1190.

Map IV–4. Sites where Scarlet worked.

Map IV–5. Sites where Rose worked.

74

Map IV–6. Sites where Ruby worked.

Map IV–7. Sites where Cobalt worked.

Map IV–8. Squat churches in which the clerestory is small and the vaults spring from within the triforium.

chartres was lucky

Fig. IV–8. Abbey of Longpont, remains of the apse and transept begun by Scarlet in the 1180s.

Fig. IV–9. Villeneuve-le-Comte, the nave.

75

76

Fig. IV–10. Relative construction effort in the three key regions of the Paris Basin between 1140 and 1240.

There are 46 of these low-vaulted churches, Map IV–8.13 This was not being old-fashioned. These churches became more and more popular as the years passed, for where only six percent were built before 1190, forty percent were built in the 1220s. For this reason, the Chapter’s choice of builders may have been aesthetic.­ They may not have liked what they saw being built just to the north of them. Yet, there may be another reason. We can now say definitely that while the cathedral was being built a construction boom was under way in the Ilede-France, dashed line in Fig. IV–10. It had begun there in the 1180s. This data has been assembled by dating every part of all the more significant 640 Early Gothic churches of the Paris Basin, and by costing each part with a common unit of construction.14 I must first say a word about dating, for the methods I have employed may be new to some of you. Dating is based on the documents (where

13 14

James, Template-makers, map 138. “XL: Funding”.

chartres was lucky

77

they exist) and on scholarly opinion — including the many buildings whose dates have been reassessed during the past ten years. However, there are few scholarly analyses and less documents for the myriad of smaller buildings, and these represent over two-thirds of the total construction of the period. I have found that by understanding the style of the foliage on the capitals we may date every part of these buildings, and usually to the decade. The way the leaves were carved can provide a reasonably secure basis for dating these works.15 This method is based on the many hundreds of capitals that can be dated with precision from the documents, such as those at Chartres, the choir of Saint-Denis, and so on. They show two things. The first is that during the decade of the 1170s, and precisely during that decade, every carver of formal capitals changed their way of working and transformed themselves into carvers of foliage, becoming more and more naturalistic as the decades passed.16 The second is that after the 1170s there was an even and readily definable evolution of foliage from one decade to the next. Each stage can be defined quite precisely. It is a process from spring and youth into high summer. We can rely on the accuracy of this evolutionary process for two reasons. First, the capitals that were laid during one campaign of construction will invariably confirm to one decadic fashion. Second, there is not one building where the majority of the capitals in an upper story are stylistically from an earlier decade than those underneath — for if they were, this method would be useless. Returning to Fig. IV–10, the solid lines shows the significant decline in building work. In the Laonnais it was far worse, even catastrophic, as nearly all construction stopping after 1200.­In the west around Paris, on the other hand, construction continued to escalate for anther two decades or more, as indicated by the dashed line. This was sufficient to compel most builders to move out of the Soissonaise after 1190 and into the west. This is why we find Chartres being constructed by men from the other side of Paris. Building in the Royal Domain was progressing on a very large number of minor churches, with relatively small amounts being spent on each. This

15 16

James, Ark of God, ch. 4. James, Ark of God, ch. 5

78

suggests the buildings were being funded more by local burghers and vintners than by the royal court — though their prosperity may have had a lot to do with Phillip’s conquests and­the advantages he acquired for the tradesmen of the Royal Domain over those of the conquered areas. One reason for the economic decline may have been the climate. The weather had been getting hotter and hotter and the winters less and less frosty for a century. Fig. II–2 shows the decreasing rainfall and the increasing temperature during this period. The growing population and the better food supply made possible by the hotter seasons, formed the basis for the expansion of commerce and the growing power of central government. ­But from the 90s onwards the climate turned into a prolonged drought. The droughts lasted, on­and off, for fifty years. They may have been severe enough to burn the famous vineyards of the ­north. As wine production was one of the few landed investments available to the merchants, any risk to the productivity of the vineyards would have undermined a major source for the funding of church construction. For this reason, the Chartres chapter’s choice of builders may have been practical. Bronze and Scarlet and the others may have been the only ones getting less commissions and therefore with open schedules, both available and willing to move a great distance from home. It is fortunate for us who love this building that the chapter chose men from the Soissons-Marne areas. Most of the inventions that created the Gothic style came from the north-east. Though thin-wall construction and its natural mate, the flying buttress, came from the Oise and Seine valleys before 1160, all the other significant creative devices were made in the northeast, Map IV–9 and Table IV–1. These are: a) The continuous triforium arcade that stretched across the entire wall between the vaulting shafts. It was possibly first used at Laon or in the transept of Noyon, and nearly all examples of the continuous triforium passage before 1190 lie in the north-east, Fig. IV–9.17 b) The full-width window stretching from one lateral wall to the other so that it occupied the entire width of the wall, Fig. IV–11. Map IV–10 indicates the churches built before 1190 with a full-width triple window in the eastern wall. Nearly all the earliest examples come from the north-east zone. 17

James, Template-makers, 116; Bony, French gothic, 105 et seq.

chartres was lucky

79

Map IV–9. Paris Basin showing the numbers of inventions listed in Table IV–1 that came from each region up to 1230.

Table IV–1. The more important inventions that led to the creation of the Gothic style, the area where most are to be found, and the probable earliest date.

80

Map IV–10. Location of churches with a triple window in the eastern wall.

c) The tall clerestory was first used in the cathedral of Soissons before 1200, years before Chartres, Fig. IV–11.18 Here the capital under the high vaults was shifted upwards, above the sills of the clerestory window sills, so the area of glass could be made much greater. d) The method of bonding the lower parts of the vault into the wall, called the tas-de-charge, made vaults over large spans easier to build. As the arches and ribs became thinner and thinner, this ensured that their junction with the wall remained secure. The first of these were built in the Soissonaise during the 1150s.19 e) The longing to build tall and­ spacious interiors. This can be clearly seen in the huge increase in the height of the main Soissons vaults compared to the transept, Fig. IV–11. This decision was made in the later 1190s.20 The height of Notre-Dame was exceptional in the Paris region. James, Ark of God, 559–566. James, Template-makers, 109–111 with map. 20 James, Ark of God, 561. 18 19

chartres was lucky

81

Fig. IV–11. Soissons cathedral, choir and south transept. The vaults of the choir are set much higher than those in the transept.

Fig. IV–12. Couvrelles, the triforium on the western wall with two windows behind it, by Bronze in the 1180s.

82

f ) From the Marne came window tracery invented by Olive, discussed earlier. g) The linking of the triforium with the clerestory, an invention that was to play a big role in later years, was probably first conceived at Saint-Remi around 1190, followed by Orbais a few years later, Fig. I–11.21 h) The fully-glazed triforium was created at Vaudoy-en-Brie in the 1220s by a master from the Marne/Reims region, Fig. I–19.22 Bronze had been the first to explore this concept in the 1180s in the west front of Couvrelles, a town not far from Soissons, Fig. IV–12.23 This raises a most important question: Would the features of the Rayonnant style that was to be created after 1230 have been very different if the inventions of the north-eastern masters had not been available. But for the decline in wealth, would the builders from the Soissonais and the Marne have travelled beyond their regions into other areas? If not, would these ideas ever have been taken up by the Court and transformed into the architecture of la Sainte-Chapelle and the Saint-Denis nave? The lower chapel of la Sainte-Chapelle is typically Parisian: low vaulted and squat, Fig. XXXIII–1. It is the upper chapel that speaks to us of the Soissonais, the Marne and the Reimois, Fig. XXXIII–4. Maybe, without the contribution of the builders from the north-east the Rayonnant of the upper chapel may not have been possible. Certainly, without these builders the classical form of Chartres would not have been possible. It is to the north-east that we must look for the inspiration and experience that created our cathedral. Whether it was the prolonged drought in the vineyards of the Aisne or some other cause that made these contractors available, it was only through the availability of these great ­and inventive builders that it was possible for the “Parthenon of ­France” to be constructed at Chartres and not anywhere else.

James, Template-makers, 179–190; Prache, Saint-Remi, 95–7. James, Ark of God, 1137; Branner, Saint Louis, assumed that the Saint-Denis nave had to be the first. 23 James, Template-makers, 58-61. 21 22

chartres was lucky

Fig. IV–13. Chartres, south porch with slot window into the stairs and the panel underneath.

83

84

Addendum I earlier described the evolution of Olive’s windows at Essômes as leading directly to tracery and also as reversing an earlier ayatemic approach to geometry by “forming openings as a consequence of the forms around it” rather then from within the shape itself. There are decorative patterns in the panels under the lowest staircase windows in the cathedral of Chartres, Fig. IV–13. Like Olive’s traceried windows and the slot windows above them, but from 15 years earlier. They show an evolution in ideas that is quite fascinating, and pregnant for the

Fig. IV–14. Panel under SW window.

Fig. IV–15. Panel under SE window.

chartres was lucky

85

future. Figs. IV–14 to 16 show that the lower stones are carved with small oculi between the tri-lobe arches. The bars between them are arranged as they would be in plate tracery. In the upper stone in Fig. IV-16 under the NW window the design is treated in the same plate-like manner. However in the lower stone of the NE window Fig. IV-17 the oculus has been settled down into the cusp of the arches, and the small triangular spaces above have been hollowed out. The same procedure was used in the thin course above that. This is bar tracery. These are precisely the same evolutionary steps previously described for Olive at Essômes where he was evolving the concept of window tracery. The process shows one way in which new ideas were developed and explored.

Fig. IV–16. Panel under NW window.

Fig. IV–17. Panel under NE window.

Toichology

V The discipline of Toichology*

W

E need a term for the archaeology of standing stone structures.­I have introduced the word “Toichology” to describe this work.­ Analysis of above-ground structures has been called Archaeology, ­but the methods of the archaeologist have advanced to such a high­ degree in the layer by layer examination of below-ground evidence­ where the previous layer is removed and destroyed in uncovering ­the next, that a separate term is needed. In an existing building­ we cannot pull out walls or remove one part to get at another.­The approach is different, and relies heavily on understanding ­building procedures and techniques. A word was needed to encompass the techniques I set out in the­Appendix of The Template-makers of the Paris Basin. My ­brother-in-law, Maurice Kelly, professor of Classics and Ancient­History at Armidale University, suggested the word ‘Toichology’ from the Greek toichos, a standing wall. Toichology is as exacting as archaeology, and the evidence is­capable of being verified as in any observation-based discipline, ­such as geology, with which there are many methodological­ similarities. It is NOT, as Russell claimed, “nothing more nor less than connoisseurship dressed up in pseudoscientific­clothing”. Connoisseurship is to do with style. Toichology, which ­is to do with the analysis of blocks of masonry, their sizes, the­ templates issued to determine their shapes and the location of ­the joints between them, has nothing whatsoever to do with style. The major skill required (besides understanding the process and­ procedures of building), is an impeccable care in observation. Anyone

* From Avista Forum, vii 1994, 9.

90

who intends to deal with ­masonry and the masons who erected it has to understand the trade ­as well as the style.1 Without understanding the church in three dimensions and being able, either on paper or in the mind, to visualize the building evolving in space­ and time, it is extremely difficult to comprehend the process of architectural construction. It would be useful if courses in stone-reading was included in­ the training of all architectural historians. The articles that follow illustrate a range of information that can be gained from this technique.

Among the growing literature on this subject, in contrast to those on style, are: Armi, Masons and sculptors; Bilson, Durham cathedral, 101–160; Davies, Standing structures, 54– 64; Gardner, Two campaigns, 584–587; James, Template‑makers, especially the appendix on Cerseuil; James, Contractors; two articles in this collection “XV: Durham” and “VII: Anomalies”; Kimpel, Développement, 195–222; Murray, Church of St‑Pierre, 533–551; and Murray, Nave of Troyes, 121–139. 1

VI Were the Chartres transept porches added? *

S

OME three generations ago scholars formed their theories on the stylistic evolution of Gothic sculpture, and to suit these ideas concluded that some figures like St. Maurice in the south west door, and some of the figures on the north porch piers, should have belonged to a later period than the other sculpture.1 Therefore it was presumed, without analysis of the building structure, that they had been added or inserted. As later historians came to analyse the architecture into which the sculpture had been built, they tended to adjust the building’s history to suit the pre-established theories about sculpture. The theories on the evolution of the sculpture began when two highlyregarded scholars, Merlet and Lefèvre-Pontalis, used the document of 12242 to show that the south porch was an afterthought added onto the face of the building after the doorways and walls over them had been completed.3 Some years later this theory was embellished by Grodecki with the idea that some of the doorways had been cut through the transept walls later on, and that the lower parts of some nearby buttresses had been deeply chopped back.4 Such ideas are easy to write about, but for the builder hard to do. Grodecki drew momentous and costly conclusions from very small misalignments,

* Extract from The contractors of Chartres, Wyong, 1979. 1 To avoid any confusion between the similar words ‘porch’ and ‘portal’, I do not use the latter word, but use ‘doorway’ for the work within the thickness of the wall, and ‘porch’ for the roof, vaults and supports which project in front of the walls. 2 Lepinois and Merlet, Cartulaire, 103. Translated by Branner, Chartres, 98–9. 3 R. Merlet, “Guide archéologique”, 56; Lefèvre-Pontalis, “Archives historiques”. 4 Grodecki, “Transept portals”, 13; Mortet, “L’expertise” was also responsible for some of these ideas.

92

and did not explain why such extensive reworking was necessary nor, in the eyes of the Chapter, worth the tremendous cost. In most of the books written on the subject since then these ideas have been uncritically accepted, though in some writings on the sculpture5 the inevitable contradictions have been picked up without influencing architectural historians, while the idea that doorways had been cut through the walls at a later date have been refuted6 without much effect on writers on scuplture.7 These many refutations, including mine contained here that was published first in 1972,8 have had little impact because the earlier theory based on the stylistic evolution of sculpture continues to dominate. Together these two theories have so obscured the scene that even after the enormous amount of research carried out this century, one expert has written that our understanding of the “transept facades of the Cathedral of Chartres is still in its infancy”.9 If we are to reassess the history of the transepts we would be advised to put aside the theories and arguments and approach the stones with fresh and curious minds. The documents give us no help about either the sculpture or the porches,10 so we must rely on the major source of evidence we have — the cathedral itself. The stones contain the full history. In the last analysis there is no other. Start by examining the supposedly added lintels supporting the vaults and roof of the south porch, Fig. VI–1. They are cut from large accuratelyshaped stones that weigh about 4¾ tons each. On either flank of the side doorways the lintels are made up of four stones each, while around the central door the lintels consist of a single stone that carries the whole weight of the main vault and roof, shown dashed in Fig. VI–2. Within the mass of the wall between the doors there is a staircase S which has windows opening onto the porch. Wide windows were built between the lintels from large stones which are butted hard up against the sides of the

Sauerlander, Strassburg, especially in his discussion of the north tabernacles. Delaporte, “Remarques”, 299–300; van der Meulen, “Recent literature”, 152 et seq. 7 Sauerlander, Strassburg; Grodecki, “Chronologie de Chartres”; Henderson, Gothic. 8 James, Contractors, 1972. 9 van der Meulen, “Recent literature”,164. 10 van der Meulen, “Recent literature”. 5 6

the chartres transept porches

Fig. VI–1. The south porch from the east. The piers support heavy lintels which are set into the buttress, and over them the arcaded gallery of kings Fig. VI–2. The plan of the SE porch staircaset showing the lintels dashed.

93

94

lintels. In one case, the lintel continues into the window reveal to form one side of the jamb. Thus, in one single piece of stone is joined together the lintel, the wall, and one side of the window. From inside the staircases the coursing is even, continuing unbroken from the porch to the north face. There is no sign of reworking and the jointing is regular all round. Considering the manual techniques of the thirteenth century, any added work would have been recognisable on the inside. The coursing within the southern staircases is regular from the floor level to the triforium, unusually even for Chartres. If you want to see what broken jointing looks like, just examine the stairs inside the northern porches, particularly around the level of the Job tympanum in the north west stairs. Many stones on the roof over the porch had deteriorated over the centuries and had to be reinforced. Starting in 1896–7 restorations were prepared by the architects Selmersheim and Mouton, and then begun in 1898. Most of the documents for this work are in the Archives d’Eure-etLoir. They show the extent of work, the details for the new stones, and of old stones reused. In new work the original shapes were carefully followed, though the jointing of some of the stones in the gables was regularised, as can be seen in contemporary photos in Beaux Arts, Paris. From within the stairs the ends of the iron reinforcing placed during the 1989 restorations is visible. The stone type and surface shows that this was the limit of the reworking in this area. Thus all the lintels are original. When examining the stones at close quarters it is obvious that the lintels were laid first, and that the stones of the window were laid up to them later on. All the mouldings on the lintels exactly meet their neighbours on all sides. There is no way in which the work could have been erected in the opposite order. The theories that the lintels for the porch were added into the building after the doorways and the transept walls above them had been completed is, from this evidence, untenable. The mortar joints between the lintel and the stones of the archivolt show this, for they are as small as the others. Yet we are expected to believe that the builders of 1200, working without sophisticated moving cranes, could manhandle so many almost five-ton stones into slots in the wall with a tolerance of only a centimetre or so either way. It could not have been done. The lintel might have been place on a wooden platform that could be moved on rollers over centring below, and slid into place. But we are not dealing with the insertion of a small stone, but with a heavy unwieldy

the chartres transept porches

95

weight with an enormous inertia, supposedly resting on timber strutting that must give to some extent under a large moving load. It could not have been simply ‘slipped in’ and still finish with such a perfect alignment to older work on all its sides. The centring and lifting gear would have been stupendous. Massive timber members, braced and strutted, would have entangled the whole area of the porches, both below the lintels to take the weight, and above to take the moving and lifting gear. The sheer bulk of the support work and crane structure would have encumbered the whole site with a real danger of damaging the older delicate sculpture, and of the lintels bumping into the archivolts as they were being ponderously moved into place. Further, mortar shrinks as it sets, and no amount of juggling can prevent a heavy stone from settling once the platform is removed. Yet at each joint the moulding around the bottom of the lintel is in line with its neighbours. It is cut to exactly the same template as the impost over the wall figures inside the doorway recess. This shows that there was no differential settlement from mortar shrinkage, again affirming that all stones were installed together. At Noyon the west porch was added over the old doorways, but with separate piers set against the older walling to support the new lintels without disturbing the old work at all. If the southern lintels had been added in later at Chartres, either they would have used this sort of technique, or they would have left signs of having done so. Thus, the southern lintels were carved with the adjoining imposts and window, laid with them, and together all the mortar was allowed to set. Later when the lintels were loaded with the weight of the stones of the archivolt and roof no differential settlement occurred because the mortar under the load-bearing stones had been given a chance to harden to the same extent as the mortar under adjoining stones. The same series of arguments can be applied to the side lintels where they butt into the corner buttresses to show that they were placed after their capitals below. The adjoining stones of the buttress were placed hard against the sides of the lintels and were therefore laid afterwards. Above the southern porch roofs there are two deep-set doors that give access onto the roofs. Their stonework rests on top of the lintels below. If the porches were an after-thought, these doorways would have been hacked through an existing wall, and we should expect to find rough edges, and random angled joints where the plane of the doorway wall would have cut the plane of the joints at an obtuse angle.

96

However, the joints are true to the doorway, they are neat, and there is no sign of reworking. The surface texture, and the treatment of the edges of the masonry is identical to that inside the stairway. The doorways were therefore planned for the porches, and built to suit them. It might be argued, that the doors were built for porches intended from the beginning, though the actual porches were not added until later. Taken on its own without looking at the other evidence this would not be impossible, though I know of no other medieval building where they did this sort of thing. From these doors we can get onto the porch roofs, and stand among the statues and their covering aedicules. There are eighteen similar if unspectacular statues each about 2½ meters high, Fig. VI–3. In the plan of the statues on the eastern side, Fig. VI–4, the bas-relief is marked B. The ones on the outside can be reached by climbing up the series of projecting flashing blocks that were inserted in the façade wall in 1900, like a giant’s causeway. If the porch had been added after 1224 these statues and the arches over them would either have been added onto the older work then, or the springing of the arches may have been placed ready for the future arcade when the buttresses were being built, as you can see in the north. But in the latter case we should be able to observe some differences in the form of the columns and arches, and the carving in their capitals, for those built into the buttresses would be in the style of 1206, while the rest would have been carved almost a generation later. In fact they are in all respects the same. On the western side the unbroken coursing round the corner of the buttress into the recess behind the kings shows that the whole of this corner of the buttress above the lintel was laid together: see A in Fig. VI–5. The original face of the buttress was set back only slightly from the edge of the lintel at B, and was probably intended to be faced with a shallow arcade like those in the north. But above B, the recesses for the statues were built with the pilasters between them and the adjoining buttress, as in the course A. Above the arches the face of the buttress was moved out in line with the lintel at C. The courses above the arches were laid together with the head of the transept aisle window, and the nearby roof. There is no sign of reworking in the last two zones. The arches are cut from large stones built into the buttress masonry. The end of one of these arch stones D hangs past the face of the buttress, and is set hard up against the underside of the masonry over it. Therefore the arches, resting on their columns, were built in with the buttress. Further,

the chartres transept porches

Fig. VI–3. The kings on the east side. The right figure is a bas-relief.

Fig. VI–4. Plan of the kings on the east side, the bas-relief is marked ‘B’.

Fig. VI–5. The kings on the west side. The arrangement of the surrounding stones shows the order of construction.

97

98

it was built with the rest of the aedicules over the south porch, as the same capitals, the same column bases, and the same type of arch is carried round the porch. Therefore the arcade and it surrounding buttress was built after, and onto, the lintel and at the same time as the aedicule and their columns were continued out over the porch roofs. There is one additional feature on the eastern side of the porch which is conclusive. The first arch against the buttress is sharply-pointed, squashed as it were, into the corner, Fig. VI–3. A photograph taken before the renovations shows that the original arcade arch that projects past the face of the buttress had a change in the shape of the arch in the middle.11 The watershed of the buttress over the arcade rests neatly on the arch. It had not been cut into, and all the coursing over the arcade continues unbroken from the buttress to the nearby aisle window. So the arcade rests on the lintel, and was built before the walls above it. The first arch on the west is, by contrast, a semicircular one, Fig. IV–5. Examine them carefully and you will see that the two projections in the corners between the buttresses are treated differently on each side of the porch, Fig. VI–6. The bulkier arrangement on the east forced the contractor to squeeze the arcade opening, and consequently to squeeze the statue placed under it. Both of these projections are reduced just below the statues and continue to the triforium as two small nibs K. But in the west the projection on the porch side is eliminated altogether L. On the east the bas-relief statue emerges from the wall, the back and the elbow disappearing from view into the stone. It is carved in high relief as if it was a free-standing statue, and is so cunningly made that no-one has noticed it is in fact part of the wall, not added onto its face, Fig. VI–3. The block from which it was carved is bedded firmly and neatly into the stones around it. It has the same pose as all the others in this level. Without doubt, therefore, the entire superstructure of the south porch was built integrally with the surrounding walls of the transept. Follow the walls round both sides, and in spite of the occasional breaks in the coursing there is enough continuous stonework above and below the aisle windows to show that porch and transepts were built with the aisle walls of nave and the choir.

11

Beaux Arts photograph Box 2, 2(4), 001–P318.

the chartres transept porches

99

Fig. VI–6. Plan of buttresses above the arches over the statues, with two ways of detailing the inner corners.

We must therefore look among the earlier crews, and not among later renovators for the contractors and sculptors of this porch. What then of the arguments of Grodecki and others that the porch was added? The theory is based on the fact that the two intermediate buttresses over the southern porch do not sit exactly over the doorways below. From this it is presumed that the lower one is a afterthought. The mistake is to assume that the porch was added, and therefore that the upper buttresses were built first, rather than the other way round, and that it is the doorways which are out of alignment, not the buttresses. Yet it is strange that the doorways are one of the more exactly built parts of the cathedral, while the upper buttresses are often woefully erratic. The plans of the doorways to each transept are identical, to an accuracy of less than 10mm. The overall length of the south 31,158mm and the north is 31,168mm: a difference of precisely 10mm. The levels are also accurate from one porch to the other in spite of inexact interior floor levels. Grodecki’s thesis that the side doors on the north were added later is, from style, stonework, coursing and geometry, completely untenable. Now that we can see that the doorways and the buttresses over them were each built with their walls, we can start looking for the real reason why one is not aligned with the other. The doorways are symmetrically dimensioned because they form a unit on the outside, and were set out together. But the buttresses are not symmetrical because they relate to the centrelines of the arcades inside the transept, which are twisted to one side. The cause of the misunderstanding is that the lower level was set out from the outside, while the upper one was set out to suit the inside which is not readily visible. This issue is discussed in detail in The Contractors of Chartres. This story illustrates the dangers on relying on simple visual evidence without looking for possible geometric or constructional causes.

VII Anomalies in the Ascension and Incarnation Portals of Chartres Cathedral

T

HE Royal Portal on the west façade of Chartres Cathedral has been studied with more care than any other assembly of medieval sculpture.1 Since some of the carvings had been altered before installation, it has been thought that the sculpture was originally erected further to the east, then dismantled and moved to its present position.2 Fels’s excavations showed this was not possible,3 but a detailed and consistent explanation of the many anomalies in the portal has, so far, not been offered. The general opinion might be summed up in Katzenellenbogen’s words, that they “are far too drastic to be explained by original miscalculations about the dimensions of the present site. It seems likely, therefore, that the façade was first planned for . . . another location with different dimensions.” 4 In 1979 I outlined the story that follows,5 and in chapter XIII discuss the relationship between the erection of the portal sculpture and the two adjacent towers.6 Because the sculpture of the portal had been recently cleaned,7 every stone and junction can now be examined with precision. Analysis shows that the shape of certain stones was skilfully adjusted to compensate for earlier discrepancies, and that these same stones were subsequently reworked to * Adapted from Gesta, xxv 1986, 101–108. 1 Sauerländer, Gothic Sculpture and Das Königsportal. 2 Lefèvre-Pontalis, “Façades successives”; Aubert, “Façade occidentale”. 3 Fels, “Fassade der Chartres”. 4 Katzenellenbogen, Chartres cathe­dral, 5. 5 James, Contractors, 220. 6 “XIII: Building the narthex”. 7 Niko, “Portail Royal”, 5–15.

anomalies in the portals of chartres

101

suit later changes in construction and design. The sequence of adaptations shows that the architectural design was altered a number of times while the portal was being erected. The changes are truly “drastic” and suggest that, rather than being the work of one team which could not make up its collective mind on the design, the portal was erected in a series of small campaigns by independent workshops working towards one iconographic scheme, but implementing their own ideas on detailing and dimensions. The evidence for this is complex, based as it is on six specific observations of construction anomalies on the Incarnation portal, and ten on the Ascension portal. They show that the adjustments were not made at one time during erection, but were made at different times in response to layout changes that were occurring during the process of erection. The design of the layout was therefore not finalized before erection began, and many stones were not carved until the erection was well advanced. The Incarnation Portal The junctions between the south tower and the right doorway show that the two were erected at the same time. The masonry of the wall behind the sculpture is bonded integrally into that of the tower, Fig. VII–1. This shows that the jambs of the doorway are contemporary with this level of the tower.

Fig. VII–1. Chartres, west doors, Incarnation portal.

102

In addition the bases of the wall statues are repeated on the south side of the tower, which suggests that they were carved by the same crew, the corner sundial is protected by a canopy matching some in the doorways, and the internal columns behind the doorways relate geometrically to the columns on the north side of the south tower. 1. On the southern portal with the Incarnation, the centres which establish the radius of the outer arcs of the tympanum do not coincide with the bottom of the tym­panum, but are placed somewhat below the tympanum. They are level with the tops of the lower archivolt course, over Gemini and Music, Fig. VII–2. This condition is unique in Gothic portals of France. Normally the radii of the curves of the upper archivolts and the sides of the tympanum on which they rest are struck from centres which line up with the top of the lintel and the base of the tympanum, as can be seen in the central door Fig. VII–3.8 The foliate frame around the sides and along the bottom, especially in the lower corners, make it very clear that the centres of the arcs had been placed below the top of the lintel when the tympanum was carved. The lowest stones in the archivolts have two registers, Fig. VII–1. The bottom ones were reduced in height after they were carved.9 Unless the original design for the lateral sculpture were to be unique in France, we must conclude that the top of the upper registers was intended to be level with the top of the lintels. Only then could the perimeter arc of the tympanum have been set out in the normal way. Since the tympanum was carved with the centres of the arcs placed below the top of the lintel, it must have been 8 9

Sauerländer, “Westportalen”. This can only he seen from the level of the impost for the legs of the figures and their seats have been arbitrarily truncated — Houvet, Portail Royal, fig. 61. The first archivolt on the right was reduced only at the bottom, as the carvings on Music would seem to have extended to the top of the stone, whereas Grammar was reduced both top and bottom, for the platform between the two figures is higher than its neighbour, and the head extends into the base of the stone above it. I do not find any evidence that the vertical joints between the three stones that make up each lateral tympanum were altered. On the Incarnation tympanum the tips of the wings clearly turn up at the ends to finish at the edge of the joint. The feathers turn with the silhouette of the wings showing that the edge of the wing was not adjusted after it had been carved. Similarly on the Ascension tym­panum, the folds of the cloth across the angel’s elbows continue unbroken from one stone to the next, while underneath the elbow and at the bottom of the tympanum the edges and the folds of material and clouds continue across the joint without any of the displacement that would have occurred if the tympanum had been narrowed.

anomalies in the portals of chartres

Fig. VII–2. Schematic elevation of the Incarnation tympanum showing the arcs struck from centres lying within the upper lintel. Possible lengths of the lower archivolts are shown in dotted lines. Fig. VII–3. Standard tympanum arrangement with the arc centres lying along the top of the lintel.

103

104

Fig. VII–4. Incarnation portal, left side lintels and archivolt of Gemini

Fig. VII–5. Right side lintels and archivolts of scholars, Music and Grammar

anomalies in the portals of chartres

105

designed at the time the archivolts were truncated, or just before. Therefore, the two lowest archivolts were carved before the tympanum and for a taller design. 2. Had the curved upper archivolts been carved with the tympanum, their lengths could have been adjusted for the height of the tympanum so that the height of the bottom registers would not have had to be reduced. There­fore, they, too, must have been carved before the tym­panum on which they rest. 3. The lintels of the Incarnation door, as has often been recognized, are narrower than the tympanum, Fig. VII–1. Yet if the half‑figure on the right of the lower lintel had occupied its full width, that lintel would have been wider than the tympanum by several centimetres, Fig. VII–1.10 There­fore, the lower lintel was carved before the tympanum. 4. Both lintels are set against the side of the archivolt on the right, while there is a gap on the left, C in Figs. VII–2 and 4. The misplacement may have been the fault of the erection gang, due to one side of this portal being more advanced than the other. There is evidence for this in a stepped joint in the coursing over this door. The whole of this side of the portal from the bases to the archivolts must have been erected with the southern tower.11 The ashlar to the left has been rebated over the stones to the right, showing that the former was laid after the latter. The centre of the portal was therefore, at that time, a little less advanced than the adjacent tower on the right. The reduction to the right figure of the lower lintel is so beautifully executed that it was probably done in the workshop before erection, the stone being delivered to the erecting gang already reduced. If the right side of the portal was a little further advanced than the left — even by one course — the erecting gang would have first installed the two right‑hand archivolt figures, Music and Grammar, and then, assuming that the lintel was the correct size, have butted it against those archivolts. Only when the time came to lay Dialectic and Gemini on the left side would the error have been realized. Part of the tympanum may have been placed by then, and the erecting gang

Both lintels were carved for a tympanum about 3 meters wide. They were then cut back to suit a tympanum about 2.75 meters wide. The Incarnation tympanum, which is about 2.87 wide, leaves the gap on the left of the lintels, while on the north, the Ascension lintel is about 2.9 meters wide. These measurements were estimated from underneath, as there was no scaffold. 11 James, Contractors, 220, 10

106

may have decided that a gap on one side was better than pulling it all down and starting again. As the cut‑back lower lintel is narrower than the tym­panum, it was truncated before the tympanum was carved. It must have been carved in an even earlier design phase for a wider tympanum, perhaps at the same time as the lower archivolts were carved, or before, but not after. 5. In the joints between the upper stones of the outer row of archivolts, Geometry and Arithmetic, and on either side of the keystone to the inner row, the adjacent faces of each joint are not parallel to one another but are slightly further apart from the bottom, Figs. VII–2 and 6. This shows that the width of the arch as erected is larger than was originally intended and confirms that the archivolts were carved for a narrower tympanum. It looks as if the truncated lintel could have been the right width for these archivolts, sug­gesting that the archivolts were carved at the time that the lintels were truncated. 6. Normally, lintels have an extension at each end which passes behind the lowest stones of the archivolts, presumably so that they may rest more securely on the imposts. One such extension may be seen exposed on the left of the upper lintel of the Incarnation door, Fig. VII–4. Once again, this normal condition may be viewed on the central door. But the right ends of these lintels do not pass behind the archivolts because the extensions were removed when the lintels were cut back, B in Fig. VII–7. Instead of the archivolts having a recess at the back to receive the end of the lintels, they extend back into the wall past the lintels to encase them, Fig. VII–5. This confirms that these archivolts were carved after the lintels had been cut back. The Ascension Portal The misalignments between the walling of the north tower and the adjacent doorway jamb show that the doors were butted up to the tower after this level had been finished. The fact that there are no jambs to the openings on the south side of the north tower should have alerted historians to their intentions to have the portal built where it now is, as open and unprotected access would never be allowed into the interior of a tower. 1) On the northern Ascension portal in Fig. VII–9, the arcs of the tympanum are also struck from points below the top of the lintel, demonstrating the same conclusions as in the first two points concerning

anomalies in the portals of chartres

Fig. VII–6. Incarnation portal, joint between upper archivolts, the outer row.

Fig. VII–7. Plan of the Incarnation lintels and archivolts, showing the cut back portion of the lintel at B, with the mortar infill at the north end. Fig. VII–8. Plan of the Ascension lintels and archivolts, showing the lower lintel in solid line J-K, and the upper one dotted line M-N

107

108

Fig. VII–9. Ascension portal

Fig. VII–10. Ascension portal, bases set into the buttress of the north tower.

anomalies in the portals of chartres

109

the Incarnation portal. However, above this tympanum, the joints between the uppermost outer archivolts, Virgo and Sagittarius, are not splayed, but parallel. They were, therefore, probably carved with the tympanum. This may confirm the right to left progression of the work. 2) The lintels of the Ascension door are not on the same plane. The upper lintel is parallel to the wall of the portal, and to the plane of the archivolts, J and K in Fig. VII–8, but the lower lintel is set back on the northern end, M and N in Fig VII–8. Yet the distance between the lower lintel and the outer edges of the imposts over the historiated capitals is the same on both sides. This means that the imposts are also out of line, I in Fig VII–8, and so are the jambs and embrasure figures underneath them. Once again, the erection gangs may have been responsible, for when they placed the lintel, they may have set it evenly over the imposts without checking their placement against the other doors of the portal. The reason for the misalignment of the jambs may be deduced from the junction of the lowest courses with the north tower. This tower was started before the portal. It had a buttress on the corner that projected into the space now occupied by the jamb. This buttress was cut back and the jambs inserted into it. At floor level, there is a 5mm. infill between the western face of the portal bases and the wall of the tower, Fig. VII–10. As this infill is made from the fine‑grained grossière used in the portal, rather than the berchére used in the tower, it must have been inserted with the portal. If the portal bases could have been inserted against the cut‑back buttress without this infill, the mis­alignment above the impost would not have occurred. If the error had been picked up at this stage, the erection crews had not been told when they came to lay the lower lintel. As is visible in Fig. VII–10, the portal plinth projects slightly to the west of the portal bases and the rest of the jamb, yet when the tower was cut back to receive it, the lower course was not given that additional trimming needed for the plinth, but was just cut in line with the edge of the jamb higher up, with the result that it was misplaced to the east by the amount of this projection. Under the plinth, there is a threshold which has upstands at each end to support the bases. The plinth was laid exactly over these upstands, showing that the real origins of this misplacement lay in the threshold. This crude installation suggests that once again the erecting gangs were at fault. Once the threshold was installed, the next courses were set out from it, without

110

checking, and the errors were left for those who came later to resolve as best they could. 3) Because the lower lintel was erected out of line, all the adjacent stones had to be adjusted as they were carved to realign the arches with the plane of the wall. The upper lintel with the four heralding angels, Figs. VII–11 and 12, has one projection along the top edge and another along the bot­tom. To compensate for the different inclinations, the outer faces of these two projections are not on the same plane. On the left the bottom is set in from the top, N in Fig VII–8 and in Fig VII–13, so it is flush with the face of the lower lintel, J in Fig VII–13. The set‑back of the lower projection from the edge of the archivolt with the personification of April is clearly shown in Fig. VII–11. On the right there is no set‑back between the upper projection and the lower, both of which line up with the lower lintel, Fig. VII–12 and M–K in Fig VII–13. The upper lintel was therefore adjusted to disguise the misplacement of the lower lintel while being correctly positioned to support the tympanum. The lower projection of the upper lintel has not been remade, as the edges of the angels’ wings hang over the face of the projection. The whole upper lintel was therefore carved after the lower lintel had been placed. 4) The right archivolt of Janus was rebated so it would lie over the face of the lintel in the normal way, I in Fig. VII–8. But the carved portion of the lower lintel can be seen continuing a little behind Janus, Fig. VII–12. This shows that the lower lintel is a little long for the doorway, but not so much that it had to be truncated like those in the Incarna­tion portal. It would therefore have been carved after the Incarnation lintels. 5) As Janus covers part of the carving of the lower lintel, the recess on the back which receives the end of the lintel is deeper than required. Thus the archivolt was not carved at the same time as the lintel, but after it. 6) The carving sequence is confirmed in the side face of the left hand archivolt of April, R in Fig. VII–8 and Fig. VII–11. The thick­ness of the archivolt is about 4 cm. deeper than the return in Janus, L in Fig. VII–8. As the additional depth was an adjust­ment for the misplacement of the lintel, April would also have been carved after the lower lintel had been placed. 7) On the left, the projections of the upper lintel butt onto April and the huntsman archivolts, Fig. VII–11, and the angels are exactly the right width for the space between the archivolts. This suggests that the upper lintel was carved either after or with either of them, but not before, when the lintel would have been carried behind the archi­volts in the normal way.

anomalies in the portals of chartres

Fig. VII–11. Ascension portal, left side lintels archivolts of April and the huntsman.

VII–13. Sections through the two lintels of the Ascension doorway. The left section is taken at the north end, looking north; the right section is taken at the south end, looking north.

111

Fig. VII–12. Ascension portal, right side lintels and archivolts of Janus, Capricorn and February.

112

Fig. VII–14. Junction between the upper lintel o the Ascension doorway, and the archivolts on the right.

Fig. VII–15. Diagrammatic elevation of the doorways with the campaigns numbered from 1 to 5.

anomalies in the portals of chartres

113

8) On the right the upper lintel is carried behind the archivolt, but because the archivolts were cut down, the lower part of the upper lintel continues behind Capricorn, as we would expect, while the upper part has been cut away so that the next stone, February, passes behind it, Figs. VII–12 and 14. This complex junction suggests that the upper lintel was carved after February. Thus it could have been carved with the tympanum of the Ascension, or perhaps in a phase between the archivolts and the tympanum. 9) The April archivolt is also slightly narrower north to south than Janus. The archivolt above with the hunts­man widens towards the top, Fig. VII–14, so that the rest of the archivolts could be the same width as Janus, Fig. VII–9. These were conscious modifications. The reason may be that the normal rebate for the lintel seen in Janus, marked L in Fig. VII–8, was omitted to permit the deeper return mentioned in item 6 above. As the Ascension lintels were later than those in the Incarnation doorway (item 4), and as April was carved after the Ascension lintel (item 6), April must have been one of the last of the lower archivolts to be carved. But since, like the lower archivolt registers in the Incarnation doorway, it has been reduced in height,12 the reduction and the carving of the tympana in both doors must have occurred after April was carved. 10) These manifold minor changes to width and depth show up in the relationship between the edges of the archivolts and the jambs under the capitals. While all those in the Incarnation portal are identical, suggesting that they were erected in one operation, on the right of the Ascension portal, Janus is placed closer to the door jamb, which may be why it overlaps the lintel, L in Fig. VII–8. If April had also overlapped the lintel, it would have had the same relation to the jambs. The change in this relationship, combined with the different junction to the lintel, suggests that the design may have been changed just here. Conclusion There are no anomalies similar to those on the side portals on the central door. Since the upper, lateral archi­volts could not have been completed without some of the central archivolts being in place, the gangs working on the side

12

Houvet, Portail Royal, fig. 36.

114

doorways would have also been at work on the central doorway. Most of the southern anomalies come from adjustments made to lintels, while most in the north come from the misplacement of the northern jamb, and both were affected by changes to the tympana. None of these three factors would seem to have applied to the central doorway, which appears homogeneous. A sequential arrangement of these anomalies, with the Incarnation lintels being carved before those under the Ascension, the Incarnation archivolts before its tympanum, and Janus with the lower archivolts of the Ascension portal, produces only one logical ordering of these elements into five phases of construction, Fig. VII–15. During erection, the southern end of the portals appears to have been a little more advanced than the north, perhaps by a course or two, though the lower Ascension lintel was placed before the Incarnation lintels. The evidence for this accumulates from the placement of the lower Incarnation lintel to the right, the stepped joint in the walling over this door, W in Fig. VII–7a, the joints in the apex archivolts, the mis­placement of the leftmost jamb of the Ascension portal, the order in which all four lateral lintels and archivolts were carved, and the relationship between the archivolts and the jambs. However, it is possible that the left portal was erected from left to right, just as the right portal had been erected right to left At least three designs had been prepared for the Incarnation portal. The first is evident in the wide lintels which had to be cut down to fit the second scheme. The second design was for a narrower and taller tympanum, as revealed in the archivolts. In the third design, the tym­panum was made shorter and wider than was intended in the archivolts, so it was wider than the lintels which had been cut down for the previous design. The upper archi­volts were carved before the tympanum was changed, at the same time as the two lower registers of archivolts in phase 3, or in the one following. These lower registers were truncated, the lintels installed, and the joints between the uppermost archivolts opened up when the portal was erected and the tympanum carved in phase 5. This last phase, in which so much of the sculpture was erected, would have been contemporary with the adjacent stone­work of the southern tower. The erection of the Ascension portal began earlier, in phase 3, with the installation of the lower lintel on mis­aligned imposts. The left archivolt with the huntsman may have been carved in phase 4, if the error had been recog­nized part‑way through the campaign, or it was carved with the upper Ascension lintel in phase 5.

anomalies in the portals of chartres

115

The order in which some of these stones were carved seems extraordinary. One would expect the tympana to have been carved before the archivolts, as even in the workshop the prior definition of the outline for the tym­panum would be useful in setting out the archivolts. This was the procedure in the transept doors sixty years later. Yet in the Incarnation portal at least, the order seems to have been reversed. Is it possible that other taller tympana had been carved in phase 3 and, while they were still lying in the workshop, rejected? Was there a dispute among the scholars of the Chapter on subject matter, resulting in a radical change to the program after most of it had been carved, and before it had been erected? If so, were these tympana thrown out, or will we find them some day? The problems left by the erection gangs suggest that they were not under the same control, and perhaps not even working at the same time, as the carvers. The complex coordination required to erect the portals was not unique to Chartres, so I would be more inclined to attribute some of these errors to changes in the personnel on site than to ignorance. The carving and erection of the portal were certainly complex. Three errors in erection occurred in the right lintels, the left plinth and lower lintel. Most of the adjustments in the archivolts were to compensate for these errors, at no inconsiderable cost and incon­venience. One would expect the master in charge to have avoided them if he had known. But if the masters had been changed, there is no reason why the next one would have known. The presence of these errors is prima facie evidence for the changes in crews. Excluding these erectional errors, there were three significant changes to the dimensions during the five phases. In some cases, sculpture was adjusted twice, once for an earlier anomaly, and then later altered again to adjust for later changes. These occurred in the course of construction and, all being major changes to the architectural layout, they suggest something more fundamental than erectional errors or adjustments to individual stones. Every time the dimensions and templates change, it implies a change in the master who designed them and the team which used them. As it is not reasonable to presume that sculpture and architecture of this quality would be produced by men who understood so little of their trade that they were constantly changing their minds, we must infer from these changes a number of crews, possibly three in the upper sections of the portals, each led by a different master, working one after the other rather than at the same time. Each crew seems to have worked towards one iconographic scheme, but to have addressed it in its own individual way. While the control of

116

subject matter and its arrangement was the responsibility of the clergy, the man­ner of execution was in the hands of the masters in charge. It will be seen that if some sculpture was carved after others had been erected, then the portal could not have been intended for some other place, at least not from the campaign in which the north jamb was misplaced. From this moment onwards, all the sculpture was carved to occupy its present position, and all the anomalies stem from changes to the layout and from control on‑site, rather than from the portals having been planned for some other place.

VIII Multiple contracting in the Saint‑Denis chevet *

I

T is often claimed that the chevet of the abbey church of Saint‑Denis was the first coherent example of the Gothic style, Figs. VIII–1, 2 and 3.1 Yet when we search for contempo­rary buildings that might contain the same format, we are at a loss to find them. Jean Bony’s comment is a recurring theme: “Is there no way of detecting something definite about the origins of the Saint‑Denis master?” 2 The chevet is isolated in its uniqueness: no other contemporary building possesses the qualities that have made this building so famous. It seems curious, if not astonishing, that those who created a work of such revolu­ tionary nature did not repeat this design elsewhere. I will argue that we do not find the Saint‑Denis design anywhere else because there was not one designer for the chevet, but two.3 The evidence suggests that each man interpreted Suger’s requirements in his own way to produce a unique building that contained something of both of them, and was therefore in its whole of neither. Sumner Crosby sensed this when he complained that the design shows “no sequence of specific details, no major structural or aesthetic system betray training or experience in one particular geographical lo­cality,” The qual­ities that each espoused are of course found elsewhere, but the particulars that made this chevet so interesting to us arose from a unique combination of talents in a combination that did not recur.

* Adapted from Gesta, xxxxii 1993, 42–62 1 Crosby, Saint-Denis; Clark, Saint‑Denis, 105–24, itemizes the scholarly work done so far. 2 Bony, Saint‑Denis, 131–42, 3 Crosby, Saint-Denis, 252.

118

Fig. VIII–1. Saint Denis, exterior of chevet.

the saint-denis chevet

Fig. VIII–2. Chevet ambulatory, interior of south side

Fig. VIII–3. Crypt, interior of north side.

119

120

The great contemporary buildings, or those built within a decade either side of Saint‑Denis, like the apses of Ferté‑Alais and Château‑Landon, or the ambulatories of Senlis and Saint‑Germer‑de‑Fly, do not have the same combination of lightness, geometric clarity and integrated vaults as Saint‑Denis. It is not until the am­bulatories of Notre‑Dame in Paris and Saint‑Remi in Reims in the 1160s that this is found again, over twenty years later. The arguments in this article are based on changes to the templates and the architectural elements that were made part‑way through construction, not to documents that tell us nothing of the architects. Unlike Gervase of Canterbury, Suger is tantalizingly silent on this subject. He writes about his craftsmen, his mosaics and the furniture, but never mentions the master masons. The plan of the Saint‑Denis chevet shows an open, spacious layout of nine chapels with wide windows, Fig. VIII–4. The middle row of ambulatory columns can be seen as either separating two aisles or marking the boundary to chapels that would not otherwise have any significant depth. The lightness of the design, in both the aisles and the walls, is apparent in the plan. We will begin our analysis by examining six changes to the architecture and the profiles in the chevet,4 and will later seek confirmation for our conclusions in con­temporary structures. First, over the window shafts and responds the impost extends continuously around the wall like a frieze. However, the imposts over the transverse arch shafts have been raised so that they rest on top of the frieze Fig. VIII–5. Second, both the profile and the height of the continuous impost left in Fig. VIII–6, are different from the upper impost, shown on the right.5 Different templates were therefore issued for these two courses, and the upper one was laid after the lower. Third, all the en-délit wall shafts flanking the windows and at the corners of the wall are made from two cylindrical stones with a slender band at the join like that in Fig. VIII–18. Though rings carved like torus moulds and placed at a constant height are quite common in the Paris Basin, small bands set at different heights are rare. The six shafts under the transverse arches are exceptional in being turned and installed in one piece. They are also larger in diameter.6 Fourth, the ribs Some consider these alterations the natural consequences of an exper­imental era. For example Henriet, Saint‑Etienne, 81–174. 5 Continuous impost 310 mm, transverse arch impost 350 mm high. 6 201 mm average compared to 180 mm. 4

the saint-denis chevet

Fig. VIII–4 . Saint Denis, plan of chevet after Crosby, Saint Denis. Fig. VIII–5. Chevet chapels, junction of imposts over wall and transverse arch shafts.

121

122

VIII–6 . Saint Denis, chevet chapels, profiles of imposts.

VIII–7. Chevet chapels, corbel under central rib.

Fig. VIII–8 . Saint Denis, dia‑ gram of the relation between the vault cells, ribs and plane of wall.

the saint-denis chevet

123

over the single shaft between the windows are supported on corbels which, like the transverse arch imposts, also sit on top of the con­tinuous impost Fig. VIII–7. Some of these corbels are simple and unadorned, as if finished in a hurry, for they were not carved with any of the delicate intricacy of the capitals un­derneath. Fifth, the face of the impost over this single shaft does not project beyond the curved wall‑plane, as an impost under a rib normally would. The rib, however, does project beyond this plane and, but for the corbel, would not have had support.. A similar situation in the ambulatory of Sens cathedral was finally explained as being “because no provision was made for ribs in the first design of the ambulatory.”7 Sixth, the torus under the wall shafts is also con­tinuous, like the impost. It passes behind the torus under the transverse arch shafts. It is significant that in some cases this continuous torus has been chiselled away to allow the transverse arch shaft to pass in front of it. Also the bases and capitals of the transverse arch shaft are the only ones replaced or restored in the nineteenth century, suggesting they may have been carved in a hurry, like the corbels, or carved so badly that the restorers felt compelled to re­make them.8 Taken together, these six details (the raised impost and their altered profiles, the bandless shafts, the corbels, the relationship between the wall shafts and the ribs, the cut‑back tori and the location of the restorations) suggest there had been a major change in the design just above the continuous impost level, in which the transverse arch shafts with their torus and impost blocks, and the corbels, were added. It was a natural place for such a change, for this is where the workmen would have set up the formwork for the voussoirs of the inner window arches. There is a major difference between laying stones in walls and laying vous­soirs in arches: the mortar joints in the walls may contract without destabilising the structure, which is not the case with arches. Any settlement in an arch would leave a gap between the arch and stonework supported on it. No stones could be placed on top of the arches until the voussoirs had settled into their final position, and this could only happen after the mortar had set and the formwork been struck. As medieval mortar was slow to dry, the formwork underneath every arch was left in place for about three months

Clark, Saint-Denis, 123, n. 66. The question was first raised by Severens, Early campaigns, 97–107. 8 For a similar case, see the Cerseuil apse shafts in James, Tem­plate‑makers, 226–28. 7

124

after the voussoirs had been laid.9 Even in a great monastery like Saint‑Denis where funds seem to have been abundant, the technical requirements of arch construction would have forced a pause in the construction. The work the masons could have turned their hand to during the setting period would have been limited, and in small projects like the Saint‑Denis chevet, men from other localities may have left the site and would not necessarily have returned after the setting period was over. They could have carved stones for use once the formwork was struck, but for carvers to be three months ahead of the layers would have caused intractable problems in the yard, such as where to store the finished stones and how to recognize where each was meant to go. If the masons were three months ahead of the layers after the crypt vaults, they would have been six months ahead at this level and a year ahead of the layers by the time the roof was reached. I do not believe it would have been a practical way to run a workshop. The addition of the six transverse arch shafts implied a change in the design. What was the new scheme? The wall shafts lie within the plane of the wall, and their imposts project only slightly in front of that plane. They were therefore intended to support arches lying within the plane of the wall, not arches projecting beyond the wall. The face of nearly all vault cells, be they groin, domical or ribbed, were usually aligned with the wall. As ribs project below the cells, they also project in front of the wall in Fig. VIII–8. Therefore shafts supporting ribs must also be placed in front of the plane of the wall. Over the corner A in Fig. VIII–9 the wall shaft can support the rib only be­cause the adjacent cell has been twisted inwards. But the single shaft between the windows lies within the arc of the wall and, without the corbel, would have supported only the outer arches over the windows as can be inferred from Fig. VIII–5. This implies a radically different scheme from the am­bulatory we see today Figs. VIII–2 and 10. As the wall shafts were designed to support wall arches rather than ribs, rib vaults could not have been intended when the walls were being built. Rather, they would have supported half‑domes as used in nearly every contemporary apse in the Paris Basin, such as Saint‑Martin‑des‑Champs, Fig. VIII–11, Saint-­Pierre in Montmartre, Saint‑Pierre in Chartres, Château­-Landun and the destroyed Temple Church

9

“XIX: Medieval mortar”. Mark, Gothic structure, 77, notes 13–15.

125

the saint-denis chevet

Fig. VIII–9 . Saint Denis, proposed first plan with semi-domical vault over chapel and compound piers.

Fig. VIII–10 . Saint Denis, isometric of chapel vaults and passages in the first plan.

Fig. VIII–11. Saint Martin des Champs, south ambulatory.

126

Fig. VIII–12 . Saint Martin-des Champs, ambulatory, plans of all the compound piers Fig. VIII–13. Morienval, chevet, plan of eastern compound piers.

in Paris,10 as well as in the crypt of Saint‑Denis itself, Figs. VIII–3 and 24. In changing the half‑dome into a rib vault the master had to add corbels so that the ribs would sit securely on the walls. The corner shafts at A in Fig. VIII–9 are the same diameter as those that flank the windows. Their function would therefore have been similar: to support arches that lay within the plane of the wall. The arches resting on these shafts would have covered passages between the chapels, and would have required matching shafts on a compound pier in the ambulatory. Clark was correct to comment that segment AA looked “more like a compound pier than a wall,” for that is precisely what the corner shafts were in­tended for.11 The arch resting on these shafts would proba­bly have been the same width as AA which was wider than a normal transverse arch. The pier would have had cor­ner shafts CC opposite to and matching AA, and would have terminated in a cluster of larger shafts for the adjacent ambulatory vaults. The passage AC would have allowed the clergy to walk between the chapels without being crowded by the throng of pilgrims that Suger describes as 10 11

Boussard, Philippe Auguste, 175–77; and Clark, Saint‑Denis, 120, n. 34. Clark, Saint‑Denis, 112.

the saint-denis chevet

127

filling the abbey.12 Passages supported on compound piers are found at Saint‑Martin‑des‑Champs13 and Morienval,14 Figs. VIII–12 and 13. When Saint‑Martin is compared to Saint-­Denis, it is significant that even though a number of masters set out this varied collection of piers (some have curved backs, some are square, some have keel shafts and some chamfers) none chose drums. The imposts over the Saint‑Denis ambulatory drums are higher than those over the windows, though the same height as those over the transverse arch shafts.15 Both drums and transverse arch imposts also have similar com­ plex profiles of roughly the same thickness Fig. VIII–2. They would therefore have been installed at the same period, af­ter the walls — a construction strategy that may have enabled Suger to preserve his earlier church a little longer. The latest moment for demolition would have been when the outer walls of the chapels had reached the impost height, during the pause between those two campaigns. The process of erection in which the walls are built first, and the interior drums set up later, was a common construction strategy at this time.16 Though con­siderable time usually elapsed between the completion of the walls and the insertion of the ambulatory piers and vaults, there was no perceptible pause at Saint‑Denis: The transverse arch shafts and the vaults were set up as soon as the mortar in the window arches would allow. Taken together, the design change from domical to rib vaults, the introduction of transverse arch shafts and cor­bels, the alteration of the

Panofsky, Abbot Suger, 87–89. A document suggests that the pilgrims were as uncomfortably crowded in the narrow spaces at Chartres as at Saint‑Denis — Lepinois and Merlet, Cartulaire, 260, which may be why the traditional deep chapels of the first plan were opened into a dou­ble ambulatory — James, Contractors, 166. 13 Saint‑Martin was begun perhaps ten years before Saint‑Denis: Lefèvre‑Pontalis, Saint‑Martin, 106–26; Ache, Prieuré royal, 5–15. The concept of a necklace of connected radiating chapels was first manifested at Saint‑Martin: Bony, French gothic, 49–52. 14 Berland, Morienval. 15 The present drums do not sit directly over the piers in the crypt, but are offset some 700 mm to the interior. If the pier had been placed more directly above the crypt supports, AA and CC would have framed a narrow opening between the chapels. Calculated from the exquisite photogrammetric studies published in Crosby, Royal abbey of Saint‑Denis, 247 and 258. 16 Called “Add‑a‑Chapel” in James, Template‑makers, 50–54. Examples within a decade or so of Saint‑Denis are Notre‑Dame, Paris with a ten-year pause, Noyon 15 years, Orbais more than 20 years and Saint‑Leu‑d’Esserent almost 30 years. 12

128

moulding profiles and the location of the banded shafts suggest a major change in direction. That all these occurred during a pause in the construction of some months while the mortar set suggests there could well have been a major reassessment of the design during this time. Whether Suger changed his masters and/or the brief is not clear from the evidence presented so far. How­ever, whether Suger or the builder was responsible, the evidence given later suggests this is the more likely alternative. Therefore I will call one the “Wall‑and‑Window Master” and his successor the “Rib‑Vault Master”. Imagine the process of construction at this point. The Wall‑and‑Window Master had built the curved walls of the chapels to the impost, including the banded shafts and the capitals that secure them. The impost course probably continued across the full thickness of the wall to protect the rubble infilling from the rain. The arches over the windows rest on this course. The Wall‑and‑Window Master would have set up the formwork for these window arches and laid the first row of voussoirs over them, and then left the site while the mortar set. Some three months later the next mas­ter, possibly stimulated by Suger himself, changed the plan by employing rib vaults. The Rib‑Vault Master naturally sets his imposts and corbels over the stones already in place, which was much easier than hacking out the existing imposts just to keep the new ones at the same level. This is as much as the present evidence will tell us. To discover more about how these changes happened we need to examine other buildings to see whether the characteris­tics of the Saint‑Denis walls can be found without

Fig. VIII–14 . Plan of windows at Saint Denis and Senlis.

the saint-denis chevet

129

those of the vaults and, conversely, whether the characteristics of the vaults occur without elements from the walls. Subject to certain conditions, we can say that if we can find the Saint-­Denis windows, wall shafts and attendant details occurring together as a group elsewhere they are all likely to have been erected by the same master. Similarly with the vaults, if we can find these templates and design concepts being used in other buildings they too may have been designed by one master. The crucial distinction that would allow us to separate the two stages of Saint‑Denis into the work of two masters would be if the Wall‑and‑Window Master’s ar­rangement were seldom, if ever, found in the same build­ings as the Rib‑Vault Master’s. The Wall-and-Window Master I have called the collected characteristics of a master a dossier.17 We know the Wall‑and‑Window Master from his generous windows, the shafts that flank them, his continu­ous tori and imposts and the men who carved his capitals. His dossier includes: • wide windows with the low sills and flanking shafts • sides of windows aligned towards the facing col­umns, Fig. VIII–14. • multiple shafts to articulate the wall surface • shafts en délit with small bands at different heights • continuous imposts and torus moulds18 • continuous external string‑course • intricate capitals19 • external buttresses turned into octagonal shafts • relieving arches outside and above the windows20 The chevet windows at Saint-Denis average 2,017 mm clear width between jambs. Similar generous and shaft-­flanked windows turn up in the chapels of Saint‑Leu­-d’Esserent,21 and Senlis Cathedral,22 Fig. VIII–15.

James, Template‑makers, 143–44. There is a subtle rhythm in its arrangement that may be a working procedure, Clark, Saint‑Denis, 112. Clark, Saint‑Germain-des‑Prés, 348–65. 19 Perhaps laid in pairs: see Clark, Saint‑Denis, 122, n. 43. 20 Relieving arches are not common. Also at Courmelles. Perhaps a dec­orative device derived from castle gateways. See Gardner, Castle building, 109–16. 21 Racinet, Saint‑Leu‑d’Esserent, 17–25. 22 Vermand, Notre‑Dame de Senlis. 17 18

130

They also turn up on the north wall of Notre‑Dame at Etampes.23 in the Glennes apse and in the Sens Cathedral ambulatory. Some, such as those in the ambulatory of Sens cathedral, are wider than those in Saint­-Denis.24 The windows in the central apse at Château-­Landun25 are 1,600 mm wide and designed just like those at Saint‑Denis, while those in the chapels of Saint‑Lomer in Blois26 are 1,700 mm wide but with much heavier flanking shafts.27 As the Sens ambulatory, the Saint‑Martin and Blois chapels, and the Château‑Landun apse may all be earlier than Saint‑Denis, builders were constructing wide windows before Suger began his. That groin vaults were built or implied in all of these earlier buildings shows it was consistent for them to have been planned for Saint‑Denis. At Saint‑Denis, Glennes and Senlis the sides of the win­dows have been aligned in a most curious and individual way, Fig. VIII–14. Normally these sides would be parallel to one another or radial to the centre of the chapel, but here they are aligned towards the centres of the opposite piers. It is an idiosyncratic detail that bespeaks an individual and eccen­tric designer. Low sills are rare, but are found in the chapels of Saint‑Leu and Senlis. Also, all the bases of their window shafts rest within a recess just below the level of the sill, as in Saint‑Denis. Continuous im­posts were used in the Sens ambulatory, at Glennes and in the Ferté-Alais apse,28 both of which have exceptionally wide windows. Sens also has a single shaft between the windows like Saint‑Denis. Window and vaulting imposts are continuous inside and out in the Saint‑Leu chapels, but only on the outside in the Senlis windows, Fig. VIII–15. From the foliate style of the capitals I am inclined to place the vaults of Saint‑Leu before those at Senlis, but after Saint‑Denis.29 In all of these chapels the vaulting imposts are at the same level as the window im­posts. Shafts in two lengths with slender bands at the join are rare in the

Houlet, Etampes. For a list of large windows see James, Template‑makers, 106, to which should be added the transept windows at Winchester which are 1,620 mm wide, the lateral chapel windows at Saint‑Lomer, Blois which are 1,700 mm wide, and one of the chapel windows at Saint-­Martin‑des‑Champs which is 2,100 mm wide. 25 Deshoulieres, Notre‑Dame, 242–58 26 Everett, Saint‑Lomer‑de-­Blois. 27 Some capitals in the Blois chapels are also like those in Saint‑Denis. 28 Aubert, Ferté‑Alais, 367–68, probably started before Saint‑Denis. 29 Saint‑Leu was an Add‑a‑Chapel scheme with walls of 1150±, while the Senlis walls were closer to 1155. 23 24

the saint-denis chevet

131

Fig. VIII–15. Saint-Leu-d’Esserent, chevet chapel

Fig. VIII–16. Etampes, Notre Dame, south porch, shaft on west.

132 Map VIII–1. Paris Basin showing those churches with some of the profiles and elements from the dossier of the Wall and-Window Master; 1 Sens Cathedral. 2 Blois, Saint Lomer, 3 Château Landun, Notre Dame, 4 Ferté Alais, 5 Saint Denis, 6 Etampes, Notre Dame, 7 Provins, Saint Ayoul, 8 Saint Leu d’Esserent, 9 Jouy le Comte, 10 Senlis Cathedral, 11 Villeneuve sur Verberie, 12 Glennes.

Table VIII–1. Buildings with characteristic walls found in the ambulatory walls of Saint-Denis

the saint-denis chevet

133

Paris Basin, whereas rings formed from thick torus‑like profiles were more common. Shafts with thin bands were installed at Saint‑Leu, the Senlis choir and western doorway,30 the Sens ambulatory wall, flanking the porches of Notre‑Dame in Etampes, Fig. VIII–16, and Saint-­Ayoul in Provins, in the Villeneuve‑sur‑Verberie apse and the north arm of Jouy‑le‑Comte. Many of the capitals in these buildings are similar to those carved below the con­ tinuous impost at Saint‑Denis.31 These buildings are listed in Table VIII–1.32 They comprise a group containing, in one part or another, characteristics from the Wall‑and‑Window Master’s dossier. There are not always enough correspondences to show that this master definitely worked on them all, but what we have is indica­tive that he may have. Map VIII–1, showing the location and approximate construction order of these places, suggests that he came to the Ile‑de‑France from the south, perhaps central France, and gradually worked his way northwards. Rib-Vault Master Turning now to the Rib‑Vault Master, we know him mainly from his uncomplicated vault profiles and the boss with a small flower and a hole through the centre. His dossier includes: • vault arches with simple, slender profiles • ribs erected over semi‑circular formwork • encasing arches pointed and stilted so that crowns are level • arches set out from a radius of 2,070 mm33 • boss with small hole drilled through it surrounded by tiny petals34 • lightweight drum piers

Men building the chapels also worked in the west­ern bays, “X: Cathédrale de Senlis.”. None of the Wall-­and‑Window Master’s capitals have been recarved or replaced, which was Crosby’s concern, Saint‑Denis, 499, n. 8. 32 Table 1 in­cludes only those parts of Saint‑Denis where elements from this dos­sier occur. For a full dossier analysis the table would be broadened to include other zones such as vaults and footings, as in James, Template‑makers, ch. 8. 33 This master may have used the Roman foot, for this radius, which is also used in the crypt arches, is exactly 7 of these feet. James, Contractors, 74–76. 34 This would allow the centring to be struck by pulling out wedges from above the ribs with a rope passing through this hole. Found in less than 12 percent of rib vaulted churches before 1150. 30 31

134

•

square plinths, with the edges of the corner chamfers maintained parallel along the inclined slopes.35 • bottom corners of imposts over the drums chamfered • tall torus moulds and imposts • tops of capitals notched Some of the Saint‑Denis vaulting profiles are found in the south transept vaults of Saint‑Etienne in Beauvais,36 in the Cambronne nave aisles,37 Saint‑Christophe just north of Senlis,38 the Senlis Cathedral chapels and ambulatory, the Jouy‑le‑Comte apse, the Condécourt north arm, the Coulanges aisles, the Marolles‑en‑Brie apse,39 Airaines north of Beauvais40 and the Saint‑Germer‑de‑Fly choir.41 The miniature boss with a simple floral ring and a tiny hole drilled right through is a fairly rare item in the Paris Basin. It is found in the choir of Saint‑Germer‑de‑Fly, at Saint‑Christophe, Marolles and Condécourt, and in the Senlis ambulatory. Two rows of thin drums in ambulatories were repeated only twice in the next fifty years, suggesting that people were not comfortable with the idea. This reinforces Bony’s sug­gestion that these drums reflected Suger’s wish rather than the builder’s.42 What pier type this master would have other­wise used cannot be ascertained from Saint‑Denis alone. Similar capitals may be found in many of these buildings, and their foliage is generally unlike the more complex carv­ing used by the Wall‑and‑Window Master. Also, nearly all the Rib‑Vault Master’s capitals have notched upper faces, called énchancré, which the other master almost never uses: A notched capital may be seen on the left in Fig. VIII–5. Nearly all the places where we find the Wall‑and-­Window Master’s dossier are different from those where we find the Rib‑Vault Master’s. Apart

Normally, the chamfer gets wider as the stone gets larger, for they are set out radially from some point within the stone. But in these bases the corners of the chamfer along the splay are parallel as the chamfer has been measured from the outside of the stone. The geometric pro­cess is unusual, as is the mind‑set behind it. 36 McGee, Saint‑Etienne, 20–31. 37 Plaignieux, Cambronne. 38 Racinet, Picardie, 199–230. 39 Gardner, Saint‑Denis, 7–31. 40 Bony, French gothic, 35–37. 41 Lambert, Saint‑Germer, 47–63. 42 To recreate the Constantinian basilica, Bony, Saint-Denis, 140, Suger even wanted to import marble columns from Rome — Panofsky, Suger, 213. 35

the saint-denis chevet

135

from a few peripatetic capital carvers who seem to move from master to master, only once do groups of items from the Wall‑and‑Window Master’s dossier appear in buildings where we find the Rib‑Vault Master’s: in the Senlis Cathedral ambulatory. The presence of both dossiers at Senlis occurred for the same reason as at Saint‑Denis: the Wall‑and‑Window Mas­ter worked on the windows and the Rib‑Vault Master, at a later stage, helped to build the vaults.43 In none of the other buildings do we find both dossiers together. The windows at Beauvais, Saint‑Christophe, Marolles and Condécourt are nothing like those in the Wall‑and‑Window Master’s dossier, and vice versa. When comparing buildings such as Saint‑Gemer‑de‑Fly with Saint‑Denis I have restricted my observations to the vaults. The ground plan and the design of the piers have little in common, and it is only the vault zones which show similarities. Just as the Rib‑Vault Master set the vaults of Saint‑Denis over walls designed by the Wall‑and‑Window Master, so the same Rib‑Vault Master may have built the vaults at Saint‑Germer over piers and walls by some other crew. Here is a group of buildings with some of the charac­teristics found in the Rib‑Vault Master’s dossier, though there are not always sufficient correspondences to indicate that he definitely worked on all of them. These buildings are listed in Table VIII–2, with the major items that are also found at Saint‑Denis noted. Map VIII–2, showing the location and approximate construction order of these buildings, sug­gests that this master moved into the Paris Basin from Picardy, or maybe even further north from England.44 A word on this lateral suggestion: The Saint‑Denis profiles and the form of the vault with its almost level crown are not part of the Italian/west­ern French tradition with their domical cells, nor are the profiles as complex as the majority of French rib vaults. The Saint‑Denis arch profiles are similar to those used in Durham and Peterborough prior to 1120 and in fifty percent of English rib vaults before 1150. Only four percent of Paris Basin ribs are similar. The technique of drilling holes through the boss originated at Durham in 1100 and occurs in 38 per­cent of English vaults compared

At Senlis various discrepancies in the choir suggest three campaigns in which the Wall‑and‑Window Master began the windows, carved some of the capitals and the banded window shafts, and the Rib‑Vault Master built the vaults with the same bosses found at Saint‑Denis. 44 Crosby, Royal abbey of Saint‑Denis, 252. 43

136 Map VIII–2. Paris Basin showing those churches with some of the profiles and elements from the dossier of the Rib Vault Master. 1 Airaines, 2 Beauvais, Saint-Etienne, 3 Cambronne lès Clermont, 4 Marolles en Brie, 5 Saint-Martin-des-Champs, 6 Saint Denis, 7 Coulanges, 8 Jouy-le-Comte, 9 Saint Christophe, 10 Senlis Cathedral, 11 Provins, Saint Quiriace, 12 Condécourt.

Table VIII–2. Buildings with characteristics found in the ambulatory vaults of Saint-Denis

the saint-denis chevet

137

Figs. VIII–17 and 18. Saint-Denis, crypt, exterior buttresses so south chapels; and to all the others to the east and north.

to half that percentage in the Paris Basin. Of nine English buildings with similar rib profiles under construction before 1140 eight have drum piers rather than the compound piers almost universally used in France. Therefore the Saint‑Denis ambulatory drums and vaults are closer to the English vaulting tradition than any other. The details and profiles in the entry porch to the Bristol Cathedral Chapter House are so close to those at Saint­-Denis that this may be the Rib‑Vault Master’s last English work be­fore coming to Saint‑Denis Returning to Saint-Denis We have now isolated the junction between two cam­paigns at Saint‑Denis, formed a dossier for each campaign, and then found each of these dossiers separately present in other buildings. Does this confirm the proposal that each campaign was the work of a separate master? Or is there some other interpretation? Before addressing this ques­tion, we should consider the interesting, if less convincing, evidence for three other junctions in the chevet crypt at Saint‑Denis. First, the crypt buttresses on the south are different from all the others, Figs. VIII–17 and 18. Where the other buttresses are flanked by small

138

pilasters to support the reliev­ing arches over the windows, these buttresses have none, Fig. VIII–19. Instead the buttress is a plain rectangle, without any projections at all.45 Furthermore, Fig. VIII–20, the axes through the northern buttresses are set toward two centres in the middle of the apse, whereas the axis through the southern buttress has been shifted to the east of both these centres. Lastly the width of the five northern buttresses is 650 mm, whereas the southern buttress has a width of only 560 mm: a noticeable difference of some four inches. The combination of width, altered axes and the omission of pi­lasters indicates there was a change in the templates used for this location. The omission of the pilasters was a major alteration, as without them there were no supports for the relieving arches. There are no arches at this level, and therefore no technical reason for a break in the construc­tion. Is it really possible that the abbey ran out of funds and work was delayed until more could be raised? Second, the reinstatement of the relieving arches over the pilaster‑less buttresses in the south suggests either the presence of someone who wanted to put them back, or a realization by the master that they were still an appropriate idea. As the corbels for these arches in the south coincide with the mortar‑drying pause that would have occurred over the crypt window arches, there may have been a change in masters at this level. Third, the layout and dimensions of the wall shafts differ in the two upstairs lateral chapels to the west of the chevet Fig. VIII–2. It was normal for the diameter of shafts to form a hierarchy from the largest under the trans­verse and arcade arches to the smallest under the formerets, and it is therefore tempting to presume that shafts of similar dimension and location were intended to support arches with a similar purpose. Different functions, be they in piers or in buttresses, are almost universally distinguished by being supported on shafts of different diameters. So rib shafts will almost never have the same diameter as formeret or trans­verse arch shafts. The direction of the arches that would have suited the sizes of these shafts is marked with arrowed lines in Fig. VIII–21. Both lateral chapels have small respond and window shafts, and larger ones that were, presumably, for ribs. However, in the south chapel there are 45

First noted by Crosby who thought they reflected the difficulties of aligning the centres of the ambulatory shafts with the offset centres of the chapels: Clark, Saint‑Denis, 120, n. 32 suggests these changes might have been necessitated by a nearby building.

the saint-denis chevet

139

Fig. VIII–19 . Saint Denis, plan of crypt showing width of buttresses. Dots on the detail of the southern chapel indicate the absent pilasters visible in Fig. VIII-17.

Fig. VIII–20 . Saint Denis, plan of crypt showing inclination of buttresses.

less respond shafts than in the north, and the sizes of the northern chapel shafts are markedly different from those in the southern chapel, marked on the figure.46 46

On the north, the window and respond shafts are 166 and 178 mm in diameter, while the larger rib shafts are 247 mm, which is 40 percent larger than the former. In the south the formeret shafts measure 187 and 231 mm, only 24 percent larger. Compared to the north, the southern responds are larger and the rib shafts smaller. Variation in measurements of diameters ±1.5 mm. Maximum range in chapel en délit shafts are 160‑166 mm and, excluding a few mavericks, 174­–184 mm.

140

Fig. VIII–21 . Saint Denis, plan of lateral chapels. Arrowed lines indicate largest shafts and actual ribs shown dashed.

Lastly, the bases are different in profile and in height, Fig. VIII–22. The base profile on the left is found under all the shafts to the northern and eastern chapels. The flatter but similar profile in the centre was used for the continuation of the torus profile along the wall. Both profiles are the same height. The profile on the right, on the other hand, applies only to the bases to the west of D in Fig. VIII–21. It is not as high as the profiles on the left, and has not been continued as a frieze along the wall, but is limited to the base of the shaft. These changes of shaft layout and diameters, and of torus profiles indicate that the templates from which the stones were cut were changed at D. If the construction program in the northern end of the building was a little ahead of the southern, then the two lowest courses in the north and east would have been laid in the earlier campaign, and this crew would have proceeded

the saint-denis chevet

141

no further in the south than the floor level.47 The thick line in Fig. VIII–23 indicates the location of this break on the north and east. It lies just above the torus mould on the interior. On the exterior it coincides with the top of the keystone to the relieving arch over the crypt windows Fig. VIII–17: a natural place to pause while the mortar between the voussoirs was setting. Was there a change in master masons? It is now time to reconsider the earlier hypothesis that the major change in design between the walls and the vaults of the chevet and, to a lesser extent, the junctions underneath should be attributed to a change in master masons. With one exception, the campaign breaks coin­cide with arches over windows or under vaults just where the slow‑setting mortar would have imposed unavoidable breaks in the construction process. If changes had occurred in a random manner we could interpret them as evidence that the master modified his design as he went along. But with changes coinciding almost exclusively with pauses in the work while the mortar set we need to examine other possibilities. The role of the template‑maker is crucial for interpret­ing the evidence. Shelby and others have shown quite clearly that the master mason in charge of the works was also re­sponsible for designing the templates.48 If the profiles and details are changed, then it is certain that the templates from which they were cut were also changed. If those changes represent different aesthetic attitudes to design and different geometric processes, then it is likely that different template-­makers were involved. However, this is not the same as saying that the master in charge also changed: it is not as simple as that, for there are alternative explanations. Let us examine some of these alternatives. The evidence for two, and perhaps five, sets of discon­tinuities in a work that was completed in less than four years raises difficult questions about the

Crosby, Saint‑Denis, 248 and Clark, Saint‑Denis, 112 both argue that the construction of the north side was more advanced than the south. If reversed there would be only minimal modification of these conclusions, except in the crypt where, if the southern buttress was an experiment that was altered in later buttresses, my first postu­lated break in the construction could be eliminated. 48 Shelby, Medieval templets, 140–52 and Shelby, Master mason, 393–5; Bucher, Design methods, 37–51; Harvey, Yvele, 51–60; and Harvey, Medieval architect. 47

142

control of the workmen and the input of ideas. All analysis is limited by the observation that the changes are not random, but occur in bands, and that they involved major alterations, not mi­nor adjustments. Since the design of a major work like Saint‑Denis could have been subject to many alterations, some of them radical, we need to consider which methods of employment and site control would involve breaks in construction and changes in design every six months or so. Either the master had absolute control over the templates, or those under him were able to issue their own, or some­thing in between. In the first scenario where the master had absolute control over the templates, the changes must indi­cate either that one master was replaced by another, or there was a group of consultant masters who were all involved at one time or another. In the second case in which some, or many of the men under the master, could issue their own templates there are a number of possibilities: either each sub‑master template‑maker was changed from time to time, and usually during pauses in the construction, or there was a fairly liquid situation in which different gang foremen would issue templates to their men which might be put aside on departure, leaving the capo‑master in sole charge. Suger was well‑travelled and observant, if conservative.49 He selected his workmen carefully, and had the connections to bring people from afar. One interpretation of the evidence is that Suger may not have had the time to decide what he wanted until the work was waiting on his attention. He may have hired the Wall‑and‑Window Master to give him big windows, knowing this man was capable of building them,50 and, realizing that this master would have to leave the site once the window arches were in place, he may not have realized that he was preparing to build compound piers. When employing the next man for his acknowledged skill in vaulting, Suger told him why he wanted drums and then left him to solve the resulting

Van der Meulen suggests that Suger was an aesthetic conservative as the, “only reference to his actually influencing the design is his demand that a mosaic be placed in the northern tympanum”, though contrary to modern custom. He then asks “was this uniquely con­ servative approach likely to have been instrumental in the genesis of a new style?” — Van der Meulen, Architectural context, 531. 50 The unique wide windows in the crypt were unheard of before, and were probably installed on Suger’s instructions. 49

the saint-denis chevet

Fig. VIII–22 . Saint Denis, chevet chapels, profiles of the bases.

Fig. VIII–23 . Saint Denis, chevet chapels, section through the external wall showing tori and window sills.

143

144

Fig. VIII–24 . Saint Denis, section through chevet, with five zones controlled by different masters marked on the right. Number 5 includes the ambulatory piers, and is the Rib Vault Master.

Fig. VIII–25 . Saint Denis, sketch of southern crypt in Fig. VIII–17 showing the likely location of the stepped joint “D” and the breaks between the campaigns. Number 4 is the Wall and Window Master.

the saint-denis chevet

145

build­ing problems as best he could.51 Alternatively, the abbot may have retained a stable of masters and consulted them whenever a major decision had to be taken. If so, then the evidence suggests that the acknowledged favourite of the moment took charge, and from then on his templates sup­ planted those of his predecessor. Or the abbot may have suddenly decided to change the brief and only thought through the problem during the quiet time when there were no workmen on the site. One capo‑master may have prepared a general overall plan, but he may have been absent most of the time leaving the day to day detailing and template‑making to either a subordinate or a sub‑contractor master.52 Or a master may have been a consultant with some influence over many stages of the design, but little input into the templates which may have been prepared by many men. Or, while the mortar was setting, some of the layers and masons may have stayed on while the master may have gone elsewhere, leaving them in control of the templates for a while.53 There is evidence for all these possibilities. It is even possible that the site was divided between two masters according to the number of men in each crew, with the smaller crew constructing to the west of D. This may explain the vertical junctions at D but not the horizontal junction at the impost level, nor the lack of communication between the two that permitted the ‘misaligned’ southern buttress and the changes to the lateral chapels. Buttresses and shafts have a structural function and it is doubtful that a master would be content to let a co‑ or sub‑contractor alter their sizes and orientations at will, especially where the work was experimental. This suggests a more complex picture than I have ear­lier argued for, in which there was constant interaction be­tween client and master, between both and visiting masters, and between masters and carvers. I have written that there was no architect at Chartres Cathedral, but only client and current

“Conceptualisation and execution were largely independent. Planning was done by a learned adviser‑either the patron or head of a monas­tic or secular shop; realisation was the work of the craftsman”: Kessler, Medieval Art, 181. 52 At Chartres, Ruby may have been the master for the whole of the up­per parts of the north transept including the rose, but employed as sub‑contractor a crude local team I called Green to do the bulk walling. Green pleased himself about profiles for stair windows and drip moulds, but Ruby controlled the overall form and the details of the rose window: James, Contractors, 491‑95. 53 This happened in the north porch at Chartres, James, Contractors, 44–45. 51

146

builder, and that the role played by the modern ar­chitect was divided between them so that the client had an intimate control over the finances and the builder had sole control over the site work. As is evident here, I no longer believe that the situation was so simple.54 Nevertheless the issue is crucial. As the survey I have conducted has shown, there is not one building among the 1,400 churches of the Paris Basin from the 1100–1240 period without similar changes. Even in the well‑funded Sainte‑Chapelle the profiles were created by a number of template‑makers, though there may have been a capo‑master in charge of the overall design.55 Whatever possibilities we examine, the discrepancies discussed here show two things: first, that client and/or mas­ter mason altered the design a number of times and, second, that the same template‑makers were not in charge through­out these three years. My personal opinion is that more than one master/contractor worked on the site, and that their authority and Suger’s were both tempered through their in­teraction with one another and with the opinions of other masters, both outside consultants and those who had previ­ously worked on the site. Together these factors resulted in the five distinct sets of design changes in the chevet, summarised in Fig. VIII–24. For clarity in discussion I shall continue to refer to each campaign as being the work of a different master, though with the reservations just discussed that decision‑making was influenced by complex interactions. The first master laid out the foundations of the crypt, starting on the north interior with octagonal pilasters, Fig. VIII–3, and perhaps the Suger‑inspired wide windows, which are three‑fifths the width of those in the chevet.56 He may have laid three or four courses above the pavement in the north, stepping down to ground level at D on the south.57

James, Contractors, 544–50. 1 am particularly grateful for the discussions with Caroline Bruzelius, François Bucher, Bill Clark, Michael Davis, Stephen Murray and Lon Shelby on this issue. See the perti­nent arguments in Shelby, Chartres, 173–76, and in James, Template‑makers, ch. 9. 55 James, Template‑makers, ch. 4. 56 Their low sills, virtually set at floor level, were the work of restorer Jules Formigé in the 1850s. They may have been a little higher. 57 This master may have built the upper central chapel in the west narthex, for it was the last part con­structed before the consecration that produced the holy water for the chevet footings. Gardner, Saint‑Denis, 587, n. 44, lists some of the connections. 54

the saint-denis chevet

147

The next master left out the pilasters for the relieving arches, continued the crypt walls, carved the capitals in­side,58 arched over the windows and began the arches for the groin vaults. If all these arches had been built together the crypt would have been a thicket of props and scaffolding: a sufficient inducement to leave the site, quite apart from the wet mortar. After the mortar‑setting pause over these crypt arches, there may have been a third master, because corbels were installed where the pilasters had been omitted to support the relieving arches, Fig. VIII–18, or the second may have realized his mistake in leaving the pilasters out. He completed the groin vaults and began the outer walls of the chevet. This included the layout for the northern lateral chapel and the torus bases as far as D. He did not start the string‑course or the sills and jambs of the up­per chapel windows, for these follow the same template all around the chevet. These campaigns in the crypt are sum­marized in Fig. VIII–25. This master was succeeded by the Wall‑and‑Window Master who detailed the wide windows and their framing shafts, continued the walls up to the level of the frieze‑like imposts and probably built the first row of arches over these windows. He was in his turn succeeded by the Rib­Vault Master who inserted the transverse arch shafts, the drum piers and set up the ribs of the vault.59 The final stage of the vaults, the cells and the uppermost courses of the wall to cornice level, may have been the work of one fur­ther master after a pause while the mortar in the ribs set.60 We have to conclude that the seminal originality of the Saint‑Denis chevet lay not in the genius of one master, but in an accidental concatenation, a fluke of appointments and opinions that were influenced by Suger’s passion for light and for re‑creating the past.61 Though his choice of masters may have been deliberate, it seems more likely that masters were chosen from whatever pool of men were currently unoccupied. In this case the greatness

Heavily restored by Debret in the early nineteenth century. Clark may be correct in suggesting Merovingan sources, Clark, Saint‑Denis, 121. 59 There are no bands on the external window shafts from the middle of the second southern chapel westwards, and internally from the middle of the first chapel westwards: lying over junction “D” this suggests that these were installed by the Rib‑Vault Master. 60 Suger states that there was a time when neither the cells nor any scaffolding was in place, nor anything that could support the ribs, Panofsky, Suger, 109. Crosby, Saint‑Denis, 257– 58, suggests that “forms could be moved from bay to bay”, but where mortar required months to harden such reuse would be limited. 61 In keeping with Bony on the origin of Gothic in Bony, Genesis of Gothic, 17–31. 58

148

of Saint‑Denis stems from the happy conjunction of three men: the fourth master from the south who opened the walls with gigantic windows, the fifth from the north who lightened the spaces with drum columns and covered them with a unified system of rib vaults, and their client, Abbot Suger. The intellectual clarity of the chevet, unlike the mani­fest confusion at Saint‑Martin‑des‑Champs, comes in part from the horizontal location of the junction between these two crucial building campaigns. No such clarity existed at Saint‑Martin, where vertical breaks occurring in nearly every bay ensured that every pier would be evolved from different templates. At Saint‑Denis, on the other hand, nearly the entire wall from north to south was the work of one team, and the whole of the interior space from drums to vaults was the work of another. Much of the satisfaction we feel in this building stems from the architectural integrity each master was able to retain within his own sphere of influence. The later works by these masters betray little of what they may have learned from their time at Saint‑Denis or their interactions with Suger and his ideas. In the decades that followed Saint‑Denis the Wall‑and‑Window Master’s influence was negligible compared to his successor’s, for windows became smaller in the 1160s and 1170s than they were in the 1140s.62 It was his successor, the Rib‑Vault Master, who introduced those space‑creating elements that we so admire: the two rows of slender drum piers and the integrated rib vaulting. If any of the masters employed on the abbey can be called the first Gothic architect, it is he, even though the windows that created its luminosity were the work of his predecessor. It is the Rib‑Vault Master’s rib forms, with their Anglo‑Norman level crowns and round ribs resting on drum columns, that inspired the next generation of masons.

62

James, Template‑makers, 107.

IX Could Suger have built the choir of Saint-Denis in four years? *

I

T has generally been argued that Suger constructed the whole choir from crypt to high vaults in the four years between 15 July, 1140 and 12 June, 1144, though he himself claimed it was only three years and three months. This is based on the reading of the text that the work was completed “ . . . from the crypt below to the summit of the vaults above, elaborated with the variety of so many arches and columns, including even the consummation of the roof.” This has been taken to mean the summit of the high vaults “above” the altar, and the “roof ” assumed to be the main roof above the high vaults. Sumner Crosby printed a well-known section that showed an elevation not unlike Saint-Germain-des-Prés with a gallery, clerestory and high vault.1 Yet Crosby was not totally satisfied by this, for he wrote that “even my own enthusiasm for Suger’s abilities questions the possibility of his erecting such a complex structure, especially one so novel, in such a short time”.2 I feel he went along with the belief from his loyalty to the French histography and his colleagues, an impression I gained when I spoke with him on the subject at Yale in the 1980s. For myself, I have not been happy with this belief. My gut feeling from my experience as a builder and architect tells me it could not have been finished in that time. Here is the evidence that convinced me: 1. There no stonework above the aisle vaults from the 1140s. The twelfth-century stones visible within the roof over the vaults all stop at the * From Avista, x 1998, 23–25. 1 Crosby, Saint-Denis, 264. 2 Crosby, Saint-Denis, 217–18.

150

same level. There is no sign of old mortar or other remains of masonry that had been removed. 2. Lime mortar needs a certain time to set before the formwork under arches can be struck. This is generally six to eight weeks, depending on temperature. During the winter more time is needed. Even eight weeks would be conservative under snow and frost. Once struck, the arches need to be left for at least two weeks to settle before being loaded. That is, the small cracks between the voussoirs need to be given time to close up under the weight of the stones so that there will be no further creep when they cells and walls over them are laid up. Failure to do this will result in the arch settling after the stones above have been placed, leaving a gap between the two.3 3. The more obvious delays would have been (a) In the crypt the formwork for the window and vault arches could have been set up at the one time, limiting the delay at that level to 8–10 weeks. (b) The cells for the lower vaults would have needed similar setting time, but as they are solid, probably no more than 6 weeks. (c) The design in the ambulatory is such that the window arches would have been constructed first, and when completed with the additional stones over the arches, ready to support the vaults. Delay 8–10 weeks. (d) The ribs were next, and as we know from the description of the storm, were “not supported by any scaffolding nor resting on any props”. They would have just had the formwork under them removed and were waiting to settle without props before having the cells placed onto them. Delay at this level, another 8–10 weeks. Since the storm was in winter, January, the longer time is the more probable. This is a total of 30 to 36 weeks. 4. We also need to allow time for setting out, demolishing the old building, and laying the foundations. Some work had already been accomplished by July 15, 1140 when the first stones were laid. Let us presume a further eight weeks for the other preliminary tasks, including bringing the foundations up to ground level. 5. Finally we have to allow for the erection of vaulting formwork, arches and cells. Though some of this could be prepared while the walls were being erected, some would have taken all available masons and erecting

3

“XIX: Medieval mortar”.

the choir in four years

151

gangs away from the walls. Lets say four weeks in the crypt, eight in the ambulatory and ten in the high vaults. 6. Adding these delays together, the work had to pause for fortyfour weeks minimum if only the ambulatory were built, or eleven months. BUT if the gallery was also built, with the clerestory windows and high vaults and the cells over them, we would have to expect further delays of at least another twenty-six weeks. In a four-year project this leaves only 35 months to complete the walls of the lower building, and 29 to 30 months for the taller. 7. To see if this would have been enough to construct the whole fabric of a three-storey building plus crypt, let’s count courses above the ground. From the crypt to the top of the ambulatory there are 43 courses, and to the top of a hypothetical high vault (scaling off Crosby’s drawing) there would be 80. This would allow sixteen working days for each course in the first instance, but only seven-and-a-half days in the second (allowing for five working days each week, to except Sundays and the manifold saint days as well). It does not improve matters that the taller the building the more time is lost in erecting scaffolding, constructing cranes, lifting stones and the general difficulties in working at great heights. No matter how many people were employed, to carve and lay one course, often with complex mouldings, there is a physical limit to how many men can work in a given space, how many cranes were available, and so on. I personally have to conclude that it is extremely unlikely, if not impossible, for Suger to have completed the whole choir to the high vaults in the space of four years. So, what of the words he wrote? Take each phrase separately:4 1. “ . . . to the summit of the vaults above” — above what? The altar or the crypt? If the former he meant the high vaults, if the latter the vaults of the ambulatory. 2. “ . . . with the variety of so many arches” which words do not readily describe the simple four-part arrangement of the ribs of a main vault. 3. “ . . . and columns”. Now, where are the columns under the high vault? Surely he would have referred to the gallery or the clerestory windows. To refer to columns suggests he was speaking about the ambulatory. 4. “ . . . including even the consummation of the roof ”. If he is referring to the main roof this “even” is curious, as roofs were erected before the high 4

Panofsky, Suger,101–3.

152

vaults to keep out the rain and carry the lifting gear, so it is more likely he would have used “even” to refer to the vaults. But if he was referring to a temporary roof at aisle level so ritual could continue while the upper parts were being completed, the word “even” would be absolutely appropriate. There are many precedents for temporary roofs. From this I paraphrase what I think Suger may have meant: “ . . . from the crypt below to the summit of the ambulatory vaults above those of the crypt, elaborated with the variety of so many arches and columns, including even the consummation of the temporary roof.” If we can accept this argument, then the nature of Suger’s upper stories remain unknown. He may have intended a simple wall and small windows like Saint-Martin-des-Champs, or a wooden roof, or any other contemporary solution. He may have intended six-part or four-part vaults, but we will never know which. Whatever it was to be, the choir would have remained, for almost a century, a low-roofed and relatively small building without much light, with a temporary roof at the same level as the roof over the ancient nave and that over the recently completed narthex. I wonder if the low ceiling over the choir would explain why no other architect copied the wide windows and spaciousness for another twenty years? After Suger’s death when the monks withdrew the monastery from active participation in the political and administrative affairs of the kingdom, they seemed content to remain within the unfinished structure. Keeping noisy workmen off the site would, as we all known from our own building operations, ensure a contented and peaceful life! Their decision to build a tall clerestory over the choir in the 1230s was then a natural one, dispensing with the temporary roof and keeping the same ceiling level over all parts of the new building.

X La construction de la façade occidentale de la cathédrale de Senlis *

L

ES deux travées occidentales de la cathédrale de Senlis, au niveau des bas-côtés comme à celui des tribunes, diffèrent, par de nombreux détails, des autres travées de l’édifice, Fig. X–1. Une analyse de la construction assise par assise montre qu’un constructeur a été engagé pour bâtir ces travées avec l’instruction de suivre le rythme des travaux des parties orientales sans jamais devancer ceux-ci. Un donateur paraît avoir mis son architecte à la disposition du chantier pour la construction de ces deux travées, et ce donateur semble avoir été Louis VII. Bien que ce maître n’ait jamais travaillé à l’est du pilier XII, en deux ou, peut-être, trois occa­sions, les hommes opérant à l’est sont intervenus à l’ouest, dans les fenêtres des bas-côté et les portails occidentaux. Il est possible qu’il faille également leur attribuer la sculpture du portail central. Toute la sculpture à été réalisée en même temps que les murs adjacents et était en place lorsque commençait l’édification des tribunes. Les simili­tudes entre les chapiteaux des trois niveaux de ces travées occidentales montrent que la construction progressa rapidement. La comparaison de ces chapiteaux avec ceux des quelques rares monuments datés suggère que la totalité des parties occidentales, ainsi que l’est de la construction, avec laquelle elles sont liées, furent bâties en moins de deux décennies, la sculpture du portail central étant en place vers 1160, sinon avant. Les deux travées occidentales de la cathédrale, les piliers qui les soutiennent, leurs voûtes et tous les éléments qui leurs sont associé montrent

* Adapted from La cathédrale Notre Dame de Senlis au XII e siècle, Paris 1987, 109–118.

154

un aspect plus massif que les travées de l’est, Fig. X–2. Les dou­bles colonnes engagées dans les piles à la retombée des arcades, les bases, les impostes, les ogives et les chapiteaux sont tous plus importants; ces mêmes éléments massifs se retrouvent dans l’ensemble de l’élévation de ces travées, tant au niveau des bas-côtés qu’à celui des tribunes, ainsi qu’à la seule voûte du XlIe siècle encore conservée sur le vaisseau central.1 Au premier abord, ces deux travées sembleraient n’avoir pas été bâties en même temps que le reste de la cathédrale. Pourtant, un examen niveau par niveau peut démontrer que les travées occidentales et leurs voisines sont contemporaines. Alors que l’est semble le résultat de plusieurs campagnes de travaux – peut-être douze – jusqu’aux chapiteaux des voûtes, la partie ouest paraît presque entièrement le fait d’un seul constructeur, bâtissant au même rythme, ou peu s’en faut, que les travaux en discon­tinu de l’est. La démonstration selon laquelle les parties orientales furent construites par niveaux successifs, et l’identification des constructeurs, réclameraient une étude plus poussée, réservée à l’avenir. Cette technique de construction a été mise en évidence à Chartres ou, par l’examen attentif des changements de profils et de la géométrie mise en œuvre, les points de jonction entre les différentes campagnes de travaux purent être localisés et les travaux intermédiaires – le fait d’un seul constructeur – dûment étudiés.2 A Senlis, ces changements montrent que l’édifice à été construit par strates dont la hauteur n’atteignait pas le même niveau sur tout le site, les parties orientales progressant avec quelques assises d’avance sur celles de l’ouest.3 Au moins trois constructeurs – indépendamment du maître res­ponsable des deux travées occidentales – sont res­ponsables des bases des piliers qui subsistent au rez­-de‑chaussée et l’on peut distinguer, dans les tribunes, cinq groupes de bases. Il en est résulté que les voûtes du chœur furent terminées à peu prés au moment ou étaient sculptés les chapiteaux des par­ties occidentales. La Fig. X–2 montre la différence entre les piliers ouest et ceux de l’est. Les doubles colonnes engagées sous toutes les arcades et arcs doubleaux Aubert, Monographie, 59‑71 et 99–121; Brouillette, Senlis Cathedral, 74–111; et Vermand, Notre­-Dame de Senlis, 17. 2 James, Contractors. 3 La rupture entre les travées V et IX, où s’insère le tran­sept des XIIIe et XVIe siècles, ne permet plus d’étudier les contacts entre le chœur et les parties occidentales. Nous ne pouvons donc pas être assurés qu’il n’existait pas de discontinuité verticale dans ces travées disparues. Les raisons exprimées en fin d’article donnent à penser qu’une rupture complète est très improbable, la construction s’étant effectuée sur une courte période. 1

la façade occidentale de senlis

Fig. X–1. Façade occidentale de la cathédrale.

Fig. X–2. Plans comparés d’une pile forte de la nef (S. IX ou S. X) et d’une pile des travées occidentales (S. XII).

155

156

Fig. X–3. Plan schématique des travées occidentales de la cathédrale.

Fig. X–4. Profiles comparés de bases des piliers est et ouest

Fig. X–5. Schéma de correspondance entre les chapiteaux de part et d’autre de la travée XII.

la façade occidentale de senlis

157

des travées occidentales présentent une disposition plus articulée qu’à l’est, comme l’illustre schématiquement la Fig. X–3. Les hauteurs des assises les plus basses diffèrent également, ainsi que leurs profils, Fig. X–4. Tou­tefois, sur le côté oriental du pilier XII,4 les doubles colonnes n’apparaissent plus, alors qu’elles sont présentes sur le côté occidental de ce même pilier; c’est une colonne engagée unique qui correspond au pilier XI. Le pilier XII fut conçu pour répondre à la préférence du constructeur des parties occidentales pour les doubles colonnes, mais réalisé pour correspondre an pilier XI, situé à l’est. Par conséquent, le pilier oriental avait déjà été commencé, ce qui démontre qu à ce niveau l’est à précédé l’ouest. La travée XI est étroite, non par suite d’une diminution des axes passant par les centres des piles, mais parce que les piliers XIII et XI sont très importants. Si le parti de l’alternance commencé à l’est avait été poursuivi, ce pilier aurait été remplacé par une pile circulaire. On peut se demander ce qui a occasionné ce changement. Les murs extérieurs des bas-côtés n’existent plus, contrairement aux colonnes et chapiteaux, toujours en place. Il existe de fortes correspondances entre plusieurs chapiteaux des deux côté de la tra­vée XII, Figs. X–5 et 6 a, b, c, d, ce qui suggère que plusieurs sculpteurs opérant principalement à l’est ont aussi travaillé à l’ouest, sur les seuls piliers XII et XIII. Toutefois, les chapiteaux de l’ouest consti­tuent, dans leur majorité, un ensemble distinct mar­qué par une plastique abstraite et un aspect massif caractéristiques du travail du constructeur des par­ties occidentales, dont les formes architecturales – comme par exemple les feuillages – sont simpli­fiées à l’extrême. Ces chapiteaux, qui permettent d’identifier son travail sur les trois niveaux de la construction, n’apparaissent jamais à l’est. Mais la présence, dans les travées occidentales, de quelques chapiteaux semblables à ceux des parties orientales montre, qu’à ce niveau, l’est et l’ouest furent contemporains. La jonction entre les piliers est et ouest est visi­ble sur la voûte des bascôté comme sur les piliers car tous les éléments ont été taillés dans des blocs plus grands. La Fig. X–6 montre que, dans les deux bas-côtés, la jonction 4

Les piliers et les travées à l’ouest de ceux-ci sont indi­qués en chiffres romains en comptant à partir du premier pilier voisin de l’hémicycle. Les colonnes engagées au revers de la façade occidentale sont numérotées XIV. Les piliers au sud sont numérotés S.XII, et les colonnes enga­gées dans les murs s.XII. Les voûtes entre les piliers XII et XIII sont indiquées XII et la fenêtre voisine s.XII. Les portails occidentaux sont désignés ainsi, du nord au sud: Wn, Wc, W.s. Je pense que le contexte et les dessins aide­ront à éliminer toute confusion.

158

s’effectue au‑dessus du pilier XII. Les clefs de voûtes des ogives des tra­vées X et XI du bas-côté méridional ne sont pas décorées alors que celles des deux travées voisines XII et XIII sont ornées de quatre minuscules pétales. Ceci suggère que les voûtes orientales n’ont pas été terminées par le même constructeur que celles de l’ouest. Pourtant la Fig. X–6 montre, qu’au nord, les clefs non décorées sont présentes dans la travée XII. Ceci suggère que l’équipe qui procédait de la sorte pour la finition des voûtes a continué de travailler à l’ouest. Un des constructeurs peut avoir travaillé sur le site avant un autre ou s’ils ont travaillé en même temps, l’équipe de l’est peut avoir fourni des pierres à celle de l’ouest. Quelle que soit la marche au jour le jour du chantier, ces trois éléments démontrent qu’à ce niveau les deux zones furent construites en même temps. De l’intérieur de la nef, nous pouvons suivre les assises des murs au‑dessus des arcades. En examinant les joints de l’appareil et les raccords entre les différentes hauteurs d’assises, nous pouvons voir que, de chaque côté des arcades, les murs furent construits avec leurs piliers adjacents. Dans la travée XI, il existe trois ou quatre assises entre le haut de l’arcade et la corniche qui marque la base de la baie de la tribune. Au sud, les assises sont appareillées d’une manière continue sur toute la longueur de la travée. Mais au nord, les assises ne correspondent pas entre elles. On peut voir que la partie ouest du mur fut montée après la partie orientale, si bien qu’à ce niveau, l’est précéda l’ouest, mais de peu, car aucune des assises de la partie orientale n’aurait pu être montée si le pilier XII n’avait pas été terminé. La corniche à la base de la tribune est sensiblement plus haute et plus plate dans les travées occi­dentales. Les piliers des tribunes occidentales reprennent les mêmes caractéristiques que les piliers inférieurs, avec des doubles colonnes engagées aux arcades, excepté à l’est du pilier XII où il existe, comme au rez‑de‑chaussée, une colonne unique en correspondance avec celle de la pile adjacente (pile XI). Les socles des piliers XII, XIII et XIV sont plus hauts et plus lourds que ceux de l’est. Mais, puisque ceux de la pile XII reposent sur une cor­niche plus petite que celle employée dans les parties orientales, l’est est antérieur à l’ouest d’une assise ou deux. Dans la tribune, l’utilisation de doubles colonnes engagées, la forme et les dimensions des tores, des impostes, des ogives et des clefs, montrent que ces deux travées furent construites selon les mêmes gabarits qu’au rez‑de‑chaussée. Leurs chapi­teaux sont dûs aux mêmes sculpteurs. L’entendue du travail de ces constructeurs est facilement identifia­ble, assise par assise, car tons les éléments provien­nent de blocs plus épais. Il n’y a qu’une seule

la façade occidentale de senlis

159

Fig. X–6. Indication schématique des différences entre les profils utilisés de part et d’autre de la travée XII.

excep­tion, le pilier n. XII, engagé dans le mur. Si son imposte est d’un type qu’on trouve à l’ouest, comme la moitié des chapiteaux situés vers l’ouest, le style des deux chapiteaux placé à l’est s’apparente en revanche à celui des chapiteaux des travées XI et X. Ces chapiteaux sont taillés en outre dans des blocs plus petits que ceux de l’ouest, leur partie inférieure étant au même niveau que les chapiteaux de l’est. Les deux chapiteaux du côté ouest sont de même style que d’autres chapiteaux des travées des tours mais leur hauteur est la même que celle des chapi­teaux de l’est. Ils ont donc été taillés après les deux chapiteaux situés à l’est pour correspondre aux cha­piteaux déjà en place. Là aussi, l’est à précédé de peu l’ouest, l’une des équipes de l’est étant venue travailler sur le mur de la travée occidentale.

160

La diversité de conception des fenêtres dans cette zone, des deux côtés de la tribune, reflète le nombre de constructeurs engagés dans ce travail et révèle qu’au niveau des appuis de celles-ci, juste au­-dessous des chapiteaux précédemment décrits, l’équipe de l’est a continué de travailler à l’ouest, jusqu’au piédroit de la porte de l’escalier tournant conduisant au rez‑de‑chaussée. Pourtant, au‑dessus des chapiteaux, les voûtes des travées XII et XIII, y compris l’arc doubleau au‑dessus du pilier XII, furent taillées selon les profils plus lourds du constructeur de l’ouest. Il plaça aussi le linteau et l’arc de décharge au‑dessus de la porte de l’escalier tournant, qui est identique à celle qui existe à l’en­trée de ces mêmes escaliers an rez‑de‑chaussée. Le linteau est différent de ceux des portes d’accès aux escaliers de l’est. Au‑dessus de la tribune, où il n’existe plus de contact avec les parties orientales du XIIe siècle, dis­parues,5 les chapiteaux, les ogives et les doubleaux de la partie haute de la travée XIII sont conformes à ce qu’on trouve dans les deux premiers étages des deux travées occidentales. Il reste assez d’indications pour démontrer que les trois piliers ouest (XII, XIII, XIV), à tous leurs niveaux, doivent avoir été, à une assise ou deux prés, contem­porains des travaux de l’est. Tout se passe comme si un constructeur avait reçus la mission d’ériger les tra­vées occidentales au même rythme que les travaux de l’est, sans prendre de l’avance – et sans même, à aucun moment, poser de pierres au-delà du pilier XII. Les hommes qui travaillaient à l’est n’avaient aucune restriction de cette sorte. D’une manière générale, le constructeur de l’ouest suivit le plan et l’élévation conçus à l’est mais – et ceci est un point important – à sa manière. Il y a tribune, formerets, arcs brisés, ogives en amande, fenêtres uniques, etc., et l’apparence reste en général homogène. Mais, dans le détail, son intervention est spécifique. Ses piliers, par leur plan et leur impor­tance, différent de ceux de l’est; ses assises ne corres­pondent pas avec celles des parties orientales; ses ogives et ses arcades sont plus fortes; ses chapiteaux et ses impostes sont plus épais. En réalité, il a construit un autre bâtiment. Ainsi que je l’ai montré pour Chartres, cette indépendance n’est pas exceptionnelle. Sur un chan­tier, un nouveau constructeur suivra, dans

5

L’incendie de 1504 ne semble pas avoir endommagé les parties occidentales, ni même le pignon d’origine, ce qui suggère que le feu a surtout sévi dans les parties orientales.

la façade occidentale de senlis

161

l’ensemble, les dispositions antérieures prévues par ses pré­décesseurs ou continuera sur la base des plans et modèles approuvés par son client. Mais il appliquera ses propres gabarits et ceci comprend des éléments majeurs tels que fenêtres, géométrie des piliers, etc. Un tel individualisme peut nous sembler aberrant, mais il suffit d’observer n’importe quel bâtiment contemporain pour le voir se manifester de la même manière. Comment pouvons‑nous expliquer la présence d’un seul constructeur pour l’intégralité des deux travées occidentales et de plus d’une douzaine pour l’est? Un seul donateur se serait‑il engagé à construire ces travées – et celles‑là seulement – dans leur totalité et, au lieu de remettre au clergé les fonds nécessaires pour les utiliser à son gré, leur aurait‑il fourni le service gratuit de ses propres maçons? Dans ce cas, ces hommes étaient proba­blement occupé ailleurs en temps normal et dirigés sur la cathédrale lorsque le travail avait atteint les niveaux requis. Serait‑ce le cadeau du roi au chapitre? Nous connaissons l’intérêt que Louis VII portait à la construction des églises et nous savons que Senlis était une ville royale, bien située au centre d’une région giboyeuse, et que le palais du roi se trouvait immédiatement à l’ouest de la cathédrale. Il aurait été naturel que Louis finance la partie des tra­vaux qu’il pouvait voir depuis son palais. Toutefois, les hommes du roi ne construisirent pas la totalité de ces deux travées. Au niveau de la sculpture du portail occidental, une observation attentive révèle la présence d’autres constructeurs ajoutant chacun deux on trois assises. Tout comme le roi fit venir ses hommes dans une séquence de la construction réalisée verticalement, le clergé semble avoir introduit les siens à sa convenance dans les travaux du roi ou bien les mit au travail à l’ouest en attendant que les hommes du roi se libèrent d’autres activité. La présence d’autres équipes aux portails occi­dentaux alors que l’essentiel du travail parait typi­que de celui des hommes du roi, suggère que les maçons royaux puissent avoir été à court de sculp­teurs. Les maçons du roi firent preuve d’audace et d’autorité dans la taille de leurs chapiteaux mais il est possible qu’ils n’aient pas été du niveau requis pour les tâches délicates à exécuter sur les portails. En fait, le maître occidental nous donne l’impression d’avoir été un constructeur de grands ouvrages, châteaux ou remparts de ville. En tant que maître des travaux royaux, il est possible qu’il ait eu davan­tage à construire pour la défense que pour la prière. C’est la raison pour laquelle d’autres furent appelés pour ce travail, peut-être à trois reprises. La sculpture de l’ouest fut mise en place en même temps que les murs qui l’accompagnent, comme le démontre une analyse, bloc par bloc, des

162

Fig. X–7. Plan et élévation partiels des fenêtres des bas-côté nord et sud de la travée XIII.

séquences de construction.6 Pourtant, la subtilité des détails et la sculpture des chapiteaux indiquent qu’une autre équipe a pu intervenir dans ce travail puisque les chapiteaux simplifiés des travées occi­dentales n’apparaissent pas dans les portails.7 Mal­gré cela, l’architecture de la façade occidentale en Fig. X–1, austère, presque totalement vierge de glacis ou de moulures décoratives, montre la même simpli­cité et la même netteté du parti que les autres travaux de ce constructeur. Si nous suivons les assises à partir de la fenêtre n.XIII en continuant sur le mur occidental, il apparaît que la pose semble avoir commencé avec les chapiteaux du portail latéral nord ou immédiatement avant les chapiteaux intérieurs du bas-­côté. La fenêtre du sud (s.XIII) montre un dessin complètement différent de celle du nord, Fig. X–7. Elle est placée trois assises plus bas que cette

Une démonstration semblable à pu être faite pour Chartres – James, Contractors, ch. 16. A Senlis, la méthode de construction est, excepté pour les porches, identique à celle de Chartres, les pierres supérieures et les claveaux s’appuyant sur l’appareil situé en‑dessous et étant donc mis en place après celui-ci. 7 Les chapiteaux originaux du portail central (W.c) et plusieurs du portail latéral nord (W.n), sont décorés de feuilles d’un dessin rigide, différentes de celles des chapi­teaux des travées S.XIII et S.XII dûs au maître royal. 6

la façade occidentale de senlis

163

dernière.8 On y remarque un socle formant glacis au droit des colonnettes de l’ébrasement extérieur, en P. Un détail semblable existe dans la fenêtre de la travée n. XIII mais le glacis est associé à un dosse­ret carré et non plus à une colonnette. Considérant la discontinuité des assises autour des bases des colonnettes de la fenêtre sud ainsi que les assises des murs jusqu’aux angles des portails, et estimant qu’un constructeur qui utilise un profil anguleux dans ses portes et dans ses arcades sera tenté de faire de même autour de ses fenêtres, j’en conclus que le construc­teur de la fenêtre nord (n.XIII) est aussi responsable de la fenêtre sud (s.XIII). Les changements au‑des­sus du socle montrent qu’il n’est pas intervenu au ­dessus de P. Un autre constructeur intervint à ce niveau dans les travées ouest. Le point de jonction entre le travail de l’architecte royal et celui du nouveau constructeur se place juste au‑dessus de ce socle. Si l’on suit ces assises vers l’ouest, on voit bien que ce maître doit avoir sculpté les corbeaux qui soutiennent le linteau du portail sud (W.s) car ceux-ci sont arrondis au‑dessous. En suivant ces mêmes assises jusqu’au portail central, on peut situer la jonction juste au‑dessus du calen­drier, sous les bases des statues d’ébrasement. Ceci est confirmé par un petit détaille: sur le côté nord, le tore de la première base à gauche se prolonge dans la colonne sous la forme d’une scotie courbe continue, alors que dans les autres bases la scotie sous la colonne disparaît dans une base rec­tangulaire sous la colonnette intermédiaire. Cette disposition indique que la première de ces bases a été posée par le maître occidental et les autres par le constructeur qui a changé la fenêtre du bas-côté sud (s.XIII). Il est intéressant de remarquer à quel point le second maître à suivi le premier pour le dessin du tore et de l’arête contigus à son bord extérieur, et à pourtant changé la base des colonnettes intermé­diaires. On sent qu’il à imité le tore afin de mainte­nir l’unité du portail et on suppose aussi qu’il a changé la base simplement parce que ce n’était pas dans sa manière. Ces changements dans les bases et les fenêtres nous permettent de suivre l’histoire de la construction alors qu’un travail plus homogène ne l’aurait pas permis.

8

Le côté nord tout entier semble avoir été construit en avance sur le côté sud, comme le montrent notamment les corbeaux des portails latéraux, les formerets et les clefs de voûte. Ce fait pourrait s’expliquer par le déclivité du ter­rain vers le sud et la réduction corrélative de la profondeur des fondations au nord.

164

Au portail nord, cette jonction pourrait se situer une assise plus haut, à partir du tore gauche, et comprenant les corbeaux. Ceux-ci sont biseautés en‑dessous, ce qui suggère un constructeur différent. Comme ils sont semblables à ceux du portail central, ils sont sans doute du même architecte royal. C’est à ce niveau que celui-ci est revenu sur le site aprés qu’un autre constructeur ait posé deux assises, changé la fenêtre sud (s.XIII) et les profils des tores du portail central, et peut-être sculpté les chapiteaux des bas-côtés qui ne sont pas dûs au constructeur occidental. Ses hommes auraient‑ils aussi sculpté les statues des montants du portail central? L’architecte royal revint poser trois ou quatre assises à l’intérieur9 et peutêtre seulement deux à l’extérieur, ainsi que les autres chapiteaux des bas-­côtés, les corbeaux du portail central (W.c) et les claveaux inférieurs des voûtes.10 Comme les piliers furent peut-être construits plus vite que les murs,11 les ogives et les doubleaux ont pu être commencés avant que les murs des travées occidentales aient atteint la hauteur des impostes. Nous en trouvons un élément de confirmation dans les profils des arcs formerets. La Fig. X–6 montre un changement de forme notable aux deux portails latéraux, à la fenêtre sud (s.XIII) et aux formerets des deux travées s.Xll et s.XIII. Ce changement indique la présence d’un autre maître à l’ouest. Afin de modi­fier la moulure de l’arcade tout entière, il lui suffisait de modifier l’assise inférieure et il est peu probable que quelqu’un d’autre aurait changé de méthode en cours de construction.12 Dans le bas-côté nord (travée n.XIII), il est également présent au‑dessus des chapi­teaux ouest, mais non pas au‑dessus des chapiteaux nord oû existe une moulure plus simple.

Bien que paraissant peu important à priori, ce nombre de trois on quatre assises est en fait loin d’être négligeable lorsqu’il concerne le développement total d’un édifice de la dimension d’une cathédrale. Ce nombre moyen d’assises est semblable dans l’autres grandes constructions (voir les dessins isométriques dans James, Contractors). 10 Les voûtes sont lentes à construire car le cintrage doit rester en place durant le temps – 3 on 4 mois – de durcissement du mortier. Ce n’est qu’ensuite qu’une nouvelle étape de construction peut intervenir. Alors que les claveaux inférieurs pouvaient être montés sans cintrage, les autres devaient attendre que les charpentiers aient construit les formes nécessaires. 11 James, Contractors, ch.11. Ce décalage s’explique par le moindre volume des piliers par rapport aux murs. Les arcades, les ogives et les formerets pouvaient être lancés à partir des piliers avant que les murs n’aient atteint la hauteur des chapiteaux. Les bas-côtés inachevés de Marizy­-Saint‑Mard (Aisne) l’illustrent bien. 12 Il existe de notables exceptions, comme à Mantes, par exemple, où des modifications mineures du profil des grandes arcades montrent clairement le processus de construction. 9

la façade occidentale de senlis

165

Comme le chapiteau à été placé par le maître occidental, la jonction avec léquipe suivante se trouve au‑dessous de l’imposte. Et pourtant, immédiatement à l’ouest, au‑des­sus du portail nord, la même moulure a été commencée deux assises plus bas, au niveau infé­rieur des chapiteaux du bas-côté. Comme ces murs sont extrêmement épais, il n’aurait pas été dif­ficile d’élever le parement intérieur sur une on deux assises de plus que celui de l’extérieur. La construc­tion des tympans latéraux aurait pu y contribuer car ceux-ci sont formés d’un mur partiellement masqué par une arcature. La différence de niveau aurait pu être rattrapée entre ce mur et celui monté en même temps que les chapiteaux intérieurs et l’arc surmon­tant ces portes. Les arcatures devant les tympans et les arcs au‑dessus pourraient bien avoir été posés par le constructeur suivant, ainsi que les arcs supérieurs associés au changement de profil. Les lourdes dou­bles clefs utilisées pour renforcer les arcs au‑dessus des linteaux sont de dimensions monumentales et rappellent davantage une construction militaire plutôt qu’écclésiastique.13 Tout ceci suggère que le maître occidental quitta le site après avoir sculpté les chapiteaux intérieurs des bas-côtés et avant l’érec­tion des arcatures devant les tympans latéraux. Ainsi, au niveau des voûtes des bas-côtés, pou­vons‑nous postuler la présence d’un second constructeur, intervenu au milieu du travail des architectes royaux. Si l’on suit les assises jusqu’au portail central, on peut voir que son travail continue derrière les statues d’ébrasement et, peut-être, au linteau. Toutefois, bien que je n’en aie pas la preuve absolue, je suis tenté d’attribuer le linteau, le tympan qui le surmonte et plusieurs, sinon toutes, les sculptures qui y figurent, à un troisième maître, principalement parce que l’élégance raffinée de ces statues s’accorde difficilement avec la simplicité austère du travail des architectes royaux. Ceux-ci se sont peut‑être contentés de mettre en place ces sculptures et de construire l’arc au‑dessus lorsqu’ils sont reve­nus construire la tribune. Il n’est pas déraisonnable de penser que le clergé ait choisi pour la sculpture les meilleurs artistes travaillant à l’est, on encore qu’il ait fait venir de l’extérieur des spécialistes de haute notoriété. Quelque soit la marche de la construction, cette sculpture est contemporaine de celle des chapiteaux des bas-côtés et des tribunes. Quelle 13

La disposition de la clef, au sud, est différente de celle du nord. je pense que celle-ci doit entre attribuée au second maître car il est improbable qu’un seul constructeur ait eu recours à deux solutions différentes lorsqu’une seule suffisait.

166

en est la date? La parenté de style de tous les chapiteaux occiden­taux sur les trois niveaux, et l’homogénéité relative de la taille démontrent la relative brièveté de la construction à l’ouest. Si l’ouest, jusqu’à la voûte, a donc été bâti rapidement, l’est – lié à l’ouest – doit avoir suivi le même rythme. Si la cathédrale était en cours de construction en 1154,14 l’intervention de l’architecte royal ne doit pas être très éloignée de cette date. Il y à peu d’autres édifices bien daté auxquels on puisse se référer mais, entre le chœur de Saint-­Denis, consacré en 1144, et le début de la construc­tion de Saint‑Frambourg de Senlis, les chapiteaux de la cathédral s’apparentent davantage au premier,15 et il est difficile d’imaginer que les chapiteaux des voûtes occidentales puissent être contemporains de ceux de Saint‑Frambourg ou de ceux du transept sud de Soissons, ou encore de Canterbury.16 Il leur man­ que l’élégance, la réalité immanente, la richesse de la décoration des années 70. Bien que nous n’ayons pas de construction datée pour les années 50 ou 60, une évolution se perçoit à partir de la rigidité formelle, un peu glacée, des chapiteaux des façades occiden­tales de Saint‑Denis et de Chartres, vers le natura­lisme des années 80. J’ai photographié tous les cha­piteaux de cette période dans le bassin parisien et, dans une séquence chronologique, ceux des bas-cote de Senlis sont juste avant 1160. Par conséquent, la cathédrale aurait été entièrement construite all cours du règne de Louis VII et toute la sculpture mise en place dès 1160, sinon auparavant. On peut en trouver quelque confirmation dans les chapiteaux associé à la sculpture des portails, comme dans la sculpture du portail central, au tym­pan, sur le linteau et dans les feuillages des vous­sures de l’archivolte. Une gaucherie de certains per­sonnages, leur agitation extatique, l’attitude de certains autres, une tendance à la torsion des atti­tudes et des étoffes rappellent une esthétique antérieure. En envisageant une date aussi précoce pour cette sculpture, il faut tenir compte du fait que les rares œuvres sculptées bien datées sont les quelques sculptures qui demeurent aux porches occidentaux de Saint‑Denis et la sculpture du transept de Chartres.17 Vermand, Notre­-Dame de Senlis, 47–49. Vermand, Notre­-Dame de Senlis, 101. 16 Bianchina, “Saint‑Frambourg”, n. 22, 19. Des ter­rains ont été donné pour la construction du transept sud de Soissons en 1176: Bibliothèque Nationale, Collection Baluze, xlvi, fol. 467. Le chœur de Canterbury fut reconstruit par le Français Guillaume de Sens entre 1174 et 1178 – Willis, Canterbury. 17 Pour la datation de la sculpture du transept de Char­tres, voir James, Contractors, ch. 3 et 16, ou la sculp­ture est datée en séquences précises entre 1198 et 1215. 14 15

XI A Northern Master for Clermont-Ferrand? *

H

AVING read Michael Davis’s work on the cathedral of ClermontFerrand and his review of Christian Freigang’s article,1 I offer the suggestion that the Jean­ Deschamps mentioned in the documents of 1248 was not a local man, but came from the area north of Paris where he­had been involved in the construction of the choir chapels of Amiens cathedral, the­ parish church in Nogent-sur-Oise and the lower level of la Sainte-Chapelle. My observations of the stonework at Clermont, and the sutures in the coursing ­and changes in profiles, show that the first campaign extended to three or­four courses above the sills in the five eastern chapels, then dropped to just­above the plinth in the next three bays.2 The footings for these eastern walls, the plinths and the torus courses over it, and the walls over that are all part of the first campaign, Fig. XI–1. This includes the window sills around the chapels. The­ nature of the stone and the manner of cutting, sizes and mortar joints are enough to show precisely where the end of this campaign ­occurred. An elevation showing every course and each change in profile would­make this abundantly clear. The torus moulds are typical of the second quarter of the thirteenth century. ­The profile has a buckled upper roll and a squeezed-in scotia, Fig. XI–2. Though Branner argued that the cathedral was not commenced in 1248, but fifteen years later,3 I know of no datable torus anywhere in France

* Adapted from Avista Forum, viii 1994, 5–7 1 Davis, “Deschamps”. 2 This happened in stages, In bay N4 to six courses above the plinth, N2 & 3 to one course dropping to none; S4 and 3 to four courses, ­S2 from two to none. 3 Branner, Saint Louis, 142.

168

Fig. XI–1. Clermont-Ferrand, interior to east.

Fig. XI–2. Section through pier torus moulds. Fig. XI–3. Pier bases in apse.

Fig. XI–4. Section through post-1250 torus moulds.

a northern master

169

with such a scotia that could be attributed to the later 1260s or 1270s. This torus profile is consistent with the date of 1248 for the­ commencement of the works. This work included the foundations and some of the most easterly walls up to the sills of the windows. Medieval masters tended to be up-to-date in the style of their profiles, as­ they were with the foliage on their capitals. The fact that most masons ­updated profiles and foliage from current fashions, even when extending buildings which were ten or fifteen ­years old, can be attested by examining cloned buildings (such as Chars) and ­the later parts of long-term projects (such as the western bays of Laon). Under the Clermont torus mould there is a small plinth carved from the same­ block of stone, Fig. XI–3. It has one most unusual feature, a tiny moulding along the lower edge, like a miniature drip mould. The tori in the Amiens chapels have the same splay, as do the bases at Nogent. At la Sainte-Chapelle this little splay is found at both levels, though the plan of ­the support to the torus is different in each case. All three buildings were ­under construction in the decade prior to 1248. This detail is, for me, the clincher. It is such a rare device that I do not ­know of any other examples in the north. It bespeaks a very personal approach ­to the nature of the torus and the shaft it supports. It is so idiosyncratic I­ doubt the idea would have been copied, let alone used by a later generation of ­architects. It is the mark of a very individual genius. The torus profiles used in the east and their plinths are unlike any in used ­during any later campaign. In all the other tori in the building, including­the piers around the choir, the scotia has been eliminated. This simpler­ profile became common practice after the 1240s, as in Fig. XI–4. The later date for ­the rest of the building is confirmed in the universal use of this flattened profile without a scotia. The torus profiles show that Jean’s work in the first campaign did not include ­the piers around the ambulatory. Building the ring of chapels first, and returning twenty years later to erect the rest of the choir, was common ­practice: Notre-Dame in Paris, Orbais, Saint-Leu-d’Esserent, Saint-Quentin, to ­mention but a few — a procedure that may have been adopted for funding purposes.4 ­

4

James, Template-makers, 50–54.

170

Fig. XI–5. Apse piscine capital.

Stopping the work around 1250 at the level of the chapel sills meant that the­forms of the tracery, the inner piers, and the ambulatory vaults would be left ­to the next campaign of the 1270s. Further connections between the Clermont chapels and the northern builders of ­the 1240s lie in the piscine, the profile used for the window sills, the ­arrangement of external buttresses and the layout of the wall shafts. The ­trefoil arch over the piscine is like those in the dado arcades of Amiens ­and la Sainte-Chapelle, and the profile is not nearly as attenuated as they were ­to become in the 70s. The little capitals in the piscine (the only ones carved in this campaign) are typical of the 1240s, and very like those in the Amiens choir dado arcade, Fig. XI–5.5 One other item needs to be considered, though I do not have the data to ­complete it. I notice on the plans of Clermont and Amiens that the external­ walls of the straight bays next to the ambulatory do not seem to be parallel,­ but are inclined inwards towards the east. The same situation is to be found at la Sainte-Chapelle, where the walls are ­out of parallel by 48 mm. This is from the use of a 7:11:13 triangle.6 5 6

James, Ark of God, 1334–40. “XXXIV: Sainte-Chapelle”.

a northern master

171

This is not quite a true right angle triangle, ­but just under 90o. At Amiens, using only the very approximate scaling from a plan, the proportion of the width to the distance from the crossing to the axial chapel forms the same 7:11 ratio. The inclination of the lateral walls came about because the master ­used this triangle to set out the ground plan with the narrower end next to­ the choir. Such a triangle would not have been a common way to set out a ­building, and if it occurred in the initial geometry for Clermont and Amiens ­as well as la Sainte-Chapelle, this would establish the connections between these buildings more­firmly. This ratio may have been a signature for this family of masons, passed on from ­father to son, for Vivian Paul has shown that a similar triangle was incised­into the paving at Narbonne, where there seem to have been family connections with Clermont-Ferrand.7 In my study of funding of church building in the Paris Basin there were two periods of stagnation.8 During the first around 1170, the masons from ­the north-eastern area that was in recession moved into the more prosperous Parisian region looking for work. During the second, which began in­ the 1230s, all areas in the Basin stagnated into what was to be a terminal­decline, and many of the masters again took to the road. It would not ­therefore be unexpected for one of the more inventive, as Jean Deschamps ­undoubtedly was, to move elsewhere as his services may have been called for. Being able to offer his­ new clients not only his skills, but also the latest style from the capital, he­would have been doubly welcome wherever he chose to go.

7 8

Paul, “Narbonne ­cathedral”. “XL: Funding”.

XII The End of Discontinuous Contracting *

P

ERMANENT workshops at the larger abbeys and cathedrals under the leadership of the best masters is often seen as the most common practice. The evidence for the later Middle Ages, when there are many documents, is quite compelling. But for the period before the mid-thirteenth century the evidence in the buildings themselves contradicts this view. Over the past thirty-five years I have not only visited, but obtained the keys for and climbed over thousands of medieval churches in France and England, in Italy and Spain and Germany. Even the quickest toichological survey has shown me that prior to 1250 every one of these buildings, without any significant exception, has been built in many small campaigns. They show that almost every time that arches and vaults were finished and had to be left for the mortar to set, there was a change in contractors. Obviously there were circumstances which produced this form of contracting at that time, and there must have been others which discouraged it some time during the thirteenth century and led the industry into the permanent and specialised roles we act out today. We cannot say that just because it is not the way we have been used to seeing things done since the Renaissance it was a less viable way of producing great works of art. We can only say that it was different. During the period when the buildings themselves show that they were constructed through discontinuous contracting, there is some evidence that has been the subject of much discussion in the hope of unravelling the names of the master masons, these are the labyrinths in Amiens and Reims, both

* Adapted from The Contractors of Chartres, chapter XXII.

the end of discontinuous contracting

173

of which commem­orate the architects of the buildings they rest in. Both were laid in the last years of the thirteenth century sixty or more years after these two buildings were begun. The information in them may be read in a number of ways, but one shows that there is no conflict between my concept of discontinuous contracting and the post-1250 practice of employing permanent masters. The one in Amiens was designed by Regnault de Cormant and in it he names two of his predecessors. Historians have naturally assumed that the first, Robert de Luzarches, was also the first architect, but the inscription does not go so far. It reads: Chil qui maistre yert de l’oeuvre Maistre Robert estoit nommes Et de Luzarches surnommes. This simply translates as “He who was master of the works was Robert from Luzarches”. It does not state that Robert was the first master, nor that he began the works, but just that he was the master. When I studied the cathedral I found evidence for 26 separate campaigns from the ground to the top of the triforium, with the same innumerable changes as at Chartres and elsewhere. So neither Robert nor any other one person could have had control over the bottom two storeys. The first inscription refers to the bishop and the king who were alive at the time the cathedral was begun in 1220. Does it mention them as a substitute for the many itinerant masters whose names had accumulated over the early years of the work? So should the inscription quoted above be taken to mean “the name of the first master to be permanently engaged by the Chapter was Robert” implying that, up to then as at Chartres, there was no full‑time professional supervision? The inscription names only three architects for the 68 years to 1288, and it is hard to believe that so few men could have controlled a job like this for so long a time. It was one of the largest buildings being constructed in France, and the Chapter would have naturally chosen experienced and proven men for it. While it is true that they could have been appointed in their forties and each of them could have spent twenty‑five years on the site, the average working span of senior appointees, as with kings and popes, is a lot less than that. However, if Robert had been appointed towards the middle of the century each master would have been in charge for a more realistic

174

period. Further at Luzarches the little church has a small chapel dating from the 1250s with details very like some in the clerestory at Amiens. If Robert had come from Luzarches in the 50s, the other two masters mentioned in the labyrinth could have completed Robert’s work, including the transepts and the laying of the labyrinth itself. We have to conclude that the wording on the labyrinth is not specific enough to show that either interpretation is correct. The situation at Reims is somewhat similar, though we have been given the names of those masters who were responsible for the most significant parts. There is incontrovertible evidence for at least thirteen campaigns from the floor to the triforium, and considering the depth of the footings there could easily be another one or two in the foundations. As at Amiens there is nothing in the Reims labyrinth that would contradict this.1 One of the four, Jean d’Orbais, is shown with a compass designing the apse, and the inscription reads ‘qu’il encommencea la coiffe de l’eglise’. He is the only master whose duration on the site is not mentioned. Jean le Loup is recorded as having worked for sixteen years, Gaucher de Reims who ‘ouvra aux vossures et portaulx’ for eight, and Bernard de Soissons for an enormous thirty-five. A text of 1287 mentioned Bernard, and if this had been his last year on the site his first would have been 1252.2 Subtracting Jean’s sixteen years and Gaucher’s eight from that takes us back to 1228, at the earliest. This is seventeen years or more after the start of the work. This seventeen years is quite enough to encompass the thirteen‑plus campaigns of the ad hoc period. Jean’s long term appointment may mark the moment when the old contractual system was being replaced by the new. As with Amiens, my conclusions are the same: there is nothing in either of these inscriptions to contradict the thesis of discontinuous contracting before the mid-thirteenth century. What, then, were the advantages of a system of discontinuous contracting in the earlier years? It would have grown naturally out of the circumstances. During the eleventh century there were few localities which could have supported large groups of masons for long periods, though I would expect that the larger quarries maintained permanent accommodation for the men

Salet, “Chronologie”, has also suggested that more than one architect is to be found in the early years at Reims. The quotations which follow come from this article. 2 Branner, “Labyrinth of Reims”. 1

the end of discontinuous contracting

175

who gathered there. Smaller local builders in stone with five or ten skilled men could have been kept constantly at work on farmhouses, chapels, fortifications or repairs, but the monster teams of fifty men or more plus assistants and labourers — which was the size of the team brought from Toulouse to build the eastern end of Compostella in 1077,3 and the large teams at Chartres — could only have found work on the biggest jobs where the funds were adequate. Over the years the very large jobs would have encouraged some crews to become larger, and then, because they were larger than their own locality could support, they were compelled to look further and further from their home base or quarry in search of work. As the boom developed their ties with their homes were weakened until they became the totally itinerant contractors of 1200. Working quickly and competently they seldom stayed in the one place for more than a year, and this spread their ideas and their inventiveness across a huge area. I have shown that some of the larger teams from the north-eastern region of the Marne and the Soissonais worked almost exclusively in their own region until the 1190s, and after that were to be found in many other regions as well.4 The carvers of capitals moved prodigious distances to find work. Some builders specialised in bridges, others in fortifications, and would work with their men wherever they were called, even from the south of France to England for work. Movement was endemic in the period, in pilgrimages and crusades and in emissaries between all the great rulers and church leaders. It was a peripatetic age. The masters worked as if they were going to be employed on each site forever, even though their experience told them it would be for a few months or, at the most, a year or two. For example at Lagny the western bays of the choir were built with a door out of the stairs for a walkway that could only have served an extension through the middle of the old westwork that is still there today. At St Pierre in Chartres the western end of the triforium continues into the tower on the same assumption. At Gallardon the windows of the choir were continued behind the older tower in anticipation of being extended to the west through the eleventh-century nave. One is struck by the confidence of these masters, their assumption that expansion would continue for ever. Even in work which had only recently been completed we find the same restless search for bigness. The master of 3 4

Harvey, Gothic World.. “IV: Chartres was lucky”.

176

the nave at Soissons was planning no more than 40 years after the charming south transept had been completed to remove at least the upper storey so it could be raised to the same height as the new work. It was the same with towers. None of them were capped off during the boom unless they had reached the topmost peak of their spires. in the lantern tower at Laon the staircase continues another ten risers above the roof as if expecting to leap into the next storey before even the one below was finished, feverishly piling even more levels of precarious masonry over the already soaring stones below. Everything was in anticipation of the next move. To stop was unthinkable. One wonders where it would have ended if the boom in building had not petered out. The economic bust put an end to that glorious dream, Fig. IV–10. Suddenly, during the 1230s, the amount of building work dried up. Compared to the 1190s less than a third was spent.5 Where did the thousands of unwanted skilled men go? What happened to the sophisticated organisation of contractors? Did they simply down-size, or did they run for cover? It would have been most natural for the teams to reduce the numbers employed, and let the men go back to whatever jobs they could find, or back to the farms and any labouring work available. But what of the most precious and skilled part of the teams — the foremen and clerks, and what of the masters themselves? The little evidence we have suggests that they looked for security, for some sort of tenure by being employed in a more permanent capacity at the great sites where some work was still going on, such as Reims and Amiens. Branner wrote that “the fluidity of the 1220s lasted into the next decade”, but that afterwards work became regional in character.”6 This is equivalent to writing that before 1230 the masons moved about the countryside spreading their ideas but that afterwards they stopped moving and settled down. Branner goes on to say that inventiveness and experiment, which in architecture is nearly “always a sign of confidence, began to dry up — perhaps firstly in the east where the economic vitality of the Champagne fairs may have been sapped by competition and debts. Reims itself began the retreat from innovation as early as 1225, and over the next decade the north‑east in general gradually followed suit.”

5 6

“XL: Funding”. Branner, Saint Louis, 24–9.

the end of discontinuous contracting

177

The Cambridge Economic History asserts that “at the end of the twelfth century and during the first half of the thirteenth, the Champagne fairs were indeed the centre of the international commercial activity of the western world” 7 and were more important than any in Italy or Flanders, but that the middle of the next century this activity had been greatly diminished, the nobility was chronically in debt, municipal revenues “barely sufficed even for normal purposes”8 and a large proportion of inde­pendent husbandmen were being forced to sell out to become rentiers. The shortage of work would have tempted the masters to settle down to assure themselves of a secure income. The ‘diffusion’ of the Gothic style into Burgundy, Germany and elsewhere after 1230 may have been as much a consequence of lack of work at home that drove the teams further afield, as it was a conscious choice of the client to use the new Gothic style. Prior to the great bust, the great buildings of northern France were produced in an organic way. Not being under continuous professional supervision they were more of a process than a project. The buildings read as a continuum of ever‑changing direction and methods. Their enormous appeal lies in being like a natural growing thing, for the final appearance is not predetermined in all of its particulars, but only in its basic genetic form. These buildings are the culmination of an organic evolution that aimed towards the commonly held image of the Heavenly City. They reflect the fascinating individuality of the men who led and worked in the great contracting teams, and the depth of their faith. More than anything, they reflect the creative individuality of the many men who worked on each one. In one sense it was as if, for a short two hundred years, there was an opportunity for the very best architects to make some contribution to every building, inserting their brilliance into one campaign or another in every site. The alternative in later years was that some sites languished under the control of mediocre men for such long periods that there was little flash of genius in them. I know which contractual system I would prefer, though it makes our job as historians incredibly more difficult.

7 8

Miller, Economic History, ii 132. Miller, Economic History, 533. Kraus, Mortar, for the role of Philippe‑Auguste’s conquests on the financing of churches.

XIII Building the Narthex at Chartres *

T

HE western narthex of Chartres cathedral, with its two towers and the Royal Portal between, are often referred to in the documents, though without the precision of Saint-Denis. Where Suger provides a clear date for the completion of the upper chapel over the Saint-Denis narthex, we have eight documents that in some way or another relate to Chartres. Two refer to a fire that destroyed the town on September 5, 1134. Fulbert’s church, then little more than a hundred years old, was not touched.1 Nothing tells us whether the westwork was commenced immediately afterwards or not, nor whether work had already been begun before that date, but we have one reference to archdeacon Ansgerius “who gave 20 sols for the building of the tower”.2 It has been much remarked that the Latin word used is turris, which refers to one singular tower.3 There is no date to this document, but we know that Ansgerius died in January 1139 and can presume that the gift was made at the latest at this date.4 In 1145 we have two stories of people dragging “with their own shoulders wagons filled with stone and wood, grain and other materials, to the work site of the church”. Two of these refer to towers, using the Latin plural turres.5 Lastly there are references to gifts of money made some time after 1149 and * Adapted from Bulletin de la Société Archéologique d’Eure-et-Loir, lxxxvii 2006, 3-20. 1 Merlet and Clerval, Un Manuscrit Chartrain, 84–85. Sancti Fulberti Carnotensis Episcopi Epistolae; and Merlet and de Lépinois, Cartulaire, i, 18. 2 Merlet and de Lépinois, Cartulaire, iii, 130–131 3 Branner, Chartres, 74–6. 4 Merlet and de Lépinois, Cartulaire, i, 147–148 5 Morte, and Deschamps, xiv, 318–319, 319 and 290. Roberti Abbatis de Monte, Appendice ad Sigbertum

building the narthex at chartres

179

after 1150. In the first the Latin plural is used opus turrium for all three gifts, while in the second the singular opus turris.6 Altogether this tells us only that work was begun in the west after the fire, was in progress before 1139 and was still under construction ten or more years later. If we can trust the scribe’s grammar, one tower was under construction in the 30s (the north), both under construction in the 40s and only one (the south) still being completed in the 50s or later. And this is, indeed, the commonly accepted view adopted by all writers on the subject. It is from here that opinions differ. The original scheme for the west It has been argued that the north tower may have been designed as a campanile, or as one part of an enclosure for an open atrium. Most scholars had thought, as with Branner, that “between the towers there was to be a sort of covered courtyard or portico, open on the west” 7 with the portal sculpture placed on the eastern wall. Without arguing the unusualness of such an idea, there is one observation that resolves this issue. There are two doors leading out of the north tower facing south. Today one passes through them into the cathedral shop, and from there directly into the crypt. There is no evidence that closable doors were planned to shut off either of these entrances. Security was essential in the troubled times of the eleventh and early twelfth centuries. Precious objects were kept inside the churches as theft was not uncommon, and raids by dissatisfied groups were to be feared, as those in Conques and Toulouse only fifty years earlier. Openings in medieval buildings needed to be secured with doors, and doors require jambs, places to locate the hinges and recesses for locking devices. These openings on the ground floor of the north tower show that doors were never intended. The jambs are plain, and the arched openings show no sign of having been closed in so that doors could be placed. The problem here is that when a door with a curved head opens, the top swings back against the arch unless it is placed on the face. If the door is placed on the face there is no rebate to stop the door being shoved inwards. This is why

Merlet and Lépinois, Cartulaire, iii, 80, 189, and 200 for the first three, and iii, 17, 33, and 205 for the second. 7 Branner, Chartres, 75. 6

180

lockable doors have level heads under level lintels unless they are flush with the face of the wall. The lack of frames, remains of hinges and recesses for locks shows these openings were designed as part of the interior, and not as an external opening. This shows that the north tower was intended, from the first, to flank a closed-in narthex in which the outer security doors would be placed where the Royal Portal now is. The Royal Portal The portal and its sculpture were erected with the south tower, but not with the north. South tower, portal and narthex were all constructed at the same time. Examination of the stonework between the tower and the portal shows that the masonry of the wall behind the sculpture is bonded integrally into that of the tower, Fig. XIII–1.8 This shows that the jambs of the Incarnation doorway are contemporary with this zone of the tower. Also, above the archivolts, the coursing is continuous between the two parts, though it does step down over the apex of this door. From what the rest of the cathedral shows, this evidence is sufficient proof that the erection of the sculpture is contemporary with the lowest stage of the south tower. This can be confirmed in a number of other items. The bases used under the wall statues are repeated on the south side of the south tower, suggesting they were carved by the same men: the sundial on the south-west corner is protected by a canopy matching some canopies in the portal; and the internal columns of the doorways match and relate geometrically to the columns on the narthex side of the south tower. Concerning the Royal Portal itself there has been endless discussion, but no documents. Much of the discussion where it affects chronology has centred around the misalignments in the sculpture. I have analysed this in an earlier chapter,9 and concluded that there were five major changes to the architectural layout of the doorways, Fig. VII–15. Together this evidence showed that the sculpture was carved by a number of crews, each led by a different master, working one after the other rather than at the same time. When this analysis of the portal erection is co-ordinated with the following description of the stages in the building of the north tower, it 8 9

James, Contractors, 220. “VII: Anomalies in the Incarnation and Ascension portals”.

building the narthex at chartres

Fig. XIII–1. Chartres cathedral, junction between the south tower and the adjacent door of the Royal Portal.

181

182

Fig. XIII–2. Lower two levels of the north tower.

Fig. XIII–3. Lower two levels of the south tower.

building the narthex at chartres

183

suggests that each of the five phases of carving on the portal may be slotted into the construction campaigns of the north tower. It shows there was an intention to erect sculpture around the western doorway from the beginning, though the number of doors, the detailed measurements and the selection of the sculptural program would seem to have been altered with each new crew. Masons’ marks There are thousands of masons’ marks in both towers, and on nearly every level, though not very accessible except within the staircases. They form clusters of similar marks, and the divisions between them are often clearly visible. Each level of change coincides with exactly those constructional elements where we would expect there to have been delays in the work – at arches and vaults where time was required for formwork to be built or mortar to shrink and settle.10 During such periods, on the small working area of each tower, there was no room for the men to work and it would seem they had to leave for other employment. Altogether there is evidence for eight crews constructing the north tower, and for nine in the south to the base of the spire. This includes the extensions of the crypt and their junctions into the bases of the tower. The masons’ marks provide the clearest evidence, and this is backed up by changes to mouldings, to elements like windows and forms of vaulting, and by the capitals.11 The different levels and the work that went into them can cause confusion, especially when attempting to align work seen from the outside with that from the inside, Figs. XIII–2 and 3. I refer to the three levels of capitals and arches in the ground floor as being part of the aisles, and will call them (a-) for the doors from the narthex into the towers, (a) for the vaults over the tower rooms and (a+) for the external decorative arches that cap the first storey. Above that each level is referred to by number, there being two remaining in the north after the destruction of the spire, and six in the south to where the octagonal base meets the inclined faces of the spire. As the vaults and external arcade in the south contain more than one level of

10 11

“XIX: Medieval mortar”. These capitals will be analyzed in Vol. 3, Ark of God.

184

capitals and arches I have referred to them as (1) and (1+). These levels are noted in Fig. XIII–4. The masons’ marks and changes to profiles and elements like doors and window frames allow us to divide the twelfth-century section of the north tower into seven campaigns from the crypt level to the completion of the domical vault over level (1), Fig. XIII–5, where they are marked on the left as campaigns ‘A’ to ‘I’. The first two campaigns could have been carried out by one crew, and the large campaign ‘D’ could have been the work of two crews. More detailed analysis of profiles and the geometry used in the templates would be needed to resolve this question. The most significant connections between masons’ marks in the two towers are those that connect the crypt and lowest courses of the south tower with the section over the vaults of the north that culminated in the paving of level (2).12 This is the work in campaign ‘G’. It include the entry from the nave into the tower, with its two openings and complex capitals, Fig. XIII–6 and Table XIII–1. There is another large section of work in the south from the present cathedral floor level to tread 23. This is campaign ‘H’, and the masons’ marks in that section are shown in Fig. XIII–7. The sides of the royal portal from the bases to just below the level of its capitals was erected in this campaign. However, in the north the ‘H’ campaign was only a small one with little in the stairs. It may have included the cornice under the large windows of level (2). It was clear that the major effort at this time was being put into assembling the sculptural work. In the south between treads 24 and 37 there is a campaign with few masons’ marks. The masonry is in unusually long stones. This was campaign ‘I’ and included the large capitals under the ribs, along with the lower three courses of the ribs themselves that were laid with horizontal bedding — an elementary tas-de-charge. The next campaign in the south, ‘J’, included the completion of the ribs and cells over the ground-floor room, and the blind arcade on the outside. All the sculpture was in place by this time, and the walling over it finished off to the horizontal cornice above the portal, along with the sills to the triple windows, Fig. XIII–8.

12

The southern crypt extension may have been set out during the previous two campaigns ‘E’ and ‘F’.

building the narthex at chartres

Fig. XIII–4. Westworks with nomenclature for levels. Fig. XIII–5. Westworks with building campaigns.

185

186

Table XIII–1. Campaigns in the westworks at Chartres, with achievements of crews and their approximate dates.

building the narthex at chartres

Fig. XIII–6. Masons’ marks for campaign ‘G’. Fig. XIII–7. Masons’ marks for campaign ‘H’.

187

188

There is a small reduction to the width of the buttresses above the springing of the arcade that indicates the precise location of the joint between campaigns ‘J’ and ‘K’. The latter was another large campaign of some twenty treads that included the jambs for the triple windows. The fill over rib vault was completed, and the walls taken up to the cornice under level (1) and its openings. It may also have included the arches over that. It is interesting that some marks are found at every level, suggesting these are local masons who joined in with each team. There are other marks that match those found by Crosby at Saint-Denis, and though he does not detail their locations it would not be unreasonable for some of the teams working at Chartres to have also been working under Suger.13 The five phases in the sculpture could have coincided with the work of campaigns ‘D’ to ‘H’, which are noted as numbers 1 to 5 in the earlier chapter and Fig. VII–15. The extent of this work is described in Table XIII–1. The dates shown in Fig. XIII–5 are based on the only marginally secure assumption that each campaign took one year. Though unlikely the campaigns could have been longer than a year each, or they could have been less. This shows that the royal portal sculpture was fully in place by 1142±, though the first panels could have been carved by 1137 or earlier. In any case, the carving of the stones themselves would, in the main, have been done after that in the portal doors of Saint-Denis where three rows of arches and an additional storey lay between them and Suger’s consecration in 1140. There are small indications that some crews returned more than once, though more analysis would be needed, and where possible comparison with work by the same crews in other locations. The indications are that ‘C’ who built the openings into the north tower room returned around 1138 to set out level (1) in ‘E’ and again some four years later to spring the vaults over the south room. There is some possibility that ‘G’ who set out the south room may have worked earlier in another large campaign, ‘D’ and an even less certain connection between the crews in ‘A’ and ‘H’. Extensions to the nave after the 1134 fire At the same time as the towers were built, the space between them and the older Fulbert church were filled in. The form of this addition can be followed from evidence on the eastern side of the towers. 13

Crosby, “Masons’ Marks”

building the narthex at chartres

189

Fig. XIII–8. Junction between triplet windows and south tower.

Within the triforium roof space on the southern outer side just under the line of the present roof there is a deep groove with short lengths of sheet metal still let into it. Mortar and pebbles round the metal show that the stones were laid after the groove had been cut into them unless the joints had been pointed up to a considerable depth at a later date. Also, the inclination of this groove is flatter than that of the present roof, and would meet the outside wall somewhat below the upper face. The edges of the groove, and in places the mortar itself, has been scorched by fire. Hence I concluded that this groove predated the fire of 1194. Higher up on the same wall of the south tower, closer to the nave, there is a heavily carved storm‑mould, largely destroyed, but with a clearly cut angled recess under it to take the top of the flashing to an earlier roof. It is about 400 mm below the existing roof at its apex and therefore could not have anything to do with the present roof. So it must have been put in for the side roof to the earlier building, which was almost at the same height as the present triforium roof. Further, the door from the tower into this roof is

190

Fig. XIII–9. Section through Fulbert’s church (estimated) and the cathedral today. Fig. XIII–10. Section through narthex showing floor levels and vault over entry.

building the narthex at chartres

191

an original opening and gave access onto a gallery. The earlier roof may have been kicked up slightly over the head of the door as it projects above the line of the old roof. There are similar markings on the side of the north tower. Above the roof on the eastern side of both towers there are scorch marks on the stonework where the intense heat of the 1194 fire has permanently affected the limestone and burnt a ruddy tint into its surface. The shape of the burnt section looks as if a flare of flame had burst out of the wall below. If the flame had come from the roof the flame marks would be vertical rather than angled outwards. The pressure of superheated gases escaping out of the building burst outwards as they were escaping through a window. These marks consequently locate the position of the Fulbert windows. On the south side the original twelfth century stonework butts against the new clerestory in the upper part, but below the level of the gothic lancet arches. Some 800 mm of the tower next to the window has been refaced. The extent of refacing and the burnt walling nearby suggest that the new stones covered up the join between the upper part of Fulbert’s church and the tower. If the top of the facing marks the top of the old building it would have finished some 6 metres above the roof of the gallery. For the roof to have been as high as this the original church would have had an aisle plus an open gallery, which would mean that the height to the underside of its clerestory windows was about 18 metres, not unlike other galleried churches such as St. Etienne in Caen, St. Remi in Reims, or St. Etienne in Nevers, Fig. XIII–9. In most respects the size and arrangement of the present cathedral is not vastly different from that of its predecessor. This is not to say that the eastern part of Fulbert’s cathedral was necessarily this size or shape. For these comments only relate to the westernmost bays. At Tournus the narthex bears little relation to the rest of the cathedral and it may have been the same here. All we can say is that the western bays of the cathedral next to the towers had a three storey elevation. The de Mici drawing, made around 1130 and illustrated in Branner,14 shows the Fulbert church hovering over the clergy, and some have interpreted it as showing the cathedral in the background. If so then there are two rows of windows suggesting a two storey elevation. However, if the action of the drawing is happening inside the building the it lies over and around the participants rather than behind them. Though it was drawn from the outside it may show only the floors above those being filled with people, and hence would illustrate a three storey elevation. The column supporting the apse behind 14

Branner, Chartres, Plate 4.

192

the main figure reinforces this idea, as do the twelve windows in the upper floors which correspond to the twelve bays in the present crypt. The paving and old narthex entry The paving slopes upwards today from the portals into the nave. These levels look original. In the earliest plan we possess, Felibien’s of 1678 printed by Challine,15 we see the same levels and steps as we have today. The fact that there are steps between the aisles and the nave, and that the footings are so indecently exposed, suggests that this was not what was originally intended, and though there is not a great deal of evidence, I think that the floor may have been laid in a hurry to suit the urgent needs of the clergy when the lower temporary roof was being built over the nave in 1200.16 The aisle paving is higher than the nave because the vault over the crypt prevented them lowering it any further. The reason for lowering the central section of the nave is to be found just inside the western door, the threshold of which lies just over a meter below the thirteenth century footings, Fig. XIII–10. As Mayeux pointed out there are signs there had once been a vault just inside the western entry, set between the two towers.17 These marks are still visible on the inside walls, and they coincide with the bases for the small wall shafts still visible at D in Fig. XIII–11, and with the larger shafts attached to the transoms in the entries into the towers E. Along the inside of the west wall the two shafts marked C delineate a central bay wider than those on either side: a three‑bay arrangement in the west with one larger bay corresponding to the larger central portal. This could have formed a three-by-three bay narthex. There could have been a wall at W, possibly with three openings through it. However the Lefévre-Pontalis excavations did not unearth footings in this position, but just to the east of E, and to the east of W.18 He dated them to the twelfth century, but there was some dispute with Mayeux.19 The records of the excavations are far from complete. Challine, Chartres, 135. James, Contractors, 107. 17 Mayeux, “Façades successives”. 18 Lefévre-Pontalis, “Façades successives”. 19 Mayeux, “Façades successives”. 15 16

building the narthex at chartres

193

The columns C on the west are bonded into the portal wall and their centres coincide with the centre of the external projections of the portal, showing that both were set out together. The measurements between the columns on the south side of the narthex exactly match those in the west, confirming the evidence of the exterior stonework that the portal was built with the south tower. In this scheme the east wall W would have projected beyond the eastern face of the towers, leaving a small space between the towers and the nave in which there may have been stairs to allow access into the upper chapel. The plan of the crypt shows that its westernmost bays do not adjoin the towers, but that there is a similar gap there too, laid in solid masonry, the east face of which coincides with the east face of W. The wall W lies neatly over the footings found by Lefévre-Pontalis, the only difficulty is that there are no footings under the positions where I have drawn axes to show the location of the four isolated columns in the plan. Instead the 1901 excavations found a strip footing with engaged piers which would have supported two piers in this narthex, in line with those attached to the inside of the west wall, and with the engaged piers in the footing under the wall that would have separated the narthex from the nave. Was it finally built only as a three-by-two bay chamber, with twisted bays along the sides? Or was there a staircase to the upper chambers? We may never know. Narthex chapel We have accepted the fact that the western portal, the three lancets and their towers survived the fire, yet how would such an area have been preserved, especially the glass? It might be argued that the glass had been taken out to be re‑leaded, but that asks for too much coincidence for my taste. The scorch marks on the eastern walls of the towers show the fierceness of the fire and that it was perilously close to the lancets. Not only was the western roof alight but also the roofs of the aisles and galleries. The inferno must have been immense: so how under those conditions could the lancets have survived? Only, I believe, by being protected, as in modern fireproof construction, by vaulting and firebreaks, with walls of sufficient height and thickness to deter the spread of flames. There was probably an opening between this upper chamber and the main body of the church as there

194

was enough to affect some of the lancet glass, which shows signs of being slightly burnt. But as the glass would not have been easily seen from the nave this chamber may have been built as an Easter Resurrection chapel, as suggested by Libby Parker.20 Since there are signs of the 1194 fire in the north west corner of the uppermost floor of the south tower (changed capital on north-west corner, altered mouldings around the pinnacles) the roof over the narthex must have burned too. It did not collapse into the space between the towers, or the glass would have gone. One explanation offers itself: not only was the last bay between the towers cut off from the rest with a wall, but it was also separately vaulted. It was, in short, a fireproof chamber. This might explain some of the anomalies in the remains between the towers — the short columns, the curious arrangement of openings and so on. If this theory is correct, then this wall and the eastern bay of the narthex would have been demolished in 1194–5 to make way for the end piers of the new work, shown dotted in Fig. XIII–7. However, they did not demolish the whole of the narthex, for there is evidence on the inside walls of the towers that the western two thirds of the floor of the upper chamber remained until the sixteenth century, or later. These walls show the signs of a partly completed demolition job. Starting from the springing capitals of the high vault and working downwards, the shafts under these capitals had been installed in the thirteenth century, with the same backing wall, and the same supports and en délit shafts in between. The metal dowels which held the en délit sections are still in place, projecting through the key stones, showing that at one time the whole assembly had once been complete. The few remaining portions of en délit work indicate that demolition was interrupted before they had finished pulling out all the shafts. Lower down, just under the line marking the edge of the narthex vaults, post‑Gothic egg‑and‑dart mouldings had been carved into the stonework. The columns in the western corners of the narthex are also of this date, so it seems that at some time contemporary with the egg‑and‑dart moulding and before Felibien drew his plan in 1678, the old narthex was removed. The purpose may have been to make room for a new organ loft across the western wall. There are many examples of this being done elsewhere at this time, with new organs invariably blocking the light from the western windows. It 20

Parker, Descent from the Cross.

building the narthex at chartres

195

Fig. XIII–11. Plan of narthex as it may have been before the fire of 1194.

is tempting to imagine the factious controversy which may have raged within the Chapter over this decision, and that it continued until those who loved the lancets forced another dramatic change in plan which left the western end of the building permanently disfigured. The remnants from the demolition, and the slope on the paving, suggest that they demolished something that looked like Fig. XIII–11. The presence of this narthex in 1200 would have forced the builders to lower the paving when the temporary roof was built over the nave to give headroom at the entry to the nave.21 It was the easiest and quickest solution. In 1194 they 21

James, Contractors, 107–8.

196

may have had the intention to pull out the narthex vault, but left it as it formed an entry into the crypt and protected pilgrims from the building work overhead as they passed into the temporary chapels down below. So the vault was reprieved for the time being. Temporary things have a tendency to become permanent: it is still with us today.

Vaulting

XIV The Evolution of Structure between 1080 and 1150

O

PINION is hardening that rib vaults were first employed before 1100, and not in northern England nor Lombardy. I believe that the first ribs were erected in a series of coastal towns from Aversa in southern Italy, along the Atlantic seaboard in Saintes and Lessay, to southern England at Christchurch and Exeter. It has been thought that ribs were not employed in the Ile-de-France until some ten years later, and that when they were, the impetus came from Anglo-Norman and Italian sources. I no longer hold to that opinion.1 The basis for this is a survey of the 376 projects with rib vaults built before 1150. The many smaller and hitherto neglected buildings in England and France provide many of the missing links that show the continuity within a gradual evolutionary process.. The ribs at Durham are usually considered to be the first. They show such exquisite planning and erectional skills it is hard to believe the master had not had some experience elsewhere.2 It appears as if every eventuality had been thought out prior to construction, not worked out on the fly, which is more normal in experimental works. It is therefore easier to credit that ribs were to be found in earlier buildings. There were three waves of invention. The dates of the first coincide with the introduction of the first pointed and peaked arches into Europe. All three seem to have reached Europe at the same time and from the same

1 2

“XV: Peaked arch”, “XVI: Rib vaults of Durham cathedral”,

200

source, the Middle East, and most likely Persia. This was during the 1080s and 1090s by my reckoning.3 The second wave of invention took off from this, between 1100 to 1120 in two areas. One was England and Normandy. These vaults had level crowns and flattened segmental ribs, and were constructed as if they were groin vaults with the ribs acting as a cover mould. The other was in the Oise and Valois. These vaults were domical and also constructed as if they were groin vaults. They are found in a few little-known small churches.4 Around 1120 there was a major structural invention, when a neat constructional device became one of the major generators of the new style. The Oise builders began to erect the rib before the cells, rather than building both together as if it was a groin vault. This halved the cost of formwork and considerably reduced the filling needed over the vaults. By 1135 these techniques had supplanted all others. It was a period of structural and spatial inventiveness that culminated in the ambulatory of Saint-Denis.5 These techniques became the basis for all rib vaults for the next two hundred years. There are details that suggest that considerations of formwork may have been the most important factor in the evolution of the rib. The third wave from the later 1110s to 1135 took these ideas and transformed them. In England and Normandy up to the chaos of the late 30s a coherent and sophisticated technique was developed. Huge vaults with ribs were built over great spans at Durham and Caen and Montivilliers, though they were still being constructed in the manner of groin vaults. Some understanding of their structural potential can be seen in the use of ribs to reinforce groin vaults in the transept at Winchester and the crypt at Gloucester. In Lombardy the earliest vaults were not begun until after the earthquake of 1117. They were groin-style, and had level crowns like those in England and Normandy. It was only in the later vaults built after 1130 like San Ambrogio that the domical form was used.6 Except for Bilson’s and Aubert’s studies written long ago, no modern work has attempted to make a major study of the rib. Yet it is of primary importance. It can be argued that from its diagonal placement a new and James, Ark of God, vol. 7, in preparation. “XV: Peaked arch”, 5 “VIII: Saint-Denis chevet” 6 “XVII: Rib vaults in Italy”, 3 4

the evolution of structure

201

dynamic attitude was developed that profoundly affected all aspects of architecture. Instead of seeing the building as a solid block framed in thick walls, it came to be visualized as thin axes connecting centres of support. The whole concept of the Gothic building as a skeletal structure seems to have been derived from this change in perception. Flying buttresses may have been stimulated by the realisation that ribs graphically demonstrated that lateral thrusts could be focussed on the piers.7 They were invented only after the ramifications of structurally independent ribs had been absorbed, around 1150.

7

“XVIII: Evidence for flying buttresses”

XV The Peaked Arch, and the Earliest Domical Rib Vaults in the Paris Basin

M

EDIEVALISTS often prefer a mixture of French and English terms for the arches that encompass a vault: the shorter French ‘doubleau’ is used more often than the longer English ‘transverse arch’. In addition, the following words are used in particular ways in this article: • level or domed refers to the three-dimensional form of the vaulting space; • round, segmented, pointed or peaked to refer to the shape or outline of an arch in the vertical plane and therefore to the shape of the centring which supported it; • torus, scotia or circular refers to the cross-section through the stones, being the profile marked on the template. Pointed arches and rib vaults were first employed in European architecture shortly before 1100.1 The pointed arch stemmed from Mesopotamia, probably via Sicily, and was rapidly accepted in France, though not so quickly elsewhere. The long-term impact on the evolution of Gothic architecture was profound. The round arch, whether with stilts or not, will be struck from a single centre; the pointed arch from two so that the arcs intersect at the top. There has been much discussion on how these centres were located along the line of the springing, but I have found no discussion of the use, at this early time, of arches constructed from a number of curves from more than two centres, Fig. XV–1. 1

Bony, French gothic, 17–18 and 468. The earliest true pointed arch is that of the 6th century, Qasr-ibn-Wardan in Syria: Creswell, Early Muslim architecture, 102–04. Photos in Pope, “Iranian contributions”; Dieulaloy, L’art antique; and Harvey, “Further thoughts”.

THE EARLIEST DOMICAL RIB VAULTS

Fig. XV–1. Airaines, nave to east.

203

204

Fig. XV–2 . Peaked arch with joints between voussoirs marked.

Fig. XV–3. Chartres cathedral, nave vaults.

THE EARLIEST DOMICAL RIB VAULTS

205

I understand this motif not being noticed. It is difficult to observe if one is not looking for it, and one often has to stand in a certain position to see it. I will call it the ‘peaked’ arch. Though historians of Islamic architecture refer to the same arch as ‘keeled’, where it is found in a multitude of buildings between Persia and Egypt from the ninth century, in the west the word ‘keel’ is used to describe profiles with curved faces meeting at a point, not unlike a pointed arch in section. The outline of the peaked arch being entirely different, referring to them as ‘peaked’ will prevent confusion. In them the lower voussoirs are curved to a short radius as if for a much smaller span, and the upper to a much longer radius that looks as if it was almost straight. The upper curvature is often so slight it gives the impression of acting like a ‘beam’, Fig. XV–2. This multiple-centred pointed arch occurs in parts of the Paris Basin at virtually the same time as those with two centres, some time before 1120. It continued to be popular for the next century and a half, occurring in more than one third of all churches, especially in the high vaults. Also, my far from complete observations suggest that the peaked arch is relative rare in other parts of Europe, suggesting that there may be reasons for its almost exclusive adoption in the Royal Domain. In the clerestory of the cathedral of Chartres, Fig. XV–3, most of the doubleau and many ribs are peaked, including those around the crossing. There is little consistency across these great spans of the high vaults where few arches follow exactly the same outline. The amount of peaking was dependent on the curvature and stilting established in the bottom courses. In some churches the double curvature may have resulted from distortion with settlement or damage over the centuries, and this could tighten the curvature in the haunch. However, in the examples we shall be discussing are part of rib vaults in which the space above the springing has been filled with mortar to such a height that such distortion would be unlikely. For greater certainty we would need to know the history of each building’s restorations backed up with photogrammetric or similar measured surveys. It seems not insignificant that they occur mainly in the Paris area, that they are contemporary with the first pointed arch, and nearly always occur under some of the earliest rib-vaults in the area, all of which are domical in form. These characteristics may identify one master mason who introduced the motif into domical rib vaults. Though I believe these building should be dated to the 1110 to 1120s, no precise argument can be mounted until the forthcoming analysis of dating through the capitals has been published in Part B of The Ark of God.

206

The advantages of a peaked arch. With the double-curved peaked arch I have the impression that the masons envisaged that the short-span arch in the lower section supported a beam producing minimal lateral thrust. If this were the case, then it shows an understanding of arch construction and lateral thrusts that seems more sophisticated than needed for the pointed arch. The concept of using only a slight curvature to maintain stability was later used to stabilise the cells between the ribs. There are four reasons for its use: engineering, formwork, constructional and contracting procedures. Engineering: In a normal arch the load shifts from one voussoir to the next, and slips slightly downwards at the junction between each stone. This continues from the apex to about two-thirds of the way down the arch, where the thrust has developed a greater lateral vector than can be contained within the turn of the arch, and therefore exits beyond the thickness of the voussoirs, Fig. XV–4. We still don’t understand its precise dynamics, but photoelastic analysis suggests that pointing the arch does appreciably reduce the outward thrust.2 If an arch is not to collapse at this exit point it must be buttressed in some way, both externally and by filling the space between the arch and the adjacent wall to the exit point or higher with a stone and mortar mass that turns the lower part of the vault into a solid block. The situation in a peaked arch is slightly different. In the upper part the loads are contained within the depth of the beam. It is only in the curved lower section that the line of thrust exits the curve of the arch. As this section is seldom more than one third the height of the arch, the lateral thrust leaves the voussoirs at an invariably lower point than where it would have been if the arch had a single curve, Fig. XV–5. Peaking lowered the line of thrust as if it had been a smaller opening, thus increasing the stability of the building. Formwork: Slightly curved formwork can be assembled on straight beams with different sized chocks under each stone, whereas curved centring has to be made up from many specially cut and shaped pieces of wood, which is an expensive and time-consuming operation. In peaked arches the lower portion can usually be erected without any centring, and formwork

2

Mark, Gothic structure; Heyman, “Masonry arches”; and Abraham, “Probléme de l’ogive”.

THE EARLIEST DOMICAL RIB VAULTS

Fig. XV–4 . Pointed arch with line of thrust shown dashed.

Fig. XV–5 . Cambronne-lès-Clermont north transept from crossing.

207

208

Fig. XV–6. Morienval, choir vaults. Map XV–1. Paris Basin showing churches with peaked arches before 1130.

THE EARLIEST DOMICAL RIB VAULTS

209

under the straight section was readily reusable.3 The advantages of this system would have been considerable, especially as serviceable timber was becoming scarce in some regions around Paris.4 Constructional: To take one example, in the clerestory of the cathedral of Chartres, Fig. XV–3, most doubleau and many ribs are peaked, including those around the crossing. There is little consistency across the great spans of the high vaults where few arches follow exactly the same outline. The amount of peaking was dependent on the curvature and stilting established in the bottom courses. Contractual: At Chartres four masters, Rose, Scarlet, Bronze and Ruby, cut and placed the tas-de-charge in the nave and choir, each with their own ideas on what the height of the crown of the vault was to be.5 Rose was preparing for a boss height of 36,700 mm. from the floor, his successor Scarlet for a height of 120 Roman feet at 35,450 mm., and Bronze, who actually set the bosses in place, and was therefore in charge of the final centring and placement of the straightened upper stones of the ribs, at the somewhat lower 34,360 mm. With different teams cutting the lowest stones, the two-meter difference in the heights would have affected the curvature required by each master’s design. Peaking was a practical solution to these varied conditions. Early peaked arches in the Paris Basin. To my knowledge, the earliest peaked arches in Europe appear in the Paris Basin, all north of the Seine and Marne rivers, Map XV–1. They are in ten buildings that were erected prior to 1130, Fig. XV–5. It is of the utmost significance that all ten buildings have rib vaults, they are always domical in form, and the ribs are often pointed. All except Airaines have ribs carved to a similar profile, and six have capitals angled precisely in the direction of the ribs, Fig. XV–6. Table XV–1 describes their more telling characteristics. Outside the Paris Basin I am aware of only one peaked arch of this period, the crypt under the Ripon Chapter House after 1110. Around 1130 there are others in the naves of Worms cathedral and San Ambrogio, Milan. Whereas the peaked arch is found in half the vaults in northern France before Fitchen, Gothic cathedrals. Roblin, Le terroir de Paris; and Devèze, La vie de la forêt. 5 James, Contractors, ch. 18, 458–59 and 478–85. 3 4

210

1160, there are only two among 80 English ribbed vaults — the doubleau of Kirkstall and St. Peter in Oxford — both after 1150. It was a north French phenomenon, and by the 1180s the peaked arch is found in some part of every major church in northern France. I have visited almost every European rib-vaulted church built before 1145 gathering information on the evolution of the rib to be published in Part C of The Ark of God. All the vaulting statistics quoted here come from this survey. Prior to 1130, 34 rib vaulted buildings in Western Europe have domical crowns, while 54 are level, though with some small undulation. In level crowns the bays form a connected sequence. By contrast, domical vaults emphasized each bay at the expense of the whole. It is therefore particularly interesting to note that within the Paris Basin this proportion is reversed, as 24 buildings are domical compared with only 11 that are level. Of the ten domical vaults not in the Paris Basin, only San Ambrogio in Milan has pointed ribs. These figures show that the conventional view that the north preferred level crowns and segmental ribs while the south preferred domical crowns and pointed ribs is not necessarily the case. Identifying the Master of the peaked arch In the Ile-de-France eighteen vaults form the core of this study. They are listed in Table XV–1, in approximate chronological order with their characteristics. The first three columns refer to the spatial form of the vault, Very Domical, Domical or Level. Ones I call ‘very domical’ are those that appear, from visual observation, to be more than one-and-half times as high as they are wide. Level may include those in which the cell is level in one direction and very slightly sloping in the other. The peaked arches, marked ‘^’ in the fourth column, were employed in the doubleau, ribs and, once, in formerets. The next column marks the form of the doubleau, rib and formeret, in that order. The symbol ‘^’ is peaked, ‘r’ round and ‘p’ pointed. Though there was considerable variety and experimentation, nearly every arrangement raises the volume of the interior. Most of the peaked arches occur in smaller buildings, suggesting that its inventor was either a smaller builder or recognised that it was less risky to experiment on a small scale than on a larger.6 The largest spans in the 6

Vermand, “Voûte d’ogives”, 137.

THE EARLIEST DOMICAL RIB VAULTS

Table XV–1. Details of Paris Basin buildings with peaked arches.

211

212

Fig. XV–7. Rib profiles used in churches with peaked arches. Fig. XV–8. Rib profiles used in other contemporary vaults.

THE EARLIEST DOMICAL RIB VAULTS

213

Beauvais transepts and the Morienval choir are among the latest in this group, Fig. XV–6. The profile used in the ribs of these buildings was not unusual at this time, left Fig. XV–7. It is marked ‘OvO’ in column 6. It occurs in a number of English and Norman churches in the 1120s, Thus its use in the Paris Basin at the same time is not exceptional, except in so far as it appears in more buildings than anywhere else. There is a wider version of this profile in which a short flat section separates the lower splays, marked ‘O_O’ in Table XV–1, and another in which the pointed middle section has been rounded, marked ‘OuO’, shown centre and right Fig. XV–7. The spatial form of the vaults with the ‘O_O’ profile is mostly rounded arches and level cells, while the latter is nearly always with pointed arches and somewhat domical cells, but never peaked arches. Together, over half the rib-vaulted churches in the Basin built prior to 1130 have these rib profiles. The profiles used in other churches are quite different, Fig. XV–8. It had been normal practice before the introduction of intersecting ribs to set imposts and capitals square to the walls or piers. After 1110 capitals were at times angled to face the direction of the rib, and were therefore placed either at 45o to the wall or — more subtly — directed tangentially at the boss.7 These three arrangements are marked ‘#’, ‘