The Byzantine monuments and topography of the Pontos, Vol. 2 9780884021223, 9781597406772

Citation preview

DUMBARTON OAKS STUDIES

XX

THE BYZANTINE MONUMENTS

AND TOPOGRAPHY OF THE | PONTOS

wee Nene NN ~. SS ‘s — SS ~ Le 3a REN . > . a oN ; ‘ . . - = - * nS ~ a WZ ad a

q\ NO By.Ang, (RABZON~ QQNSRNENS FON pmP whereas the plateau village is a small nucleated inhabitants of each side of the mountains. The coastal peo- huddle of houses not much different, until very recently, from ples have always included traders, whereas on the other side the settlements of six thousand years earlier. of the mountains subsistence farming dominated the normal On the seaward side goods were transported by horse, pattern of life, and in addition to this economic difference mule, donkey, or woman, since the steep valleys were only there existed until 1922 religious and ethnic differences large traversed by narrow tracks unsuited to carts, whereas on the enough to foster a division of peoples for which nature had plateau transport was for the most part by the solid-wheeled

already provided so apt and rugged a barrier. ox cart, admirably suited to the conditions in which it operClimate is responsible for the contrast in landscape. The ated, and little changed in design since the Bronze Age. On typical plateau landscape is an arid plain bounded by ochre- the coastal side carts or waggons are only to be seen at the ous hills, with a watercourse running through it where pop- western end of Pontos where gentler hill slopes and delta lars and willows provide a relieving touch of green, the plains make the use of them practicable. whole being encompassed by a clear blue sky. The conscious- Sea fish are not eaten by the inland peoples, and the area ness of change comes along the Pontic mountain ridges within which villagers partake in the annual autumn feast of where the divide is marked by dense masses of billowing hamsi* is still a good rough guide as to the boundaries of the cumuli trying in vain to spill southward over the plateau coastal region. Water buffalo are the commonest cattle of the country, and breaking up into a few thin cloudlets, thence to plateau, whereas they are hardly to be seen on the Pontic become mere wisps which dissolve into the clear plateau air. coast except on the deltas. The Pontic village has always Northward of the watershed the heavy rainfall sustains a lush farmed hazelnuts and walnuts as cash crops, and grown little green landscape of crops or forest-lined narrow valleys which in the way of cereals, whereas the plateau village grew cereal for the most part extend northward from the watershed

down to the sea. a 2. The most notable accounts of zarzaka (tezek) are by Leo of On the seaward side of the mountains, the traditional — sypnada, in J. Darrouzés, Epistoliers byzantins du X® siécle (Paris, houses are either constructed of wood with a masonry fire- 1960), 198-99; Curzon (1842), 119-14: E. R. Huc, trans. W. Hazlitt, place and chimney, or else of a timber frame with an infilling Travels in Tartary, Thibet, and China during the years 1844-5-6

(London, n.d. [1851]), II, 89-90 (on the four Tibetan varieties of 1. The importance of the region’s singular geography to Pontic argol); and L. Robert, “‘Les Kordakia de Nicée, le combustible de history was first, and most attractively, demonstrated by Fall- Synnade et les poissons-scies. Sur les lettres d'un métropolite de merayer, Trapezunt, 286-312. Among modern commentators, de Phrygie au Xe siécle. Philologie et réalités, 1," JSav, (1961), 115-66. Planhol’s work is the most challenging (and, sometimes, mislead- 3. The churches no longer mark village centers since the Christian ing): ‘‘Chaines pontiques,” 2-12; and Fondements, passim. population has left. The modern village center is usually on a new A.A.M.B. has made three attempts to relate Pontic geography to road and consists of a tea house and a shop or two, with a new history: Thesis (1967), I, 31-84; H, 7-27; Neo-Hellenika, 1 (1970), mosque. The older mosques are few in number and seldom coincide 33-36; and DOP, 29 (1975), 93-96. The text of this chapter rep- with the modern centers. resents D.C. W.’s turn to present the subject, to which A.A. M.B. In East Pontos the settlement pattern might have a Caucasian

has largely contributed the footnotes as commentary. origin and is in contrast to the high nucleated villages (e.g., Santa) We owe the suggestion for the derivation of **Halt’ to Professor established later: see Bryer, DOP, 29 (1975), 120.

O. R. Gurney. On the ancient Chaldaioi (later confused with 4. The hamsi, which are anchovy, appear in shoals in the autumn Chalybians—see note 6), see Xenophon, Anabasis, IV, 1n, 4; V, v, 17; and become the occasion of feasting after the manner of those and Strabo, Geography, XII, m1, 18-19. Cf. Bryer, DOP, 29 (1975), marked by the arrival of sprats on the east coast of England or

73. grunion on the coast of California.

2 CHAPTER ONE extensively. Common to both sides of the mountains are area in the east which in the middle Byzantine period coinsheep and goats, but different again are the dogs which guard cided roughly with the Georgian principality of Tao. Our them. On the plateau it is the aristocratic karabas, a large area 1s thus a very large tract of country and we have ceranimal of the mastiff breed which fights off the wolves, tainly not explored all of it thoroughly. We make this clear in whereas on the coast the dogs are smaller; typical of them is our more detailed coverage of the regions. the zerdava, a thoroughbred animal somewhat resembling a The main geographical feature of the Pontos is a range of

collie, which is bred in the Tonya valley.° mountains running from the hinterland of Themiskyra These few impressions may serve by way of introduction (Terme) in the west to Apsaros in the east. This east-west to a more detailed survey of the land and its people, revealing chain of mountains forms the spine of the Pontic Alps and that while contrasting impressions are first and strongest in from it lateral ranges branch out to north and south forming

the mind of the traveler, there is a blurring of distinctions an intricate pattern of ribs. This elevated mountain backwhen the land ts studied in more detail, and no clear line can bone with its diverging ribs is the determining factor in the

be drawn between the Pontic coast and the Anatolian character of the major and minor features of the region. The

plateau. line of the central spine is irregular, with the watershed now nearer and now farther from the coast; south of Trebizond-

GEOGRAPHY Trapezous (Trabzon), where the river Philabonites (Harsit) In defining the limits of our survey we have triedtomakea CUS its way deeply into it, this central chain doubles back on reasonable compromise between the claims of geography itself. West of Themiskyra the Iris runs southwards through and history. Historically, we have taken the Empire of 7? 84P which cuts off the mountain spine, while the Akampsis Trebizond at its greatest extent, and its peripheries, as our at the eastern end divides the Pontic Alps from the Caucasus. boundary. Ethnically, it represents, very roughly, the area of The remaining coastal strip at the west end between ancient, medieval, and modern Greek settlement or in- Themiskyra and Sinope resembles the coastline of Bithynia fluence. So, geographically we have set our westward limit at more than the Pontos proper. The land rises steeply southCape Karambis, most northerly geographical point of ward from the sea to 1,000 m or more, reaching the mean Anatolia and most westerly of Trebizond’s medieval out- height of the Anatolian plateau, so that while a watershed posts, and our eastward limit falls at Bathys (Batumi), the exists to divide the coastal valleys from the inland valleys, it 1s historic border where the coastline turns northward and the not the spectacular feature that it becomes in the Pontic Alps. river Akampsis (Coruh) breaks through the mountain bar- The geological skeleton of the Pontic region took on its rier to force its way to the sea. We have landed at both Cape _- Present form in the last era of great earth movements which Karambis and Bathys, but the Soviet border limits investiga- threw up the Alps, the Himalayas, and the Andes, and detion of the final, eastern, stretch of the Pontos. Southward ‘termined the general shape of sea and land as we know them our boundaries are not marked by the watershed of the first. | '0day. The major part of the chain consists of Upper mountain chain, as would appear to be good geographical Cretaceous volcanic rocks, while at the eastern end the

sense, but by the east-west valleys of the rivers Lykos | Mountains rise to a height of nearly 4,000 m, south of (Kelkit), Iris (Yesil), and Akampsis, all of which turn north —- Rhizaion (Rize) at Kackar, and this massif of the Tatos to flow into the Black Sea and are vital to the history of our mountains consists of intruded granites and diorites. Lesser region. But we cannot always keep strictly to these river formations of the same intruded rocks are to be found at high valleys, and our southern limits are in fact as untidy as isthe | Points westward along the mountain chain. The great easthistory of these regions. A fourth river, the Halys (Kizil) west valleys south of the Pontic Alps mark faults which flows into the Black Sea within our area at Paurae (Bafra), developed when the mountains were thrown up, and they are but it is the greatest of the Turkish streams and pursues a still subject to earthquake; majestic in scale, they are far too course through central Anatolia which takes it through a _—~VaSt to have been formed by mere erosion, even Ona geologhistorical picture of wider dimensions than ours. It plays ial time scale. The most important of these rifts now form little part in the history of the Pontos, but a major one in __ the river valleys of the Akampsis flowing eastward and the Anatolian history, where its course is inextricably threaded _Lykos flowing westward. On a secondary scale are the rift into the major epochs of the past. Its headwaters are east of Valleys of the upper reaches of the Philabonites south of Sebasteia (Sivas) and not far south of the boundaries of | Trebizond; the upper reaches of the Melanthios, south of central Pontos; from there it winds its way south westward | Kotyora (Ordu); the valley of the thrice-named river which ts into Cappadocia and central Anatolia and thence makes a a tributary of the Iris and starts as the Bag, continues south of

great bend northward to the sea near Paurae. Oinaion (Unye) as the Karakus, and becomes in its upper Within the geographical limits defined above we take in reaches the Bakircik; and the valley of the river Amnias most of the old provinces of Hellenopontos and Pontos (GOk) south of Sinope and Paurae. The aspect of these river Polemoniakos, parts of Armenia, and an undefined extra valleys varies from that of a wide and fertile valley bottom, with sloping hills on either side of a meandering river, to

. gorges of cliffs, containing raging torrents. The Akampsis o wih Caspian regi ondeme mh pas um aero orate out that these and the Philabonites have far to fall in their short courses and areas of ‘‘foréts refuges littorales” have a certain geographical iso- BOTBES are frequent, while the Lykos and the Iris are larger lation and cultural tradition which made them long resistant to the rivers and tend to run through stretches of gorge which widen

influence of Islam. out into valleys and even into large hospitable basins of

TOPOGRAPHY OF THE PONTOS 3 fertile land. This pattern is important in determing the lo- There were probably silver deposits at classical Argyria,

cation of towns. near the mouth of the Philabonites'® where Hamilton saw Much of the basic geological structure is still bare to the traces of mines,'' and silver was probably mined somewhere eye but the valley bottoms, coastal deltas, and some coastal in the mountains to the south of the upper Philabonites terraces have been modified by deposits of more recent sed- valley in the medieval period. If so, it was possibly between

imentary rocks, clays, and gravels. Tzanicha (Canca) and Paipertes (Bayburt) and probably out The volcanic period of Pontic geological history produced of the hands of the Grand Komnenoi of Trebizond; at all quantities of mineral-bearing ores of different kinds, and events these mines should not be confused with the later ones some of these have been worked since the earliest ages of | of Giimtishane.!? metalworking. Indeed the Chalybians are credited with the Among other minerals the red earth of Sinope was famous

invention of ironworking,® and gave their name to steel in for its quality in antiquity, and in the medieval period medieval Greek; from then it passed into medieval Latin as ‘‘sinoper” become a synonym for red earth. Probably other ‘‘Calibs’’ ’ and the name of the mineral chalybite is derived earth colors were also produced in the region in medieval from them. The work of smelting iron continued down into times, since they are not difficult to find even today. the nineteenth century when Hamilton was excited to find the According to Pliny, alum was mined in Pontos in antipeople of the Oinaion region practicing their craft in much quity; in the Byzantine period it was mined near Koloneia the same manner as they had done when Xenophon observed (Sebinkarahisar) and this was probably the source of alum

them.® known to Pliny.'* It was an important export of the region.

Skilled metalwork is still a living tradition along this coast, The climate of the Pontic region is dictated by the land where fine knives and daggers, and good copies of factory- forms outlined above, and falls into two distinct categories. made revolvers are produced in illicit village workshops, with Along the coastal strip and inland as far as the watershed of

only the simplest of tools. | the Pontic chain it falls between the mild temperate and the

The number of mines listed by Cuinet in 1890 is large. In warm temperate type of climate with considerable variations the sancak of Trebizond, which comprises the modern in temperature and rainfall. The high rainfall along the whole Trebizond, Giresun, and Ordu districts, there were twenty- coast is caused by the prevailing northwesterly winds sweepone mines of argentiferous lead, thirty-four copper mines, _ing across the Black Sea and precipitating their moisture as three of copper and lead, two of manganese, ten of iron, and they hit the Pontic land barrier and rise with the mountwo of coal. In the sancak of Gimushane there were thirty- tains.'* There is no great seasonal variation in the rains and seven mines of argentiferous lead and six copper mines.’ The

ch mj Id B t d d by th PP | t 10. Arrian, 24: see p. 139.

rich mineral deposits evidenced by these mines are almos I1. Hamilton (Researches, 1836). 1. 259. absent at the western end of our region, where the sancak of 12. The question has been bedeviled by misidentifications. The Samsun had only one mine of argentiferous lead. But there is facts seem to be these. In 1294 Marco Polo noted silver mines near

no direct evidence that any of these mines were worked in Bayburt and Erzincan. These seem to be identical with the mines

Byzantine times noted by Al Umari as active in 1332/33 at Bayburt and Kumish (Gimus ?), for Ibn Battutah found the Kumish mines west of Erzincan in the same year. In the 16th century Tzanicha (Canca) was

6. On the effects of silting in Byzantine times, see C. Vita-Finzi, a silver mint, closing down in the period 1574-1644. It is possible The Mediterranean Valleys. Geological changes in historical times that its silver came from mines to the south, toward Bayburt, and (Cambridge, 1969), 77-88, 116-20. On the Chalybians, add to the that these were the ones recorded by Marco Polo and Al Umari, but otherwise exhaustive references in Magie, Roman Rule, II, 1068-70: they were flooded and abandoned before 1661. However, new mines Clavijo (1404), ed. Estrada, 73; trans. Lestrange, 108; Lazaropoulos were opened close to Tzanicha before 1644, when Evliya Celebi in Papadopoulos-Kerameus, FHIT, 61; Panaretos, ed. Lampsides, records for the first time an alternative name for the settlement below 63, 73, 80; and X. de Planhol, *“Geographica Pontica: Les noix des Tzanicha: Gumiushane. The Greek name of this place, Argyropolis Mossynéques; II: Les Khalybes, Nom de peuple ou qualificatif pro- (a simple translation of Gumishane) was only adopted in the 19th fessionel?”’, JA, 251 (1963), 293-309. De Planhol maintains the non- century. Thus Yule and Gibb are mistaken in identifying Kumish geographical nature of Chalybia, but ignores the medieval evidence, with Gumishane (a place which did not then exist); on p. 26 we Curiously (apart from Stephanus Byzantinus, who draws upon anti- argue that Kumish is at Maden Dere or Gimiulsakar, 70 km west of que sources) the district of Chalybia is not mentioned in Byzantine Erzincan. Vryonis was still more mistaken in stating that “‘it is sources (i.e., before the 13th century), but there is no doubt of its interesting that Marco Polo still refers to Gumushane by its earlier survival in Trapezuntine times thereafter. Is it possible that Chalybia Byzantine name, Argiron” (i.e., Argyropolis [sic]): an alternative

and the Halys share the same root? reading of Arcingan demonstrates that Argiron was Erzincan. See

7. Theophilus, De diversis artibus, ed. and trans. C. R. Dodwell Polo (1294), I, 46 and 49 note 3; Marco Polo, The description of the (London, 1961), 162, giving the derivation of the word according to world, ed. A. C. Moule and P. Pelliot (London, 1938), I, 21-22, II,

Latin tradition. p. vi: Ibn Battutah (1332), I], 436-37; C. Defremery and B. R.

8. Hamilton (Researches, 1836), 1, 273-78. Cf. Xenophon, Sanguinetti, Voyages d'Ibn Batoutah (Paris, 1949), II, 293; Al Umari Anabasis, V, v, 1; Apollonius Rhodius, Argonautica, book I, line (1342-49), 337; S. Vryonis, Jr., ““The question of the Byzantine 1323; bk. II, lines 375, 1475. Clavijo’s description of ironworking in Mines,”’ Speculum, 37 (1962), 8-9 and notes. On Gimishane, see the same region (see note 6) provides the only direct evidence for p. 303. mining of any sort except alum in the Empire of Trebizond. On 13. On sinoper, see Pegolotti, ed. Evans, 296, 431; on alum, see copper, see Hamilton, (Researches, 1836), I, 259; Cuinet, Turquie Pliny, Natural History, XXXV, 52, and p. 148 below. d’Asie, 1, 56-58, 68; V. J. Parry, ““Materials of War in the Ottoman 14. Monthly Bulletin for Statistics, Devlet Istatistik Enstitust IEmpire,” in Studies in the Economic history of the Middle East from IV (Ankara, 1971), 6 ff. The monthly rainfall figures for Trabzon the rise of Islam to the present day, ed. M. A. Cook (London, 1970), include: January, 90 mm; April, 56 mm; July, 37 mm; October,

225. 109 mm. In these statistics, the yearly average is about 875 mm, but 9. Cuinet, Turquie d’Asie, I, 17-18. see following note.

4 CHAPTER ONE certainly no dry season, and overcast grey days seem to the in the previous two millennia. Up to a height of about 1,200 resident there to be almost as frequent in the summer as in the meters there are broad leaf forests, with beech, oak, chestwinter. While the seasonal variation in rainfall is not so very nut, maple, alder, elm, hornbeam, lime, and plane trees—all

great, there are considerable differences in the amount of native to these mountains. In the hinterland between precipitation along the coast; and it is to be noted that, while Themiskyra and Kotyora, beech and hornbeam predorainfall is heavy over the whole region, there is less of it at the minate in the remaining forests, but centuries of felling in the

western end and in the Trebizond area; more of it in the hinterland of Kotyora and Oinaion have reduced the forest Kerasous (Giresun), Tripolis (Tirebolu), and Kotyora to a few stands in the remoter valleys. These were the regions (Ordu) areas; and most of it at the eastern end from Rhizaion inhabited by the Chalybians and the need for charcoal onward to Bathys, which French sailors describe as ‘le pissoir for smelting must be one of the prime reasons for deforesta-

de la Mer Noire.’ These variations are due to the lie of the tion. In the hinterland of Trebizond the forests are again land. Stretches of coast which face the prevailing windshave thin, perhaps because of the relatively dense population the highest rainfall while those at an oblique angle to the of the mountain valleys, but eastward of [SouJsourmaina winds and to some extent protected from them are subject to (Stirmene) the forest grows denser again. The principal trees

a lesser precipitation.'° of the broad leaf belt at the western end are beech, oak, Temperature variations are of course primarily dependent maple, hornbeam, and alder. Above the 1,000- to 1,200-m on the height of the land and there is no simple way of contour line the forest becomes primarily coniferous, with reducing them to sea level equivalents. In general the coastal spruce, firs, and Scots pine predominating. Towards 2,000 m climate is humid with moderate temperatures,'® while the the forest thins out to give way to patches of ground scrub northern slopes of the mountains remain humid right up to and the rich summer pastures which play an important part the summit ridges, but the range of temperature becomes in the economy of the region.

greater with the rise in altitude. The main undergrowth of the forest is Rhododendron The second type of climate is to be found on the southern Ponticum and Azalea Pontica which have impressed many a slopes of the Pontic chain; this 1s the cool, temperate con- traveler in the spring with the brilliance of their purple and tinental climate which 1s typical of the Anatolian plateau as a yellow flowers. The flower of the azalea may be responsible whole. The rainfall is low and confined to the winter months, for the intoxicating “‘mad honey” which caused such havoc leaving long dry summers with frequent years of drought."’ among the Ten Thousand.'® Paphlagonia was famous for its The seasonal temperature variations are extreme, with frost masts, while eastward of Trebizond the denser forest has an and snow in winter and relentless summer sun; the tempera- undergrowth of box, the wood of which is used for making ture variations between night and day are also marked. spoons. Thus a variety of timber was available along the Climate makes the coastal side of the watershed very fertile whole length of the Pontic coast for building, ship building, and the vegetation becomes more lush as the traveler moves and furniture making, or for export, and as a fuel supply.

eastward along the coast, so much so as to suggest a sub- Until recently, game was abundant in the mountains, tropical climate rather than a temperate one. In eastern where wild boar and bear inhabit the forests, ibex live close Pontos the primordial landscape of mountain, forest, and sea above the line of the summer pastures, and hare dart along

is but little marred by the encroachment of humanity and the slopes of the valleys south of the watershed. Partridge provides not only aclear notion of its ancient aspect, but also and pigeon are to be found everywhere, while the pheasant a vision of the land before men ever attempted tochangeit.It — (Phasianus Colchicus) is native as far west as the delta of the

is only in the years following the Second World War that the river Iris and may have had a wider habitat in Byzantine propagation of tea and the availability of modern road- times. The rivers and mountain streams on both sides of the making machinery have set in motion a process of physical watershed are inhabited by trout, and there are coarse fish in change more momentous and rapid than anything achieved the large rivers. The edible snail flourishes in the lush coastal undergrowth. Among migratory birds, the quail provides a 5amsun, 15. The yearly Zonguldak, 1,245 mm; item of contemporary diet.and 720rainfall mmaverages or 713are:mm; Trabzon, 830i snificant mm or 875 mm; Rize, , ; seems , to have been 2.415 mm: Batum, 2,423 mm. See n. 14 above: J. C. Dewdney, eaten in Byzantine times; in lesser quantity there are woodTurkey (London, 1971), 40-43 (rainfall and climate maps for cock, snipe, and varieties of duck, and, no doubt, the smaller Zonguldak); Black Sea Pilot, 81-82. 16. The average centigrade temperatures are:

Town January August 18. See Xenophon, 20-22. for Sull“mad available to the ° unwary as deli bal;Anabasis, Thasos isIV, alsovin, notorious honey.” For Zonguldak 6.8° 23.3° more conventional kinds of honey, neighboring Mingrelia was better Sinop 7.1° 21.2° known than Trebizond in the Middle Ages. Wax, on the other hand, Samsun 6.8° 23.3° was exported from the Pontos as a Venetian cargo 1n 1406 and 1434.

Trabzon 7.2° 23.2° The extensive literature on ‘“‘mad honey” is summarized in Th. Rize 6.8° 22° Pastiades, To patvopevov péAt, AP, 9 (1939), 43-62; for later Pontic Batumi 7° 25.5° apiculture, see N. Topalides, ‘H peAtoooKxopia ot Lavta, AP, 29

(1968), 332-40. Hills (1961), 108, is the latest published account of 17. By contrast with the figures in note 15, those for inland towns the stuff, which A. A.M. B. finds sickens rather than elevates. On include: Kastamonu, 644 mm; Sivas, 411 mm; Erzincan, 311 mm; Imerethian honey, see Klaproth (1813), 405; and on wax, see Thiriet,

Erzurum, 478 mm. Reéegestes, nos. 1237, 2349.

TOPOGRAPHY OF THE PONTOS 5 birds suffered the twice-yearly massacre that is their fate as and the autumn crocuses remain in the mind of the traveler.

they migrate across the Mediterranean world.'? One species of the latter is locally known as “‘vargit,” freely Since antiquity, walnut, hazelnut, and chestnut trees have translated ‘There is a going away” because its appearance in all been grown along the coast, and among the fruits the fig is the snow pastures is the signal for the summer villages to plentiful and the cherry a native of the Kerasous region, from close down and the herdsman families to start the long trek whence it 1s said to have been brought to Europe by Lucullus. down to their winter villages. In late spring, the native arisThe flora of the coastal region is much too rich and varied to tocrat, Lilium Ponticum rears its single- and many-headed

be described here in detail.*° Among the flowers, Colchicum varieties in golden glory above the surrounding flora. Vegetables must have been cultivated in abundance, with the 19. Much of the Chalybian and (especially) Chaldian forests have bean and pea families amen the dieta ry staples in Byzantine

been lost to charcoal-burners for smelting, but some of the east times, as indeed they still are today.*' Cereal crops can never Pontic rain forests remain primeval. In the discussion which fol- have been easy to produce because of the high rainfall and

lowed Rickmer Rickmers’ address to the Royal Geographical humidity, although no doubt some were grown, while Society, Dr. T. G. Longstaff, who had also visited Lazistan, said: “I Procopius mentions that the Laz grew millet.?? It is however

would call your attention to the very remarkable and little-known . fact that in Lazistan and also in Adjaristan, but in Lazistan par-

ticularly, there is the nearest virgin ‘‘Himalayan’” forest that is left in OT

existence.” See Rickmers (1934), 480. In classical times the Pontic cherry, half-grape, black with the stone of a cherry. The English forests provided exports, and in the sixteenth and seventeenth cen- expedition of 1292 (see note 19 above), spent an average of three turies Trapezuntine timber was sent to Egypt; the trade is flourishing aspers a day on “‘fructes divers.” But one of the few major and again today. See Magie, Roman Rule, I, 179; H, 1068. Timber is continuously recorded Pontic exports, from classical times to the comparatively rare on such a scale in Anatolia and it must have been present, is hazelnuts, for which references are provided in Bryer, an asset to the Grand Komnenoi. They were modest shipbuilders DOP, 29 (1975), 122 note 26. Here Dr. V. Ménage kindly points out and their coastal towns must have been largely timber-built, and that the proposition that findtk (the Turkish word for nut) is derived such they remained until the widespread use of concrete. There is, from the Pontos is probably mistaken: see A. Tietze, ‘““Griechische however, only the most meager hint that they exported wood: a Lehnworter im anatolischen Turkisch,” Oriens, 8 (1955), 204—57 wooden bow which Manuel I gave St. Louis. See Sire de Joinville, (No. 220).

Histoire de Saint Louis, ed. De Wailly (Paris, 1874), 324. 21. Xenophon, Anabasis, IV, vit, 23: “And the Trapezuntines The English embassy of 1294 had falcons at Trebizond and ate supplied a market for the army, received the Greeks kindly, and gave partridges on five occasions. See Langley (1292), 590-608, to be cited them oxen, barley meal, and wine, as gifts of hospitality.”” In passim. John Eugenikos devoted stanza 12 of his Ekphrasis to the Medieval Technology and Social Change (Oxford, 1964), 69-76, abundance and excellence of Trapezuntine game, equaled only by Lynn White, Jr., makes much play of the importance of the wide-

the skill of Pontic huntsmen—an imperial court official was a spread introduction of pulses from the 10th century as an explaProtokynegos: see ed. Lampsides, AP, 20 (1955), 3-39; Iorga, N&E, nation for demographic and economic growth in the West. The I, 273. The spectacle of the annual flight of quail up the Pyxites, one Pontos, by contrast, had probably always had a wide and, by of the great migrant funnels, excited nineteenth-century and modern medieval standards, remarkably well-balanced diet, for which there observers: see Spencer (1836), 195-96; and M. Q. Smith, ‘‘Notes on are hints in the Acts of Vazelon: see Bryer, DOP, 29 (1975), 120; the birds of the Trebizond area of Turkey,” The Ibis, 102 (1960), Vazelon Act 134. 576-83. From 9 to 12 September 1967, A. A.M. B. witnessed great 22. Procopius, Wars, VI, xin, 18. On Laz millet, see Bryer, BK,

clouds of quail arrive from the sea, a scattering as far west as 21-22 (1966), 176, 186. since the eighteenth century, American Tirebolu, some going up the Harsit, and numbers flocking to Boz maize (sweetcorn), has commonly provided a flour in Lazistan and Tepe above Trebizond. Their apparent tameness is in fact exhaus- (after the Greeks left) in Matzouka: see J. Humlum, Zur Geographie tion, which makes them easy prey. Odoric (1318), 98-99, has a tale des Maisbaus (Copenhagen, 1942), 29, 90; Vazelon Acts 3, 64, and which may be derived from the phenomenon and Is interpolated in 108. Here, A. A. M.B. is in some disagreement with D.C. W.’s text. some versions of Marco Polo: “In this land I beheld with great Basically speaking, the Grand Komnenoi faced the same problems delight a very strange spectacle, namely a certain man leading about as the Palaiologoi and early Ottomans of Constantinople in feeding with him more than 4,000 partridges. The man himself walked along their capital, although on a much smaller scale, for theirs was not the ground and the partidges flew in the air. These he led to a certain really an urban economy. They exported hazelnuts and wine, but castle called Zauena (Zigana ?), being three days’ distant from had to import cereals and salt fish which (as in Constantinople) was a Trebizond. The partridges were so tame, that when the man desired principal source of cheap protein. In both cases they relied, like the to lie down and rest, they would all come flocking about him like Palaiologoi, on Italian entrepreneurs from the Crimea and the Sea of chickens. And so he led them to Trebizond and to the palace of the Azov. So far as cereals went, this is a curious reversal of the last emperor, who took as many as he pleased, and the rest the man words of Constantine Porphyrogenitus, DA/J, I, 286: “If grain does carried to the place whence he came.”’ Similar tales were later re- not pass across from Aminsos and from Paphlagonia and the Bou-

ported from Chios and Grasse: Tournefort (1701), I, 172-77: kellarioi and the flanks of the Armentakoi, the Chersonites cannot Busbecquii epistolae (Amsterdam, 1660), 164; The Turkish Letters of live.’” Nevertheless, substantial cargoes of wheat, barley, oats, and Ogier Ghiselin de Busbecg, trans. E. S. Forster (Oxford, 1968), 103. millet passed from the Crimean ports to Trebizond in 1289 and 1290

20. See P. H. Davis, Flora of Turkey (Edinburgh, 1965-); and (mostly in March and April). As for fish, the Pontos enjoys the K. M. Guichard, ‘‘Flowers of the Black Sea Coast,” Gardeners’ gyration of the tunny and the shoals of hamsi in early September, Chronicle, 147 (1960), 184-85. The quality of Pontic orchards only when the quails come. But much of the catch is wasted, for, as became widely known from the late nineteenth century, when apples, Procopius pointed out (/oc. cit.), the Pontos has little sea salt, and plums, medlars, filberts, grapes, and apricots could be exported any matters are not helped by the fact that the salinity of the Black Sea is distance. In stanzas 13 and 15 Eugenikos (/oc. cit. in note 19 above), about half that of the Mediterranean. The salt pans lay around the claimed to have known a thousand different and delectable Pontic Crimea and in the sea of Azov, which in April, May and July 1290 fruits, although, when he came to it, he could name only grapes, exported quantities of salt fish to Trebizond; in June and July of that walnuts, perfumed lemons, and olives. Orchards are named in year there were at least seven shipments of salt. Significantly, John I Vazelon Acts 10, 23, 75, 100, 104, 105, 108, 115, 134, 135, 143, 161, himself bought cargoes of Genoese Crimean corn and salt in June and 172. One curious modern Pontic fruit is the karavemis, half- 1290: a final delicacy came in August, with 50 to 65,000 Ibs. of Azov

6 CHAPTER ONE significant that, according to Xenophon, the Mossynoikoi From Amisos (Samsun) to Sinope the coastal region 1s used a flour ground from nuts for their bread-making.?* This differentiated from the inland valleys in the same way as would almost certainly have been chestnut flour which pro- described above, but the contrast is less striking. The coast ts vided the staple bread of Corsica and certain parts of Italy fertile and there is a greater extent of arable land than further until recently, and no doubt in the Pontos in Byzantine times eastward where the mountains come right down to the sea. it was a staple substitute for cereal flours. The cereal crop of | The land between Amisos and Alacam is now devoted to maize, and the potato, tomato, tobacco, and tea plants have tobacco and fruit farming but would have provided ample all been introduced since the medieval period; their culti- space for the growing of cereal crops, vegetables, and fruit in vation has transformed the eating habits and theeconomy of — the medieval period. From Alacam to Gerze the land rises large sections of the coast, and caused a considerable change steeply from the coast, with broad leaf woods and villages in in the appearance of the landscape. The vine has been known the clearings, and then again the mountains fall back in the since antiquity, when wine was among one of the gifts given hinterland of Sinope to leave a wide area of easily cultivable to Xenophon’s men, and the Venetians exported it from rolling lands. Trebizond in the Middle Ages.** It is now only made in any For the whole coastline, the sea provided salt and fish in significant quantity in the plain of Tokat. Olives still grow in surplus quantities sufficient for trade. There are a great the region between Koralla (Gorele) and Trebizond and number of varieties of fish which are caught in large quanboth the olive and the grape were among the more important tities, among the most popular are: horse mackerel, grey products of this part of the Black Sea coast. Flax and cotton mullet, red mullet, gurnard, tunny, whiting, and anchovy.”°® are grown in small quantities, and the mulberry flourished,

so that linen, cotton, and silk could have been home pro- tn oa , ; .

ducts. while hemp is grown in the mountain villages for rove- the “‘blattia’”’ and “kylichartia” of the Italo-Trapezuntine treaties

og PIS8 8 P probably came from Persia, while in 1289 Trebizond was certainly making. importing Chalons cloth. But the linen which practically every member of the English expedition bought there in 1292, and prob-

TT ably the camelot, should have been locally produced: Balard, sturgeon, sold at 55 hyperpers a hundredweight. Cheese, salt pork, Sambuceto, nos. 87, 191; DVL, II, 128; Heyd, Commerce, II, 94: and hog’s lard were also imported from the north. See: Magie, Zakythinos, Chrysobulle, 67-72; Langley (1292), loc. cit. Bordier Roman rule, 1, 182; Bratianu, Actes génois, Nos. 152, 203; Balard, (1609), 121, noted that the chief commodities in the Trabzon bazaar Sambuceto, nos. 7, 107, 119, 184, 404, 409, 410, 411, 412, 419, 423, were “cloths which are sent all over the country and to other lands,

430, 438, 480, 501, 502, 505, 586, 615, 616, 618, 625, 626, 639, 703, which are called Trabzon cloth, being very tenuous, light and

740, 788, 797, 903. strong—more so than any other—and the trade in these cloths is 23. Xenophon, Anabasis, V, 1V, 27-30. wonderful in this town.”’ Cotton, linen, and raw silk were all spun or

24. Xenophon, Anabasis, 1V, VIII, 23. With hazelnuts, Trape- woven until recently at Rhizaion (Rize) and exported throughout zuntine wine was an important medieval export. Zamora was ex- Turkey, an industry which has now succumbed to synthetic fibres. ported to the north-west: Pegolotti, ed. Evans, 24, 434; Bratianu, As for the appearance of Pontic stuffs, the fine purple, black, and Actes génois, 127. Many travelers report on the enormous Pontic gold striped cloths which first appear in 19th-century engravings are

grapes, on untrellised vines which grew on olive trees, in the probably much older in design. Until 1923 each Greek valley was Byzantine fashion. Trapezuntine wine was not cheap, but highly distinguished by a slightly different striped tartan. Town and market appreciated: in 1292 the English expedition’s expenditure on wine clothes were, and are, black. They probably were in the Middle Ages, rose daily from 15 aspers to 23, 24, 414, 42, and 48 aspers (by far the too, for the Trapezuntine imperial color of mourning was white: largest item of its budget, dropping back to an average of 15 aspers Panaretos, ed. Lampsides, 73; Koukoules, Bios, IV, 243-44. Hemp when it reached the interior); in 1438 Tafur found that in Caffa the (cannabis) was presumably grown at Kanaborge: Vazelon Act 4 of

going rate was two virgins for a measure of wine (presumably the 15th century (cf. Zerzelides, AP, 24 [1961], 262). But, although Trapezuntine)—two years later Venice decided that Trapezuntine Ibn Battutah (1332), II, 467, noted the excessive consumption of wine merchants had an unfair advantage at Tana by paying no dues hashish at Sinope, Trebizond hemp was probably for rope-making. and declared all wines exempt; Genoa had been enjoying a vast See H. Godwin, ‘‘The Ancient Cultivation of Hemp,” Antiquity, 41 indemnity paid by Alexios IV in wine and hazelnuts since 1418. In (1967), 42-49. 1471 Barbaro found that, by contrast, a cask of Italian wine was 26. Whereas the Pontos had exported grain in the 10th century worth less than a ducat in Trebizond. As late as 1609 Bordier and fish in the 11th century to the north, the position was reversed reported that the trade was still flourishing in ‘‘all corners of the by the 13th century, perhaps through more aggressive Genoese Black Sea, and they drink no other wine in Caffa and in other places marketing, possibly because a population growth outstripped local in the Cimmerian Bosphoros.” He did not find it agreable, however, resources; in the case of fish, certainly because the Pontos lacked and 19th-century travelers were divided about its quality. Today it 1s quantities of salt: in a glut, hamsi had to be used for manure. In the Crimea that exports wine, while D.C. W. claims to have drunk 1292 the English bought an average of 10 aspers of fish (including the last bottle of Trapezuntine wine in 1958. See Schiltberger (1402), sturgeon) a day in Trebizond. They were even able to buy 45 aspers 41; lorga, N&E, 1, 274; HI, 246-47; Thiriet, Régestes, No. 2532; of fish in Bayburt— possibly trout from the Akampsis. Trapezuntine Clavijo (1404), ed. Estrada, 245; Tafur (1438), 134; Barbaro (1471), fishing was conducted from a peculiar kind of boat and there seems p. 48v; Bordier (1609), 134; Langley (1292), loc. cit. in note 19. to have been an imperial tax, or license, for professional! fishermen. 25. Bordier (1609), 129, 134, was especially impressed by the The local fish market was lively in Evliya’s day, when the hamsi’s gigantic olives of Trebizond. There is slight evidence for silk negoti- burnt head was used to scare snakes, and its flesh reckoned to be an ation in Trebizond, before and after 1461, but the chief markets lay ‘aphrodisiac of extraordinary potency.’ See Evliya (1644), II, 48— to the west, especially in Bursa—see A. Bryer, ‘““The Latins in the 49; Mynas (MS 1248, 1844), fol. 119a; Polish Janissary, 261; Jaubert Euxine,” XV¢ Congrés International d'Etudes Byzantines, Athens, (1805), 139; Deyrolle (1869), 23; J. Ray, A Collection of Curious 1976, Rapports et Co-Rapports (Athens, 1977), I, forthcoming. The Travels and Voyages (London, 1705), II, 17-18; Koukoules, Bios, V, Book of the Prefect mentions linen from the Pontos and Kerasous: 331-43: Lamberti (1650), 48: Bratianu, Actes géenois, 190-91, 196; The Book of the Eparch, ed. 1. DujCev (London, 1970), 39, 166, 247, Laurent, AP, 18 (1953), 266 and note for line 121; Bryer, Mariner's 273, 289. There is ample evidence for trade in local cloths thereafter. Mirror, 52 (1966), 11; the same, BK, 21—22 (1966), 185 and note 74: The ‘‘panni aurei de sirico, veludi, camocha, sendadi, bocrani,” and Langley (1292), loc. cit. in note 19; K. Devedjan, Péche et pécheries

TOPOGRAPHY OF THE PONTOS 7 The latter, called hamsi, have given their name as a nickname Anatolia, is the raising of sheep, goats, and cattle. In the to the people of the eastern half of the southern shores of the Pontos this is rarely the main farming activity and is usually Black Sea. Fish, to which may be added molluscs and crus- organized on the basis of transhumance. Shepherds of the taceans, must have always been a staple in the diet of the lower valleys of either side of the Pontic mountains take their coastal peoples, and salt fish provided an important item for flocks in the spring and drive them up to the summer pastures

export. above the tree line where they have yavlas, which are a simple

The inland valleys south of the coastline between Sinope form of summer village. The chronicle of Panaretos provides and Themiskyra partake of a similar climate and character to evidence that pastoral life was organized in this way during the coastline because these valleys and basins are formed by the Empire of Trebizond,”’ and the tradition of transhumthe rivers Iris, Halys, and their tributaries. The rivers form ance may well be much older in origin since it is a logical

gaps in the mountain barrier through which a certain means of making the maximum use of natural resources. *°® amount of moisture-laden air can pass, thereby increasing

the amount of rainfall in these valleys. The basins and valleys COASTAL TOWNS one a ren unastamon a oyabat) ane Phazimonites The Pontic coastline provides very few natural harbors, iM erzi on ava adi ‘i ‘ wanerora ( asoum.! masela with the notable exception of that of the city of Sinope, where

( the vn d u on h t ht so elativels ( i at), nd the peninsula provides ideal protection against weather; it is Th, sy un ae i, in re tan "1 atively ll : aan a tribute to the wisdom of the Milesians that they founded oe nen y ti to ae th ture aba an van there their first Black Sea colony. Along this coast a good nod bt ah rul a done Brow , ere In abuneance, an harbor must provide deep enough water free from rocky Ou vinvine di ne beck t h vhetecty he hte hazards, and protection from the prevailing northwesterlies

race a conlinuous pedigree Dack to the classical world, while which often develop quickly into gale force winds sufficient prehistoric mounds suggest an even earlier ; i d cultivatihabitation to endanger shipping of small tonnage. Theoccuharboroealso

P Furthe st ard. th h of Ker: Trebizond needed to be defensible and situated near to, or at the ter-

4 Rhine eas ore b ° “ oo ile wormed 6 in Izone, minus of, a route across the mountains so that it might an h ih, 1. kK asins a va hot ormed by | _ d the function as a center of commerce or as a military supply base snc AG © LYKOS AL ns trl ee satan _ © and not just as a refuge for shipping in bad weather. A further val war cheih . im gel(AUMIAN)

Oo ak. ASPIRE teOe | gees f aSs { on Fon {asin RN oF ' oP ner ASPIR “ITARCONYD X «ah grin [s)BAYBURT & ALLAS? OSIRAN ST A}iN oea:fs {

we ue () SADAK i , aR ° PA ° ERZURUMI] / ‘ ELESKint”"

t o? ee nee (ALLEQUIX, ALESQUINER) "

BERZINCAN PAUNIGUE) ow AT pel oS EIR ere

O 50 100 150 200 KILOMETERS Blwacere rows poe SAKEN a? CoAN TO ————————— i sevens POSSIBLE ALTERNATIVE ROUTES 3. Pontic Sections of Clavijo’s Outward and Return Routes

Chapter Two

(FIGURES I, I], and 3)

°‘;..,.°.1 INTRODUCTION same time the town of Kelkit began to supersede Germurt

Neither author has traveled all the routes described in this 48 the posting station. The town of Kelkit appears to be an chapter. We have therefore supplemented our own ex- Ottoman foundation and to have succeeded Satala, Sadak, perience by the accounts of travelers. Few of these accounts “° se ministrauvecene for the hepa. ra Kelkit, region.

are of medieval date, but we have thought it fair to use later We be e the old oe t eee "1 Pontos i. detail ones since the physical and geographical conditions of travel since IU Is f ecoming more and more cl 1 rh, visualize the

remained largely unchanged in Asia Minor until the nine- pattern 0 ancient and Byzantine roads. The slowness of teenth century when the advent of the steamer along the premechanized travel must constantly be remembered. It is coasts, and the railway and road blasting activity on land too easy to overlook the fact that modern blasting and earth drastically changed an immemorial pattern of movement. moving has enabled recent road builders to ignore, for the Where several place-names appear in succession, the first first time, the features which dictated the old patterns. The name or names will be those from the classical Itineraries ifa ™Odern pattern of roads overlays the earlier system and name is suggested for a site. The second name or names will causes confusion since the student always prefers to think be those given by a traveler or other authority who is that he is traveling along some old and hallowed path. We quoted; or a traveler’s name will come first when there is no have rarely used the simple tabular form for indicating the classical name. The last name or names will be the current _ Stations along a route. The tabular method imposes a misTurkish ones in so far as we have been able to ascertain them. __ '¢ading certainty, whereas the message of this chapter is to

It has proved impossible to be any more consistent with point out how little we are certain of at the moment and to place-names. The reader will be well advised to use the maps indicate the fact that there were different roads for use in for a clarification of place-names which may cause difficulty W/7t€r and summer. This has often been ignored in the in the text. Before 1928 the spelling of Anatolian place- endeavor to fit old itineraries into a neat and unified pattern. names presented foreigners with problems. Two particular So also has the means of travel been ignored. North of the difficulties should be borne in mind. First, most early trav- Pontic Alps, wheeled traffic may be discounted until recent elers, few of whom knew Arabic script, transliterated place- Toke, bulky goods went by sea. South of the Pontic Alps the names from oral information, usually obtained from illiter YKOS valley allowed solid-wheel ox-carts, but their range ates Second, informants were of a variety of tongues. Up to was not wide. On both sides even pack animals (of which the early nineteenth century, travelers seem mostly to have there were 15,000 on the Trebizond-Tabriz route by the midused Moslem dragomans or guides and stayed in regular nineteenth century) would always be less often encountered caravansaries or at Turkish houses. Consequently, these than human porterage. One of the first to choose a horse earlier travelers usually adopted Turkish names for places. ©” aba (waggon) to cross the Pontic Alps seems to have been Later, asa result of the nineteenth-century concern for Chris- Weeks (1892), and probably the first to take a motor car was tain minorities, European travelers switched to Armenian or Rawlinson (1919): both regretted their decision. Greek dragomans and often stayed among Christian minor- Unless otherwise stated, the map references are to the ity communities, thus learning the Armenian or Greek ver- Turkish 1: 200,000 sheets, which date from the turn of the sion for place-names. An example of a complete switch of century but have been frequently revised. Where the Kiepert name is that of the river Lykos. With the exception of Evliya maps are mentioned without specific reference, they are the Celebi, travelers up until the nineteenth century commonly | :400,000 sheets for Asia Minor. Another map, frequently referred to this river as the Carmili, Guermili, Germeili, or referred eee na Map, is that of Nazim Tarhan, Tarihte

Germeli, naming it after the old posting station of Germirii Turkiye (Ankara, 1962). a

which has now declined into an insignificant village. Begin- The short introduction by Munro is still the most adning with the early nineteenth century, the river has usually been referred to by its Armenian name of Kelkit (= Wolf 1. For Kelkit, see p. 171. An early example of controversy over River, a translation of the Greek), which in its turn has been transliteration is that between Blau and Mordtmann; see Mordt-

adopted as the modern Turkish name. And at about the mann (1859), 427 ff.

18 CHAPTER TWO | mirable general summary of Pontic routes, and the relevant carried on through anchorages at the village or small town to

itineraries in Taeschner are very useful.” be found at the delta of nearly every river valley. With the building of the new coast road in the 1960’s it was superseded

SEA ROUTES by lorry traffic, but until the 1950’s rusting steamers of

The main means of communication along the coast of indeterminate age and kayiks of up to 200 tons were still to be Pontos has always been the sea, since the valley system with Seen anchored off river deltas, selling cheap hardware and its numerous valley torrents makes a landward journey along _‘Stuff’s from Istanbul and taking on local products in return,

the coast extremely difficult. while kayiks of a moderate size conducted the same type of The military use of the coastal sea route is exemplified in trade over the shorter distances between the villages and the the Roman period by Arrian’s Periplus which is essentially ©™Poria. Heavy products carried by sea could or did include

the report of a tour of inspection of coastal depots and COPPer from the mines at Mourgouli, Murgul, which might garrisons carried out by a senior officer.? Of the thirty ex- have gone downstream to the mouth of the Akampsis, peditions recorded by Panaretos of the Grand Komnenos 0ruh; the silver of Argyria going out through Tripolis, Alexios III (1349-90), twenty-four were made by seaandhis _ | irebolu; the alum of Koloneia, Sebinkarahisar, going out

six land journeys were performed in the immediate hinter- through Kerasous, Giresun; the building stone of the land of Trebizond.* The coastal sea route was also in- | O!naton, Unye, region carried away from anchorages near valuable as a safer and shorter means of communicationand ‘he quarries; and the red earth pigment exported through

supply for the inland towns of eastern Asia Minor. A direct 5!0pe, Sinop. .

overland journey from Constantinople, Istanbul, to Together with the coastal trade, both local and interTheodosioupolis, Erzurum, took about twenty-five days, national trading was carried out ACTOS the Black Sea with whereas a journey by sea with a favorable wind to Trebizond _—‘the Crimea, part of which was briefly included among the and Sousourmaina, Siirmene, or Rhizaion, Rize, and then __ titular territories of the Grand Komnenoi. over the mountains on horseback, might take a third of that

time. International trade by sea was conducted from the RIVER ROUTES

larger coastal towns which acted as emporia in the carrying Allexcept three of the Pontic rivers are steep and torrential trade to and from Europe and Asia, but portulan maps in their descent to the sea, and totally unsuited to navigation, show that western captains also had a considerable knowl- _ but many still serve to float logs down to the sea where the

edge of local anchorages.” Genoese merchants knew the timber is used for shipbuilding, houses, and furniture. Black Sea coast well, both by sea and land, and they have left Timber may also have been used and exported by this means their mark both in the name ‘“‘Ciniviz,”’ occurring on the in the Byzantine period.® Turkish maps of Pontos, and in the invariable reply of vil- At the western end of the Pontos the Halys, Kuzil, is lagers when they are asked about the builders of an ancient _ navigable inland as far as Celtik which is situated well into

church or castle, “‘Cinivizlerin Zamanindan kalma.” © the mountain chain, and tiles are made and brought Among later accounts of the sea journey along the coast are downstream from the village of Kurugay. It is possible that the careful itinerary given by Pitton de Tournefort, and the this industry was already in being in Byzantine times and that

journeys by Evliya Celebi and Rottiers.’ it used the river route, since the means of getting the boats Local trade involved the shipping out of whatever cash upstream is simple. The empty boast are sailed or towed crops, manufactured goods, or minerals that were produced upstream. Four or five tons of tiles are brought down by each in the mountain valleys, and the supply, in return, of the few boat in the summer, and double that amount in the winter necessary imported goods. This kind of small-scale trade was when the river is deeper.’ It may have been along this Halys route that the survivors of the battle of Phazemon, Merzifon, 2. Munro, JHS, 20 (1901), 52-55. Tarhan, Map, is not good for escaped to the sea in 1101;'° it forms a natural alternative to

roads, but draws upon regional reports to the Department of the overland route from Amisos, Samsun, for communiAntiquities, Ankara, and is very useful for unpublished sites. cation with the inland cities to the south and for the export of Taeschner, Anatolische Wegenetz, should be treated with reserve in produce from the regions of the Dazimonites, Kazovasi, and

identifying stations. An example is his confused identification of Amaseia, Amasya. The existence of the classical ThemiPontic Aksehir, restated s.v. ““Ak Shehr” in E/?; JA wisely confines , a itself to the more important towns of the same name in central skyra, Terme, and the medieval Limnia, Taslik (?), on the Turkey. For Aksehir, see p. 25. Semseddin Talip, Le strade romane delta of the Iris, Yesil suggests that this river also may in Anatolia (Rome, 1938), has nothing to contribute in the Pontos. have been navigable for a considerable distance inland, and if

3. Baschmakoff, Synthese, 80-107. our conjecture about the identity of the medieval Kinte is

4. Bryer, “Shipping,” 4. 5. Kretschmer, Portolane, is not used in Bryer, AP, 24 (1961), 97-127, which therefore needs modification. See also the Vicomte de 8. Strabo, Geography, XII, 11, 12, on Bithynian and PaphlagoSantarem, Atlas composé de mappemondes, de portulans et de cartes nian timber; and XI, nm, 17, on floating timber down rivers to

hydrographiques et historiques ... (Paris, 1849), for the Feduci Colchian shipyards.

d’Ancona map of 1497 (unnumbered). 9. E. Akkan, “‘Kizilirmak’in asagi Kesiminde Kayikla nakliyat,” 6. Evliya (1644), II, 36, 40, already has references to castles sup- Dil ve Tarih Cografya Fakultesi Dergisi, 20 (1962), 263-70; Idrisi, ed.

posedly built by the Genoese. Jaubert, II, 393; ed. Nedkov, 96-7 and note 297, mentions that the 7. Tournefort (1701), I, 40-85; Evliya (1644), IT, 36-52; Rottiers river was navigable.

(1820), 175-305. Peyssonnel, Traité, I] 13-130, describes con- 10. Albert of Aix, Historiae, in Recueil des Historiens des

temporary port facilities and trade. Croisades, Historiens Occidentaux, TV (Paris, 1879), 570.

LAND ROUTES, EAST-WEST 19 correct, it is by this route that John Komnenos’ ill-fated the way, but there was certainly a track of some sort along expedition pushed inland to Neokaisareia, Niksar.'' We most of the coast. Idrisi, writing in the twelfth century, seems have not however explored this river, and a route southward to describe a land route along the coast from Trebizond to

would need further verification. Constantinople taking twenty-eight days, but the places

At the eastern end of Pontos the Akampsis is navigable — mentioned all appear to be on the coast and it is possible that about as far as the reaches above Artvin, and goods could this is a sea route.'® There is still no coastal road in the west have been shipped from the city of Ardanoutzion, Ardanuc, between Cide and Inebolu, but, among nineteenth-century

to the mouth of the river near Bathys, Batumi, for transit travelers, Preusser skirted the coast eastward from Amasra westward by sea and imported merchandise shipped up- to Inebolu, and Hirschfeld traveled a shorter stretch of it stream. The Akampsis may also have served for theexportof | between Aeginetes, Hacivelioglu Iskelesi, and Tshatal surplus fruit, wine, honey, and timber from the province of Zeitun, Catalzeytin.'’ From Paurai, Bafra, eastward as far Tao and for the transport of the copper ore of Mourgouli as Polemonion, Fatsa, there are no great obstacles to a before the opening of the road via Borcgka and accross the coastal track, providing that we can assume ferry services mountains to the sea at Hopa. Guarracino writes that the across the rivers. From Polemonion eastward the mountain journey upstream took about two days and the journey ribs become increasingly high and the rivers more torrential downstream about nine hours. The region of Borgka made making the landward journey difficult, and the track seems to bricks and earthenware jars which were exported by river have cut across the larger capes such as Yasun Burunu and and sold along the coast between Rize and Batumi. This Cam Burunu. The stretches between Ordu and Giresun, and industry, like the similar one at Celtik on the Kizil, may date between Platana and Surmene present no great obstacles to a back to the Byzantine period. The boats recorded by Guar- coastal track and there would always have been short easy racino were about 15 m long by 1.25 m wide and carried a paths around many of the bays, but in general it may be cargo of 6 to 8 tons. Guarracino notes, however, that traders remarked that the mountains close in to the sea at Trebizond, from Trebizond or Constantinople with goods for Artvin making a landward journey very arduous. Evliya Celebi preferred to land them at Hopa for transport overland. This traveled the whole length of the coast to Batumi by sea, took eighteen hours and was clearly safer and easier than the although he records that a detachment of one hundred men

river route upstream. !? went from Giresun to Trebizond by land.'® Kinneir, who

Brant’s figures are rather different for the river and land was a determined traveler, got from Samsun as far as routes. He gives the journey from the mouth of the river to Giresun by land, but his account makes it clear that this was Artvin as eight to ten days, and the downstream journey as unusual. At Unye “‘the Mutsellem wished us to perform the three days, but Guarracino seems to have been the better remainder of our journey to Trebizond by sea, adding that it

informed, and himself to have made the downstream was not customary for travelers to go by land,” and at Ordu,

journey.'° “in an interview we this morning had with the Aga of the place, he stated that, ‘as it was madness to think of us

LAND RouTEs, EAST-WEST traveling by land, he had ordered a felucca to carry us to Kerasoun.’”’!°

Coastal Routesfrom i, Consul Brant, writing 1835, is quite categoric about Travelers, Xenophon to theinRussian invaders of «“ ; ; the ;

, 1a TL: eastern end: ““From Trebizond to Batum the distance is 60 found progress along the coast difficult by land.'* ,is1915, .because , ours, or as many leagues. It caused can only besucperformed in This boats;h of theare numerous indentations byHowever, the 1 20the sea ; . there no practicable roads. could cessive river valleys which until recently made the coastal prove up, dangerous, road into .a always series of zigzags down, and and acrossmerchants the spines : .seem to have

oa avoided it when they could. Tavernier refers to the sea route of mountains, reaching inland toPersia the first available ford over f ; .; a rom Istanbul remarks the river torrent. aThe existence of ainto coast route and is however ; that, although it is

1S 4: , shorter, it is unpopular with merchants because of the bad attested by the;aPeutinger Tables'> ; along , weather, and hewhich later mark givesstations the bestbeports the Black Sea

tween Cape Karambis in the west andasBathys in theOnuye, east.Samson, The Trebisonde, ;;; Quitros, Sinabe, and route appears to run for the most part along the sea coast and

it is quite possible that it represents a se te f h of Gomme. He reckons the distance from Istanbul to Tre-

q P P Sa sea Foule tor muel O bizond as 970 miles and from Trebizond to Goniya as 200

11. See p. 99.

12. Guarracino (1841), 296-305. Koch (1855), 96, gives the dis- 16. Idrisi, ed. Jaubert I], 394; ed. Nedkov, 94—99. Idrisi’s stations

tance from Batumi to Artvin as 16 hours. are: Trebizond to Bersenda, 2 days; to Kendia, 5 days; to Ania, 3

13. Brant (1835), 187-223. Burnaby also traveled downstream days; to Astinoboli, 2 days; to Amastra (on the sea), 5 days; to from Artvin and gives a lively account of the journey which took 9 Herakla, 3 days; to Constantinople: 8 days. These stations present hours, a time which agrees with Guarracino, but not with Brant. considerable difficulties, and the time taken can only be explained as

Burnaby (1877), II, 301-9. an estimate for sailing in very bad weather, whereas it could rep-

14. See p. 61. resent a fast overland journey.

15. Miller, 7R, cols. 631—51. For the Pontic roads in general, see 17. Preusser’s route is marked on Kiepert’s | :400,000 maps but pp. 629-83, figs. 210-13, and table at the end of the chapter. The we have found no account of it. G. Hirschfeld (1890), 76-208.

Ravenna Cosmography and the Guidonis Geographica also give 18. Evliya (1644), I, 41. confused lists of coastal towns and there is no indication of whether 19. Kinneir (1813), 319, 324. these are sea or land itineraries; J. Schnetz, Ravennatis Anonymi 20. Brant (1835), 222; and Koch (1855), 104, on the coast west of

Cosmographia (Leipzig, 1940), 134. Hopa.

20 CHAPTER TWO miles.2" Pococke remarks: ‘“‘No caravans go to those parts The upper route is the one that concerns us since the [Sinope], the Euxine sea being dangerous and the ports of it eastern half of it ran through southern Pontos. The way ran are bad, which is the reason why there is little trade that way; from Constantinople in the west to Theodosioupolis in the

and if the Black Sea was much navigated it would hurt east, and from thence onward into the Caucasus or central Constantinople and Smyrna, though the danger of it must be Asia.*° Along much of the distance there were variant routes

the principal reason why goods are carried such a long which we shall discuss below. The mountain ranges of journey by land from Constantinople to Tocat, which cannot Bithynia and Pontos run in a west to east direction and be above four or five days journey from the sea.” 7? between them are the great fractures which form the rift valleys of the rivers. There is thus a natural framework for

Inland Routes Lo.

lateral routes such as does not exist for communication from

General Remarks north to south, except in the case of the Euphrates valley. It

Asia Minor is a bridgehead between Europe and Asia, and must be remembered that there would have been seasonal the means of crossing it were by three main trunk roads, each variations in the course of the road since protection from the with variants. One ran diagonally from the northwest across weather and snow blockages governed winter travel, whilst

to the lower or more southerly parts of the Anatolian speed, safety, or the finding of fodder for armies governed plateau, a second across the center, and a third across the summer travel. The evidence for the course of the roads may upper or northerly part of the country.?> Anna Komnene be divided into five categories. First, the literary evidence of refers to the southern route, taken by the Crusaders, as the the ancient geographers, and the Antonine, Peutinger, and “quick route”’ and she mentions the middle route “‘straight to medieval lists. We have regarded this literary evidence provi-

Horosan,” which the Franks wanted to take in 1101.74 sionally as of equal value; it is however apparent from the reater number of stations listed in the Peutinger Tables that

21. Tavernier (1681), 18, 273-76. Villotte (1685), 20—21, refers to “he ften d ‘bi ds diff tf ° those of the

the land route as less dangerous but longer and passing through very y are ° en escribing Toads Cumeren FO .

agreeable country; and to the sea route as more dangerous but Antonine Itinerary. A good example of this is offered by the shorter. He gives the sea route as about 300 leagues or 10 days travel two lists of stations between Satala, Sadak, and Trebizond

from Istanbul to Trebizond. 7 which almost certainly represent the summer and winter

22. Pocock (1740), II, 91. By contrast emphasizing the trade of rgadg respectively.2° The Byzantine chronicles seldom if ever Tokat with the sea: Peyssonnel, Traité, II, 91—92. . . 23. Ramsay, Asia Minor, 197-221. Ramsay’s description of the give account of roads as such, but the place-names which road systems of Anatolia remains the most comprehensive study of they mention in the course of describing a campaign are at the subject, but is seriously misleading in some respects. His view of least a help in establishing the fact that roads existed between the Royal Road is not now generally accepted and his account of the the towns which they mention. The second class of evidence

Byzantine military road suggests that there was only one road ‘ded bv the milest Third. th ‘st f ent

whereas there must have been several routes from Constantinople to IS Prov ea OY Me milestones. ANT, Me existence Of ance

the East. The aplekta of Constantine Porphyrogenitus need not ruins. Fourth, the accounts of travelers whose routes, so long necessarily lie on one route; see note 28 below. There is a short as they traveled on horseback, are likely to have differed little general survey in Vryonis, Decline, 30-33, but it suffers from some from the routes used by the Byzantines or the Romans. And

inaccuracies, and in general follows Ramsay’s view of roads. A anh;

general view of the classical roads of Asia Minor is given by W. M. fifth, our geographical observations.

Calder and G. E. Bean, Revision of Anderson’s Classical Map of Asia The western part of the upper or northern route across Minor (London, 1960). This is at present under revision for an | Anatolia seems to have followed the course of the modern edition on a larger 1:1,000,000 scale. The routes of the Classical motor road from Constantinople to Ankyra, Ankara, as far Map may be usefully compared with the routes given in 1 9th-century as the town of Kratia Flaviopolis, Gerede. A northern road editions of Murray’s Handbook for Travellers in Turkey in Asia, and ran from Kratia Flaviovolis into the Amnias. Gok. valle

with the annual editions of the Turkish Highways Map. Early maps an ITO rata PlaviOpous | O as, , y

which are useful for routes are to be found in Murray’s Handbook and then across the Halys River, through Andrapa, and in Saint-Martin, Asie Mineure. A rather confused impression of Vezirkopru, and the regions of the Phazemonites, Merzifon,

Chinatchefl 1s referred to without qualification the reference 1s to : : .

ancient and medieval roads is found in Tarhan. rin Where the Havza, Ladik, and the Phanaroia, Tasova, and along the the map of his itineraries: P. de Tchihatcheff, Asie Mineure. oe nena neo the ~ Kos M alley to Ko oneia and mheodo

Description physique, statistique et archéologique de cette contrée SIOUPOHS. IS road was aimost certain y in existence w en (Paris, 1853-59), Atlas (Paris, 1860). His itineraries are important | Pompey established new towns along it; Munro remarks: “It

for travel by horse and are summarized in Tschihatcheff (1863). is tempting to see in Pompey’s colonies, Pompeiopolis, Where Tschichatschof (1858) is cited, the reference is to C. Ritter and

H. Kiepert, “‘Itinerar der kleinasiatischen Reise P. von Tschichat- 25. In the Byzantine period one of these upper routes is fully schof’s im Jahre 1858,” Zeitschrift fiir allgemeine Erdkunde, 6 (1859), attested in the Armenian itinerary of the 10th century, commented

275-343. on by Manandian, Trade and Cities. The western parts of another 24. Anna Comnena (Komnene), ed. Leib, III, 19. In the 9th route, starting from Mous, Mus, but going north to Koloneia,

century this 6E0c dpdpoc was important enough to have its own Sebinkarahisar, and then westward through Neokaisareia, Niksar,

Chartularioi, who are listed by Philotheos and in the Escorial and Pimolisa, Osmancik, are listed by Al Muqaddasi, in E. Tak tikon; N. Oikonomides, Les listes de préséance byzantines des [X¢ Honigmann, A/PHO, 4 (1936), 263-71. The upper route between et X° siécles (Paris, 1972), 233, 273. Neither Oikonomides, 311-12, Osmancik and Erzurum is listed by Hadji Khalfa (Kiatib Celebi) in nor Bury, /mperial Administration, 91-92, recognize that the ref- the Djihan Numa of about 1640; translated by Armain in Saint-

erence here appears to be to a particular road. For a different Martin, Asie Mineure, II, 687. Pococke (1740), 91, refers to “‘the interpretation as a reference to the first postal service, see D. A. great road from Persia which is by the way of Tokat, Amasia, and Miller, ““The Logothete of the Drome,” Byzantion, 36 (1966), 443. Tocia to Constantinople.” It is still to be found as route 79, For some account of sections of the central routes, see J. G. C. “Constantinople to Erzeroom,” in Murray’s Handbook for TravyAnderson, “Exploration in Galatia cis Halym, II,” JHS, 19 (1899), ellers in Asia Minor, 4th edition (London, 1872), 426-31.

52-134, 280-318, and pl. tv. 26. See p. 51.

LAND ROUTES, EAST-WEST 21 Neapolis, Magnopolis, Diospolis, and Nicopolis, a series of miles.*’ The journey with a fully-laden merchant caravan

Stations on a great trunk road through Bithynia and seems to have taken about two months, whereas a fast

Pontos.” 2’ messenger might cover the whole distance in about nine days. A more southerly and direct route ran along from Gerede Morier mentions that near Cheriana, Sheyran, Ulu Siran, he

via Tosya, then Pimolisa, Osmancik, and Amaseia, through met three Tatar postmen for Erzurum. They had left the Dazemonites (military assembly point at Kazovasi)?8 Istanbul seven days previously, and would need about two and Komana Pontika, Gomenek. From there it followed up more days to complete the journey.*? There is no reason to

the Iris valley and might either join up with the northern suppose that these times would have been appreciably difroute or continue directly eastward to the plain of Nikopolis, ferent in the Roman or Byzantine periods. Other early trav-

Purk. An alternative road eastward branched south at elers using the route, or part of it, were Ibn Battutah, Dokeia, Tokat, to go to Sebasteia, Sivas. From there, routes d’Aramon, Newbery, Evliya Celebi, the Sieur de la Boullaye radiated northward through Nikopolis and Koloneia to the la Gouz, Melton, Tavernier, Tournefort, Ouseley, Morier, Black Sea coast, or eastward through these same towns and Fontanier, Kinneir, Ker Porter, and Fraser. It was in the through Satala to Theodosioupolis, or eastward through course of his return journey from India that the Reverend Tephrike, Divrigi, and along the line of the Euphrates to Henry Martyn died on the same road at Tokat in 1813.°

Eriza in Acilisene, Erzincan, and Theodosioupolis, Karin, We have not attempted to trace the exact course of (Erzurum).?? Another route which falls outside the bounds branches of the trunk road eastward from Bithynia into the of the Pontos, went southeastward to Melitene, Malatya, Pontos, but they are well known and marked on our general and from thence across the Euphrates and on to the northern map.°+

shores Lake Thospitis, Van,Eastward at Chliat, Ahlat.Dazimon, Th , ; Tokat; ; PIS: , ° of ah TE MESe WETS The Roads from Dokeia,

still the normal trade routesPontika, from Istanbul to Tabriz and the . _ ,; introduction . Komana Guimenek; andthe Neokaisareia, Caucasus until the of the steamship and .; . —— qe Niksar; to Nikopolis, Purk; and to Koloneia, building of a carriage road in the nineteenth century diverted .; Sebinkarahisar

the greater part of the trade to the ports of Samsun and Trebizond. Perhaps the last completely recorded journey The course of the road north of Dazimon via Komana along the northern route for business purposes was made by Pontika and Neokaisareia is clear.**> From Komana Pontika

the American missionaries Dwight and Smith in 1833.%° and from Neokaisareia eastward, the question remains open Their account is invaluable. Many of the regular staging as to how far the roads lay in the Lykos or Iris valleys, or ran posts which they used go back to the Roman or Byzantine along the flanking mountains—all of which have been reperiods, and they give a time of twenty-five days (262 hours’ corded as routes by travelers along this way. The existence of riding) and an estimated 786 miles for the distance from a Byzantine route to Koloneia, Sebinkarahisar, is indicated Istanbul to Erzurum. A tenth-century Armenian itinerary by Arabic sources. One of these states that Koloneia was the seems to follow the same route, but only gives the more sixtieth post station on the road from Constantinople, but important intermediate stops, computing the distance as 675 gives no information as to the actual course of the road.*° 31. Manandian, Trade and Cities, 168—69.

27. Hogarth and Munro (1891), 739. The position of the Hittite 32. Morier (1808), 333; Fraser (1835), II, 386. In April 1835 capital at Bogazkoy, well to the south of Amasya, suggests that at a Fraser made the journey from Erzurum to Constantinople in 11 much earlier period the main route ran to the south of the Byzantine days, of which 2 days were enforced rest because no horses could be

route. This earlier road is Ramsay’s ‘“‘Royal Road,” Ramsay, found. He seems to have averaged 2 to 3 hours’ sleep per day, and

Historical Geography, 27-35. laments of his journey: “‘I know that it has been done even quicker by 28. See Constantine Porphyrogenitus, De cerimoniis, Bonn ed., I, European gentlemen as well as Tatars, and it was not my fault on this

444-45. Ramsay, Asia Minor, 202-3, reinterprets this important occasion that it was not performed in 7 days; but mud, and water, passage. His view was severely criticized by J. B. Bury, ““The Aplekta and want of horses are things which cannot be contended with.”’ of Asia Minor,” Byzantis, 2 (1911-1912), 216—24; the whole passage Caravan times are given by the Anonymous (1807), I, 43; II, 485: has more recently been clarified by G. Huxley, “*A List of Aplekta,”’ from Constantinople to Tokat 20 days, and from there to Erzurum GRBS, 16 (1975), 87-93. Bury concludes by observing that marches 15 days. along the northern road are omitted; this list of aplekta 1s certainly 33. Ibn Batuttah (1332), II, 434-38; Chesneau (1548), 68-78, for

incomplete, and to take it as the only basis for considering the the stretch from Laodikeia to Theodosioupolis (Erzurum), where regular gathering points for Byzantine armies would give a very false the geographical footnotes of Chesnau’s editor, Schefer, are some-

impression of Byzantine campaigns in Asia Minor. See note 23 what misleading; Newbery (1581), VIII, 470-76; Evliya (1644);

above. Gouz (1647); Melton (1670); Tavernier (1681); Tournefort (1701); 29. If Baynes’s interpretation of the first campaign of Heraclius Ouseley (1810-12); Kinneir (1813); Ker Porter (1817-20); and

against the Persians is correct, then we may add that the site of the Fontanier (1827). For page references see below, where relevant. For great battle was almost certainly in the region of Sebasteia, Sivas: N. Henry Martyn, see Smith and Dwight (1830), 44. H. Baynes, *“The First Campaign of Heraclius against the Persians,”’ 34. Ramsay, Asia Minor, 23-88; Munro, JHS, 21 (1901), 52-56 EHR, 19 (1904), 701. It must have been through Sebasteia that and pl. 1v. Munro points out that the importance of the trunk road is Basil I conducted his campaign against the Paulicians, although the indicated by the milestones along it. These show that it was repaired

town is strangely not mentioned. For an interpretation of this cam- on at least twelve occasions between the last quarter of the Ist paign, see J. G. Anderson, ““The campaign of Basil I against the century and the mid-4th century a.p. Anderson, JHS, 17 (1897), Paulicians in 872 a.p.,” CR, 10 (1896), 136-40. Romanos IV 22—44; Anderson, SP, I, 73-104, Maps vu-1x; Cumonts, SP, II, marched through Sebasteia on his way to Theodosioupolis in 1069: 121—48, maps x1, xu; Wilson, Thesis, 365-69.

Michael Attaliates, Bonn ed., 147. Burnaby (1876), I, 319, noted a 35. Hogarth and Munro (1891), 732-735; Anderson, SP, [, body of troops marching from Sivas to Erzurum via Sebinkarahisar, 60-67, map v; Cumonts, SP, II, 254-58.

taking one month. 36. G. W. Freytag, ‘““Geschichte der Dynastien der Hamdaniden 30. Smith and Dwight (1830). in Mosul und Aleppo,” ZDMG, 10 (1856), 467.

22 CHAPTER TWO And another, Al Muqaddasi, gives it as a part of a route from Murray also traces this stretch of road across the mountains Mous, Mus, to Constantinople, with the section Koloneia to north of the Lykos.*° He gives the stations as Alma; Ermenu

Neokaisareia taking four days.*>’ Romanos IV returned (both unidentifiable); Kotanis, Kotani; and Iskersu, Berealong this road in 1069, when the stations mentioned from ketli, or a village on the river Delice. Here his tracks divide:

Theodosioupolis westward are Koloneia, Melissopetrion, the main one proceeds eastward through Kaledibi, and and Dokeia, Tokat.*® It is attested in the Ottoman period by Kiziljeuren, Kizilcaviran, to Yaghsian, Yagsiyan; a second

Hadji Khalfa who gives a road from Osmancik to Erzurum. loop runs south to Kassoba, Kuzbagi (?). This route This is obviously part of this main road from Constantinople may represent the track down to rejoin the Lykos near

eastward; the stations are discussed below.*° Resadiye. Murray’s placing of the central stations is inacFrom Neokaisareia, Niksar, to Nikopolis, Purk, and to curate, but his last two stations of Afan, Afan, and Chardak,

Koloneia, Sebinkarahisar. We can reasonably assume that Cavdar, make a recognizable route into Koilihisar, there was a Roman and Byzantine road up the Lykos valley Koyulhisar. eastward from Neokaisareia to Nikopolis even though there Evliya Celebi traveled from Niksar to Sebinkarahisar. is no literary evidence for it. The archaeological evidence Between Niksar and Koyulhisar, he lists: Kariebash, Ciftlik, consists in Munro’s record of sections of paved road which Iskefser, Shakhna, Tekine, and Chadar. The first and last of he thought were of Roman date, and the remains of an early these stations may be identified with the villages of Basciftlik bridge or bridges recorded by him,*° and by Grégoire,*! and Cavdar respectively, but we have been unable to locate about halfway between Neokaisareia and Nikopolis.*? the intervening names. If our two identifications are correct, Among archaeologists, only Munro and the Cumonts trav- Evliya’s route ran over the mountains to the north of the

eled along the mountain ridges on the north side of the Lykos, but much farther to the north than the Cumonts Lykos.** The Cumonts only came down into the valley itself | route, and he may have traveled indirectly making excur-

at Resadiye, about halfway between Neokaisareia and sions to the north and south.*’ Anniaca. From Anniaca, Koyulhisar, they followed the river Ouseley traveled westward across the mountains. He menvalley, which is wide and has gentle slopes up to the point at tions leaving the Lykos valley at Kuilhissar, Koyulhisar, and which they left it and turned north to Koloneia. At more or sleeping in the forest of Eider Urmani, Igdir Ormanlari, on less the same point as the northern turning, a side valley to the first night. On the second day he passed the Isker Su and the south leads up to the plain of Nikopolis. Eastward of this its castle, which unfortunately does not now appear on the

junction the Lykos enters gorges from which it does not free map. His resting place on the second night was Kutani, itself until Zagpa. This stretch of valley is not suitable for a Kotani. If he was traveling on a direct road, the castle should road; hence the two major stations of Koloneia and perhaps be on the upper reaches of the river Delice around Nikopolis stand well apart from the Lykos valley, to the Hasanseyh or Ulukoy. On the third day he passed through north and south respectively. Hadji Khalfa gives a northern Ermenli (not now identifiable unless it be Elmacik), and route between Osmancik, Pimolisa, and Erzurum, part of Boschiftlic (Basciftlik) to reach Niksar. He makes the useful which is now relevant. He lists Nighsar, Niksar, followed by observation that “‘besides the summer or forest road, we Telmesseh, Hadji Murad, and the plain of Achkar. If learned that there was another, but very bad, along the river Telmesseh is Tilemse, then it seems likely that Hadji Khalfa’s side.” 48

route may lie directly along the banks of the Lykos, since Ker Porter’s description of this section of the road is Tilemse is just short of Resadiye, and Hadji Murad must be perfunctory, but he appears to have followed Ouseley’s the ruined han at Asagi Kale Koyu, equivalent to Anniaca, route. He mentions Issa-Cossar, Isker su (?), Armari and Koyulhisar. The plain of Achkar is probably the plain of | Alma, one of which should be Elmacik; and Bachi Chifflick,

Nikopolis, Susehri Ovasi.*4 Basciftlik.*°

The Kiepert route eastward from Neokaisareia to Morier, traveling westward from Carahissar, SebinkaraNikopolis, collated from travelers’ accounts, runs on the hisar mentions Kuley Hissar, Koyulhisar; Isker Su, Bereketli

mountain ridges to the north of the Lykos.*° In his map a Antiqui (Berlin, 1903); the 1:800,000 wall map; and the final

37. Honigmann, A/JPHO, 4 (1936), 261-71. 1:400,000 sheets published during the First World War.

38. Skylitzes, Bonn ed. 701-2. 46. Murray, Handbook, 428-29 and map.

39. Saint-Martin, Asie Mineure, II, 687. 47. Evliya, II, 104, Evliya’s Iskefser could be the Isker Su of 40. Hogarth and Munro (1891), 730. Ouseley and Morier. It seems highly unlikely to be the Iskefer

41. Grégoire (1907), 28-33. Findicak placed on the Turkish 1:500,000 survey in the Yesil or 42. For our own observations on the valley, see p. 118 for the Tozanh Irmak valley south of Resadiye. It could however be the north to south route which crosses this stretch of trunk road. The Findicak, north of the Kelkit on the Niksar to Koyulhisar route. position of the Mithridatic stronghold Taularon is also discussed on 48. Ouseley (1812), III, 482—83, whose observation about the p. 42. Mithridatic castle sites were often reused by the Byzantines; river route is partially confirmed by Fraser. On his outward journey, therefore, its position is relevant to the problem of routes in this area. Fraser ({1835], I, 210-14) slept at Eski soor (Isker su of Ouseley),

For the possible Byzantine sites along the hill route, see p. 22. came down to the river valley and continued along its banks to the

43. Cumonts, SP, II, 273-95. branch north for Sebinkarahisar. On his return journey (II, 352-58), 44. Saint-Martin, Asie Mineure, II, 687, and map. For a discus- he followed the same route, but this time his stopping place between

sion of Achkar, see p. 27. Koyulhisar and Niksar is called Iskee Soor. The point at which he 45. The Kieperts’ great series of maps of Asia Minor were de- left the river is not specified except that it was eight hours from

veloped over the latter half of the 19th century and the first decade of Koyulhisar. It was perhaps by the bridge at Kundur.

the 20th century. We use three editions: those in Formae Orbis 49. Porter (1818), II, 697—98.

LAND ROUTES, EAST-WEST 23 (?), Hasansey (?), Ulukoy (?); and Kizil Taveran, K1zil- ched north-eastward from Afan to Matuasco, Mesudiye, caviran. It is clear that he also kept to the mountains north of and thence east through Sisorta to Koloneia. the Lykos. He notes: “‘About five miles from Isker Su on the A southerly version of the mountain route left Neokaileft of the road is a rock completely isolated among green sareia and kept closer to the heights above the Lykos valley in

fields. The substance is a hard grey granite in which is ex- the region of Resadiye. It may then have continued on cavated, certainly with great labor, a chamber nine feet through the Mithridatic site of Taularon, Tavara (?)°’ and square, with a seat and two recesses. On the left of the inside from thence through Eski Koy and Igdirormani to Anniaca. on entering is a figure which, from its resemblance to a cross, A direct alternative route is to keep along the line of the induced me to suppose that the spot in which it appeared had modern motor road in the Lykos valley, for the whole disbeen the retreat of some of the primitive Christians.” °° This, tance between Neokaisareia and Anniaca. This appears to be with Ouseley’s castle of Isker Su and a stone with crosses at the Hadji Khalfa route, but the gorges and Ouseley’s stateMesudiye, noted by Hogarth and Munro,”' appear to be the ment suggest that it was not a much used track. only mentions of possible Byzantine antiquities along the From Komana Pontika, Gomenek, to Nikopolis, Purk, or

mountain route. to Koloneia, Sebinkarahisar. The Peutinger route runs eastSmith and Dwight kept to the northern mountain route, ward from Komana Pontika and not from Neokaisareia. stopping at Kotely, Kortelos (?), and Koylisar, Koyul- The first station on this road at 16 m.p. from Komana hisar.°* Consul Brant traveled on the same route as far as Pontika is Gagonda, which Miller gives as between Georek, Kulehisar, Koyulhisar, after which he kept to the Lykos Gevrek, and Kavaklyk, Kevaklik; the second is Megabula,

valley.°* given by Miller as Almus; the third at 25 m.p. is Danae, given Tchihatcheff also traveled along these northern moun- as Samail, Semail; the fourth at 25 m.p. is Speluncis, given as tains, and gives a geological section of them. From Niksar he Kostenjazy (?); and the fifth at 12 m.p. is Mesorome, given as

names Basch Tchiflik, Basciftlik; Elmenek, Elmacik (?); Ortock or Tschiftlik. This station according to Miller marks Kotanis, Kotani; Kossaba, Kuzbagi; in the valley of the the junction with the Sebasteia, Sivas, to Nikopolis route.°8 Diledji Sou, Delice Su, Yaghsian, Yagsiyan; Afan, Afan; Grégoire followed this route and identified Megabula with Tchardak, Cavdar (?); and Koili Hissar, Koyulhisar.>°+* the village of Meghelle, Megelli. He then went to Samail, Taeschner’s account is somewhat confused, but adds Semail, and from there crossed the mountains to the north to

Asardjyk, Asarcik which he wrongly wished to change to come down into the Lykos valley at Tchermik, Cermik River Hisardyjik, Tshaqras, Cakraz, Ermanigq (?), and Qyzyldjoren, at Resadiye. Between Tchermik and Madasoun, Mudsun, Kizilcaviran. He makes the point that the early equivalent of Muday, he found the remains of what he considered to be an Koyulhisar was the now ruined han of Haci Murad at Asagi early bridge over the Lykos.*? Kale, and that this may well have been the Roman and Early Edward Melton appears to have traveled along the direct

Byzantine station of Anniaca.°> A consensus of travelers route from Tokat to the plain of Nikopolis, Susehri Ovasi. suggests that the road from Neokaisareia, Niksar, to His first station, Charkliquen, is not recognizable, but the Anniaca, Koyulhisar, took two to four days. The castle and second is Almous, Almus; the third station, Karabehir, the cross carved in, or at, a cave near Isker Su, reported by sounds correct as a Turkish name but does not appear on our Ouseley and by Morier, are perhaps indicative of a Byzantine maps; the fourth at Adras is Endiryas, Susehri; and the fifth site. We suggest that Isker Su may be either the river Delice in at the eastern end of the plain of Nikopolis is Aspidar, the region between Ulukoy and Hasanseyh, or the modern Ezbider.°° Bereketli.°® A route across this stretch of country is therefore The way ran north from Tokat, and then eastward up the likely to have followed a line through Basgiftlik, Elmacik, Iris, Yesil, valley to its headwaters. Thence it crossed the Kotani or Ketenigi, Uluk6y or Hasanseyh, Findicak or watershed formed by the Kizil and Karacam mountains, to Kizilcaviran, Afan, and Cavdar, to reach Koyulhisar. come down by a winding road into the plain of Nikopolis. The castle at Sisorta may indicate that a direct route from Tavernier also went along the direct way from Tokat to the Neokaisareia, Niksar, to Koloneia, Sebinkarahisar, bran- plain of Nikopolis.°! His first stop was at Almus. His second place-name, Chesmebeler, is no longer clearly recognizable, 50. Morier (1808), 304, who appears to have reversed the orderof —_ but Bel indicates a mountain pass, so the station may be at

the stations Iskersu and Kiziltaveran. the headwaters of the Iris, near the watershed. His third Sionh and Dwight ( $30). 44 St. Station is Adras, Endiriyas, Susehri, and his fourth is

53. Brant (1835), 187~223. Izbeder, Ezbider.

54. Tchihatcheff, Asie Mineure (see note 23), IV, 121-28; A Tchihatcheff itinerary®’ gives this route in more detail. Tschichatschof (1858), 277-80. Other itineraries for this route, not seen by us, are in Nepveu, Voyages en Perse (Paris, 1813), III. 55. Taeschner, Anatolische Wegenetz, II, 13—17, pl. 41. Chesneau 57. For a discussion of Taularon, see p. 42.

(1548), 70, 71, gives no details of his route, but refers to Assarguict, 58. Miller, /R, col. 675. Tarhan, Map, marks ruins between presumably Asagi Kale, and to the castle of Coyouassar. He crossed Findicik and Ficak and identifies them as Danae. This accords the Lykos about two miles below the castle and traveled on to roughly with the Miller identification, although Semail is a little to

Asebids, Ezbider. the west of Findicik. 56. Both identifications present difficulties and must await verifi- 59. Grégoire (1907), 28-33. cation on the spot. We suggest Bereketli, since the Turkish “Iska 60. Melton (1670), 254—57.

etmek” means “‘to irrigate’, and the Turkish map marks water 61. Tavernier (1681), 9-15.

wheels on the unnamed river which runs by Bereketli. 62. Tchihatcheff (1858), 324—29 and map.

24 CHAPTER TWO Traveling east from Tokat, it crosses the hills to Almous, and met the Sebasteia to Nikopolis road in the region of Almus. This is a shorter traverse than that of the road which Serefiye, whence the two roads continue together, probably

follows the Iris and turns eastward at Komana Pontika. along the line of the modern one, as far as the plain of From Almus he traveled on the north side of the river Nikopolis. The location of the Peutinger stations remains through Terzi, Tiyeri, close to Grégoire’s Megelli. At Samail, uncertain. We prefer, with Grégoire, to place Gagonda at

Semail, he crossed the river to the south bank and went via Almus, which is frequently mentioned by travelers; and Yumbelet, Tepumpelit, over the hills to Hipsala, Ipsile.°° Magabulla at Megelli, where both the distance from From there he went up the headwaters of the Iris through Gagonda, Almus, and the form of the present name favor the Khamarly (?); Mourassy, Morvasit (?); Geusten (?); and identification. The last three stations before Nikopolis are Kourdkeui, Kurdkoy (not marked on any map) to Enderes, Danae, Speluncis, and Mesorome.®’ If the latter marks the

Susehri. junction with the Sebasteia road, we suggest that it should The Tavernier and Tchihatcheff itineraries are sufficient to fall in the region of Serefiye.

establish the existence of this direct route eastward from The route from Tokat to Koloneia follows the track that

Tokat or Komana Pontika. has just been described, up the Iris valley, but probably Pitton de Tournefort appears .to have returned from crossed over into the Lykos valley somewhere below Koloneia along the Lykos valley and followed Grégoire’s Anniaca. This move from one river to the other is suggested route in reverse. He writes that he crossed the river Lykos by the routes followed by Newbury, Tournefort, and Barth. and went on to the camp at Almous. From there it was Strecker’ and Taeschner list this itinerary, which was taken

another nine-hour stage to Tokat.°* by Gregoire. It could even be that followed by Newbery and John Newbery, returning from the east in 1581, traveled Barth.°? from Andre, Susehri, to Yeoltedder, Koyulhisar (?): ‘‘This Perhaps there was a road directly along the valley for fast day wee passed by a very great castle to the north of the travel and winter use, and summer roads along the mountain water.’ This makes it fairly clear that his first stop was ridges, where fodder for man and beast were more easily

Koyulhisar. From there “‘wee passed over a very high obtainable in season.

Mountaine ... met andFrom at the foot of Purk, this Mountaine ; - ne , ; ; ae wee Nikopolis, or Koloneia, me Sebinkarahisar,

; , ,; to Satala, Sadak of these routes eastward to

with another river,” and stopped at Longo. Weare unable to

identify Longo, but it .looks as ifchoice he crossedfor the the mountains to ; The course

the south and came down Iriswhere valley, so that Longo ; between a; . ; Satala andinto the the point there is a junction them

should bethe onefrontier of the villages on thenorth upperfrom reaches of this .Malatya, . . ; , and road running Melitene, river. His next station of Prassa should be well down the ows . ,;and ey ~ it andmight with roads eastward from Magabula, Sebasteia, Sivas, is less ;certain. valley be Grégoire’s Megelli. ,; . Kiepert saw thatand there were two roads. from Nikopolis to Newbery then passed by Nannous, Almus, Manec, 9 ; 65 RamsayTokat.°° marks theThis Byzantine routeroughly to the north Mamo (?),TL: to Satala; arrive,;and atthe Tocat, was also ,; ; of river Lykos on his road map, presumably on the basis

the route followed byreports.’° Barth, traveling westward. went 70 ;of ; ; of travelers’ Yorke contributed theHe firstand analysis from Koloneia, Sebinkarahisar, to Enderes, Susehri, , ae i op: ; . the stations and comment on Koyulhisar them, pointing out that “the thence to Koiluhisar, Koyulhisar. From he con; 71 ; stations on these roads are evidently much confused.” ’' The tinued down the valley as far as Modassu, Mudsun; between

-ruin Cumonts traveled over be whatAsagikale. they assumed was the the two he records a castle which may Heone, .ofsta; ; roads, and commented extensively on the Antonine then, crossed the‘*mountains tounpublished the south of him to drop into . 45 . ; ; ;has made , , tions; and in an thesis T. B. Mitford the Iris valley at Samail, Semail, and continued down the Furth tions.73 Our own tribution is to add furth valley to Almusch, Almus, and thence via Komana to UPMICT SUBBESIONS. COMMMOUTON IS TO ade further

Tokat."" 67.evidence Grégoire 59.59.Ptolemy, G hy, ed The sum of this ests th (1907), . Gregoire (loc. ), Joc.cit. cit. ini note Ptolemy, Geography, ed.

Tokat and K IS Pontik SUBE she a the route vom Miller, 874. Miller, 7R, cols. 675, 731, whose account of the

OKat ané Komana Fonuka Fan up the Aris valley about as Sebasteia to Nikopolis route is very misleading. Adontz, Armenia, far as the reaches below Ipsile. The Peutinger route to 62—64, discusses Mesorome in connection with this road. Nikopolis probably then continued up to the Iris headwaters 68. W. Strecker, ‘“Topographische Mittheilungen uber Hocharmenien,” Zeitschrift fur allgemeine Erdkunde, 10—11 (1864), 355. The Strecker articles are fundamental for a study of eastern Pontic

63. N. Adontz, Armenia in the period of Justinian, trans. N. routes. Taeschner, Anatolische Wegenetz, 19-24, pl. 44. Garsoian (Lisbon, 1970), 62, 68 and note 26, 202, 404, gives this 69. Cumonts, SP, II, 292-93, make this suggestion. For Barth Ipsile as the Byzantine “YwnAn of Theophanes Continuatus, Bonn and Newbery, see above, notes 65 and 66. There is a general account ed., 354; Cedrenus, Bonn ed., II, 250; and Notitiae, II, X, XIII. In in Ritter, Erdkunde, 216—24. this he follows W. Tomaschek, ‘“‘Historisch-Topographisches vom 70. Ramsay, Asia Minor, map opposite p. 23.

oberen Euphrat und aus Ost-Kappadokien,” in Beitrdge zur alten 71. V. W. Yorke, ““A Journey in the valley of the Upper Geschichte und Geographie, Festschrift fiir Heinrich Kiepert (Berlin, Euphrates,” The Geographical Journal, 8 (1896), 465-68. Mordt1898), 148-49. The identification is important if correct, since mann (1850), comments, “‘Diese ganze Strecke bedarf noch einer Kedrenos connects Hypsele with Charsianon, thereby giving some recht griindlichen Untersuchung.”

clue as to the position of this important stronghold. 72. Cumonts, SP, II, 318—42, and maps xxul-xxvi. Cumont notes

64. Tournefort, II, 294—98. (p. 321), that he does not use the Peutinger Tables which he regards 65. Newbery (1581), 470-76. as too confused. See also Ramsay’s view of Peutinger, in Asia Minor, 66. Barth (1858), 1-105. Account also in Mordtmann (1859), 96.

453-59. 73. T. B. Mitford, The Roman Frontier in the upper Euphrates

LAND ROUTES, EAST-WEST 25 geographical information and records of sites, but the final to Sinn Nuhas; one day from Sinn Nuhas to Samuqmus; and _ location of the stations and of the road remains to be verified. two days from Samuqmus to Koloneia, Quluniyat-al-Aufi. ’°

However, the evidence for these sections of road is quite Mustawfi, writing in 1340, describes a route from Sivas via extensive. We do not attempt to trace the road per ripam Acilisene, Erzincan, to Erzurum. His stations, with distances southward from Zimara, Zinegar, to Melitene, Malatya, in leagues, are: Sivas, 4; Rubat Khwajah Ahmad, 10; Zarah, since this area is already well south of the limits of the 8; Akarsuk, 5; Aq Shahr, 8; Arzancak, 7; Khwajah Ahmad,

Pontos. 5; Arzanjan.”’

Lists of Stations. The Antonine and Peutinger lists of The seventeenth-century geographer Hadji Khalfa lists stations may best be set out as they are given in Yorke’* two routes between Erzurum and Sivas, one from Erzincan (where the Roman figures represent milia passuum). The to Sivas, and one from Osmancik to Erzurum. The route whole of the road Melitene-Satala is given in the Antonine Sivas to Erzurum is given in march times as follows: "8 Itinerary. It is described as having led per ripam:

la. Satala XVII; Suissa XVIII; Arauracos XXIV; Carsagis Distance Hours XXVIII; Sineruas XXVHI; Analiba XVI: Zimara XVI; Si f the Adii-Sou’ 5

Teucila XXVIII; Sabus XVI; Dascusa XXXII; Ciaca Vas fo source 0" © Agr OU sree eee e renee

,.,t

XVIII: Melitena Adji Sou to Kodj-Hissar ...................... 5

A considerable portion of the same road is given in the i Pt aa ne 71 Peutinger Tables, but it starts from Draconis, the second Ar anout Eusi wo lailsk d’Aiach Cee 6 station from Satala on the direct road Nikopolis—Satala (see ae ee

below Ila. IIb.) lalak d’Aiach to Chahneh Tchemen............. 4 Ib. Draconis XVI; Haris XVII; Eregarsina VIII; Bubalia Themen to AKehar seen i XXVII; Zimara XVIII; Zenocopi XVIII; Vereuso XIII; > to lar azi Binari. ee 71

Saba XVIII; Dascusa XVIII; Hispa X VIII; Arangos VIII: Baal © a

Ciaca XXVII-: Melitena. largazi Binari to lailak Tschemen .............. 4 A direct road Nikopolis—Satala is given both in the Itinerary lailak Tchemen to lassi Tchemen............... 43 and the Tables. lassi Tchemen to Kara Boulour . Lecce eee eeeee OD IIa. In the Antonine Itinerary: Nicopolis XXIV: para Rowlour fo stenir Sarass. ae :

Olotoedariza XXVI; Dracontes XXIV; Haza XXVI: en JANI ernest ee eee

Satala Dyjanik to Toloslar ............... 0.000202 e eee OF .Draconis . - Ak Deghirmen to Mama Khatoun.............. 54 figure); XIII; Cunissa X; Hassis XIII; Ziziola ); XII: Satala. Mama Khatoun to Penek ..................... 45

IIb. Inthe Peutinger Tables: Nicopoli XIV; Caltiorissa (no Toloslar to Ak Deghirmen (crossing Euphrates) .. 4;

Another road Nikopolis—Satala which, if Carsagis is taken aie Cs ; to be the same as Carsat, joins road Ia at that point, is given 4 JOM poser e sree seers essences es

in the Itinerary. Ilidjeh to Erzroum ... 2... eee eee ees 4

IIc. Nikopolis XXIV; Olotoedariza XXIV; Carsat XXIV: ; oo. ; Arauracos XXIV: Suissa XX VI: Satala His second route is given from east to west in much less rae , detail. It names Erzroum, Ech Kalah, Terdgian, Ghelghis, A road from Nikopolis which cuts into road Ia at Analiba lain of Chir. and Kara-Hissar.°

and Ib at Zimara, is given in the Peutinger Table. pain ° . ld. Nicopoli XXX: Ole Oberda XV: Caleorsissa XXIII His third route from Erzincan to Sivas runs: Erzendgian;

r) ° “D: ., ar on 80

Analiba XV: Fimara , “Erzendjik de Khavadgiah Ahmed,” 3 hours; Sourzadeh;

There are several medieval lists which are relevant to this oak ce Khavadenan Ahmed : Sivas. .. stretch of road. An Armenian itinerary of ca. A.D. 971-81 is fourth route from Pemolisa, Osmandjik, Osmancik, to gives the road from Dvin. the Armenian capital, to Rome Theodosioupolis, Erzurum, lists the following names on the

via Karin, Erzurum; Ankyra; and Constantinople. The part Muted Serine trom ee he rae Cova bar: Had which concerns us gives from Karin to the Frontier Ditch as Kemakh: an “thence to Erzroum 81 6

100 miles, and from the Frontier Ditch to Koloneia as 90 ° —_— , miles. 7° Finally there is the Pegolotti route which runs from Ayas The Arab itineraries of Al Mugaddasi and Mustawfi only on the Cilician coast to Tabriz, via Caesareia, Kaysert;

, , Sebasteia; Eriza, Erzincan; and Theodosioupolis. Pegolotti partly concern us, since they are largely outside the Pontos. .;; Coming from the east, Al Muqaddasi names Mous, Mus: gives the stations between Sebasteia and Eriza as follows: then one day to a place unrecognizable; one day from there Allentrare in ; Salvastro di a Laiazo; Dudriaga, Greboco, Mughisar, Arzinga.

from the Black Sea to Samosata (Oxford, 1972) (unpublished D. Phil. 76. Given in Honigmann, A/PHO, 4 (1936), 263-65. thesis). We are most grateful to Dr. Mitford for allowing us to make 77. Mustawfi, trans. G. le Strange, chapter XV. use of his thesis, which he will be publishing as a continuation of SP 78. Saint-Martin, Asie Mineure, II, 685.

and as a study of the Euphrates frontier in the Roman period. 79. Saint-Martin, Asie Mineure, Il, 655. 74. Yorke (1894), 465. We follow Yorke’s schema, with the excep- 80. Saint-Martin, Asie Mineure, II, 655.

tion that we assume Analiba to be identical with Bubalia. 81. Saint-Martin, Asie Mineure, I], 687.

75. Given in Manadian, Trade and Cities, 169. 82. Pegolotti, Pratica, 28-29.

26 CHAPTER TWO The Milestones. Five Roman milestones lie more or less at Koruyolu Koyii, now Tavginler.8° Tarhan marks imporon the Cumont’s route from Nikopolis to Melikserif, via tant ruins at Zagpa®’ on the river Lykos which he identifies

Ezbider and Refahiye (see below).°? as Olotoedariza, and other ruins at Refahiye, to which he The Sites. The number of ancient and medieval sites attributes no ancient name. He also marks ruins at Buldur which the Cumonts were able to report in this area can now Harabeleri, southeast of Nikopolis, which he identifies as be improved a little. Yasar Paksoy mentions ruins at Aksehir, Mesorome; at Yenik6y Harabeleri, which he identifies as about 20 km north-east of Refahiye, between the villages Caltiorissa, Caleorsissa; and at Distas, which he identifies as of Kiirtbaloglu and Hacirke. He also speaks of ruins of a Subalis.°° church and castle in the area between the Melikserif valley Our own researches have yielded sites along the Lykos and the village of Horun.®* The Sipdigin milestone of the river at Asagi Haydurrtik, where there are the foundations Cumonts and Boré must be associated with one of these of a large church and a considerable area strewn with pottwo sites. Boré was the first to report these ruins, which he sherds. Below Kalur are twin castle sites in a defile, one on believed to be a part of a temple. He was told that the former either side of the river. There is a stretch of paved road, and name of Melikserif was “Erzez” or “‘Anourgia,’’ which further downstream is another castle site at Avarak, ancient might possibly be connected with the Eregarsina of Peutinger Arauraka.®? There is a reported castle site at Balikhisar, route Ib. To the south of Refahiye, Paksoy has a rather above Asagi Hayduruk. Churches were also reported by a confused report of two city sites one of which appears to be villager of K alur to exist south of thé fiver at Beskilise, which on the Maden Dere, perhaps near modern Akarsu, and the means “Five Churches.”’ other in the region of Distas K6yi, perhaps near the modern North of the Lykos are Alansa, where Clavijo stopped for Gumisakar, but the reported situation does not appear to a night on his way between Trebizond and Erzincan, which tally with the villages marked on the map. He identifies the might be an old station.?° Going westward, there are the two sites as Sinoria and Subalis respectively, and Maden sites of Tarsos, Asagi Tersun; Cheriana, Ulu Siran; Mumya; Dere K oyu, Akarsu (?), as Elegarsina. He also notes that they Cirmis; Govatha, Kovata; and Koloneia.”!

are at the fourteenth post on the old caravan route from The Medieval Lists and Those of Hadji Khalfa. The Melitene, Malatya, to Paipertes, Bayburt.®° In the area tenth-century Armenian route given by Manandian is unnorth of Kamacha, Kemah, he notes three old churches; Isa specific as to stations and we are not even sure of the identity Vori¢ near Gokkaya, Tasdibi Kilise, and Meryemana Kilise of the “Frontier Ditch,” but the mention of Koloneia in-

83.. Mitford. Thesis. 1. 166-70: 500. 202-5 this route the upperK valley the river Mitford, Zhesis, I, HI —7/0; UI,dicates nos.that; —5. L went bly down via Cheri ta. ofnorth of

1) At Kiicik Gizel, 5 mi northeast of Nikopolis, Purk, noted by nos ene then probably via Cheriana and Kovata, n Grégoire (1907), 37, unpublished but recorded in Mitford. 2) At ucta.

Sevindik, noted by the Cumonts, SP, II, 319, and Grégoire (1907), The other medieval routes present considerable difficul33; unpublished but recorded in Mitford. 3) At Agvanis, noted by ties. The Mustawfi route, taken from Sebasteia, Sivas, eastCumonts, SP, I, 320; unpublished but recorded in Mitford. 4) At ward, gives the second station as Zarah, Zara, and so the

Sipdigin, noted by Cumonts, SP, II, 324; unpublished but recorded .

in Mitford. 5) At Melikserif, first inaccurately noted by Boré (1835), road must have followed the river Halys along an easy stretch 369, in CIL, III, 306, then noted by the Cumonts, SP, II, 327-28, and of valley between these two towns. The first station, Rubat

published by F. Cumont in “Le gouvernement de Cappadoce sous Khwajah Ahmad, is therefore probably the equivalent of les Flaviens,” Bulletin de l’Académie de Belgique (1905), 218 ff. The Camisa of the Antonine Itinerary and Comassa of the unpublished Cumonts and Grégoire inscriptions will be published Peutinger Tables. preserved as Kemis. next to Hafik or

by Mitford together with new inscriptions found by him. 6 > P »»

84. Yasar Paksoy, ““Tarthi Kaleler,” Turkiye Turizm (Ankara, 86. Paksoy, ‘Tarihi Kaleler.”” One of these may be the site of September, 1964); Cumonts, SP, I, 326—30; see pp. 318—36 for their Tortan, where Gregory the Illuminator spent his last days, according

finding and comments on this whole section of the route. Bore to the Geography of Vatabied Vartan; see M. J. Saint-Martin,

(1835), 369. Mémoires historiques et geographiques sur l’Arménie, I] (Paris, 1819), 85. Paksoy, “Tarihi Kaleler.” The celebrated mines of the town of 43] —33.

Madden, Maden, are referred to by J. von Hammer, trans. J. J. 87. Tarhan, Map. Evliya (1644), II, 105, speaks of a cave full of Hellert, Histoire de l’Empire Ottoman (Paris, 1836), IV, 211, 435. treasure at Tilismat Za’aba. Ruins were noted by Boré (1835) and by Hellert marks this site in the atlas which he added to the French Strecker (1855), 356-57. They were also reported to D.C. W.

edition, together with a number of other ancient sites, but the 88. Tarhan, Map. identifications are of little value since both the place-names and the 89. See Asagi Hayduruk, p. 175; twin castles, p. 169; Avarak, routes are very inaccurately placed: J. J. Hellert, Nouvel atlas de p. 177.

l’Empire Ottoman (Paris, 1844), pls. x1, xx. (The volume has a 90. Clavijo (1404), ed. Estrada, 83 (where the name is handsome title page worthy of better contents.) The site at Maden is Alangogaga), and ed. Le Strange, 120. almost certainly to be identified with Ibn Battutah’s Kumish: see Ibn 91. See Asagi Tersun, p. 174; Ulu Siran, p. 173; Cirmis, p. 176; Battutah (1332), II, 436-37. The recent renaming of the town as Koloneia, Sebinkarahisar, p. 145. Govatha, Kovata, is mentioned Gumisakar or “‘Silver flows” is indicative of its productivity. This only by Ghevond as one of the regions raided by Yezid, Governor of may also be the upper mine referred to by Al Muqaddasi as Ma-adin Armenia, in the 770’s. The name survives on the Turkish map for the an-nuhas and identified by Honigmann with the station Sinn Nuhas region northwest of Alucra and D.C. W. was given a report of ruins

on the Mous, Mus, to Koloneia, Sebinkarahisar, route. See Honig- there. Ghevond also mentions Castillon and the province of mann, A/PHO, 4 (1936), 263, 266. Van Lennep (1864), 216-17 has a Marithinesse which we have been unable to identify. Ghevond, few words on mines east of Tokat. The basis of further research on trans. G. V. Chahnazarian, Histoire des guerres et des conquétes des this subject is being provided by twenty-one geological maps, scale Arabes en Arménie (Paris, 1856), 152. Also see p. 46, on the route 1: 500,000, with explanatory booklets, which will cover the whole of from Esbiye southward.

Turkey. The booklets contain brief historical notes and are published 92. Manandian, Trade and Cities, 168-69; see p. 35 for the by Maden Tetkik ve Arama Enstitusii Yayinlarindan, Ankara. Frontier Ditch.

LAND ROUTES, EAST-WEST 27 Koghisar. From Zara eastward, there are three possibilities this assumption, the third station, Akarsuk, would fall at the for Mustawfi’s route, depending upon our identification of village of Aksu on the north slopes of the Kosedagi moun-

the stations. tains, and Aq Shahr would fall at Eskisar, with the rest of the

The first possibility takes the line of the modern road road following the course described above. around the western and northern slopes of the Kosedagi The third possibility depends upon a radical change in the mountains to the plain of Nikopolis. This would put the third siting of Akarsuk and Aq Shahr. Modern Akarsu exists as a station, Akarsuk, in the region of Serefiye, and the fourth township to the west of Refahiye, near Basgercenis, on the station, Aq Shahr, in the area of the plain of Nikopolis, line of the most direct route from Sivas to Erzurum or Sadak. formerly known as the plain of Askar.?° A site for this And Aksehir is the name of some ruins on a site between station may plausibly be put at the village of Eskisar, called Hacirke and Kurtbaloglu on the same direct route.?’ If we by the Cumonts Eski-Sheir, Eskisehir, where they found move these two stations to their new sites, then we have a ruins. They may represent a middle Byzantine and Turkish route running directly eastward from Zara up to the headlocal successor site to the Roman and Justinianic Nikopolis, waters of the river Halys, and then across the Kuzildagi to which, for practical administrative purposes, had retreated come down the river Binasor Dere to the third station of to Koloneia.?* The fifth station of Arzancak cannot now be _— Akarsuk, Akarsu. The fourth station then falls at the Aksehir

found on the map, but may tentatively be placed at the ruins. This leaves two stations, Arzancak and Kwajah Melikserif site, which according to Boré was formerly known Ahmad, to be found along the Cimen and Sipikor mounas “‘Erzez” or ‘“‘Anourgia.”°?° The sixth and last station, tains. The distance is short for two further stations, but the Kwajah Ahmad, is also unknown but may be the village of problem may be resolved by the reference to the third of Ahmediye on the road which runs from the Acilisene, Hadji Khalfa’s itineraries,?® which clearly describes the same Erzincan, valley across the Cimen Daglar: mountains. road, but unites these two stations into one, as ““Erzendyik de The second possibility for this itinerary would be to iden- K havadgiah Ahmed.”’ This station should fall at the modern tify it with a summer route running along the river Halys to Ahmediye, described above.

the second station, Zara, and then northeastward, more or On the basis of the first and second possibilities, the less directly across the Kosedagi mountains to the plain of Mustawfi route would correspond to the direct Nikopolis to Nikopolis. This route corresponds to one taken in 1838 by Satala road, along a stretch from the plain of Nikopolis, Consul Suter, who was traveling from Susehri to Sivas.?° On Susehri, Purk, Askar Ovasi to Arzancak, Melikserif, as we

- shall see below.”?

__ 93. Cumonis, SP, II, 303, has Ashkar-Ova. In a Hadji Khalfa On the basis of the third possibility, the Mustawfi route

itinerary Il, it appears as the of Achkar: Saint-Martin, Asie ; ; to Mineure, 687. would be aplain separate and direct road from Sebasteia 94. The reporting of this site is confused. The Turkish map marks Acilisene. This direct road only makes a brief junction with Eskisar just east of Pirk, on the banks of the river Pulat, at the point the Nikopolis to Satala road at or near the site of Aksehir,

usehri Ovasi. ar is about m upstream to the south in

where it Hows ou of its valley an¢ into the plain One eee Hacirke, Kiirtbaloglu, in the valley of the river Orcil.

the Pulat valley. Tarhan. Map, marks a castle at Eskihisar which may All three of these hypotheses present problems which will be his equivalent for the ruins of Nikopolis, Pirks, which hedoesnot Only be resolved by further exploration. mark at all. The village next to Piirk is Eskisar, but the Turkish map The Pegolotti route takes four days from Sebasteia, Sivas, marks no Eskihisar at all in the area. Taylor (1866), 301-8, noted the to Eriza, Erzincan, which is fast traveling. Yule suggests that remains of a castle with a tunnel cistern at Aksehr, Aksar, but his use this route went via Tephrike, Divrigi, and thence up the

of the names of the two places, on pp. 301-2, suggests that he had Euvh r Eriza. He b thi th ti

confused them (in striking contrast to his usual accurate reporting). uphrates va ey to rza. © Dases Els on € assump 10n On p. 301 Taylor refers to a milestone at Aksehr and to Akshehrabad that the station Dudriaga should read Duvriaka, Divrigi, as laying three hours to the south, which would equate it with Aksar. Tephrike,’°° which would be reasonable if the distance beHowever, the caption to his illustration of the milestone on p. 302

states that it was found at Akshehrabad. It seems most likely that a Taylor’s castle and milestone were at Aksar, and Cumont was able to to detail deserted him after the Harsit valley and he remarks “‘it is find it again at Ashkar, Aksar. He gives an illustration of it in SP, II, useless to delay the reader by a journal of my daily progress.”’ The 314. On p. 313, Cumont describes a Byzantine chapel at Eski-Sheir, travelling conditions in early January must have been appalling,

Eskigar, together with other remains. Aksehir was a city of some whichever route he took, and it may well be that he was unable to importance in the Seljuk period. G. Le Strange places it at Aksar in keep any journal. Grégoire traveled up the Pulat valley and across the valley of the river Pulat: Lands of the Eastern Caliphate the mountains to Zara (Grégoire [1907], 38) but describes it as a most (Cambridge, 1905), 147. The placing of Aksehir is further confused unlikely route, taking twelve hours, whereas the route along the by Taeschner, s.v. ““Akshehr” in E/?, where he describes it as being modern Susehri to Zara road takes only nine hours. on the river Kelkit between Koyluhisar, Koyulhisar, and Susehri. 97. The problem of siting Aq Shahr in the area of Nikopolis has This would place it somewhere west of Susehri, but he then suggests been discussed above. We have only Paks6éy’s note for the existence

it to be Guzeller or Ezbider, which would place it east of Susehri. of these ruins.

95. Bore (1835), 36. The Turkish map marks a nearby village 98. See p. 25. called Hanzar, which tnay possibly be connected with the site. Evliya 99. See p. 28. (1644), II, 205, records an Armenian village (Erzensi), and Newbery 100. H. Yule, Cathay and the Way Thither (London, 1866), II, (1581), 472, records Ardansegh. These places could all represent the 299-301. This may represent a Byzantine road: Anderson, JHS, 17

same station and be read as versions of the Peutinger station (1897), 32, states that ‘“‘this road leading to Tephrike and thence to

Eregarsina on route Ib. Zimara on the Euphrates was of great importance during the

96. Suter (1838), 437-38, and see below. pp. 46—47. Chesney Paulician revolt in the ninth century.” The existence of a road to (1831), 121-41, may also have passed this way on his journey from Tephrike from Bathys Rhyax is attested by the list of aplekta in Trebizond via Sebinkarahisar and Sivas but his customary attention Constantine Porphyrogenitus. But the fact that the road branches

28 CHAPTER TWO tween Sebasteia and Tephrike were not too long a stage for aplekton (military assembly field), before moving one day. Yule does not identify the other places, but Greboco eastward.'°4

"would have to fall at Ilic, and Mughisar at Kamacha, The first Hadji Khalfa route'®°’ includes the enigmatic Kemah. Manandian and Kiepert follow Heyd'°’ in finding a Akchar as its seventh station, but it is still impossible to more direct road. Their stations are: Dudriaga, Todirge, by identify it or decide where the route went. The second the lake between Hafik and Zara; and Greboco (Agreboce of station, Kodj Hissar, Kochisar, or Hafik, corresponds to the Pizigano map of 1367,'°? to be identified with Mihar,a | Camisa of the Antonine Itinerary and Comassa of the village north of the track about halfway between Refahiye Peutinger Tables. The route then continues through four and Erzincan on the route mentioned by Strecker and fol- unknown stations before reaching Akchar. The station imlowed by Tchihatcheff and by the Cumonts). This itinerary mediately previous to Akchar is Chahneh Tchemen. If this is has the disadvantage that its first and last stages are too short to be identified with Sahnacimen, a village which lies about and its middle stages far too long. A more sensible way would four-hours march west of the plain of Nikopolis, therefore be to follow the third possibility for the Mustawfi itinerary, agreeing with the distance stated by Hadyji Khalfa, Akchar and to place Dudriaga at Yenikoy on the upper reaches of the should be the Seljuk and Ottoman equivalent of Nikopolis. Halys beyond Imranl. Greboco would then fall in the Hence the Hadji Khalfa route would have run roughly along region of Akarsu; Mughisar would be at the Aksehir ruins the line of the motor road from Zara to Nikopolis. However, between Kurtbaloglu and Hacirke, and the road would con- station ten along this road is Iailak Tchemen, Cimen Yayla, tinue from there parallel with the Strecker, Tchihatcheff, and so it is still possible that Akchar is to be identified with the Cumonts routes. But the Tephrike to the Euphrates route Aksehir ruins, between Hacirke and Kurtbaloglu. In this favored by Yule could well be the right road for Pegolotti’s case, Hadji Khalfa’s route continued east from Zara, up the itinerary. The answer must remain uncertain until the — Halys valley, past Imranli, to the headwaters, and crossed the possible sites for Dudriaga, Greboco, and Mughisar have mountains somewhere along the line of Alakilise, Akarsu,

been examined. and Refahiye. The tenth and eleventh stations are Iailak The Al Muqaddasi itinerary is no clearer to us than it was Tchemen, Cimen Yayla, and Iassi Tchemen, Yass: Cimen, to Honigmann and needs further elucidation.'°* It is men- the famous Turkish aplekton north of Erzincan—which tioned here because one of the stations is Koloneia, Hadji Khalfa places near Ghelgit, Kelkit.1°° It may be that

Quluniyat al-Aufi, Sebinkarahisar, and a route from there to Iassi Tchemen, which means “‘flat, or wide, meadow,” refers Mous, Mus, must have followed for some way, and certainly to the fertile country around Kelkit itself, or it could be that

crossed, those under discussion. The Ibn Hauqal route given this is the area south and west of Satala where the river by Idrisi runs from Kamacha, Kemah, to Constantinople via Lykos, Kelkit, Balahu, has its headwaters, and place-names Amorion. We assume this route to run south of the Pontos such as Ovacik, “‘little plain,’ and Otluk Kaya, “grassy and do not include a discussion of it. It passes by the still rocks,” are suggestive of good pasture. In either case, the

enigmatic site of Charsianon. twelfth station of Kara Boulour is probably Pulur at the We have already seen that three of the Hadji Khalfa eastern end of the mountains of that name. routes, which run east-west, pass through Sivas. It 1s likely The second Hadji Khalfa route is not given in detail,'°’ that the pattern of roads which he gives is essentially that of | but it is important in that it establishes a route from Erzurum the Byzantine period when troops met at the Bathys Rhyax to Sivas via Sebinkarahisar. This was still used as a military route in the 1870’s, when Captain Burnaby was told that a

brigade of recruits would take a month to march from south a few kilometers before Sebasteia may account for the factthat Erzurum to Sivas.'°® The stations mentioned are: Ech Sebasteia itself is not mentioned in connection with it. Basil I must Kalah, Askale; Terdgian, Tercan; Ghelgis, Kelkit; Plain of have marched it on his personal campaign against the Paulicians, Chir (?); and Karahisar, Sebinkarahisar. The section which probably in 871. From Tephrike he turned south toward Melitene. Ay, concerns us is from Kelkit, which must be regarded as

e Lemerle, ‘“‘L’histoire des Pauliciens d’Asie Mineure d’apreés les , ;

sources grecques,” TM, 5 (1973), 98-103; T. S. Brown, A. A.M, __ the Ottoman equivalent of Satala, Sadak. It seems likely that Bryer, and D. C. Winfield, ‘Cities of Heraclius,’” BMGS, 4 (1978), the intermediate station Plain of Chir is Cheriana, Siran, so 21-22, and Anderson, CR, 10 (1896), 136-40. The route receives _ that this route went across the Camlibel pass and through the confirmation from Burnaby, who traveled from Sivas to Divrigi in plain of Alucra, Kovatha. mid-winter, and thence south to Arapkir, possibly following Basil I’s The third route,!°% from Erzincan to Sivas, is clearly the

route. From Arapkir he turned north to Erzincan via Kemah. ? ee ; Burnaby (1876), I, 319-52; II, 1-58. The railway follows a direct same as the Mustawfi one discussed above, but is given in

route: Sivas, Divrigi, Kemah, Erzincan. less detail. The first station is ““Erzendjik de Khavadgiah io, . Heyd, Commerce, I, 113-19. Manandian, Trade and Cites, | Ahmed.” The second station of Sourzadeh may be Akarsu or 102. The original portulan by Francesco Pizigano of Venice is in

the National Library at Parma; there is a copy in the British Library 104. On Bathys Rhyax, see T. S. Brown, A.A.M. Bryer, and Map Room. Few inland towns are shown in Asia Minor, and the D. C. Winfield, “‘Cities of Heraclius,” BMGS, 4 (1978), 19-22. position and identity of Pizigano’s Agreboce must remain entirely 105. Saint-Martin, Asie Mineure, II, 685. speculative. It is however reasonable to assume that he put it in 106. Saint-Martin, Asie Mineure, II, 652; Brown, Bryer, Winfield, because it was on some important trade route known to Venetian BMGS, 4 (1978), 19; and the Excursus on p. 62.

merchants and that it is likely to be identical with Pegolotti’s 107. Saint-Martin, Asie Mineure, II, 655. Greboco. 108. Burnaby (1876), I, 319. 103. Given in Honigmann, A/PHO, 4 (1936), 263-65. 109. Saint-Martin, Asie Mineure, II, 655.

LAND ROUTES, EAST-WEST 29 Susehri, depending upon which course we favor for this road; Strecker as a part of the normal course of the road from

and the third station is “Ribatk de Khavadgiah Ahmed,” Erzincan to Sebinkarahisar. On the way he mentions

Kochisar. Sadagh; Kelkit, Karadscha, Siran; Ulu Scheiran, Ulu Siran; The fourth route''® is a trunk road from Istanbul to Kirinte Cirmis (?); the pass of Fyndykly Bel, Findikli Bel;

Erzurum, which Hadji Khalfa only gives from Osmancik — andthe Kovata Alydschora, Alucra district.''’ Barth, travel-

eastward. The section which concerns us runs from Hadyi ing westward from Ulu Scheran, Ulu Siran, has Korssyk, Murad, Koyulhisar, to the plain of Achkar. This mentionof —_Kersut; Sycheri Tekkesi, Zihar Koyt (?); and Kara Bok, Achkar makes an identification with Susehri Ovasi or the Karaburk. Between Karabiurk and Sebinkarahisar, we are ~ Cumonts Ashkar Ova highly likely, since the next stationof | unable to recognize the place-names of Sil and Fardere, Gherdgiamis must be Basgercenis, close by Refahiye. The but Alascha, Alisar, and the general lines of his route are following station is Kemakh, Kemah, on the Euphrates. The clear.'!®8 Barth marks a summer version of this road, which road then turned south on the line of the Roman volute to cuts off a considerable distance for the traveler from TrebiMeletine.''! If we assume the identity of the plainof Achkar, —_zond by keeping along the high mountain ridges to the north

Susehri, this fourth Hadji Khalfa route is the same, or runs of Siran and then coming down at Karaburk to join parallel with, the Nikopolis, Susehri, Purk, to Satala route the main route. between the plain of Nikopolis, Akchar, Susehri Ovasi, Ritter has a summary for this section of the route.''? A _ and Gherdgiamis, Basgercenis. We might also assume that consensus of the travelers suggests twenty-four hours, or Dracones-Draconis and Carsat-Carsagis areinthe regionof three days’ travel, for a group over this part of the journey. Basgercenis, where the Roman road north from Melitene Strecker suggests two days for it.

joined the Nikopolis to Satala road. On what might be called the Lykos road, Tavernier’s route The Arab and Pegolotti itineraries almost certainly rep- eastward brought him to the plain of Nikopolis and Ezbider. resent routes which had changed little from those established Leaving the plain, he turned northeastward to Zacape, in the Roman and Byzantine periods. They are likely to be Zagpa. From there his intermediate stations to Garmeru, among the routes used by the Ottoman and Akkoyunlu Germuru, which is his nearest point to Satala, are Dikmebel armies which contended for this region in the fifteenth cen- and Kourdaga. Tavernier is unfortunately not at all clear in tury,''? when the military situation reflected that of the time his itineraries, but he describes his camp as being in a small of Vespasian, which was repeated under Theodosius, Jus- plain under the mountain Dikmebel, which would corres-

tinian, Heraclius, and Theophilos. pond well with Baglar, Mindeval. The nearby village of

The Evidence of Travelers who went over this sectionof | Kourdaga may well be Huseyinaga, and from _ here the upper trunk roads shows that there were two main vari- Tavernier’s track goes up the Lykos, crossing three sizeable

ants, depending upon whether the start was from Koloneia, tributaries, to reach Germirt.'?° Strecker also gives itin-Sebinkarahisar, or Nikopolis, Purk. Among the travelers via eraries for routes from Ulu Scheiran, Ulu Siran, to Purk and

Koloneia were Pitton de Tournefort, who on his return Sebinkarahisar. These run together as far as Zaghapa, journey westward passed by Satala to the post station of Zagpa, via Jenidsche, Yenice; Pajnik, Panik; or Mutta, Kermeri, Germuru, on the Lykos, Kelkit, and then past Mutaisekt; Taschdemir, Tasdemir; and Teschdik, Testik, Sarvoular and by the river Carmili to Chonac, Koloneia.''* Mindeval.'?' From Zagpa the road to Nikopolis crosses the Weare unable to identify Sarvoular, but Carmill is the earlier river and a ridge of hills into the Nikopolis plain, while the name for the river Lykos, down which he was traveling; it road to Sebinkarahisar turns north through the mountains.

was known as the Guermili-tchai by as late a traveler as A third traveler’s route eastward from Purk went to

Tchihatcheff. Erzincan. This is the road traveled by Newbery in reverse; Tournefort, therefore, seems to have followed the modern over the first part he may have been following our preferred road from Kelkit via Ulu Siran and Alucra to Sebinkarahisar, Mustawfi and Hadji Khalfa routes. He went from Arsingam,

as did Fontanier and Consuls Suter and Taylor.''* Morier, Erzincan, via Serperron (?) and Ardansegh (Mustawfi’s traveling westward, mentions Caraja, Siran, and camped out Arzancak) over the mountains of Chardalor, which must be in intermediate stations between Chiftlik, Kelkit, and Cara- either the pass of Cardaklibogazi or the parallel route north hissar, Sebinkarahisar.''> Smith and Dwight, traveling east- of this across the Cimen Dagi. Newbery is normally an ward, camped on the west side of Fundukly-bel, Findikl Bel, uninformative traveler, but the weather of January 14 clearly and then stopped at the station of Sheheran, Ulu Siran, on their way to Germery, Germiirii.''® The route is given by 117. Strecker (1855), 350—59. 118. Barth (1858), 1 ff.

110. Saint-Martin, Asie Mineure, II, 687. 119. Ritter, Erdkunde, XVI, 190-210.

111. Not discussed by us. 120. Tavernier (1681), I, 14-15. Melton (1670), 258, gives the

112. Zeno (1471), 22 ff.; Angiolello (1473), 84 ff., 119, 121; same stations as Tavernier, with almost exactly the same spelling. It Anonymous (1511), 181-82. D. E. Pitcher, An Historical Geography seems clear that this was a standard route at least in the 17th century. of the Ottoman Empire (Leiden, 1972), 79-82, 136-37. Maps xx, Gouz (1647), 68-69, gives a very brief summary of this section of the

xxv, have some information but are in general disappointing. road and mentions a small han Sahabha about three days or thirty

113. Tournefort (1701), H, 317-21. hours before Erzurum. This Sahabha must be in the region of

114. Fontanier (1827), H, 136; Suter (1838), 432-44; Taylor Germiru. It could even be a version of Satala, and if so, would

(1866), 281-361. provide valuable evidence that the trunk road was still passing the 115. Morier (1808), 331-35. ancient site in the 17th century. 116. Smith and Dwight (1830), 51-55. 121. Strecker (1855), 356.

30 CHAPTER TWO worried him and he ventured to complain: “And this day wee Four hours north of here is the station of ‘““Tapan Ahmed pased over the Mountaines of Chardalor with the greatest Agha,” “‘where a feast of ten days was ordained.” Evliya left

snow, frost, and wind, that ever I was in and were in danger the Ottoman army here to travel to “Shin Kara Hissar’ to have remained in the Mountaines all night.”’ From these without giving further details of the route. This is a dismountains he went on to “Shewbaning,”’ where he reports a appointing itinerary and the most that can be said is that it small church standing on a round rock. We are unable to was along a route from Erzincan to Sebinkarahisar. identify this place. It might well be Refahiye, where there is a Tchihatcheff traveled the northern version of the same small tiurbe standing on a rock. Or ‘““Shewbaning”’ may rep- route by a clear itinerary. His intermediate stations are: resent Cobanin. Newbery stopped at the village of Cobanh Mikar Yaila, Mihar; Melikscherif, Melikserif; Aghvanis, further along the road to Andre, Endiryas, Susehri, to Agvanis; Ezbider; and Endiryas, Enderes, Susehri.'?° Nikopolis. He was presumably traveling slowly in caravan Strecker’s route appears to be the same as far as Melikand he appears to have taken five days along this section of serif, but then takes a more northerly direction into the

the road.'?? Tschumen-Su, Cimen Su, Orgil, or the Zevker Dere before Evliya Celebi traveled eastward from Anderes, Susehri, to turning westward to Tschatt, Cat, and on to Agvanis.'?’

Ezbeder; Tilismat Zaaba, Zagpa, and thence by four un- The evidence of the milestones shows that the travelers identified villages to the castle of Shiran, Ulu Siran, on the quoted above were following a Roman road eastward from frontier of ‘““Shuban Kara.” '?> He mentions the pass of | Nikopolis, Pirk, via Ezbider or Agvanis, as far as the region Tekman on the way. It seems likely that he went along the of Melikserif. Thence two parallel routes lead across the river eastward to Fol Mindeval, and then north into the mountains to Eriza, Erzincan. Their regular use 1s attested Alucra valley and over the Findikl: Bel pass to Siran. After both by travelers and by a series of names of ruined hans the castle of Shiran he 1s a little wrong in giving four hours to along the southern road, which is now the motor road, but so

Karajalar, Karaca, Siran; it is at the most an hour’s ride. far we have no evidence for a Roman or Byzantine route. Between Karajalar and Tercan, his exact route is uncertain, Both are slow roads, winding over high mountains, but it but he writes of crossing the great plain of ““Kerkuk’”’ in five seems likely that the southern one, across the Cardakh hours; this is clearly the plain of Kelkit.'*+ He also writes Bogazi pass, was the winter route, while the northern one, that he crossed a difficult path before arriving at it. The place- across high snow pasture land, can only have been practiname ‘“‘Sarichalar’ may be Sarica, although this is only cable for armies in the summer months. about one hour from Siran, and not five as he states. If this is The statement of Yasar Paksoy that the Madendere Koyu correct then it seems likely that the difficult pass was over the ruins lie on the old Malatya to Bayburt caravan route is Cimen Dagi in a southeasterly direction, to come into the useful in suggesting that this site is to be identified with one of valley of the river Balahu, and so descend eastwards to pass the stations between Melitene, Malatya, and Satala, Sadak, near by Satala, Sadak. His other place-names are ‘‘Salut,”’ since the latter lies on the Bayburt road.'?* associated with the pass, the villages of ‘““Gen) Mohammed Geographical Observation shows that there are four Agha” and ‘‘Keremli,” and the small castle of ‘‘Dermeri.” possible ways of traveling eastward to Satala. The northern Evliya Celebi also traveled westward along a route from route runs from Nikopolis northward to Koloneia and then Erzincan, but in the von Hammer version which we use a eastward via Alucra, Kovata, and Cheriana. This route desection of the journey east of Erzincan and Erzurum seems to scends into the Lykos valley from Nikopolis. The river must have been interpolated, making the itinerary more or less be forded before climbing once again over gentle slopes to useless.'*° From “‘Erzenjan,” Evliya records seven hours to Koloneia at about 1500 m. From there, it is another long *““Bashkan” and then five hours to “‘Erzensi,” an Armenian descent to the Ilim Su and a gentle climb over hills to the wide village; thence it is six hours to “Sheikh Sinan,” near which is valley of Kovata which collects the headwaters of the Ilim. ‘“Baragunde.”’ At this point, he says that it is three hours to Kovata is once again at about 1,500 m, and the pass to the the “Bridge of Shepherds” near ‘“‘Hassan Kala’assi’’; these east at Findikli Bel is only about 200 m higher. The high are clearly Coban Koprii and Hasankale, east of Erzurum. ground is stony moorland, with short grass; the slopes are He mentions the important Coban family; the confusion is gentle, but travelers complained frequently at this point, annoying, since it would otherwise be tempting to equate perhaps because there was no han and they had to camp out. Celebi’s ““Coban”’ with Newbery’s “Shewbaning”’ (Cobanin, We have found no ruins on the pass. There should, however, means ‘‘of the Shepherd’). From Baragunde, Baragunide, it have been a Roman and Byzantine station at this point, and is eight hours to Ezendeler in the district of Tercan; here a castle was reported a mile or two south of the modern road, again there is confusion since Tercan is east of Erzincan. but neither of us has explored it. The road from Cheriana to Kelkit runs over a fertile region of gentle slopes, averaging 122. Newbery (1581), 471—72. Chesneau (1548), 71, traveled east

from near Nikopolis (Asebids = Ezbider (?)). He gives the place- 126. Tchihatcheff, Asie Mineure, map. names Girbanambea, Ardingiely, and Agiardacaly. The last may be 127. Strecker (1855), 357-59. Newbery’s Chardalor, Cardaklibogazi (?), but the other names are 128. Paksoy, ‘Tarihi Kaleler’’ (see note 84 above), but we have difficult to identify. It is interesting to note that both 16th-century been unable to trace the source of this statement. Barbaro (1471), ed. travelers went to and from Erzurum via Erzincan, and not by the Stanley, 93, gives the towns between Erzincan to Malatya as Cymis,

direct Lykos route taken by the 17th-century travelers. Casseg, and Arapchir. The Cymis is perhaps Madenkoy (Sinervas ?

123. Evliya (1644), II, 105-6. = Sinoria ?) and also the Kumis of Ibn Battutak. Casseg should

124. See footnotes 106 above and 129 below. fall at Egin, Kemaliye, or at Divrik, and Arapchir, Arapkir, remains

125. Evliya (1644), II, 205. unchanged.

LAND ROUTES, EAST-WEST 31 about 1,500 m, again followed by a rise to about 1750 m at at about halfway along their course by a route from north to

Satala. south which will be discussed more fully below.'*' This

| A central and direct route leads from the eastern end of the north-south route starts from the Black Sea coast at

plain of Nikopolis, Endiryas, Susehri, over a low ridge into Trebizond and runs as directly as the mountain ranges will the Lykos, Kelkit, valley to cross the river at Zagpa, and allow to the Euphrates valley and Melitene, Malatya. In the thence along the river perhaps as far as Kalur, where it would part which concerns us here, it runs from Cheriana in the have been easy to ford the river and continue east-southeast north, in a south-westerly direction to the Lykos valley, and to Satala. This route starts in the plain of Nikopolis at about then from Mindeval southward up the Zevker valley, to join 900 m and, after crossing the low ridge into the Lykos valley, route three at Koymat Kopru and southward again to join continues a gentle ascent to reach about 1,500 m at Asagi the southerly route in the region of Refahiye, Basgercenis. Hayduruk. From here the easier way was to continue up the So much for the evidence; but evidence and stations still do valley to join the northern route at Germutru. The short and not fall into an easy pattern. We suggest that the Antonine direct way, which could only have been used in summer, was route Ia running from Satala to Melitene could have run first to climb to the south out of the Lykos valley to about 2,000 m to Suissa, Germuru (?), to the west of the Lykos. Germtru over the northern flanks of the Cimen Dag1, and thence down was the posting station equivalent for Kelkit Ciftlik; Biliotti the Balahu valley; this may have been Evliya Celebi’s route. was told of Byzantine ruins there.

A more southerly direct road is the Cumonts route via From Germirt the road ran to Arauracos, either AvaEzbider and Agvanis, as far as Karayakup. It then continues rak or Asagi Haydtirtk, site of the martyrdom of St. eastward, instead of southward to Refahiye via Koymat Eustratios.'°? Carsagis would then fall at the Horon, Kopru and Cat Koyu up the rivers Zefker and Orgil, to the Melikserif, or Aksehir ruins, Sinervas at Madendere Koyu, site of the Aksehir ruins near Hacirke and Kurtbaloglu. —_Analiba in the region of Kurugay, and Zimara at Zinegar.'*? Thence a track continues up to the sources of the river Orgil The Peutinger route Ib starts from a junction between and over the Cimen Dag, by a pass which leads down the Nikopolis and Satala and runs south to Melitene.'** We Balahu Dere on the eastern side. This road presents no assume that Draconis, from which it starts (see also routes difficulty to the traveler between Nikopolis and the Aksehir Ila and IIb), is in the region of Camoluk or Mindeval in a ruins, with a gentle rise from 900 to 1,500 m. But between wide and fertile stretch of the Lykos valley. From here there here and Satala the track has to cross the Cimen Dagi by a isa natural route southward up the tributary river Zevker.'°° pass at about 2,250 m. This was almost certainly the route

used by Mehmet II on his expedition in 1461, since he 131. p. 53. _ a

camped in the Yassi Cimen.!2° It is also Hadji Khalfa’s road 132. For Germuru, see Bilotti (1874), 226. The Tarhan map, probably following Kiepert, places Arauracos at Kalur near Asagi

between Osmancik and Erzurum. Neither we nor any trav- Haydiiriik. Chrysanthos, 4 P, 4-5 (1933), map, puts it roughly in the eler that we can find, have used this route, of which Kiepert same place, but it is impossible to use his map as an accurate guide. was ignorant. Adontz makes no attempt at an exact route, | Cumonts, SP, I, 328-29, give a brief summary of its history but do but he suggests that the direct route ‘followed the slopes of NOt attempt to place it exactly, except that it lay at two days’ journey

.; Sas 130 ; or some 40 or 50 mi west of Satala according to the itineraries.

the Mountains now called the Gimen Dagi. The Turkish Honigmann, Ostgrenze, 70, is amused by other scholars’ wrongsurvey marks a track across the pass, and it 1s for this reason, headed attempts to place the site, but omits to place it better. and because it forms much the shortest way eastward to Another tempting site for Arauracos, favored by A.A.M.B., is Satala, that we advance the suggestion that it was a Roman Avarak. much ae yyastzeam to the west where there isa castle.

: : see p. eciow. VC. . even tnou a € villager wno name

and Byzantine route. The height of the Pass, however, makes the site to him called it Ararak, but the situation S30 far west of

it likely that this could only have been in use as a summer Satala as to upset the placing of too many of the other stations route. At a point where the Balahu turns north to flow into between Nikopolis and Satala. the Lykos, the village of Sipanazat lies only about 5 km away 133. Adontz, Armenia, 62, refers to “‘Zimara,” which still stands

to the west across the hills from Satala. on the banks of the Euphrates south of Erzincan between Kemah

A fourth route follows the third eastward to Karayakup and Divrigi. This is presumably the site at Pingan discussed in Yorke, ;1894), 334-35, 465-67, and chosen by Mitford, Thesis, I, and then branches south to Refahiye and eastward to 202-12. The Zinegar ruins are suggested by Yorke for Zimara and Melikserif. From there it either went across the mountains we follow him. northeastward via Ekecik, Kacakk6éy, and Riskan into the 134. The route from north to south is discussed on p. 46. Orcil valley to join route three once again; or it could have 135. We leave open the question of whether Draconis of routes

¢! y tJ 8 , . Ib, and IIb, is, or are, identical with Dracontes of Ila. Miller, 7R,

followed one of the two routes to Erzincan along tortuous Route 95, cols. 671—77, identifies Draconis with Ad Dracones and mountain tracks, rising to 2,000 m and then dropping to the Dracontes, and places it near Shamik, following Kiepert. This must great plain of Erzincan at 1,200 m. The road then turned be either Samukbala or Asagl Semuk, which are situated to the west north once again by a pass at about 2,000 m and thencefor- of the valley of the river Siran, about halfway between its confluence

. with the Lykos, and the town of Siran. Asagi Semuk ts not far from

ward across high but pleasant country to Satala. the site that we report at Cirmis, on p. 176 below. Tarhan also places

These four routes from west to east were all interconnected Ad Dracones at Samik, which he only marks as a modern village without ruins. Miller’s account is confused: in Route 98, cols. 129. See the excursus on p. 61 below. T. B. Mitford, “Some 680-82, Draconis is placed at Melikserif, and in the facsimile of the inscriptions from the Cappadocian Limes,” JRS, 64 (1974), 165-66, tables at Ulusiran. Miller’s other identifications are Haris as Cat agrees in suggesting the road over the Cimen Dagi. He was told that K oyii and Elegarsina as Kurugay. Cumonts, SP, II, 321, suggest the

the Russians used it in 1917. region of the confluence of the rivers Siran and Kelkit for Dracones, 130. Adontz, Armenia, 65. and the district of Tersun for Haza. The former is quite probable, but

32 CHAPTER TWO Haris might then fall at Horon, Melikserif, or Refahiye; and place the site in the region of Camoluk, Testik, Fol, Eregarsina at the Distas-Madendere sites; Bubalia in the Mindeval, on the Lykos; and the identity of the latter pair region of Kurucay and be identical with Analiba of route Ia; and place the site in the region of the Aksehir, Melikserif, and

and Zimara at Zinegar. Horon ruins, or Basgercenis, Refahiye. This allows for two The direct Antonine route IIa from Nikopolis to Satala junction points, one of which may have succeeded the other would run to Olotoedariza in the region of Agvanis'*® and as it declined in importance. thence Dracontes could fall around Horon, Melikserif, or The direct Peutinger road III from Nikopolis down to Kurtbaloglu; and Haza on one side or other of the Cimen Melitene seems likely to have run south, up or along the line Dagi pass. However, it is quite possible that Dracontes is of the river Pulat and over a pass into the valley of the K1zil identical with the Draconis of routes Ib and IIb; in this case Irmak east of Imranli. The station Oleoberda ought then to route Ila would run to Olotoedariza, Agvanis, and could lie somewhere on this route, and Caleorsissa in the area then either cross over the ridge into the Lykos valley and between the headwaters of the Halys, Kizilirmak, and proceed upstream to Dracontes, Camoluk (?), Mindeval (?); Kurucay, where we have placed Analiba, Bubalia.'*? The or it may have kept along the ridges to the south of the river three sites reported by Tarhan along this route are the ruins until reaching Dracontes. The other station, Haza, would of Buldur, the castle at Aksar, and ruins at Yenikoy. We have then fall in the Lykos valley around the twin castles below not traveled this route. It will be mentioned below'*® but Kalur. Or the route crossed south to the Orgil valley, and meanwhile we can only suggest that the Kilinclar or Buldur Haza must fall on the east or west slopes of the Cimen Dag. sites represent Oleoberda, and that Caleorsissa falls at The direct Peutinger route IIb might run from Nikopolis Yenikoy. The main road from Zimara, Zinegar, led southvia Sevindik, where we have the milestone, and from there ward per ripam to Melitene. along a well-used route through Elibtiyik and across the The average time for the journeys between Nikopolis or northeastern boundary ridges of the plain of Susehri to come Koloneia and Satala, or between the first two places and down and cross the Lykos at Zagpa, which could be Cal- Eriza, Erzincan, seems to have been about five days when tiorissa.'*’ The road then followed the river on the north traveling in caravan, which amounts to about 40 hours. bank past the fort at Avarak to Camoluk, the region which Smaller parties did these journeys in three days or about we have suggested for Draconis. We then have a choice of twenty-four hours’ riding, but the time could be reduced to routes either up the Lykos valley or the Cheriana valley to the two days or even less if the traveler were hurrying. north of it, with known sites at Cirmis, Mumya, Cheriana, It will readily be seen that our reconstruction is tentative Tarsos, Kalur, and Haydurtk to offer for the stations of | and hypothetical since it is based on the location of ancient Cunissa, Hassis, and Ziziola.'°8 Or we can assume that sites, the evidence of travelers, and the dictates of geography. Cunissa falls at Melikserif, or Aksehir, Kurtbaloglu, and Certainty must await the recovery of further milestones or trace the road from there on up the river Orcil and over the inscriptions. But in the meanwhile, if we use this reconstrucCimen Dagi to the valleys of the rivers Balahu and Satala. tion as an hypothesis, some points of interest may be noted. The Antonine road IIc would now run on lines already First, the Roman road system seems to leave out the great indicated if we assume that Carsat is the same place as plain and the city of Eriza, Erzincan. This is hard to explain Carsagis, but from Carsagis it would turn north over the since the city was an important holy place with a temple mountains to the Lykos valley at Arauracus, Haydtrrtk or dedicated to the goddess Anaitis, and its fertile valley should

Avarak, and then up the river to Suissa, Germtru, and have constituted an important source of supply for the Satala, Sadak. On the general question of the placing of — garrison at Satala.’*’ The explanation may be that the Draconis, Dracontes, and Carsat, Carsagis, a logical and Euphrates valley from the bend at Pingan eastward was neat solution is to assume the identity of the former pair | Armenian heartland territory, rather than a frontier district, and therefore unsuitable and unsafe for a Roman frontier the latter, is situated much too far north to serve a direct route, road. The city of Kamacha-Ani, Kemah, was the burial place although there is a site at Asag Tersun—on which, see p.174below. Of the Arsacid kings of Armenia, and Gregory the IllumiAdontz, Armenia, 66, also suggests the confluences; he gives the nator spent his later years near Tortan in the hills north of name Hapul-Kopru, which can no longer be found, but it is clear the Euphrates between Kamacha and Eriza, where he dethat both he and the Cumonts suggest the vicinity of Fol-Mindeval. stroyed a pagan temple and where Armenian churches are

136. Cumonts, SP, II, 322, equate Olotoedariza with the 142

Byzantine Lytarariza of Procopius, Buildings, III, 1v, 10. And with reported. Ala Rizena of the Notitia dignitatum, Or. XX XVIII, where a cavalry

wing was stationed. Adontz, Armenia, 65, equates Olotoedariza of 139. Gregoire (1907), 37-38, also favored this route. Miller, /R, the Itinerary with Caltiorissa of the Peutinger Tables, and assumes placed Oleoberda at Kilinglar, on the slopes of a high mountain and that routes Ila (of the itinerary) and IIb (of the Tables) are the same Caleorsissa near Bapsi—perhaps Babsu K6yiu, roughly on the line road. Yorke (1894), 465, 467, also seems to have assumed this. of march in Taylor (1866), 298 ff.

137. The Tarhan map places Olotoedariza at Zagpa. He places 140. P. 46. Calteorissa, which he equates with Caleorsissa, at Yenikdy on the 141. Cumonts, SP, II, 334-39, for Erzincan. The Cumonts trav-

direct route from Nikopolis to Melitene. eled up from Erzincan to Satala, and Lehmann crossed from 138. Miller, 7R, has Cunissa as Telme, Hassis as Aschuz (perhaps Erzincan to Bayburt but gives no details except that he crossed the Asut), and Ziziola as Melisan. This itinerary resembles the route of Sipik6r mountains: Belck and Lehmann (1899), 64. He must have an explorer of the Lykos valley, crossing and recrossing the river crossed the caravan road from Satala to Erzurum not far east of

over difficult country, rather than the route of a road intended to Satala.

travel from A to B. 142. Lynch, (1893-98), I, 294, 348; and Paksoy, “Tarihi Kaleler.”’

LAND ROUTES, EAST-WEST 33 Toumanoff states that, in the Christianization of Armenia in that it is placed so as to control the three northern routes at the turn of the fourth century, the Gregorid family ac- by which invaders might come from the east into Asia Minor; quired the temple-state of Anaitis at Erez in Acilisene, through Paipertes, Bayburt, and the Lykos valley in the Erzincan,'** and Procopius quotes the view that Kelesene, north; or directly through Satala and Refahiye along a Acilisene, was Armenian.'** In the Ottoman period it cer- central line; or through Erzincan and along the Euphrates tainly came back into prominence, and seems to have per- valley to the south. The city itself was able to serve as a formed the military function of Satala, which had by this military headquarters for half a millennium, and the same time declined in importance. Erzincan was a garrison town in legion was stationed there for nearly three hundred years.'*? the Turco-Russian wars of the nineteenth century, and con- The general continuity of military policy and knowledge is tinues to be so today. The fact that the valley lies over a borne out by Procopius’ account of the military expedition

serious earthquake fault and is frequently devastated by which showed the flag in the country of the Tzans. In tremors may be among the reasons for its lack of prominence mentioning two of the places at which the expedition made

in the Roman and Byzantine periods. '*° camp, he remarks that Bourgousnoes (sc., Burgus Novus), Second, it may be noted that the good northern route via was formerly Longini Fossatum, and the second was called Koloneia, Kovata, and Cheriana has ancient sites along it, Germani Fossatum.'*° These little passages surely indicate but cannot be easily equated with the routes given in the a continuity of information. Procopius’ informant, preAntonine Itinerary and Peutinger Tables. Perhaps the reason sumably a serving officer in the sixth century, knew where for this lies in the history of the cities of Nikopolis and his predecessors had made their camps for four centuries. Koloneia; Nikopolis is a Pompeian foundation and a great This is also a valuable indication of the nature of the eastern city of the plains which lost much of its importance in the frontier. There is no hint in the sources that there could have Arab wars.'*° In the reorganization of the themes the more been a static walled frontier, and the geography of the region inaccessible site at Koloneia took the place of Nikopolis and makes a continuous wall impracticable if not impossible.

became the military capital of the region; so the main This was essentially a frontier as detailed in the Notitia Byzantine route east-ward ran through safer country to the dignitatum, with strong garrisons at strategic points, and north of the Roman road. It was this Byzantine route which smaller detachments covering supply routes and outlying survived into the nineteenth century as the highway to the areas. It was a frontier in depth rather than a linear border, East.'*’ This more northerly route also had the advantage of and Procopius’ reference to Fossatum suggests only the difcloser communication with the sea coast, which was always ference that, while the Romans stuck to their rectangular safely in Byzantine hands. In the Turco-Persian wars of the ditched camps as strong points in whatever flattish ground fifteenth and sixteenth centuries it was a much contested they could find, the Byzantines moved upward to make route, with Amasya, Sivas, or Tokat, serving as the muster- castles on the hillstops.

ing for theSatala, OttomanSadak, armies, or likefrom the Byzantine oo, ve BPpoint ythem, From Eriza inones Acilisene, before whenever they marched northeastward. ; . ; cain. Erzincan, to Theodosioupolis, Erzurum

Third, the town of Kelkit, Ciftlik, 1s an Ottoman ad- ,

a ; From Satala, the road continued onthe toTurkish the plain ministrative center and must be regarded as ; ; ;of ; , , Theodosioupolis past the following Peutinger stations: successor Roman Satala and the Byzantine bishopric ofLucus .; ; Basaro, 148 ,toSalmalasso, Darucinte, Aegea or Elegia, Chachaiou. It is clear from travelers’ accounts that the . 151 ; , , a Sinara, Calcidava, and Autisparata. Theodosioupolis posting station for this area was Germuru, west of Kelkit, . , ; ae —

,; ;earlier , itself does not occur in the Peutinger listlooked since it for is a late and the site of Chachaiou must be .; elsewhere foundation: it is a matter of speculation as to which

wo , siting ; Peutinger is the equivalent for it.is ; ; Fourth, of thestation legionary ofcampaign Satala ; ,; ~ the When Romanos IV returnedfortress from his of 1069,

outstanding evidence of the military genius of the Romans,

149. Legio XV Apollinaris up to the composition of the eastern

143. C. Toumanoff, Studies in Christian Caucasian History section of the Notitia dignitatum in ca. 406—8. The best account of

(Georgetown, 1963), 218. the military organization of the frontier is in Mitford, Thesis, which 144. Procopius, Wars, VIII, v. will shortly be published. We do not know for certain of battles in

145. Grumel, Chronologie, 408-81, lists ten earthquakes for this area in the Byzantine period, but that it continued to be an

Erzincan between 1043 and 1457. Skylitzes, Bonn ed., 682, mentions important strategic region is shown by the battles of Yass: Cimen in Keltzene (Erzincan) in connection with Romanos IV’s campaign of 1230, Kose Dagi in 1243, and OtlukBeli in 1473, all of which are 1069. Mustawfi, trans. Le Strange, 175, places Erzincan 24 leagues within a day’s march of Satala. See the excursus, p. 62 below.

from Erzurum, via Asjah and the pass above Khuman Qubuh. 150. Procopius, Buildings, III, tv, 6. 146. Constantine Porphyrogenitus, DA/, I, 212, 238; II, 177, 190, 151. Miller, /R, cols. 671—77. This is the continuation of Route 95 has a Boilas as Katepanos and then an Armenian as strategos, of which must have run across the plain of Theodosioupolis since this is Nikopolis, apparently the strong point of Koloneia rather than the the only feasible way east to the crossing of the Araxes, Aras, at Ad

Epirote theme of Nikopolis. confluentes, Coban K oprii (?). The geography of the region rules out 147. It is perhaps significant of the link between the two towns any other possibilities. From Ad confluentes the road continued on

that in the 19th century the bishopric of Nikopolis was at to Doubios Dvin, Artasara, and Artaxata. This was probably the Sebinkarahisar and not at Pirk. See Bryer, Isaac, and Winfield, AP, route of Heraclius in his campaign of 632 against the Persians: see N.

32 (1972-73), 129 and fig. 28. See Ramsay, Asia Minor, 74-82, on H. Baynes, “The Military Operations of the Emperor Heraclius,”

the changes in the Roman road systems of Asia Minor. United Services Magazine, 47, (1913), 405; and Ja. A. Manandjan,

with a Kadilik. (1950), 133-53.

148. Saint-Martin, Asie Mineure, II, 652, gives Kelkit as a town ‘‘Marsruty persidskih pohodov imperatora Iraklia,” VizVrem, 3

34 CHAPTER TWO he marched from Theodosioupolis to Koloneia and from Route 4 of Hadji Khalfa'*’ does no more than establish there to Melissopetrion and Dokeia, Tosya, Tokat (?).'°? He the existence of a route from Erzincan to Erzurum.

was clearly using the northern trunk road and must have The tenth-century Armenian itinerary gives a shortened crossed from the plain of Theodosioupolis into the upper account, in which we have a distance from Theodosioupolis Lykos valley in the region of Satala and Kelkit. The earliest “to the ditch separating the country of the Armenians from detailed account of this route comes, however, much later. It the country of the Greeks” as one hundred miles.'°® This

is part of Hadji Khalfa’s route 1.'>° tells us nothing except that there was a tenth-century route westward from Theodosioupolis. Routes from Acilisene,

Distance Hours Erzincan, and Trebizond joined the trunk road in this sec-

Iailak Tchemen to Iassi Tchemen............... 4 tion. The Erzincan route 4 of Hadj: Khalfa may partly lassi Tchemen to Kara Boulour ................ 5 equate with the Pegolotti route from Ayas in Cilicia, via Kara Boulour to Signir Sahrassi................ 3 Kaisareia, Kayseri; Sebasteia; Eriza; and Tercan to TheSignir Sahrassi to Djanik...................0.. 5 odosioupolis. This would have struck into the Peutinger Dyanik to Toloslar ...................0..-02.. 5F route either at Derxene or at Askale. The stations from Toloslar to Ak Deghirmen, crossing the Euphrates 4s Eriza, Arzinga, Erzincan, are as follows: Gavezera sulla

Ak Deghirmen to Mama Khatoun.............. 5+ montagna, Ligurti, “ponte a Cantieri,” Gavazera fuori Mama Khatoun to Penek ..................... 4s d’Arzerone, Bagni d’Arzerone, Arzerone.'*?

Penek to Khanes.............. 0.0.0.0 eee 5 The route from Trebizond southeastward across the

Khanes to Ilidjeh ............................ 4 mountains to Bayburt and Erzurum would have cut into the Ilidjeh to Erzurum ............ 0.0... 00 00000. 4 Satala to Erzurum route in the region of Bayburt or Askale. There are no milestones and it is quite possible that they

The plain and mountains of Cimen Dag lie near Kelkit, were never set up beyond Satala in the wild mountain according to Hadji Khalfa.'°* Four hours beyond is Kara country over which the trunk road passed, since the country Boulour, Pulur (?), at the eastern end of the Pulur Daglart, was under Armenian control until ca. 389/90 when a Comes east of Kelkit. From there it is three hours to Signir Sahrassi. Armeniae was appointed to Theodosioupolis.'®° In a Sahra is rather ambiguously translatable as an open plain, graveyard near Camur, southeast of Satala, Taylor found a wilderness or field, and it seems likely that this place must be fragment of a Latin inscription, DELIV.'°! It could perhaps in the region of Lori and Otlukbeli. The next station, Djanik, have been part of a milestone but it might just as well have does not now appear on the map, but Toloslar is Asagi and been a stone brought from Satala. Tournefort reports what Yukari Tulus, and thence the route is clear.'** It must have could also have been milestones at S6kmen.!°? Biliotti!®? run down the river Dorum Dere to cross the Euphrates at found a Byzantine epitaph at Djourouzma, Corozma, northAk Deghirmen. This site is not marked on the modern maps, east of Satala, but he was told that it had been carried there so that the exact place of crossing is not known, but the next from Satala; and some stelae which he thought to be station is Mama Khatoun (Derxene, Tercan), After Tercan Byzantine at Haoudjouz, Haveis. This village lies three and a the station Penek cannot be found, but Khanes is Cinis and half hours north of Satala on a route to Trebizond, along the route must have approximated to the line of the motor which the Cumonts also recorded Roman or Byzantine road north eastward from Derxene across the mountains remains. '°* into the plain of Theodosioupolis, Erzurum. From Cinis the Among sites dating back to antiquity, Paksoy reports

final station is at the baths of Ilidjeh, Ilica. Gelengec, which he identifies with Salmalasso. He claims Hadji Khalfa’s second route follows a more northerly road that these are the ruins of a substantial town with Byzantine from Erzurum to Ech Kalah, Askale, and thence to Terdjian, churches and palaces, and mosques and fiirbes of the Tercan, probably along the Euphrates valley. The next Akkoyunlu period. '°° This site, which is also marked on the station westward is Ghelghis, Kelkit, our nearest equivalent

for Satala, but no intervening stations are given.'°° Erzurum used the northern route through Sebinkarahisar; this is probably the same as route 2 of Hadji Khalfa, representing an earlier

152. Skylitzes, Bonn ed., 701—2. The identification of Melis- Byzantine and even Roman route. sopetrion with Pemolissa at Osmangik has not been previously 157. Saint-Martin, Asie Mineure, II, 687. made, to the best of our knowledge, but it appears to make good 158. Manandian, Trade and Cities, 168-69, quoting MS 2679 of

sense. the Hermitage collection; 98-100, for his conclusions about 153. Saint-Martin, Asie Mineure, II, 685. Peutinger Table stations.

p. 62. 160. Toumanoff, Studies, 152.

154. Saint-Martin, Asie Mineure, I], 652, and for a discussion, 159. Pegolotti, ed. Evans, 28-29.

155. This appears to be the route marked by Strecker on his map 161. Taylor (1866), 286. in Zeitschrift fir allgemeine Erdkunde, XI, (1864), pl. 11. West of 162. Tournefort (1701), I, 317.

Tolos he marks a place Djanur, which is situated roughly at Camur 163. T. B. Mitford, ‘Biliotti’s Excavations at Satala,” AnatSr, 24 of the modern Turkish map. This falls on the direct route eastward (1974), 235-39.

and so Hadji Khalfa’s Djanik could easily represent the modern 164. Cumonts, SP, II, 352-54, with photographs.

Camur Koyu. 165. Paksoy, **Tarihi Kaleler”’ (see above, note 84); Paksdy is not 156. Saint-Martin, Asie Mineure, I], 655. We have suggested specific as to where the Gelengeg ruins are situated. If they are at one above that Kelkit be considered as the Ottoman successor town to or other of the two villages this would be a certain identification of Roman and Byzantine Satala as the administrative center of this the line of the route. The lack of knowledge as to their exact whereregion. Burnaby (1876), I, 319, notes that troops marching from abouts prevents us from establishing the line of a road. The name

LAND ROUTES, EAST-WEST 35 Tarhan map, is in the region of the Lori villages and of Otluk southern edge of the plain.'’? This appears to be the same Beli, where the great battle took place between Mehmed II site as that reported by Strecker at Kottur, Kotur Kopru, and Uzun Hasan in 1473. It forms the modern boundary which he thought to be a fort or a monastery,’ 7° but it could between the vilayets of Erzurum, Gumushane, and Erzincan, also be Koroglu Kalesi, a little further upstream toward which is suggestive of a Byzantine identity for the site since Tercan. Procopius states that Horonon was at the meeting place of Paksoy also reports important ruins at Sirinli, Sihkoy, three roads on the borders of Romania, Persarmenia, and Konarl: about 30 km southeast of Tercan. Here there is a Tzanika; we therefore propose that Horonon was the suc- castle of which he gives a photograph! ’* and ruins of a town cessor to Salmalasso.'®® Its position on the border of dif- with ancient and medieval remains. The same site 1s reported

ferent regions also suggests that it may be the ‘Frontier by Tarhan, who, however, places it west-southwest of Ditch”’ referred to in the Armenian itinerary,'©’ and possibly Tercan.

the elusive Ikrita.’°? Further to the north is the village of The town of Mamahatun, Tercan, which now gives its Siintiri, which is at the north-eastern end of the plain of name to the plain of Derxene is situated a little way off to the Mormusduzu. Kiepert, probably following Blau, suggests east, at a point where one of the routes to Erzurum branches,

this site as Sinoria, and Tarhan gives it as Xenophon’s with a direct eastern route leading up the Tuzla valley and Gymnias. On the eastern side of the watershed between the across the flanks of the Bingol mountains to the northern Lykos and the Akampsis is the village of Varzahan.'°? Here shores of Lake Van at Ahlat. Tercan is therefore well placed stood a group of Armenian churches, perhaps on the site of a as a candidate for a Roman or Byzantine site, and it is no halting place. It may be that this was another meeting point surprise that it was erected into a pocket theme by 951/52, as

between the west-east road and the route south from part of the new tenth-century frontier, more to pacify the

Trebizond. local Paulicians of Mananalis than fend off the Arabs. The Other sites within the area that may help to indicate the naphtha for which the area was noted still congeals there.'’* course of the roads are at Pekeric, on the eastern edge of the Tarhan marks ruins at Askale and Karabryik on the plain plain of Derxene, Tercan, where there 1s a medieval castle of Erzurum but there is no indication of what they are. We

with rock-cut tombs and a tunnel cistern, marks of high know of no reports of Byzantine antiquities, except at antiquity in the Pontos. It may possibly be a Mithridatic Theodosioupolis itself.

castle, and was the site of a pagan temple.' 7° Traveler’s reports show that there were various ways of

Ww , ;

The Justinianic township and ninth-century bishopric of traveling directly between the Lykos valley and Erzurum. Bizana is placed by Honigmann'’! near Vican at the south- There were two indirect routes as well, if we include the ern end of the plain of Derxene where the Euphrates valley possibilities of turning south from Satala to Erzincan and

narrows again into gorges. thence to Erzurum, or northeast to Bayburt and on to

Paksoy reports ruins of a town with Byzantine and Erzurum. The direct route, taken by Smith and Dwight,'’® Armenian remains at Kirogharabeleri, which he places near continued eastward from their overnight stop at Germuru the confluence of the Euphrates and the river Tuzla on the near Ciftlik, to the point near Sokmen where the Lykos turns south. They did not follow the river south to Satala but

——___-—_—— crossed what they call the Cimen Dag: to Lori, and they ee nBes translates roughly as “let those who come pass by,” _ rightly point out that this ridge separates the waters of the

Abraham Hartwell, The History of the Warres between the Turkes __‘K@Mpsis, Coruh, from those of the Lykos, Kelkit, but give and the Persians ... written in Italian by John Thomas Minadoi and it the wrong name. The ridge which they in fact crossed was translated into English by Abraham Hartwell (London, 1595), 262, the Pulur Daglar. They then proceeded across the Otlukbeli

records that Ferad Bassa, the general of “Erzirum,” built fortresses Daglari to their first stage at Karakulak. The second stage at Lory, Lori and Tomanis in perhaps 1584. Chesney (1831) I, brought them, by a route not easy to pinpoint on a map, as

the Gymnias of Xenophon.” 8 PP! 8 P y °

288—89, mentions the “plain of Lori to the town of Gemeri, probably far as an irregular stopping point on the Seytan Dere, four 166. Procopius, Buildings, III, vi. Adontz, Armenia, 51 discusses hours short of Askale which they should have reached. From

Horonon and puts it at Halane Han (apparently just east of the Askale to Erzurum was a nine-hour stage. The Smith and Vavuk pass). This position woul’ make a reasonaie fr onuier puts Dwight route seems to have been the standard northern one Trebizond, while 40 km from Hart on the Trebizond road would _‘f0llOwed by travelers. Evliya Gelebi’s pasha made a diversion bring his site to an area around Kovans which is at a junction. But to the tomb of Gagir Kanli Sultan,’ °’ which may be the tuirbe

unfortunately, as his editor points out on p. 398, no such name as mentioned by Consul Taylor at Camur.'’® Pitton de

Halane can now be found on the maps. Tournefort’s itinerary for this section does not have his usual ee Manacer Trade and Cities, 169, and 219, note 143. clarity, but is valuable for the reference to the sighting of an 169. For Sinoria, see O. Blau, ‘““Aphorismen alter und neuer 172. Paksoy, “Tarihi Kaleler.” Ortskunde Klein-Asiens,” is Petermann, Mittheilungen (Gotha, 173. Strecker (1855), 265. 1865), 252. For Varzahan, see Blau (1860), 375, and Monier (1711), 174. Paksoy, “Tarihi Kaleler.”

374. Monier describes two large churches with mosaic and a 175. On Derxene and this area, see A. Bryer, ‘““Excursus on mausoleum, and remarks that the ruins suggest that the place was Mananalis, Samosata of Armenia, and Paulician geography,” in formerly a town rather than a village. Winfield and Wainwright, Iconoclasm, ed. A. Bryer and J. Herrin (Birmingham, 1977), 83-84.

AnatSt, 12 (1962), 137-38. 176. Smith and Dwight (1830), 53-61. 170. Lynch (1893-98), I, 294, quoting Agathangelus. 177. Evliya (1644), II, 106. 171. Honigmann, Ostgrenze, 53-54, following Markwart. 178. Taylor (1866), 286.

36 CHAPTER TWO aqueduct, which seems to be the first traveler’s mention of | the Euphrates. He forded the river opposite Bagaritson, the Satala ruins. At Sukme, Sokmen, a little further north, he Pekeric, and his route probably coincides with route 1 of records two columns, one of which had much old Greek Hadji Khalfa. It runs via Mamahatoun, Yenikdoy, and writing on it. These ‘‘columns’”’ might perhaps have been Cinis'®* to Erzurum.

milestones.!7° Other routes from the Euphrates ford might follow the

Tavernier passed through Seukmen, Sokmen; Louri, Lori; modern road up the Gogdere and across the Kikiirtlu pass and Chaouqueu, Karakulak (?). He then crossed (but his to Askale'®° and thence to Erzurum, via Asagi Kagdaric,'*° itinerary seems to be in the wrong order) the delightfully Cinis, and Ilica. Or the traveler might continue along the named Aagi Dogii, Aggi Dagi, Pulur Daglari, mountains to Euphrates valley, here called the river Sahun, now following come down to the Giobanderesi, Euphrates (?) and then on the line of the railway as far as Askale, and thence to

to Achekala, Askale.'®° Erzurum.'®’

The itinerary by an anonymous colonel passes from Finally, there were the routes from Erzincan to Erzurum. Bacheiftligh, Kelkit, or Germurti to Karakulak, Tchalok—a Newbery traveled through Bettareg, Backerreg, ‘‘where is

caravansary on the Choghenderesi, Ak Khala and Iligia, great plentie of wine,’ Gebesse, Gotter, Shennar, and Ilica.'®' Morier’s journey (in reverse) took him from Pretton, taking four or five days.'®8 It is not easy to identify Chiftlik, Kelkit, to pass by Satala and thence to Karacolagh, this route but it is possible that Bettareg is Peteric in the

Karakulak, Mamakhatoun, Tercan; Purtun, Pirtin; and foothills at the northeastern end of the Erzincan valley, Ilya, Ilica. He therefore seems to have taken the southern above Altin Tepe. Gebesse, which he mentions as two dif-

alternative after leaving Karakulak.'®? ferent places bearing the same name, may be Gelmize,

A third, and more direct route suggested by a reading of | Gelmizekomu, or Gelmize komu harabeleri, which are on the map would follow the track through Sadak Hanlar1, then the upper reaches of the river Mans. Or it might be Cibice, go up the Karlankas river to cross the Pulur Daglari ridge which gives its name to a river and to a range of mountains. and come down to the Lori or Zelkeler river at Yukarithayik. In the former case Newbery went north-northeast from Altin From there a northerly branch would go through Gelenge¢ Tepe over a pass between the Kesis and the Mirpet, Murit, to join the Smith and Dwight route at Karakulak; southerly mountains. Or, if Cibice is right, he went south of the Mirpet

possibilities would go through Yukari, Gelengec, and Tolos, range. Gotter may be Koti Koprt,, and Shennar may be as described in Hadji Khalfa’s route 1; or across the moun- Cinis. But it is odd that there is no mention of Tercan. If this tains to the villages of Semek or Sosunga Pilk and thence interpretation is right, the caravan route, or one of them, cut down into the valley of the river Pulk, which flows into the off the southern bend of the Euphrates and avoided a long Euphrates at the western point of the great bend around the gorge by turning northeastward out of the plain of Erzincan

plain of Derxene. at the northeastern end near Altin Tepe. Altin Tepe seems to

The route taken by Taylor'®* (who was in fact traveling have been the capital of the plain of Erzincan until it was westward, but we reverse his itinerary) was from Kelkit, deserted in ca. 600 B.c., and there are therefore some arthrough Sadak Hanlari and then up the Karlankas River. chaeological grounds for assuming that a traditional route But instead of proceeding over the watershed to Lori or left the valley at this point.'®? Gelengec, he followed a route turning south at Camurk6éy A route across the mountain is confirmed by the and up a steep pass. High up on this route, either at Camur- Tchihatcheff itinerary and by Strecker.'°° Tchihatcheff only the text) he ° at j the tomb of a Kuzilbas rales of the distri 184. Probably the Trapezuntine bishopric of LaKaBov, TC.wion,

° _ — Tonvynn, which Bees, Byzantion, | (1924), 123, mistakenly places in

and a fragment of a Latin inscription, mentioned above. Canik: see the MS Soumela 27, fol. 1; Honigmann, Ostgrenze, 194: From there he crossed the watershed between the tributaries Chrysanthos, AP, 4—5 (1933), 160; Bryer, BK, 33-34 (1967), 166

of the Lykos and of the Euphrates and continued down the _—snote 160. . .

river Pl, via BasKiy and Gilebag to theriver Mansand 185. Askalemty he regia butno documenta o archaic fit in to the probably geographical scheme.

179. Tournefort (1701), II, 317. 186. Probably Haltoyaric; see Honigmann, Ostgrenze, 54, 226. 180. Tavernier (1681), I, 15. Melton (1670), 258, lists the same 187. Manandian, Trade and Cities, 100, 192, favored the route places as Tavernier once again, and if our interpretation is correct, in along the Euphrates for both the Peutinger and Pegolotti routes. He

the same confused order. used a Russian translation of Strecker in an Addendum to the 181. Anonymous (1826), 225, 230. Izvestiya of the Caucasian Branch of the Russian Geographical 182. Morier (1808), 325—32. He notes that about six hours north Society, III (Tiflis, 1874). For the original, see Strecker, (1855), 267.

of Karacolagh, Karakulak, there is a branch route going north to 188. Newbery (1581), 471. Chesneau (1548), 73, gives Bettaric, Bayburt. Suter (1838), 434—44, also traveled over this route; he gives Newbery’s Bettareg (perhaps Peteri¢) but his stations are no easier

details of the new town of Kelkit, Chiftlik, on p. 435. Porter (1818), than Newbery’s. They are Dibligy, on a barren mountain; then II, 672—84, traveled westward and seems to have kept to the stan- across the Euphrates; Chiobane (perhaps Tercan ?); Pertary; and the

dard route. He mentions Ashkala, Askale; Karakoulak, Karakulak; baths outside Esdron, Ilica. ;

and Lori. From Lori he followed the Tournefort route through Orgi, 189. Ten years of excavation by Professor Tahsin Ozgu¢ have Agg (?), and across the southern shoulders of the Alma-lee Dagler revealed architectural plans of great interest and a rich treasury of (through Elmal: and the Pulur Daglar1 ?) to pass by Saddock, Sadak. objects. 183. Taylor (166), 281—87. Evliya (1644), II, 199, traveled west- 190. Strecker (1855), 263-68. Abul Fida remarks rather injudiwards on the same route and names Ilija, Khinnis, Mamakhatun, ciously of this route (Erzincan to Eruzurum) that it “‘runs entirely

and Ketur. At the last named place there was a bridge: it must be through plains and cultivated fields.” See M. Reinaud and S. Kotur K6pri, which has long been the site of a bridge over the river. Guyard, Géographie d’Aboulféda, 2 vols. (Paris, 1848, 1883), II, 146.

LAND ROUTES, EAST-WEST 37 gives Kargya, but the placing of it on his map suggests that he It seems likely that the Peutinger route from Satala went traveled the same road as Strecker, who gives Karghyn. over the mountains to Gelengeg, or Lori. At the former, there Strecker gives Merteklu, Mertekli, south of Altin Tepe, then are ruins which provide a suitable site for the first station, Chan, which must be one of the two nameless han ruins. One Salmalasso, and also for Procopius’ Horonon, the **Frontier

is on the northern route, between the Kesis and Mirpet Ditch,” and Ikrita.'?* From there we are limited to specuranges, and the other on the road south of the Mirpet range lation; the second station of Darucinte might lie either at running through Cibice. The next station is Karghyn, Karakulak (or beyond) on the northern branch of this route, Kargin, and from there Strecker followed the route to or in the Piilk valley on the southern branch.'?® Aegea would Erzincan described above, through Mamachatun, Tercan:; then fall in the area of Askale, and Lucus Basaro in the region Jenikoi, Yenikoy; and Ilidsche, Ilica. If a northerly route via of Theodosioupolis. Mans and thence along the Euphrates valley to Askale was to The Pegolotti route from Arzinga, Erzincan, to Erzurum be followed, it seems likely that the track left the Erzincan is more likely to have passed Altin Tepe and then run across valley just east of Altin Tepe, to cross the Kesis and Mirpet the mountains, cutting off the stretch of gorges from Altin mountains by a pass leading northeastward over to the tribu- Tepe to Vican or Kargin. It probably also ran through taries of the river Mans. This was a fairly direct route con- Tercan, since the Seljuk hans and tirbes there indicate the

tinuing from Mans to follow the Euphrates, as outlined importance of the town as a market. There is not much above. It may have been the summer route since it requires evidence for placing the stations, but the geography of the the crossing of a high pass and the fording of rivers. If a more route would suggest one of the fan ruins around Altin Tepe southerly route was sought, there were two alternatives: it for Gavezera sulla montagna; Tercan for Ligurti; Cinis

could either have followed the modern road and railway Koprii or Askale for ‘ponte a Cantieri”; Yenihan or along the Euphrates valley to the point where it opens out’ LEvrenihane for Gavazera fuori d’Arzerone; and Ilica for into the plain of Tercan,'?' and then branched east to Tercan Bagni d’Arzerone.'?’

itself and along Taylor’s route to Cinis; or the route could The stretch of road from Satala or from Erzincan to have cut off the lower loop of the Euphrates and the rather Erzurum once again prompts the general observation that difficult river gorges, and passed into the mountains near there were probably summer and winter routes, although the Altin Tepe. The lower mountain route went via Hinzori, height of the mountains east of Satala must have made Sirihlimanastir Harabeleri, and Han, '?? to regain the river, winter crossings arduous, if not impossible for much of the perhaps around Kargin, where it begins toleave its gorgesfor — time.'? the plain of Tercan. The upper mountain route went north- The roads from Erzurum eastward are described briefly

eastward to cross the pass between the Kesis and Mirpet below.'°? They follow the same track as far as “Ad mountains and came down past Gelmize and the river Mans confluentes” *°° beyond which they normally ran outside to the Euphrates. From the confluence, it went along the Byzantine territory.7°!

Euphrates to Askale. —

The time taken between Erzurum and Erzincan or Satala inspection, on entering and on leaving. The system seems to go back

193 .; at least to Ummayad times when caravansaries were built outside

appears to have averaged about four days, which agree large cities such as Raqqa or Kasr el Heir, presumably for foreign with the Pegolotti itinerary. Theodosioupolis does not, of merchants. The Syrian merchant colony outside Kanesh, Kayseri, is course, appear in the Peutinger Tables, but if its site was then an even older example of the method of dealing with foreign trade.

marked by Lucus Basaro, four days were also allowed for the 195. See footnote 166 above for Procopius. Miller, /R, follows

; ; ; ae Kiepert in giving Karatulak, Lorilar, for Salmalasso. See Manandi-

journey. Pegolotti appears to allow SIX days, but this is an, Trade and Cities, 91—100. 169, for a reconstruction of these roads. because of the Turkish custom of making caravans stop at 196. Miller, /R, following Kiepert, gives Darucinte as Pekericg. hans outside the big cities in order to have better control over This castle site is a likely identification though rather too far from

them. !°4 Loni, if that is the first station. The Hadji Khalfa route | must have

followed down the river Dorum running from the west into the

191. This would be roughly at Bizana, Vican. Euphrates just north of the river Mans; it is just possible that this 192. The Turkish map shows two han ruins, two castle names, name preserves a memory of Darucinte. Manandian, Trade and and the Sirihlimanastir ruins on or near the likely course of this Cities, 100, gives its name as derived from that of the province of route; the ruins strengthen the likelihood that it may go back to the Derxene, Tercan.

Byzantine period if not earlier. The Tarhan map marks an ancient 197. Manandian, Trade and Cities, 192-93, interprets the castle and town site at Sirinli. Tarhan’s siting suggests the modern Pegolotti itinerary as following the Euphrates westward after Sirihli of the Turkish maps, but Paksoy reports the Sirinli site to be Askale, rather than cutting across the hills to Mamahatun, Tercan. 30 km southeast of Tercan. This would put it on a route partly Lynch (1893/98), II, 227, mentions, without describing, a summer traveled by Tchihatcheff. From Tercan it went up the river Tuzla to road to Erzincan. It is unclear which road R. Hannay took to

cross the Bingo] mountains and came down to Lake Van. The Erzurum. problem of these sites awaits solution by a future explorer. The 198. Fraser (1835), I, 231-50, made the crossing at the end of Tschichatschof, Ritter, and Kiepert map (1858), marks Sirinli south- January in conditions which few would now tolerate. Burnaby east of Tercan; this suggests that it is rightly placed by Paksoy and (1876), was also a winter traveler in seemingly impossible conditions.

that it lay on the route from Tercan to Lake Van. Others were Newbery, the Jesuit Fathers Monier and Villotte, and 193. Smith and Dwight (1830), did the Sadak to Erzurum route in Chesney, Flandin, and Curzon.

three days. Strecker says that from Erzincan to Erzurum normally 199. See p. 38. took three days but that he had managed it in two. Caravans on both 200. Ad confluentes, Peutinger Tables, xcv, is generally agreed to

routes would certainly have taken four days. be at Cobandede K opru, by the confluence of the river Araxes, Erax, 194. Tavernier (1681), 16, 19, gives evidence of this custom at Aras, with the river Kaplica. Erzurum, where he had to stop at both sides of the city for customs 201. Constantine Porphyrogenitus, DAJ/, I, 212-15.

38 CHAPTER TWO From the Lykos, Kelkit, Valley to Paipertes, Bayburt; ably took the most northerly, which runs from Vicer, to Syspiritis, Pharangion, Ispir; and Eastward into Kaleifisrik, south across the lowest of the passes over the

Tao Ferikdagi, and then south again across the shoulders of the The northern branch of the river Lykos follows an east- mountains to Norsen, and then eastward down to the river west course and to the east of the town of Kelkit goes under Tortum. Deyrolle was probably on the same track, mentionthe name of Kusmasal; the upland plateau of Mormusduzu ing Ischen, Hisen; Zagos, Zakos; Fisrik castle, Kaleifisrik, of rises to a watershed between the headwaters of the Lykos and which he gives the only known sketch; Norschen, Norsen; tributaries of the Akampsis. Tracks run across this plateau to and Tortum. The journey from Tortum to Hisen, which was join the main caravan road from Trebizond to Erzurum at three hours short of Ispir, took him two days.?°° Varzahan. Varzahan ts on the edge of the plain of Paipertes Clavijo, Hamilton, and Deyrolle make no mention of the and Charton, Hart, which presents no obstacle to travel, and monastery of Hahouli, Haho, Baglar Basi but Clavijo and tracks across it may have run in many directions. Paipertes Deyrolle mention the castle at Tortum and probably they all stands at the southern end of the great western bend of the — followed the same route continuing south from Oduk across river Akampsis. To the north of the town the river runs __ the high shoulders of the mountains to the river Norsen. The between low ranges of hills and then comes out at the north- Norsen valley led down to Kisa and the river Tortum. ern end of the bend into the northeast corner of the plain of Hamilton mentions Compor, Kompor; Yenikeui, Yenikoy Paipertes and Charton, Hart Ovasi. It is possible to continue (not on the map); Euduk, Odik; and Khizrah, Kisa Yayla along the river valley eastward to Pharangion, Ispir, but the or Oduk Yayla; and he crossed the Tortum valley by the fort valley track appears to be difficult and Stratil Sauer left it at Kaledibi.2°’ A more southerly and direct route could have between the villages of Kan and Norgah.*°? Hamilton’s crossed the high summer pastures dominated by Mescit Dagi route along this way, taken in a reverse direction, started and and come down into the valley of the river Tortum, past the

ended along the river valley, but in the middle stretches monastery of Ekek or the castle of Tortum. The Greek between Bayburt and Ispir, he traveled in the hills to the _ inscription of Gregory at Ekeki,?°® the site of the city of south of the river Akampsis. His journey between the two Ketzeon which must be somewhere in the upper Tortum towns took eighteen hours, or two days, and he stopped for valley,*°° and the castle of Tortum, which has the ruins of a the night at the village of Kara Agatch, Kara Agag, which is chapel of Byzantine appearance, may each be evidence for

not now marked on the Turkish map, unless it be the identification of a Byzantine station along this route Karakog.?°? Hamilton’s route may well be identical with the eastward from Ispir into Tao; they are tangible evidence of a

main track between the two towns which is marked on the Byzantine annexation of this region.

Turkish ;maps. This Paipertes, may also be the route Deyrolle, in , , ; ; é From Bayburt, to of Theodosioupolis,

reverse, but he is less informative on this part of his travels Erzurum than elsewhere he had been much disturbed by news , ; ; ;the . . , :;because This comparatively short journey 1s athe difficult one since of theaFranco-Prussian war. He mentions stopping night ; ; . ; travelerwhich must cross a massive mountain theand northern at Perghitisi, we have been unable to barrier, identify, ; ,; .

A A Between . slopesNorgah of which shed their waters Neurkak, Norgah. and Kosaba, Ispir, he ,

into the river Akampsis,

Coruh, the southern slopes feed theVannas Euphrates. visited thewhile monastery of Surp-Ovannes, St. John (?) , ; The

; ; 04 ; motorthe road follows the valley of the rivera Akampsis eastward which preserved little finger ofas the He gives ;. . we - from Paipertes farBaptist.°°* as the village of Maden. Thus far the

on the confusion. . i

drawing of Ispir the name ‘‘Kossaba. . 7 . castle valleyunder is a gentle one, of rising to about, 1600 m at Maden. The

Kasaba simply means “‘small town.”’ One may only speculate ; oa: road then turns south up the river Kop Dere, climbing ; ; eastward ; steeply to the pass Kopdagi Gegidi at 2390 m, and deFrom Ispir the Akampsis runs inofgorges for

; scending to theasEuphrates below Askale at about nearly ;all of its course far as the valley great bend above Artvin. . ; 1,600 m. These mountains are barren of forest and the winter Tracks to the east do. 505 not,crossing therefore, follow the river but run has always been difficult. In summer, blizzards can

well south Onetheroute, traversed ;. ,; ier ,of stillit.make passage hazardous,in butpart whenby notClavyo covered in ran through Vicer, Kaleifisrik, and Ersis, and thence across snow the high sloves provide plentiful pasturage and the

the Karadagi to join the river Glaukos, Oltu Cayi, at its S pes P P P 6 y

confluence with the river Tortum. From here, there were tes to Ard tzi Ard ‘tal of T th 206. Deyrolle (1869), XXXI, 412-14. O. Blau, ‘“‘Miscellen zur

routes to “Ardanoutzion, Ardanug, capital ol. 1a0, OF SOUUN- atten Geographie,” Zeitschrift fiir allgemeine Erdkunde, N.S., 12

eastward into upper Tao. es (1862), 296-99. Blau very plausibly suggests that Xenophon and the

A second route ran from Ispir by several variant tracks Ten Thousand came down the Tortum valley and across the moun-

eastward into the upper valley of the Tortum. Clavijo prob- tains to Ispir. He identifies Gymnias as Gimil, Cimil, in the district of Hemsin and gives Mount Theches as Makur Dagh, Makur Dag.

202. Stratil-Sauer (1934), 403-6. . 207. Hamilton (1836), I, 214-19; the itinerary took two days: I], 203. Hamilton (1836), I, 226-31. The Kiepert map marks a jour- 390.

ney by Everett along this route, but we have not traced the account of 208. E. Takaishvili, Arkheologicheskaia (Tiflis, 1952), 9, 77, and

4. Deyrolle XXXI, 414-16. Th account is al aneofofthethchurch: church:/6—/6, 76 Me 109. it13has It deen beenpulled pal ivdown vn 204. Deyrolle ; (1869), ’ —10. e€ monastery 1S also pls.ole ~—15. briefly described by Hakovb Karnetsi, trans. F. Macler, ““Erzeroum, now: see D. C. Hills, ““Turkey’s richness in old churches,” The Times

ou Topographie de la haute Arménie,” JA, XI Ser., 13 (1919), (London, 20 April, 1963).

174-75. 209. For a brief discussion of possible sites for Ketzeon, see notes 205. See p. 55; also note 395. Koch (1843), 41-83. 418, 425.

LAND ROUTES, NORTH-SOUTH 39 can be crossed by the traveler more or less where he wills. A appears to think that the two were of equal value, although less arduous and shorter route 1s to continue eastward up the the Guide is unclear. Its Route 75 goes from Baibut, Bayburt, river Akampsis, here called the Masat Dere beyond Maden, up the river Marsat Dereh, Masat Dere, and then, passing

instead of turning south over the Kop pass. The valley leads over Coph Dagh, Kop Dagi, comes to Mihmansur, up past the hamlets of Masat, Masat Han, and Masat Meymansur.?!’ This suggests a long traverse of the high Mahalle, which give their name to this stretch of the river. It country such as was made by Father Monier. Route 82 may be that one of them represents the important strong mentions Mussat, Masat, and Khooshjah Beenar, place of Mastaton mentioned by Constantine Porphyro- Kosapinar, which suggests the standard route.?'® genitus.*'° It is possible to turn south at Masat and cross the Once again, more exploration is needed, but it seems likely

mountains to Elegia (?) Askale, or continue eastward up the that the old route through Meymansur, Kosapinar and rivers Masat or Kurt, and then south into the plain of Masat is more likely to represent the Byzantine road, and Erzurum over any of a number of passes which are about as that the motor road over the Kop Dagi Gegidi follows the

high as the Kop, but less steep.*"? course of a newer route which appears to have come into The Jesuit Father Monier, starting from Erzurum in 1711, more general use in the nineteenth century,”'? perhaps berecords Chaouf, which we cannot find; then Chimaghil (one cause it calls for one quick crossing of the dangerous high

of the four Cimagil villages); Aviraq, Everek; and ground rather than the longer traverse of highland country Varzouhan, Varzahan. The situation of Cimagil and on the Masat route. Everek suggests that he chose to travel along the high pas-

tures of the Kop range rather than make a straight cros-

sing.2'* He was traveling in October, and presumably the LanD Routes, NoRTH-SOUTH snows must have been late that year. By the nineteenth North-south communications in the Pontos necessarily encentury there seems to have been some standardization, and counter mountains. Modern roads tend to follow the valley the route ran from Bayburt through Maden to Masat. From bottoms as far as possible in their courses through the moun-

there it ran southeastwar d ACTOSS the mountains to tains, but ancient and medieval routes did not necessarily do Kogapinar in the valley of the river Kagdari¢ Su, and across a this, and a mountain crossing was often best made by keep-

low ridge to drop into the p lain of Theodosioupolis, ing along a winding route on gradually ascending ridges. Erzurum Ovasi, at Meymansur. This was the route taken by This avoided the danger of being overlooked by robbers, and Ainsworth and by General Chesney, it ts the only one the necessity for steep ascents and descents from one valley to

marked on Saint-Martin Ss map. However, in the latter another. The present roads inland from Oinaion, Unye, and half of the nineteenth century, the track over the Kop Dagi Kerasous, Giresun, are examples of ridge roads which may Pass seems to have been accepted as a route. Strecker Ss map. follow the more ancient routes. The problems are lesser at the

gives it as secondary to the Masat route,*'° and Murray estern end of the Pontos where the mountain barrier is not

210. Constantine Porphyrogenitus, DA/, I, 212—13. Mastaton is so high, and greater at the a end where there are the coupled with Avnikon, Avnik, which would lead us to look for it east added difficulties of heavy rainfall and thick forest. The of Theodosioupolis, but on the other hand Constantine distin- eastern routes are essentially seasonal, giving a pleasant cros-

guishes it from Avnikion by saying that it belonged to the sing and plenty of pasturage for horse in the summer months, Theodosioupolitans. Adontz does not mention it, and Honigmann, but a bleak and dangerous journey in winter. This is well

Ostgrenze, 80, 164, makes no identification. A site at Masat along ; 8 ‘ J y ; ;

this important route near the Maden mines might also be considered NOugh illustrated by the disasters which struck Alexios III

for Procopius’ Bolon, ‘which lies very near the limits of Theo- when he carried out a winter expedition against Cheriana,

dosioupolis”: Wars, I xv, 32-33. . Ulu Siran??°, or by nineteenth-century accounts of winter 211. The region around the headwaters of the Akampsis, here crossings from Trebizond to Erzurum.22! Up to the 1960’s Masat and Kurt, through which these routes pass, was the district of ; . , Tsourmeri. It was among the lands given by Basil II to David the Sebinkarahisar in the vi/ayer of Giresun was cut off from its Kouropalates in 979 in return for military support against Bardas administrative capital for five months of the year. We list the Skleros: Honigmann, Ostgrenze, 151, 226. One or more of these routes in a west to east sequence, starting from Sinope.*?? routes from Paipertes would have joined the route from Rhizaion to

Theodosioupolis (p. 55 below) on the later stages of the journey. Sinope, Sinop, to the Trunk Road in the Valley of the

212. Monier (1711), Hf, 373-74. Amnias, Gok 213. Ainsworth (1840), II, 394-95. On p. 396 he identifies : :

Tekiyah Tagh. Rosten Dak ), on the Pon tic watershed with No evidence as to the precise route has been found and we Xenophon’s Theches. But, ina later work, Travels in the Track of the Ten Thousand Greeks (London, 1844), 247, he shifted Mt. Theches to

the Kop Dagi, and ina still later work, Sir John Lubbock’s Hundred 217. Murray’s Handbook, 415, Route 75. Books, No. 78, Xenopon (London, 1894), 326—27, Ainsworth trans- 218. Murray’s Handbook, 435-36, Route 82. posed the names and gave Mt. Theches as the Tekiyah Tagh between 219. Deyrolle (1869), XXIX, 270, writes of two roads, one over

Bayburt and Erzurum. the Kop Da§g1, and a shorter one over Khochapounhar, Kosapinar, 214. Chesney (1831), 122. It was also taken by Gamba (1822), 1, as the summer route.

416-19, who gives Massata, Masat; Kiochapoun Gar, Kosapinar; 220. Panaretos, ed. Lampsides, 77. and Hermanu-Kei, Eregmansur (?). He rightly observes the water- 221. For winter travel see note 198 above and p. 37. shed between the rivers draining to the Persian Gulf and those to the 222. For routes to the west of Sinope which run south from the

Black Sea. coast, see R. Leonhard, Paphlagonien (Berlin, 1915), pp. 82-88 for 215. Saint-Martin, Asie Mineure, map. Ionopolis, Inebolu, to Pompeiopolis, Taskopru, and pp. 92-96 for

216. Strecker (1855), pl. m1, map. Ionopolis to Anadynata Boyali (?).

40 CHAPTER TWO have not searched carefully for it. It may have followed the route for this retreat would have been down the Halys valley,

line of the modern road south from Sinope, but a road in which two castles are reported.*?? Evliya Celebi remarks mender on the pass knew of no ancient ruins along it. This of “‘Kopri,”” Andrapa, Vezirk6pru, that “the harbours of road crosses some thirty-five kilometers of flattish fertile land this town on the shores of the Black Sea are Bafra and south of Sinope and then gradually mounts along the slopes Sinope, which are but a journey distant.” This is fair evidence of a steep rift to a height of 1200 m. The high country makes for the Halys valley route to Paurae but it is unclear as to how

comfortable traveling over rolling ridges and the road de- merchandise would have been taken to Sinope or as to scends gradually through a conifer forest and finally more whether the routes to which he refers are by land or sea.*°° steeply into the valley of the Amnias, just to the west of The large Kiepert map marks land journeys by several Boyabat. However, the Kiepert map marks a route by German travelers through this region, but we have been Maercker, which branches off the modern road in a south- unable to trace their accounts.??!

westerly directi _M , , , ny rection at the atin acre ker pursued a From Amisos, Samsun, South to the Junction with

more direct course for Pompeiopolis, Taskopri,, and came

the East-West Trunk Road into the Amnias valley just toisthe east of the town. We have _ .coast ; or , oe This the easiest route from the Pontic been unable to trace the publication of this route but it might er ge ; to the , eee interior of Asia Minor and it must have been an important correspond to the Peutinger route. This is givenfrom as .;at; least as long ago as , og: ; factor in the growth of Amisos, Gangra—Pompeiopolis—Sinope, which gives but thefor vewhich , er ; ; . ; — of . the foundationBut of athe Hittite capitalus at little Bogazkoy, notification its existence. group of stations coming ; , 993 it was the natural port. South of Amisos, the road crosses the

after Sinope must have Dag belonged to it. ; at ; , ; three gentle passes of Mahmur at 840bem.onHacilardagi Wilson tentatively identifies names which could the - ,

; = 820to m Sinope. and the Karadagi at 900 m. The township of1 Kavak, route from Gangra There road, are Vicui, Kastamonu; ; on the ; ‘ :; which is on the modern was almost certainly Tyae, ;in; ancient the region of Burnth; Cereas, Findicak; and Milete, . oe road too. follows Both ancient andofByzantine antiquities: Kabali. This arrangement the line the modern ,; ; . ,; 324 . ws have beendescribes reported from it, butand it remains without an ancient road.*** Hamilton the route remarks that it is , ; . name.of Grégoire suggested a Roman might not; ener much; ;used because thetodifficulty of that crossing theroad passes; ; ;have branched off here go directly to Andrapa, but pointed but in reality it is not too difficult and the prosperity and 539 out trade of Sinope in the ancient world must have enhanced that it was not an easy route. In the region of the 4 225 Pe Phazemonites (Merzifon, Havza, and Ladik), the road must have divided, with one branch continuing south to Thermae From Paurae, Bafra, up the River Halys, Kizilirmak, Phazemonites, Havza, and across the plain of Chiliokomon,

as Far as the Region of Celtik, to Join the Trunk Suluova, while the other turned eastward to pass through

Road at the River Crossing West of Andrapa, Laodikeia, Ladik.?>°

Vezirkopru , The Route South up the Iris, Yesil, Valley to Join the There is no documentation for the use of this route in ; , , Trunk Roadstation at Eupatoria-Magnopolis Roman, times. The Roman of Helega probably lay on ; ; (Confluence of Iris and Lykos, Kelkit) or near mouth of the Halys, toPtolemy, be succeeded by Paurae. aa:in ooriver 1 the 226 This road isaslisted by and Munro must be; oa. right The is navigable far as Celtik,**° where the first ford ; . a , 327 , , suggesting that siteofof Eupatoria-Magnopolis in part is reported;**’ and there arethe ruins a bridge about one hour 334theisfact ; aor . . ; dictated by the need to guard this road.*°* But that higher up. The description of the bridge leaves it uncertain as Maenopolis seems to disappear in Bvzantine times ma to whether it was Roman or not??® but it is safe to assume enoP PP y y that the trunk road crossed the Halys somewhere at or

between he Celtik ford and the ruined bridge. The evid at aaa the Ge tik ford and the ruine ri ge. Ine evidence 230. Evliya (1644), II, 218.

for a Byzantine route down the Halys, either by land or by 231. Maerckher traveled from Karousa, Gerze, inland to the river, is provided by the retreat of the Crusaders from around northern loop of the Halys just west of Andrapa. He also went from

Merzifon to Paurae after their defeat in 1101. The natural the mouth of the Halys south to Boyabat, traveling along the west

bank through Celtik. This may represent the direct land route up the 223. Robinson, AJA, 9 (1905), 327—29; milestones nos. 75, 76, 77 valley. Von Prittwitz and Gaffron, and von Frottwell traveled from

may be relevant. Miller, /R, fig. 210, col. 642; Routes 93, 94a, 94b, Zalekon, Alacam, to Celtik on the Halys. Kannenberg and von

cols. 670—71. But Miller’s account is unsatisfactory. Prittwitz traveled up the Halys about as far south as Asar and then 224. Wilson, Thesis, 355-58. He points out that Boyabat is not turned eastward to come down to the coast region at Eusene Dagale,

known to have been an ancient site. Nor have we seen any sign of Karakoy Irmak. |

pre-Turkish settlement there: see B. Basoglu, Boyabat ve ¢evresi 232. Grégoire (1907), 7-11. Tchihatcheff had taken the same

Tarihi (Ankara, 1972). route as Grégoire, and Kinneir (1813), 307—8, seems to have traveled

225. Hamilton (1836), I, 317-20. Leaf, JH'S, 37 (1916), 1-16, this way. argues forcefully, though from the armchair, that Sinope could not 233. For descriptions of the roads and milestones, see Munro, have had any significant trade with the interior because of the JHS, 20 (1901), 53-55; 1899. Anderson, SP, I, 48-50; Cumonts, SP, difficulty of crossing the mountains. This ignores the fact that cities II, 121-23; M. E. Fountaine, ‘“‘A Butterfly Summer in Asia Minor,”

further east traded across much more formidable mountain barriers, The Entomologist, 37 (London, 1904), 79-84, 105-8, 135-37, but he is probably right in his thesis that Sinopitan wealth was based 157-59, 184-86; and Wilson, Thesis, 369-75. Miller, ZR, col. 671,

226. See p. 90. number.

primarily on the function of the city as an emporium. See p. 69. mentions the road and milestones but does not give it a route

227. Hamilton (1836), I, 327. 234. Ptolemy, Geography, ed. Miller, V, v1, 3; Munro, JHS, 20 228. Cumonts, SP, I, 84-85. (1901), 54-55; summary in Ritter, Erdkunde, XVIII, 232-36.

LAND ROUTES, NORTH-SOUTH 4] suggest that the route was then little used. However, some evidence for the existence of the route in Byzantine Anderson noted a stone with across on it from one period of __times.**?

reconstruction of the bridge over the river at this point,?*> Geographically the route is not one of great difficulty. and Hamilton noted the medieval castle at Boghaz Hissan There is a flattish coastal strip running a few kilometers Kaley, Kalek6y,?*° a little further down the valley. The inland from Oinaion, followed by a river valley, guarded by bridge may be connected with the trunk route east to west, Caleoglu castle, which leads up gently sloping ridges and but the position of the castle suggests that it may have then over rolling hilltops to a maximum height of 1,500 m guarded the route north along the Iris valley. The fact that beyond Karakkus, now Akkus. If the old route passed by the coastal Limnia assumed the metropolitical rights of Amaseia Kizilelma ruins marked by Tarhan, it followed the valley of in the fourteenth century perhaps indicates a connection the river Ceviz as far as the village of Karakkus, from which between the two towns.?°’ And if we are correct in identify- it would have been a short and steep ascent to cross the ing Kinte as Limnia, John IIT Komnenos probably took this heights to the south. South of Karakkus the road drops route on his way to Neokaisareia, Niksar, in 1140,7°8 follow- down about 700 m to cross the river Bag. The castles of ing in the wake of Lucullus, who had marched this way into Kevgiirk and Kainochorion, Mahalek or Kekir (?).74* and

the Phaneroia more than a thousand years earlier. other ruins at Ahret are marked by Tarhan along this river The only traveler’s account unfortunately lacks any detail, valley, which descends in a westerly direction to join the Iris, but the fact that this route was used as a means of traveling | and so must be considered as a possible alternative route from Amisos to Amaseia suggests that it was without any from Neokaisareia to the sea in the region of Limnia; it difficult obstacles. Colonel Rottiers stated. ““We went down would of course be longer than the direct route to Oinaion. first to a point near the ruins of Ancona, on the river Ekil The last stage in the journey south is to ascend the second [Yesil]. We followed this to the point where the Tokat Irmak, range along the slopes of Tinik Tepe to a height of about which formerly borrowed from the Ekil its antique name Iris, 1200 m and cross a wooded plateau to drop down into the runs into the last-named river. We then followed the valley in Lykos valley at Neokaisareia. The road 1s typical of those which this river runs as far as the town of this name.” **° reaching south in that it must cross two ranges of mountains,

The Route South from Themiskyra (Region of Terme) wi a urther deep valley in between. it is easier going than those east, and the mainBut characteristic of it would have

up the Thermodon, Terme Suyu b nable ride th h thick forests. Th ti

We have no evidence for this route, but the existence of cen an interminable ride Mrougn Uick forests. Nie vestiges

, Themiskyra ,; of the forest still remain on the high ground, with classical makes it probable that itashad access to ; broad. and - a , .; leafed trees in the mountains as far the river Bag, the Phaneroia. Ritter deals with the area?*° and the Kiepert . . ; conifer forest on the Tinik range. In these upland areas the map marks a journey by Hirschfeld. He traveled from Carh . ne ouses are single-roomed log cabins withee pitched roofs made

samba the valley of the Thermodon and thence to with ;;; . , ; across of largeinto wooden shingles. They are built on steep slopes Neokaisareia, Niksar, on a route which may have taken him ; . stilts supportingMahalek, the floor where ground drops past; ;Kainochorion, nowtheKekirkale (?).away; ; , the animals winter in the space underneath. One section of one

The Route South from Oinaion, Unye, to the Trunk wall only is built of masonry with a half timber frame to

Road at Neokaisareia, Niksar accommodate the fireplace: fodder for men and beasts 1s There appears to be no textual or archaeological evidence stored in separate little barns on stilts, built close by the for the existence of this route in the classical period, but the house. This type of farmstead dwelling must look much the Hellenistic sites of Caleoglu castle near Oinaion, of Kaino- same as its predecessor in the Byzantine period. chorion castle, and of Kabeira, Niksar, which was a The travels of the Patriarch Makarios of Antioch, of 1658, Mithridatic capital, suggests that a route connected these give evidence that trade from Tokat to the Black Sea was places.24! The Tarhan map marks important ruins at conducted along this route. He states that “from this town of Kizilelma, east of the path of the modern road, which may Sinope to Tocat, it is a distance of fifteen stations. Those, indicate a different course for it. Gregory of Nyssa offers therefore, who wish to pass Tocat with heavy loads go by sea to a town called Oenos, in Turkish Onia, distant from it two 235. Anderson, SP, I, 77.

236. Hamilton (1836), I, 342. The castle is marked on the Tarhan 242. Cumonts, SP, II, 260 note 2., quoting Gregory of Nyssa’s

map. Life of St. Gregory the Wonderworker 1n PG, 46, col. 897: “‘The 237. See p. 98. countryside teems with fruits, the town is a maker of the Faithful, 238. See p. 99. and the neighboring sea brings in from all parts its gifts and its 239. Rottiers (1820), 252. ‘‘The town of this name”’ presumably power.” refers to Tokat. 243. Jerphanion, Mé/USVJ, 5 (1912), 138, identifies Mahalek with

240. Ritter, Erdkunde, XVIII, 95-104. D. M. Girard, “Un coin de the Mithridatic fortress of Kainochorion in Strabo, Geography, X11, l’Asie Mineure, le Djanik,” Le Muséon, N.S., 8 (1907), 152, may be 1, 31. Jerphanion is mistaken in correcting Kiepert and making the speaking of this route in connection with the transport of flour Devchur Irmaq flow into the Thermodon, Terme. Kiepert was quite

from Tokat to Carsamba via Erbaa in the Kelkit valley below correct in making this river flow into the Iris below the Lykos

Neokaisareia, Niksar. confluence. For a different identification of Kainochorion, see Van

241. See p. 101. It is interesting to note that although the place Lennep (1864), I], 61-77, with a description, plan, and drawings of a

receives no mention in Strabo, it is certainly among the coastal fort on the summit of the Yildiz Dagi north of Sivas. However, Peutinger stations. The town has excellent natural shelter for an Brown, Bryer, and Winfield, BMGS, 4 (1978), 19, identify this with anchorage, but perhaps it slipped out of importance in the Roman the region of Herakleioupolis, and we prefer Jerphanion’s identifi-

period as a result of the rise of Polemonion. cation of Kainochorion on p. 102.

42 CHAPTER TWO hundred miles, and a skala or port to the city of Cafa.” Taularon on the grounds that it had no ancient name.?°! In Makarios did not take the direct route from Oinaion to general geographical terms this makes excellent sense, but Neokaisareia for fear of brigands, and his track over the Sebasteia lies in a great plain, whereas Mithridates’ strongmountains appears to have taken him four days, although in holds were as a rule on mountain tops. Grégoire rejected the an earlier passage he noted that it is only four days journey identification of Taularon with “Taourla.” The situation is

between Oinaion and Tokat.?** further complicated by the Turkish maps which show no

The modern route is about 100 km long and takes four exact equivalent of the ““Taourla’” mentioned by Munro. hours by motor car. The journey took two days’ riding in There is a village above the bridge on the north slopes of the earlier times, according to Evliya Celebi and to an informant Lykos called Tavara, but Cumont clearly assumes that

Niksar 4 252 Thi

of Cumont.?*° Munro meant to refer to a village on the south slopes, and from the mapFatsa, it seems clear that the route The Route from, i,Polemonion, to Neokaisareia, P which the

, Cumonts took on leaving Resadiye must have passed

; through Tavara. However they do not reportis on it.*°? The existence of aseer Roman theof two towns MsThis ; ; Tavararoad must between be the village ““Tavoura” which. the in-

attested by the Peutinger Tables, which give a total distance . oo ; .

; , and , 346 NAL Grégoirestation, decided toBartae. investigate; he found253 nothof 49; m.p., one defatigable intermediate .; ; _ ing there.*°° The Tarhan map marks ruins at Miller Mezre, which follows Kiepert in suggesting Sarkis for Bartae. If this is the ; .

he identifies as Taularon. There are five places called Mezra modern Serkes, theallroute would have the modern ; . and ; eee ; on the map, of which are in thefollowed general area of his site, ridge track as ar far; ;as the Egrikirik mountains and thence in 7a . ; on one of which is near Tavara. The village above the bridge

the same southwesterly direction to Niksar. Anthere alternative ; . ; the south slopes is named Kundur, but are two possible

route would follow the modern dirt road up the river — names on the southern slopes of the Kabasakal mountains in Boloman to the watershed beyond Aybasti and thence east- -

.the ; the valley of the Iris, Yesil.and These and Tozanl.. ward ridges the Cumonts othersare inthan :Dogla ae im ; ,along ne 247 Doglatraveled soundsbymore like ‘“Taourla” Tozanli, but the the nineteenth century, between Niksar and Koyulhisar. 354

, latter is awarded ruins Tarhan. At Aybasti the Boloman valley opens out by into a broad a: ; ,; ; If the reported movements of Mithridates frombowl Kabeirawith well cultivated land. This is a suitable place for the ar ; ; ,; ; Neokaisareia tosince Komana Pontika and thence toalong Taularon intermediate station it would be about halfway ;, . qe ; and the Euphrates are correct, then the likely site for .this on a two-day journey. The multiplicity of tracks and villages — ; ; ; Turkish ; townmap is in the upper Iriswellvalley, perhaps at the Tozanlimarked on the suggest that Taularon this was .on the .inhabited Findicak ruins. To place north the country, easy to cross by numbers of routes. Such abank na . offrom . or Lykos is to make Mithridates double back on his tracks relative ease of Pontika communication southward over the moun, _— ; an ; , - ;may Komana and takein athe verysiting indirect route, involving tains have been a factor of the city of ae , oe unnecessary of a major river and mountains. Polemonion, since itcrossing 1s without a natural harbor. ;.;_, Geographical considerations suggest that the original di-

The Route from Polemonion, Fatsa, to Sebasteia, rection of this route could have lain due south of

Sivas Polemonion, through Aybasti, and thence past the castles of

The evidence for such a route in the ancient world depends =Megdun’°° and Isker Su?°° and down into the fertile partly upon the whereabouts of the Mithridatic stronghold Resadiye area in the Lykos valley, where the tributary rivers of Taularon and whether or not it lay upon it. After his defeat Delice and Cermik flow in from the north and break up the by Lucullus, Mithridates fled through Komana Pontika, steep slopes. Here the route forms a junction with the road up

Gumenek, and Taularon on his way to the Euphrates. the Lykos valley and probably crossed it at Kundur Reinach’*® placed Taularon at the suggestively named Koprii?°’ about 15 km to the east. A junction of routes at ‘“Taourla”’ noted, but not commented upon, by Munro.?*? this point would explain the lively fair which the Cumonts?°® Cumont tentatively accepted this identification.?°° At a later saw in progress with merchants from Merzifon and Sivas. date Munro himself plausibly suggested Sebasteia for The Merzifon merchants perhaps still used a Roman trunk road; the Sivas merchants would use our suggested route 244. Makarios (1658), 429-30, 437-38. On p. 440, the old Greek 251. Munro (1899), 58-59. name for Tokat is given as K@y@ow. Peyssonnel, Traité, II, 91-92, 252. Cumonts, SP, II, 284-86. The reader may be reminded of also stresses the importance of Ounia, Unye, as a port for Tokat and the difficulty in identifying place-names by trying to trace on the

its merchandise. Turkish maps the itineraries of even so recent and careful travelers as

245. Evliya (1644), II, 104; Cumonts, SP, II, 260. the Cumonts. Bearing this example in mind, it is no wonder that

246. Miller, ZR, cols. 667-69, Route 93. Miller and Kiepert are Roman and Byzantine place-names present problems. both following Tchihatcheff, who went through Serkes. The stations 253. Grégoire (1907), 33.

are: Polemonion; 11 m.p., Bartae; 38 m.p., Neokaisareia. 254. Tarhan marks other ruins which might be relevant to this

247. See p. 23. problem at Mezre. This common place-name simply indicates

248. T. Reinach, Recueil général des monnaies grecques d’Asie “fields.” Mineure: 1,1, Pont et Paphlagonia (Paris, 1904), 106. Reinach rather 255. Cumonts, SP, H, 280-82, with photograph; Hogarth and

injudiciously refers to Taourla as a ‘“‘gros bourg.” Munro (1891), 731. 249. J. Munro, in Hogarth and Munro (1891), 730. 256. Ouseley (1812), III, 482-83. 250. Cumonts, SP, II, 284, with the reservation that someone had 257. See p. 99.

better look at Taourla. 258. Cumonts, SP, II, 282-83.

LAND ROUTES, NORTH-SOUTH 43 southwards. There was apparently no township in the been found along this road, which has not, to our knowledge, Cumonts’ time, but the earlier use of this road junction is been explored. The sites which may be relevant to it are: suggested by the nearby castle of Megdun and the reportsof | Evkaf Koy Kale,*°* Golkdéy Kale,?®° Koyulhisar?°® and tombs and pottery.?°? The development from a fair on open Sebinkarahisar.*°’ There are also several village names sugor common ground to the present township of Resaditye is a gestive of castles, and a firmly reported castle at Sisorta.7°® late example of the kind of development from fair to town The likely course of the road from Polemonion would

which is typical of medieval Europe. seem to be southward up the valley of the river Bolaman. At From Kundur Koprt the route then crossed the Kaba- some point unknown, but possibly at Catak or farther south sakal mountains to the south by a track passing the villageof | at the confluence of the Golkoy stream with the river Kundur and descended into the valley of the Iris, here called Bolaman, the route must leave this valley to cross the mounthe Tozanh, and the Tekelici. A modern track, which appears tains and descend into the valley of the river Melanthios, to be a very direct route, leads up the valley past Hypsele (7), Melet. It was necessary to cross this great rift and then climb

Ipsele, and across the Kesis mountains to come down into over the second range of mountains and descend once again the Halys valley at Comassa, Camisa, Hafik, whence it is an into the Lykos valley. From there an easy route climbs gently easy ride down the valley to Sebasteia.*°° A second and less southward into the plain of Nikopolis.

direct alternative would be to continue up the valley of the The previous guesses for the stations along this road have Iris to the village of Serefiye, which 1s at the junction of the been limited. Kiepert, followed by Miller, proposed Melet road from Sebasteia to Nikopolis. At Serefiye there were Hamidiye, now Mesudiye, for Sauronisena; a point of conByzantine antiquities, and Grégoire tentatively proposed an fluence along the upper reaches of the Melanthios for

identification with the Dagalassos of the Antonine Matuasco; and Koyulhisar for Anniaca.*©? Grégoire sug-

Itineraries.?°! gests Madasoun, Mudsun, for Matuasco.?7°

We know that a road from Polemonion to Sebasteia ex- The Kiepert-Miller location of Sauronisena is too far

isted in the Trapezuntine period, since we have the record of south for a first station, and we prefer to place it at Golkoy the fast four-day journey of the Genoese notary Federico di Kale.?’' The reasons are that it is situated on the direct line Piazzalungo along this route; unfortunately we have no de- for our road, it has an ancient site (albeit of uncertain date), tails of it.2°* The only other traveler who appears to have and the valley slopes open up gently at this point to form an traveled directly between the two places was Tchihatcheff, obvious place for habitation and cultivation 1n an otherwise

who went via Niksar. His is therefore another possible an- precipitous region. cient route, but more travel and exploration of this crossing From Golkoy we propose that the old route followed the is still required in order to-verify it. What does seem clear is course of the modern road as far as Mesudiye, which we that the Resadiye region must yield further evidence of settle- suggest for Matuasco, and that it continued along the same ment in the Byzantine period. Its geographical situation road over the Igdir mountains to Anniaca in.the Lykos halfway between Neokaisareia and Anniaca, Koyulhisar, valley. Between Sauronisena, Golkoy, and Anniaca, this and between the coast of the Black Sea and Sebasteia suggest road would precede that built by the Turkish army from a stopping place; the area 1s a fertile one capable of support- Ordu to Sivas in the mid-nineteenth century. This 1s undering some population; the castles of Megdun and Isker Su, standable since there is no reason to suppose that the geoand the possibility that Taularon lies in this area—all are graphical or strategic considerations governing the building

indicative that there may be more to be found. of a Roman road in the first, or a Turkish road in the The Route from Polemonion, Fatsa, to Nikopolis, nineteenth, century, through this region would have

Park; changed. From Anniaca, Koyulhisar, the ancient and , ; ; , modern roads diverge since their destinations were different;

the intermediate stations of this route: he Roman road probably followed up the Lvkos vallev to 33Peutinger m.p. gives to Sauronisena; 16 m.p. to Matuasco; 11 m.p. to te ., " P y P ey

Anniaca: and 18 m.p. to Nikopolis.2°? No milestones have cross it at the bridge carrying the road between Koloneia and

Nikopolis. However, the Chrysanthos map, which marks this

259. Cumonts, SP, II, 281-82. He suggests that the present ruins

are not older than the Middle Ages and quotes the normal locally 264. The authors of the present study have, however, made dif-

held view that they are those of a Genoese castle. ferent crossings of it. See p. 112.

260. For Hypsele, Ipsile, see note 63; citing Tomaschek, Kiepert 265. See p. 116. Festschrift, 148—49. Tomaschek notes this as a possible alternate 266. See p. 118.

route from Sebasteia northward and suggests that Charsianon 267. See p. 145. should be on it somewhere between Koghisar and Ipsile. Adontz 268. The name is a pleasing one: “The castle in the middle of the puts Charsianon east of Sebasteia on the main highway to Nikopolis mists.”’ These may be the ruins reported by Kiepert, ZGEB, 25(4) at Horsana, but this was rejected by Honigmann who puts the place (1890), 322. west of Sebasteia: Adontz, Armenia, 68; Honigmann, Ostgrenze, 269. Miller, 7R, col. 679. The identification of Koyulhisar with 49—50, returns Charsianon to the site at Mushalem Kale, which Anniaca seems to have originated with Bore. Ramsay, Asia Minor, 249-65, rather confusingly suggests as both 270. Grégoire (1907), 32. Grégoire’s suggestion would take the

Charsianon and Hypsele. road into the Lykos valley at Modasoun and from thence up the

261. Gregoire (1907), 39. valley to Anniaca; alternatively it could have crossed the mountains

262. See p. 112. to Dagalassos, Serefiye (?), and Nikopolis.

263. Miller, 7R, cols. 643, 675, 697, Route 97. 271. See p. 116.

44 CHAPTER TWO whole route with great prominence,?’? takes it across the Kirkharman where the modern road crosses the mountains Lykos at Anniaca to follow the modern road up the southern to Sauronisena. slopes of the Lykos valley to a point where it joins the From Matuasco there is a long climb up to the pasture modern road from Sivas near Grégoire’s Dagalassos in the land and conifers of the Igdir ridges at a height of about

region of Serefiye.?’* 1,800 m before dropping down to Anniaca and the valley of The road that we propose runs southward through lime- the Lykos, at about 600 m. The site of Anniaca might be the stone country for some miles before hitting the harder rock great ruined castle of Yukarikale at the top or eastern end of massif of the mountain spine. It is unlikely to have kept to the this open stretch of river, on a precipitate slope about 200 m

valley bottom of the river Bolaman, which runs in deep above the river and covering the bend in it. Or it might be gorges for much of its length, while the road probably ran up Asagikale at the bottom or western end, on an easily dethe eastern ridges overlooking the Bolaman. There were said fensible rock overlooking the river, where Munro observed a to have been numerous Greek and a few Armenian families —_fortress.*’’ in these valleys, but no churches were reported from the area. From Anniaca, the road either crossed the river to the People from as far away as Unye and Terme use the Bolaman south and climbed over the mountains to join the route to travel up to their yay/as south of Aybasti, which are Sebasteia—Nikopolis road, or followed a bank of the Lykos named the Persembe Yaylalari. Such seasonal use of the eastward to a point where a tributary runs into it from the valley may be some further indication of the line of an plain of Nikopolis. The valley along this stretch is modeestablished route.*’* The confluence at Catak, mentioned rately wide, with barren ochreous rocks on the north bank, above as a possible point of departure for Sauronisena, is contrasting with the fertile green slopes on the south side. marked by an opening out of the gorges, giving gently slop- The modern road runs up the north bank, but there is also a ing fertile land for tilling. There is an admirable site for a track up the more fertile southern bank.?’8 The ascent up to castle on the hillock at the confluence, but it has not been the plain of Nikopolis, which lies at about 950 m, is an easy investigated by us. It is in fact a typical site for one of the one of some two hours by horse.?’?

secondary which lie from halfKotyora, way upOrdu, many of the og: weary 375 Ptownships y up many orn The Route to Sebasteia, Sivas

Pontic valleys.’ ’° The deciduous woods of the region are still or ; ; ; Kotyora was already decaying in the time of Arrian, but it used by the charcoal burners who may observed atremains their , , are ; , re- , is ancraft, obvious harbor town, andbe Byzantine traditional but no sign could be seen of the iron; 580 ; ; eswho ported fromstill its acropolis.7°° castle is reported smelting Chalybians were at work in AHamilton’s a south of

day Ordu, but there is not much more to indicate a route south The great bowl in the hills at Sauronisena marks the from the coast except the geographical fact that this Is avery

, easy crossing of the mountains, which is why the nineteenthchange from coastal agriculture to the upland farms Lo. ; century road was built.among It may.be to the road from Kotyora,

the remnants of forest. The bowl is similarthat to that which ; . when he . ; Sebinkarahisar; Ordu, to Anniaca, Koyulhisar, Evliya refers surrounds both were formed perhaps by vast ee er land slips which have filled up and partially evened out the writes: ‘“Two stations north of this village [Koyulhisar], on SP ; P P y the shore of the Black sea, is Baihssa—Bazari”’ The identity of precipitous valleys these . ; ok to, . in Baihssa-Bazari is mountains. uncertain, but it must be ;Ordu, Piraziz, or At a height of about 1,000 m extensive cultivation begins .

. . Bulancak and the distance suggests the Kotyora and to give way to forest and pasture, marking the limit ofport theancient useJanik 381 7,on the ; ; , future Ordu.*°' Morier writes: “‘To the of of a particular varietysea, of small-wheeled oxKuley cart peculiar ; . more ; 276 ; Black the distance from Hissar istonot than

western Pontos.*’° Other signs and of change arewe thehave superseding 19 282among ,, hours, again the choice the of fat-tailed ,bytwelve thin-tailed sheep, and of water buffalo by ; ; three place-names mentioned above for the port of Canik,

oxen. Women wear different costumes, whilewhich half-timbered ae _—is the district name for this coast. Munro reports the houses are succeeded by all-wooden ones. , ; me as distance from Koyulhisar to Ordu more realistically Above Sauronisena, the ,road climbs slowly through 583

, twenty-four hours.

broad-leafed forestKotyora to a heightthe of about at the gentle pass , .; ; From road 1,500 risesmalong

valleys of Hagbeli, then drops steeply the as valley theand ; ; ;the . ; ridgesand of the hills through easy into country far asofSauronisena. Melanthios at a height of about 1,000 m. there Fortoabout km it oo , The stretch from the Lykos20 at Koyulhisar is de-

follows the valley southeastwards to Matuasco, Mesudiye; the fields along the valley bottom are irrigated by noma ~ 277. Hogarth and Munro (1891), 729. There is no sign of ancient monstrous wooden water wheels turning ponderously with remains at the modern town of Koyulhisar which is on the north the motion of the stream. We have not explored the lower slopes of the valley, and here as elsewhere we have a well-defined

reaches of the Melanthios, seems unlikely that area inatwhich the focal point ofofhuman ay is “2 asbut. .itmoved around tothere suitseograpnical varying needs different periods history.

ever was a road along it since it runs into gorges north of 278. Hogarth and Munro (1891), 728; Munro refers to this stretch as ‘‘a narrow defile,” which is perhaps a little misleading.

272. Chrysanthos, AP, 4—5 (1933), map. 279. Morier, (1808), 337, may be referring to this junction when 273. See p. 23, for the east and west roads at this point. he says, ‘“‘about three miles from our last station we saw the road to 274. De Planhol. ‘“‘Chaines pontiques,” !—12, for a description of Diarbekir and Bagdad.”

transhumance in these parts. 280. See p. 120.

275. See p. 12. 281. Evliya (1644), II, 105.

276. Cf. G. K. Chatzopoulos, “‘H Gpagéa sic tov Tovtov, AP, 33 282. Morier (1808), 338.

(1975-76), 36-50. 283. Hogarth and Munro (1891), 729.

LAND ROUTES, NORTH-SOUTH 45 scribed above. There are the remains of earlier bridges in the link with the sea, particularly when it became a theme capbend of the Lykos below Yukari Kale; it is probably at this ital, to export its alum, which may be that mentioned by point that the road continued south to Sebasteia, like the Pliny?®’ and is certainly the ‘‘allume di rocca di Colonna” 288 modern one. The ascent out of the valley is a long but gradual about 684 tons of which were exported annually on a sevenclimb of over 1,000 m. into upland pastures at around 1,800 m, day porterage to Kerasous in the fourteenth century.?°?

where lies the watershed between the Lykos and the Iris. The The bond uniting the two towns is confirmed by Evliya road then descends round the heads of the Iris catchment Celebi who wrote: ““The inhabitants of the villages along the valleys in a southerly direction past Serefiye. Munro men- shores of the Black Sea send all their best goods into the tions Kechuit (probably Keceyurt), which is over against castle [Sebinkarahisar] to protect them from the inroads of Serefiye, a little to the north.*°* Here he found an inscription the Cossacks.” ?°° The route finally seems to have declined in of Justinianic date said to have come from Sivri Tepe, three importance with the introduction of the steamship, when hours to the east. He suggests that Dagalassos may lie here. If Consul Suter wrote of Sebinkarahisar: ““The traders procure this is the Ikissivritepeleri of the Turkish maps, there was a their supplies from Constantinople. Formerly they emshort and direct road over very high country between Zara barked at Kerehsin [Giresun] for the capital; but since the and Nikopolis. However, there is a place nearer to the establishment of the steamers they generally proceed for that modern road and lying to the east of Serefiye, called Alisir purpose to Trebizond. ’’??' Meryemana, which is more suggestive of a church found- The first motor road used to run out of Giresun just to the

ation, and farther along the same track is a castle name, east of the town and climb up past Gedik Kaya along a Camlikale K6yu. The course of the old road to Sivas from seemingly endless series of gently mounting ridges, with one

Koyulhisar, and the point at which it joined the Sivas— considerable descent, to reach Yavuzkemal at a height of Nikopolis road have therefore to be further explored before about 1,750 m. Nuts and maize are now the main crops of we can be certain of the exact course of either route, of the these well-cultivated slopes south of Giresun, with wellplacing of the Peutinger Tables’ junction station of populated villages in the valleys. Yavuzkemal is a modern Mesorome, and whether this is identical with Dagalassos.?*° township mainly existing on forestry; south of it the mounThe information of Munro suggests that the course of the tains are much more sparsely populated with stretches of modern road is not the same as that of the old road between forest in the high valleys and snow pastures above them with Zara and Nikopolis, or it may be that there were different peaks rising to just over 3000 m. The road from Yavuzkemal summer and winter routes between Sebasteia and Nikopolis. descends sharply into the valley of the river Kurtin, which it Such a suggestion is made the more probable by the different follows to its confluence with the river Aksu. From here it number of stations in the Antonine and Peutinger lists. The follows the river valley southward for a few miles up through Antonine list has three intermediate stations while Peutinger cultivated clearings into a rocky and narrow valley forested has four; the names are not easy to reconcile except for the with beech hornbeam, alder, maple, and wild cherry. This first after Sebasteia, which is Camisa in the Itinerary and gradually gives way to the conifer belt which in its turn peters Comassa in the tables. From Zara to Sebasteia the course of out in the rolling snow pastures of the pass of Egribel at well the ancient road cannot have deviated much from the over 2,000 m. This is the watershed between the rivers runmodern route which runs along a natural highway formed by ning north into the Black Sea, and those running south into the upland reaches of the Halys. The intermediate station of the Lykos. The road descends steeply southward into the Camisa or Comassa 1s agreed by all our authorities to lie at valley of the river Arslanyurdu. At Katochorion, Gedahor, it the modern Kemis or Hafik. From Hafik, the Turkish map passes through one of the nineteenth-century centers of alum marks a track directly northward over the Kesis mountains mining.*’? A few kilometers further south the valley opens to Ipsele and the upper Iris valley. This could have continued out into the wide Koloneia bowl, at a height of between 1,400

north across the ridge into the Lykos valley at Mudsun and and 1,500 m. thence across yet another range to Mesudiye and on to the The route taken by Tchihatcheff appears more likely to coast at either Polemonion or Kotyora. All these possibilities approximate the old one of the alum traders.?°° It runs up

await further exploration.?°° the valley of the river Aksu, which flows into the sea about The Route from Kerasous, Giresun, to Koloneia. 5 km. east of Giresun. The track, still essayed by postal jeep,

; ; , a, runs for the most part along the eastern shoulders of the

Sebinkarahisar, Nikopolis, Purk; and the Euphrates. Aksu valley. Tchihatcheff mentions three names between

or to Sebasteia, Sivas , Giresun and Geudul, Gudil, which cannot now be traced,

route between the Kerasous and Koloneia have ; ; ye , For ; ; utthe one is probably place simply markedwe Han. AtbGidiul,

literary but not archaeological evidence. Koloneia needed a Tchihatcheff left the Aksu valley and continued due south up

284. Hogarth and Munro (1891), 729. 287. Pliny, Natural History, XXXV, Lu, 184. 285. See p. 24, and Adontz, Armenia, 62-64; the stations are 288. Pegolotti, ed. Evans, 369.

conveniently summarized on p. 62. 289. Heyd, Commerce, Il, 566. 286. According to the Kiepert map, a traveler named Austin took 290. Evliya (1644), II, 206. He maintains later that they once the road inland from Ordu, partly along the Melanthios, Melet penetrated as far as Sebinkarahisar. Irmak, but we have been unable to trace an account of this journey. 291. Suter (1838), 436. Chrysanthos, AP, 4-5 (1933), maps, Krause went inland from Abdal, Piraziz, up the Pazar Suyu and then shows the route as of major importance. crossed eastward to follow the Aksu down to Catak, where he joined 292. Bryer, Isaac, and Winfield, AP, 32 (1972-73), 238-52.

the Giresun to Sebinkarahisar road. 293. See Tchihatcheff, Asie Mineure, Mineure, Atlas.

46 CHAPTER TWO a tributary valley through Gumbetkhan, Kumbet. From dismissed since all the indications are that Mesorome lies here he may have joined the road described above and west of Nikopolis whereas Buldur lies in the mountains to the crossed the pass at Egribel; or he may have kept on adirect southeast. The identification of Aksar, or Kilinglar, castle route farther to the east and come down to join the motor — with Oleoberda makes good sense if we assume with Yorke road below Lidjese, Licese, which he mentions, near __ that this is the first station of the direct Peutinger route from

Katochorion. Nikopolis to Melitene. Both of these tracks have Han names along them, but it The identification of Caltiorissa with the Yenikoy ruins is

seems more likely that the Tchihatcheff route along the Aksu possible but unlikely; Caltiorissa is the first station east of is the old one, for it avoids an extra mountain crossing. A Nikopolis on the Peutinger road to Satala, and Yenikoy, further point in its favor is that Tchihatcheff mentions six which lies south of the Halys valley, seems too far to the places along the road, which would agree with Pegolotti’s south of any reasonable line of march to Satala. We prefer statement that it took seven days for the alum to reach the not to assume the identity of Caltiorissa and Caleorsissa, but sea. Consul Taylor gives the distance as eighteen hours which accept the identification of Caleorsissa, Yenikoy, as the would have been a two- or three-day journey, but this 1s a second station on the direct Melitene road. The third station calculation for a swift party on horseback rather than for of Analiba would then lie at Kurugay.

porterage.??* The route from Nikopolis to Sebasteia via the river Pulat From Sebinkarahisar to Nikopolis, the modern road was followed by Grégoire, but he considered it unlikely that crosses a low saddle of land south west of the town and then this was the normal ancient route between the two towns. He descends gently into the Lykos valley near the great rock pointed out that the route westward from Nikopolis and then

of Dumankaya, where Taylor noted a Byzantine cave south via Zara was shorter and easier. He followed Yorke in chapel.”?*° The Lykos widens out at this point from its course regarding the Pulat valley route as representing the trace of

through the Zagpa gorges to the east. It would have been the direct route from Nikopolis to Melitene via Zimara, possible to ford the river except when it was running very Zinegar,”?® we, too, follow this opinion. high. To the south the road continued for a short ride of two A more direct route across the mountains to Sivas was hours or so up a tributary of the Lykos, to come out in the followed by Consul Suter.??? He went up the Gemidereh,

plain of Nikopolis. Gemindere, and around the flanks of the Kose Dagi on a

The Roman city of Nikopolis and the Byzantine castle of track which would have taken him through Kumoglu. The Koloneia derived their strategic importance largely from Tarhan identification of Kumoglu with Megalassos, or their positions at major road junctions. In the case of | Dagalassos, may not be accepted, but it seems quite possible Nikopolis the roads running east, west, and southwest have that the Suter route represents an ancient summer road been described above.*?° Another route of some impor- between Sebasteia and Nikopolis, and that Kumoglu 1s a tance, which is not of proven antiquity but whose course Roman or Byzantine station along this route. suggests an ancient road, lay more or less due south. This Consul Taylor traveled the whole journey between route turns out of the plain of Nikopolis up the valley of the Nikopolis and Zimara and we regard his route as a likely river Pulat and crosses one of the minor ridges of the K1izil approximation to the direct Peutinger route III between Dagi to come down into the upper reaches of the Halys these two places. He mentions the finding of a milestone valley. From here the choices were to follow the valley west- which he illustrates, and refers to remains of paving as “*the

ward down to Zara and Sebasteia, or to follow it upstream massive even blocks characterizing Roman work.”’*°° east to Refahiye, Basgercenis, and from there on to Satala or Well south of this, near “El Khan,” which is a little south to Erzincan. Or the traveler could continue in a southerly of Kurucay, he noted some cave dwellings in which a direction to Zimara, Zinegar, and join the frontier road Byzantine coin was found. going down the Euphrates to Melitene. The section of country through which these roads pass has From Esbiye Southward been little explored, but the Tarhan Map marks antiquities at We have no warrant for the existence of such a route in the Buldur Harabeleri and Yenik6y, and a castle at Aksar.?°’ ancient or medieval period. However, geographical considerHe identifies these sites as Mesorome, Buldur Harabeleri; ations suggest that there was a perfectly feasible crossing to Caltiorissa, Yenikoy; and the castle as being at Oleoberda, the Kovata, Alucra, valley, and there are plenty of tracks Aksar or Kilinglar. He marks a fourth site at Kumoglu, across the mountains.*°! A number of place-names ending in

without any ruins, as Dagalassos or Megalassos. The identi- |

fication of the Buldur Harabeleri with Mesorome may be 298. Grégoire (1907), 37-41, esp. 38. 299. Suter (1838), 437-38. Cumonts, SP, II, assumed that the

294. Taylor (1866), 295. route up the river Pulat was the route to Zara and Sivas.

295. Taylor (1866), 297-98. 300. Taylor (1866), 301-8, and notes 94, 95, 97 above, for confu296. See p. 21. sion about this identification. 297. The castle marked by Tarhan at Eskihisar may be his equi- 301. Peyssonnel, Traité, II, 53, 83, writes that Haspie, Esbiye, is valent for the actual ruins of Nikopolis, Purk, which he does not the summer port for the great mines of Kure which supplied the mark at all. The Turkish map correctly marks the village next to Ottoman empire with copper. He says that Kure was two days Piirk as Eskisar, but gives no Eskihisar in the area. There is no journey from Esbiye and four days journey from Trebizond, where Kiliglar in the area on the Turkish map, but there is a village the copper was shipped out in winter because Esbiye provided no Kilincglar south of Aksar. Tarhan is unclear as to whether his shelter for ships. Kure, as described by Peyssonnel, should therefore

Oleoberda is placed at Aksar or Kiliclar. be south of Esbiye, and probably in the region of Alucra. However,

LAND ROUTES, NORTH-SOUTH 47 ‘“Kilise’’ (= church), or “Kale” (= castle) suggest that this possibly at Kerasous, Kireson, on the east side,*°’ but nothregion would be worth exploring;*°? but Rivet, the only ing is known of the history of either place. We tentatively traveler who made the crossing, seems to have left no account identify Vakfikebir as the Sthlabopiastes of Panaretos*°® of it save the trace of his journey on Kiepert’s map. and there is some evidence for a Byzantine route southward Tichihatcheff made the journey between Esbiye and from here. The literary evidence lies in the campaign deArdasa, Torul, traveling across country which does not sug- scribed by Panaretos.*°? Archeological evidence is provided gest a natural route, but we have not explored the greater by the church at Fol, south of Tonya and the castle at Suma part of it. It is possible that the route inland from Esbiye in a tributary valley of the Philabonites, Harsit, above served as a means of connecting Tripolis, Tirebolu, with the Kiurtiin.*'° The geography of the area encourages such a interior. It is only a short journey westward along the coast road, for at Erikbeli there 1s a lower pass than any of the from Tripolis to Esbiye and, as will be seen below, there is no others over the eastern Pontic mountains; it is also a lower natural route southward from Tripolis. The villages noted by pass than any other for a considerable distance to the west. Tchihatcheff are: Adabau (?); Aurakevi (one of the Avluca The route itself runs inland up the line of the river Fol, but is villages?); Agatsch Bashi (one of the Agbasi yaylas?): unlikely to have kept to it since there are two sections with Sarybaba, Saribaba; Eryklu, Erikli; and Emberek, deep gorges before the Fol emerges onto the wide irregular

Emrek.°°? bowl in the hills where the township of Tonya is situated, at about 750 m above sea level. At the southern end the valleys

From Tripolis, Tirebolu, Southward once again close in to form a gorge for about 4 km, where the We have no evidence of a classical route southward from track must have lain along the shoulders of the valley, and for

Tripolis and only a few indications that such a route may the last 8 km up to Fol Maden it widens out to gentle have existed at a later date. The existence of the classical cultivated slopes with even some flat land in the valley town and of its fortified Byzantine successor is suggestive of a bottom. Tonya occupies the typical situation of one of the

route inland, since there would normally be little point in a secondary townships of the Pontic valleys,*'' but has no coastal town isolated from its hinterland. In the case of antiquities save for a reported castle whose existence we were Tripolis it has, however, been suggested that the reason fora unable to verify. town lay in the need of a port for the Argyria silver mines.°°* At Fol Maden, not far from the church, there are the ruins It has already been pointed out that the Philabonites, Harsit, of a small bath and han, providing further evidence of a River valley, which might appear to provide a natural route through route along this valley. There was also a mine there, inland from its mouth a few kilometers to the east of Tripolis, as the name of the place implies. runs through an exceedingly deep gorge over much of its From Fol the track climbs up past the headwaters of the length between Ardasa and the sea. And any route south- river to the Erikbeli pass at about 1,500 m. A ridge track runs ward from Tripolis has to cross awkward mountain country. east and west from here up through the summer pasture However, the evidence of Panaretos as to the expeditions of lands, or the traveler can take the easy descent southward Alexios III in 1380 shows that an expedition could cross such past the castle of Suma into the Philabonites valley at

mountains,°°> and if the identification of Simulika as Kurttn, Deyrolle followed (in reverse) a track from Sumuklu is correct it seems likely that he may have been Karatchoukour, Karacukur, which goes over the shoulders following a route down into the Kovata valley. In the nine- of the mountains eastward to Ardasa and Mesochaldia.?!? teenth century, Kiepert’s map shows Krause making jour- Or the traveler could continue southward across the moun-

neys along the line of the Harsit on both flanks of the tains to Cheriana and the Lykos valley or to Kovata and

valley.*°° Koloneia. The Cheriana track followed the eastern branch of the river Kurtin through villages well populated by Greeks

From Sthlabopiastes, Vakfikebir (?), Southward in the eighteenth and nineteenth centuries.*'*> At Emrek it There are two small classical sites in the bay of Vakfikebir: would have crossed the Tchihatcheff route from Esbiye to

at Liviopolis, Yuvabolu, on the west side of the bay and Ardasa.*'* From there it climbed to the watershed of the

oo Trebizond to Cheriana.

Yildiz mountains and joined one of the direct routes from

there is probably a confusion here. There are mines in the hinterland The Kovata track followed up the line of the western south of Esbiye and it is to these that Peyssonnel must refer. But he is branch of the river Kiirtiin and would have crossed the

mistaken in identifying them as Kure, which is the well-known .

arsenic mine south of Ionopolis, Inebolu. Tchihatcheff route at Saribaba.°'> At Simulika, Sitim302. See p. 26 and note 91. Esbiye would have been the port for

Gouatha, Kovata: and Ghevond’s unidentified Castillon and the 307. The Kiepert map appears to mark Kerasous at Vakfikebir. district of Marithinesse may be in the area south of Esbiye. For a table of identifications of places on this stretch of coast, see 303. Tschichatschof (1858), 288-93 and map. For the country p. 154. around Erikli and Emrek, see Bryer, Isaac, and Winfield, AP, 32 308. See p. 141.

(1972-73), 221-35. 309. See p. 140. 304. See pp. 138-44. 310. For the church at Fol, see p. 159; and for Suma Kale, p. 144.

305. See p. 140. 311. See p. 12.

306. There is also a Krause journey inland from Karaburunu, 312. Deyrolle (1869), XXIX, 15. which lies between Tirebolu and Gorele. This journey took him 313. Bryer, Isaac and Winfield, 4 P, 32 (1972-73), 176 ff. eastward along the high summer pasture ridges to the Zigana pass, 314. See above.

where he joined the main Trebizond to Tabriz road. 315. See p. 140.

48 CHAPTER TWO iikli,?’© it turns southwest across the Sarryar mountains, to the Zigana pass. The pass, at 2,025 m, lies in the snow and then south. Kaledibi, “the village at the foot of the pastures a little above the tree line, but is relatively well castle,” lies on a track leading from the Sari:yar mountains sheltered. On the south side, the road descends a valley of a down into the Kovata valley. Thence it was possible to tributary of the Philabonites through forested slopes as far as continue south across the Berdiga Daglari mountains into the village of Zigana, with its small castle. From here it the valley of the Lykos at Camoluk, Mindeval, or westward continues downward through cultivated land to the valley of

he v 1S j l

to Koloneia. the Philabonites at a height of about 850 m. A short stretch of

From Platana, Akcaabat, Southward the valley eastward from this junction as far as the castle of , Ardasa, Torul, has gentle slopes now devoted to orchards, ; - 317 but westward the river enters fearsome gorges as it cuts its routes southward through the bandon of Trikomia,”*’ the ; 323

The geography of the hinterland allows for relatively easy ; .

.; , 318 through the mountains to the sea. Haska Yaylas, and Katirkaya toway the Zigana ridges.°'® aepass ; represented a . This winter route across theGood Zigana tracks lead southward up both the Kalenima and Sera val-

oe ——. lengthy detour for the on the road to Satala or leys; from.;their headwaters it is easy to strike intotraveler the main . ; 304 ; ots Theodosioupolis, Erzurum. The summer routes were

route southward from Trebizond at Dikaisimon, Macka, or ,

4s ;Fikanoy ; shorterYayla, and there were numerous choices, snow pasvia Phianoe, and Spelia, Ispela. with Panaretos ; ;for the the wee . tures provided open country few obstacles; account of an expedition into these mountain pasturesaccording andtraveler . . .to . might choose his course across the watershed then northward to the sea is evidence of the use of tracks ; , .

319part where wished to descend into the Philabonites valley, but through; this of thehecountry in .the fourteenth century. ; , the westerly summer , a common meeting point for routes,

Deyrolle followed a of route fromPylae, the Philabonites, - at ; on was theatpass the northward Pontic Gates, Kolat Bogazi, Harsit, valley Karatchoukour, Karacukur, near Kirtiin . i, .; about 2,400 m. The direct route to this point is frequently

He seems to; have gone upbytotravelers the watershed atnineteenth Erikbeli and . the century. then climbedmentioned the ridges to the east as in farfrom asDikaisimon, Katirkaya and oot ; It ran south Macka, with the winter route as far as

thence northward to Beypinar Yayla and down the Kal- , .

. 320 ; ; Chortokopion but then due southtook alongeighthe summit enima River to Platana.’*° The climbed whole journey ; ; of ; ridges through Karakaban to the Pontic Gates. A variant teen hours. He also traveled from the Zigana pass westward ,a along the watershed and then down into the Philabonites this was to follow the winter route as far south as Hamsikoy valle and then eastward through Ferganli to Gates. the Y. pass southward, one route descends thethe valley of From the river From Trebizond, Trabzon, Southward Istavriand Kurum to the confluence with the Philabonites at

The hinterland of Trebizond is described in some detail Harava Hanlari, about halfway between Ardasa and below; we give here only a brief geographical outline of | Gumushane. From here it was possible to cross the river and

routes from Trebizond into the Philabonites valley.°?! continue by a more or less direct route south to Cheriana.*?° The main road out of town climbed past the Theo- A second route from the Pontic Gates continued down the skepastos monastery and across Boz Tepe and the hilly river Istavri to 1ts confluence with the river Kurum at Istilas, country to the south for about 15 km, when it descended and from there turned east??° and south again across the eastward into the valley of the Pyxites, Degirmen Dere, at Bazbent Dag: mountains to come down to the Philabonites Esiroglu. Here a summer route to Bayburt crossed the river, valley at Beskilise, below the heights crowned by the castle of but the main winter highway seems to have followed up the Tzanicha, Canca. river valley as far as Dikaisimon Cevizlik, Macka. An alter- The third route from the Pontic Gates was to continue native route southward out of Trebizond°?? crossed the along the summer pastures of the watershed ridges toward Pyxites directly east of Trebizond and climbed the hills from Bayburt; at Anzarya Hanlari, there was another junction

there southward through Zafanos and Cambur. with a north to south crossing. From Magcka the main winter highway climbs up the Two more routes starting south from Dikaisimon foleastern slopes overlooking the Pyxites, passing through lowed up the Panagia River, Meryemana Dere, along a well Chortokopion, Hortokop, and Yanandon to come down to cultivated valley as far as KinalikOprihan. From there one the river once again near Palaiokastro (below Hamsikoy). route turned more or less due south to follow up a valley Here alternative routes present themselves, but the winter road continues in a southeasterly direction with the stream 323. Deyrolle (1869), XXIX, 15. Deyrolle traveled the short which forms the principal source of the Pyxites and climbs up stretch between Torul and Kiirtiin keeping high on the northern slopes above the river, but we find no traveler going straight through

316. Panaretos, ed. Lampsides, 44, 79. Cf. Bryer, DOP, 29 (1975), to the sea by this river route. 147 and note 138. In this identification we follow S. Papadopoulos, 324. This, as will be seen, is the suggested route of the Antonine

Avop8aoceic, BNJbb, 6 (1928), 399-400. Itinerary. The relative distances of the different routes to Theo-

317. See p. 160. dosioupolis are given and discussed by Lynch (1893-98). IT, 225,

318. Janssens, Trébizonde, 9, 20. 240. He follows Murray’s Handbook, in giving 199 mi for the winter 319. Panaretos, ed. Lampsides, 76. route, as opposed to his own calculation of 145 mi for the shorter

320. Deyrolle (1869), XXX, 258. route.

321. See p. 140. 325. See p. 165; Chesney (1831), Texier (1839), Hell (1846), and

322. Teule (1842), II, 551-58, on a journey from Erzurum to Barth (1858), were among the travelers along this route. Trebizond, but he is not very informative as a traveler. Blau (1860), 326. The river Kurum is also called the Yagli Dere, and the camp

380-81 and map Vv. Mochora, Mollaalh, may lie along it; see p. 304.

LAND ROUTES, NORTH-SOUTH 49 passing the villages of Larhan and Agursa, and a mineral are separated by the narrow band of the Vavuk Daglar1. water source, to arrive on the watershed and join the ridge Between the western end, Balaban, Yildiz, and Sariyar road at Anzarya Hanlari. From here one track led south Daglari, and the Vavuk Daglar, are the Yastar Daglar westward along the river Kurum Dere and past the castlesof | separating the towns of Kelkit and Giimtishane. The Kose Kurum and Mochora, Yagli Dere, to Gimtshane. A second Daglari separating the town of Kose and the upper Lykos, crossed the hills to the south through Imera and Tefil to come Kusmasal, valley from the upper Philabonites, or Kanis, out at Gumishane or at Tekke, higher up the Philabonites ___valley.**'

valley.>?’ Two motor roads now cross these mountains. The western

Or the Bayburt track continued along the ridges eastward, route starts south, a few kilometers east of Ardasa, Torul, passing some unnamed junctions between the Deveboynu and at present it represents the shortest direct route southand Kostan, or Komacan, mountains. A route which con- ward from Trebizond into the Lykos valley. There is at least tinued up the river Meryemana past the monastery of |— some evidence that this same route always was the shortest Soumela might either join the ridge road at Anzarya Hanlar1, way to travel southward, as we shall see below. It runs up or pass around the Karakaban mountain to Taskopru and along the river Dipotamos, Ikisu; which is a tributary of the wind its way down south to the Bayburt ridge road below Kanis; there is a small Byzantine watchtower and chapel

Mount Kostan.°?8 guarding the mouth of the valley. The Dipotamos is one of

The route which left the Pyxites valley at Esiroglu followed the largest tributaries of the Kanis. From the confluence

up the river Galiana, Kustul, past the monastery of southward, its course is along a narrow gorge for a short Peristeriota and joined the tracks mentioned above at distance, but this gradually widens out into a valley with Taskopruhani from where it is an easy ascent over the gentle some gentle slopes which are much eroded and almost bare ridges to the Bayburt track. And the route which followed up of vegetation. Up a side valley to the east is the castle of the eastern slopes of the Pyxites through Zafanos would also Godaina, Kodil,*** and on the western heights above the

have come through Taskopruhan to join the Bayburt ridge valley, a little further south are the castles of Byana and track. The traveler heading for Kelkit and Satala might leave Colosana,*** which effectively controlled the crossroads and

the Bayburt track at a junction south of the Deveboynu the confluence of the two rivers Soruyana and Ertabel with mountains and follow a track southward through Kermut to the river Dipotamos. Continuing along the course of the Tekke in the Philabonites valley. Or a little further to the east Dipotamos there is no great ascent for a further 15 km to a track went south past Kabakilise, with its early Byzantine Haskoy where the river divides into two branches among church and the Leri villages, once the seat of a bishopric, to much cultivation. The western branch splits into the rivers join the Philabonites at Zindanlar Arazi.*?° The ridge track Soruyana and Ertabil, which rise in the mountains directly to Bayburt continued eastward but split into two major north of Cheriana. Direct tracks used by earlier travelers lead branches between the Deveboynu and Kostan mountains. across the mountains and down past Mumya Kale?** on the The southerly branch went through Veysernik and Iskilas, way to Cheriana. In the eastern branch of the Dipotamos, to come down into the western end of the plain of — whichis the valley of the river Karamustafa, runs the motor Paipertes/Charton, Bayburt/Hart, at the village of Hadrak. road; shortly after leaving Haskoy it climbs steeply through The northerly branch continues along the high ridges, avoid- broad and narrow leaf forest to the pass below Elmali Dag ing villages, and it comes down into the plain of Charton, at about 2,100 m. The forest here stretches up even above the Hart Ovasi, at Niv or at Ostuk. From here it crosses the pass, although at this height it is thinning out into patches of gently rolling plain to Varzahan, where there is a junction pasture land. The descent on the southern side is a long and with the road coming over from the Lykos valley, and thence gentle one through forest, which gives way lower down to the

to Bayburt.**° scrub-covered hills left throughout Turkey wherever man, We must also consider the mountain barrier between the goat or charcoal-burners have destroyed the woodland. At Philabonites, Kanis, Harsit, and the Lykos, Kusmasal Yukari Tersun, above Tarsos, Asagi Tersun, the road turns Kelkit, valleys at this point, since it is crossed by a number of west to Siran, the Karaca of nineteenth-century travelers,

routes which must have been used in the Roman and but the old route may have continued south past the ByzanByzantine periods. Roman or Byzantine sites which may be tine site at Tarsos before forking to the west and east.

relevant to the routes are mentioned in the course of the The eastern motor road starts south at the han at

description. Pirahmet, about 17 km east of Gumishane. It runs up a The mountain barrier forms a southern flank of the main tributary valley of the Kanis. At Kirikli it joins a valley and Pontic chain which is cut off from its northern neighbors by route running westward past Ulu Kale to Kelkit. The motor

the deep rift of the Philabonites and Lykos which gradually 331. rth nai «standardized. . The names of theTh mountain ranges are not standardized.W We taper in toward each other at their headwaters, where they follow the Turkish Highways Map, and the 1 : 800,000 map in referring to the western end of the triangle as the Balaban Daglari, and

327. It was at Tekke, or eastward of it, that one of the Roman the |: 200,000 sheets of this area for the other names. The Vavuk

roads to Satala must have crossed the Kanis. Daglari at the eastern point only appear on the | :200,000 maps, but 328. Lynch (1893-98), II, 239-40, describes the route, and gives a early travelers all refer to these mountains by this name.

time of 2 days between Soumela and Bayburt. 332. See p. 308, for Kodil castle.

329. See p. 311. 333. See p. 308, for Colosana castle. Byana castle is reported only

330. See p. 352, for Paipertes, Bayburt, to Theodosioupolis, by Chesney (1831). 127.

Erzurum. 334. See p. 173, for Mumya castle.

50 CHAPTER TWO road continues more or less due south, up gentle slopes have turned north from here.**? He writes further: ‘At the through the forest to the pass at about 1,800 m. The southern meeting of these two rivers [which of them is the Korshat?] slopes of Kose Dagi again constitute a gentle descent to the exactly where the Greeks passed, one sees a square enclosure

northern headwater stream of the Lykos, Kusmasal. The which was an Odjiak of the Sultan Murad’s, according to river 1s below the modern township of Kose. The valley at Hadji Khalfeh Orlah, or Ourlah Gumish Khanah.”’ This site this point 1s about 1,550 m above sea level, and some three at the confluence of the northern tributary river Soyran with kilometers wide. The vegetation of the mountain slopes the Kanis, Harsit, has the ruins of a fort and ditches as well as around Kose, on the southern side, consists largely of scrub. those of the han described by Borit-Briot. Strecker traveled

This route may have been used in the Roman or Byzantine north from here; like Briot he regarded this as one of the periods but with a different starting point, about three kilo- main summer routes from the Kanis northward across the meters west of Tekke. This alternative is suggested by the site watershed to Trebizond. We have tentatively identified the of Kog Kale, where potsherds on the level ground below the ruins of Zindanlar Arazi at the confluence of the Soyran and castle indicate a settlement of early date,*** whereas at the Harsit as Procopius’ Longini Fossatum and BourPirahmet nothing has been found. A route passing by Kog gousnoes.°**° The place is described by Strecker as being Kale would probably have joined what is now the motor the seat of the miuidiir of the kaza of Kofias, Kovans.

road at the Kirikli junction. This statement seems to have led Kiepert to confuse

In addition to the motor roads, there are other possible Murathanogullari with the village of Kovans, about 5 km routes across these mountains. The Tarhan map marks a further east up the Kanis valley. Here are the ruins of a castle at Ulukale to the west of the modern K6se road. This is Byzantine church and the impressive Goat Castle, Keci Kale.

in fact no castle but a high and peculiarly impressive rock Kiepert confuses the two sites on his map, and even the formation. A direct track between Gimushane and Kelkit Turkish map seems to be in the same confusion over its runs below Ulukale and is in use throughout the winter for nomenclature at this point.**! carrying fuel. The route runs through Gumishane, Sorda, The shortest crossing that Kiepert notes is via Schinnik, Mavrengil, Zaganl, Venk, Kom, and Pekun before reaching Sinik, and across high mountains. The Turkish map marks Kelkit. We were told that the summer version of this road no short crossing of the mountains directly north of Sinik, runs more directly across the slopes of Ulukale. This route is but the village is very close indeed to a junction with the described by Strecker, who mentions, north of Kelkit: Bojiin, motor road at Yukari Tersun and it looks as if it followed the Pekun (?); Bolodor, which he identifies as the Porodor of line of the motor road from Tersun northward.**? His map is

Smith and Dwight; Kangeli Dagi, Avliya Dagi (?); and misleading in its proportional distances and the placing of Mavrangelion, Mavrengil. Strecker writes that this whole villages in this area, errors which were followed by Kiepert route from Erzincan to Gumishane can be covered in two and Blau, but the place-names allow of a correct interprehard days’ traveling, so that from Kelkit to Gimtshane tation of the routes. would be one day. This was also the route of Consul Most of the travelers using the short crossing were howBiliotti,**° and it may have been the return route of Smith — ever starting from Cheriana, Siran, and not from Kelkit. and Dwight. The latter mention Germery, Germurt, which Morier notes the road north to Gumusk Khoneh, Giimiiis their equivalent of Kelkit; an unnamed village two hours to shane, and gives the time for it as twelve hours, with a the north, which could be Pektn or Kom; and Prodor, further ten to Trebizond.?*? Bolodor. It took them seventeen hours to cross from Brant traveled north across this way in 1835; he gives the Germurii to Gumilshane and a further twenty-four hours to distance as sixteen hours, but no details except that it was a

cross the watershed and reach Trebizond.**’ rough road.*44 Strecker gives further valuable indications of routes across Hommaire de Hell made a direct crossing southward from these mountains.*°8 He once crossed from Kose to Gimi- Trebizond. He mentions Gumushane; Adile, Edirebasi (?): shane in winter in eight hours by an unspecified route. He Edret, Edre; Halazou, Halazara; and Dorena, Turuna (?).**° gives the summer road from Erzincan to Trebizond as run- This was a short route which would have cut across the

ning through Schirut, Surut, and across to Murad- Dipotamos, Ikisu, Karamustafa, valley motor road south of Chan-Oglu, Murathanogullari, in the Kanis, Harsit, valley. Haskoy and crossed into, or followed along the ridges over Borit (a mistake for actual name Briot), also mentions this the rivers Livene and Turuna. After crossing the eastern place. He thought that Xenophon’s Ten Thousand might flanks of the Karagol Daglari, he probably came down through Norsun and passed by Mumya Kale and Kersut. 335. See p. 310, for Kog castle; and Tschichatschof (1858), The villages of Yukari and Asagi Kersut are given by Barth 293-95 and map. At a place called Aadja along this route Tschichatschof sighted a castle on top of a hill. This could have been 339. Briot (1867), 464; Strecker (1855), 348-49. Kog castle, but is more likely to have been the village of Akcakale, 340. See p. 311.

judging by the distance of 15 hours which he gives for the ride 341. Kiepert’s footnote to Strecker (1855), 348. The Turkish

between Aadja and Gimishane. 1:200,000 map marks Kecikale (i.e., Kovans Kale), but also a 336. Biliotti (1874), 225-26. Biliotti started from Giimiishane second Kale at Kovans, a second Kovans at Murathanogullari, and

and rode on the first day over Gumush dag, Guimiis Dag, and along a ruined Harap Kale (which we have not located) further to the west.

a bad track to spend the night at Pollodor, Bolodor. On the second 342. Strecker (1855), 350 and map 11.

day he went through Pekeun, Pekiin, to Kelkit. 343. Morier (1808), 332-33.

337. Smith and Dwight (1830), 444-45. 344. Brant (1835), 222—23.

338. Strecker (1855), 346-48. 345. Hell (1846), I, 390-95.

LAND ROUTES, NORTH-SOUTH 51 as ““Korssyk” or ““Gossuk”’ and by Strecker as ‘‘Gersut.”’ Cheriana, with only one exception that crosses southeast to Kiepert relates the name to the Antonine station Carsais.**° Kelkit.*°°

A variant of the Hommaire de Hell route was taken by The following subsections are concerned with the ancient Barth. He traveled over the Zigana pass to Ardasa and then routes from Trebizond via Cheriana and the Kelkit valley along the Kanis to turn south where the motor road now down to the Euphrates.

forks up the Ikisu valley, where he marks the ruined The Ancient Routes from Trebizond to Satala, Sadak. chapel°*’ near the confluence of the rivers. His next name is = Apart from the evidence of Roman occupation at Hortokop

Kodil, suggesting that he rode along the eastern ridges above Kale, we have no certain archaeological evidence for the the Ikisu valley; and then Bulbiloghlu, Bulbuloglu. To reach ancient routes, and must largely rely upon Antonine, the latter he probably came down to cross the motor road at Peutinger, and Ravenna lists.°°' These may be conveniently the confluence at Haskoy, and then followed the valleys of tabulated as follows: the rivers Ertubel and Livene, so that he must have joined or

cut into the Hommaire de Hell route somewhere along this Peutinger = mp. Ravenna Antonine mp.

. . e Trebizond 20 ' Trebizond 20 22

stretch. He marks a castle before reaching the watershed , ;

between the Kanis, Harsit, and the Lykos; notes the ; 7 “teas cexy aos Magnana 10 AdandVincesimum

Gizenenica 18 Zigana 24/32 /;. ;;;Frigdarium Bylae 6 Bile Thia 17 8

names “‘Chan Daghdibi” and “Yaila”’ before reaching Ulu ; . ;

Schehran, Ulu Siran, but neither are identifiable. ; Texier, traveling north from Cahiran, Siran, seems to be . ,

348 Sedisca F1 Pont 24 tae ; Patara 14 Patra . og 45 349 ; Medocia 12. Medocina

the only earlier traveler who followed almost the entire ;

course of the present motor road. He went through

Terma, Telme, which is suggested by Miller as the location of ; ;

, . Solonenica 18 18 Solodicina 4. Domana Domana Domana 18

the Peutinger station ““Cunissa. He next mentions Zimo, _

Zimon, on the south side of the pass above Tarsos, Tersun, and then a Caravansary, which is probably the anonymous

. Satala Satala Satala

han or Karamustafahanlani, well down on the north side of 124 the pass. He then turned off from the valley road which runs

north and the mountains in a northeastward ,, _ ar ;to; ItIkisu has longcrossed beenviaclear thatEdra, the Antonine andHePeutinger direction to Giumushane Edima, Edirebasi (?). , . routes the same the station they have rode for.nineteen hoursare and not twenty minutes oversince two days, . only ; ; in common is Domana, although and Ad and then a further; eighteen hours toare Trebizond. His route , . ;The , ; Magnana ; Vincesimum evidently identical. Ravenna from;Zimon southward isidentical the one with described above by . . ,;route , .; , would appear to be the Peutinger. According Strecker; it probably represents the mountain crossing made . to its paragraph 12, the stations are Satala, Domana, Saloni, . Mecia, Medoia, Bile,of and an unidentified further We thus have a choice of aPatra, number routes betweenappear the , .in ; ; .; ; .; series. Two stations, Medocina and Solodicina, Philabonites, Kanis, Harsit, and the Lykos, Kelkit, valleys . i ; . the list of coastal towns in paragraph 17 next to Ysilime-

by Clavijo. ; ; .

which may perhaps be summarized as follows. The easterly ,; ; ae routes ve Susurmaina and Ofeunte-Ophis, which may indicate routes start well east of Gumtshane from the stretch of valley ;

coast inland to these stations. between Tekke and Kovans and; from leadthesouth to the Kelkit ; ;paths , .; , ow ; Geographical considerations suggest two likely region; routes from the region of Gumushane itself run both ., ; for . , . these roads, corresponding with the summer and winter southeast to Kelkit and southwest to Cheriana. Routes from 359 ; routes of travelers across the coastal watershed.’°’* The

the stretch of valley between Ardasa and [kisu run south to ; a , ; Kiepert map, which in this matter summarizes scholarly 346. Strecker (1855), 354, and p. 25. opinion up to the First World War, assumes that the

347. Barth (1858), map. Mordtmann (1850), 423-30. It seems = Antonine Itinerary followed the winter route across the probable thatrom General Chesney ee route in January hrSivas, when mountains, since e travele treodizon oOthissedinkaranlSar, an e€ppo: : it included the known station of Zigana. It Chesney (1831), 127-29. He crossed the Zigana pass to reach takes the Peutinger and Ravenna routes across the more Gumiishane and then mentions the castle of Godol, Kodil, built on direct, higher mountain country through the snow pastures. two pinnacles. From there he went on to Darnade, Dorenci (?), and

Byane-Kaleh, Bayana. He presumes these and other castles to be 350. Chrysanthos, AP, 4—5 (1933), map, marks a direct route Genoese work. His detailed itinerary stops at this point with the south from Tzanicha to Kelkit via Alantza, Alanza, but this seems to frustrating remark, “‘It is useless to delay the reader by a journal of be an arbitrary line. For routes south of Cheriana, Ulu Siran, and my daily progress.” If the journal survives it would be very useful, Satala, Sadak, see p. 165.

but perhaps the January weather was too severe for him to have 351. Cuntz, 7R, 216.4; Miller, 7R, cols. 680-81, Route 98; J. been able to keep it. Chesney favours Giaur Tagh near Bayana Schnetz, Ravennatis Anonymi Cosmographia (see note 15 above), 23,

as Xenophon’s Mt. Theches but gives Mt. Zingani, Zigana, or 29; sections 12 and 17. Karagool, Karagol, as other possibilities. (kiaovp Ntay appears in 352. Anonymous (1685), 45 records little about his journeys to Chrysanthos, AP 4—5 (1933), map, but not on the Turkish maps. It and from Trebizond across the mountains. He does however state

seems to correspond with Kelahmet Muvakkati Gol. A Byzantine that there are two routes to Bayburt, one of which he took, via equivalent to these modern itineraries may be found in Lazar- Agatch Bach: (Agag Basg1). He writes that the other route is via

opoulos, in Papadopoulos-Kerameus, FHIT, 86. Ghumich Kana, Gumiushane. He was traveling in December and 348. Texier, Asie Mineure (1839), 591. took the shorter mountain route whereas most winter travelers took

349. Miller, 7R, cols. 671-77. the longer Gumushane-Zigana route.

52 CHAPTER TWO Both routes must have followed the same course across the valley at the confluence of the river Panagia, Meryemana, hills southward from Trebizond for some distance before with the Pyxites, Degirmen Dere; it has no known ancient or separating, and they probably rejoined each other and fol- medieval remains. Hortokop Castle is not more than two or lowed the same trace for the southern section of the journey three miles from it, on the nineteenth-century caravan road across the Kose Daglar if we accept the identity of the at a point where it commands views of an extensive stretch of Antonine ‘Sedisca’”’ with Peutinger ‘‘Solonenica” and the Pyxites valley. A sequence of coins from the Hellenistic, Ravenna “‘Solodicina’”’ or ‘Saloni Mecia.”°°* Either Kog Roman, Byzantine, Seljuk, and Ottoman periods provided Kale or Zindanlar Arazi could represent the site. If Kog Kale suggestive evidence that this is a station on the caravan were accepted, then the ancient route may well have followed road.

more or less the same course as the modern motor road The Clavijo Route from Trebizond, Trabzon, to across the Kose Daglar. If Zindanlar Arazi be accepted, the Acilisene, Erzincan (see fig. 3). Clavyo took one week to ancient route may have followed the Strecker route over a travel from Trebizond to Arzinga, Erzincan, including some somewhat higher track through the villages of Perek or delay in negotiating with Kabazites of Chaldia.*°’ On the Zimon and Surut. Or these classical stations may each re- first day his route was over hilly country and took him down

present a different place. to camp in the Pyxites, Pexic, Degirmen Dere, valley at a

At Gumushane tracks west of Gumushane itself run south place now unidentifiable where there was a ruined church. up and across the Dipotamos valley.*°* One of these may On the second day he passed the castle of Palatomatzouka, represent the southern half of the Roman route. There are Palomacuga, which he describes correctly, and went on to numbers of ways of crossing the mountains between the camp in the open, probably in the region of Karakaban. On Kanis and the Lykos. The Cumonts’ choice of the ancient the third day he stopped by the castle of Sigana, Zigana, route appears to be based on nothing more substantial than which must be the small ruin below the present village on the the fact that they took it. The presence of one worked block spur of a hill, which he noted belonged to Kabazites. On the of stone, probably Roman, in the cemetery of KOse is their fourth day he passed Ardasa, modern Torul, at 9 a.m., only reason for identifying the site as Domana.*°° More __ correctly describing its commanding situation. Three leagues

exploration is required. farther on he passed a tower on a high rock and a narrow An identification of the Antonine stations, according to passage, which is probably the fort at the entrance to the Kiepert, is as follows: Ad Vincesimum, at 20 m.p. from Dipotamos, Ikisu, valley; and on the evening of the fourth Trebizond, falls near Macka; Zigana is at the village of that day his party was stopped in the valley under the castle of name on the south side of the Zigana pass; Thia ts in the Dorile, where Kabazites was in residence. This medieval region of Beskilise; Sedisca in the region of Tekke; and Torul (an area name) would correspond to a castle which can Domana in the region of Kose. With the exception of Zigana, be seen above the road up the Dipotamos valley near

the placing of these stations is hypothetical. Colosana.*>8 The fifth day was spent in haggling with An identification of the Peutinger stations, according to Kabazites about what dues the party should pay for a safe Kiepert and Miller, is as follows: Magnana in the region of passage through the mountains and protection from the Macka; Gizenenica in the region of Karakaban; Bylae in Cepni Tirkmens.**? On the sixth day Clavijo crossed over the region of Kolathanlari; Frigdarium in the region of | mountainous country, passing a valley with a castle held by Phrangkanton (a Kromniote village); Patara at Linotron, Cepni, and came in the evening to Alango gaga in the district which, like Phrangkanton, cannot now be identified on the of Arzinga. His precise route across the mountains is not Turkish maps; and Medocia at Tanera; Solenenica then falls clear, but it seems probable that he followed the line of the at Kovans; and Domana at KOse. These identifications, like motor road southward over the mountains and down the those for the Antonine Itinerary are largely hypothetical, southern slopes as far as Zimon or Yukari Tersun. From here although Bylae is evidently Pylai, the Pontic Gates, Kolat Clavijo branched south-eastward on the route described by

Bogazi. Strecker which runs through Sinik and would pass by Alansa

Thus the first and last stages of these routes are identical, which, following Le Strange, we identify as Alangogaga. but the middle stages take quite different paths. Our own Here the Kabazites escort left the Spanish party; from this contribution provides no certain new alternative for the ac- point onward they were in Mongol territory where they cepted Antonine and Peutinger routes. We only point out a traveled easily. Alansa was also proposed by Kiepert for the number of new sites relevant to the problem and a number of station, or stations, Hassis and Haza of the Peutinger and travelers’ routes which deserve further consideration. The Antonine routes from Nikopolis to Satala, but the village site of Lower Chortokopion, Hortokop, Castle*°® may be seems to lie rather too far to the north for a station on a direct noted here as an alternative candidate for Ad Vincesimum road, since it is already well into the foothills of the Gurtiz and Magnana. The modern town of Macka is right in the 357. Clavyo (1404), trans. Le Strange, 116-22; ed. Estrada, 353. See Papadopoulos-Kerameus, ed., FHIT, 12-14, 31-32. 79-84. Both the spelling of the place-names and their identification

354. See p. 310. are at times misleading in Le Strange’s version. The original Spanish 355. Cumonts, SP, II, 354. shows how carefully Clavijo transliterated his place-names.

356. However, Chrysanthos, AP, 4—5 (1933), 80, following 358. See however Chrysanthos, AP, 4—5 (1933), 81-82, who reMiller, /R, col. 681, puts Gizenenica at Karakaban on the southern gards Dorile as a corruption of Torul. This is a reasonable suggesboundaries of the Hortokop region and identifies it with the tion but makes nonsense of Clavijo’s itinerary. But see p. 302.

Chasdenicha of Panaretos, ed. Lampsides, 73. 359. Cf. Panaretos, ed. Lampsides, 68, 79.

LAND ROUTES, NORTH-SOUTH 53 mountains to the north of the Lykos valley.°©° From Alansa routes Ia and Ib, and because geography and later travelers to Erzincan was a further two days. Clavijo gives no details, support their existence.

but probably followed the route through Kelkit to a point A low range of hills separates Cheriana from the Lykos just north of Satala, where he would have joined a track valley, where the river flows in a wide valley beside the site at which was to become the nineteenth-century military high- Asagi Haydiirtik.*©’ The river can be easily forded in these way between Erzincan and Trebizond. This seems to have reaches. A track leads south up the Sifon valley to join our been the normal direct route until the first motor road was suggested course for the direct route from Nikopolis to

constructed.3°! Satala where it crosses the Cimen Daglar1.°°® Alternatively Panaretos’ Accounts of Expeditions to Cheriana, Ulu Strecker gives a route from Asagi Hayduruk east up the Siran.°©? Panaretos gives no details of routes, but mentions valley, which ran as far as Ilag. Here a direct route from

Sorogaina, Soruyana, three days’ fast mding from Cheriana comes down to the Lykos and crosses it to run Trebizond, and Golacha, Colosana.*®* The route between around the northeastern slopes of the Cimen Daglari Trebizond and Cheriana is also attested by Lazaropoulos. In through Chyzyr, Hinziri, and into the valley of the river the twelfth century, the monastery of St. Eugenios used to Balahu below Balahor. In the valley of the Balahu the route obtain butter and cheese from the monastery of St. George of to Erzincan would have joined the road from Nikopolis to

Chainos in Cheriana.*°* Satala in an eastern direction as far as Iskilor, where it would From the Lykos, Kelkit, Valley South to Eriza, Erzincan, in have branched south along the line of the modern motor Acilisene. The bishopric of Chachaion has usually been road to Erzincan. identified with the modern town of Kelkit, but we place it, An alternative route from Cheriana to Erzincan would be tentatively, in the region of Kalur°®° and Asagi Hayduruk.If to travel eastward along the line of the modern motor road our identification is accepted, there was probably a medieval from Siran to Kelkit and then south via Satala. route southward from Kalur and the Lykos valley, crossing A road from Cheriana south to the bend of the Euphrates

the northern flanks of the Cimen Daglari and coming down at Zimara and Pingan 1s attested in journeys made by the river Balahu near Iskilor at the point where the motor Hommaire de Hell and Texier. Hommaire de Hell traveled road crosses the river. From here southward such a medieval from Trebizond across the watershed, via Horducop, road may well have followed the course of the motor road Hortokop; Karakaban; and Stavri to Gumouch Hane, across the southern flank of the Cimen Daglari and down Gumushane. From this place his route to Scheilan, Siran, has into the western end of the plain of Erzincan at Yalnizbag. been described above.°°? An alternative northern half of this route is to leave the From Siran south he passed Zadik Keu, Sadik(?), north of Lykos valley at Kelkit and follow the motor road up the the confluence of the rivers Siran and Lykos and crossed the Balahu valley to Iskilor, where it crosses the river. river, passing through Jenitche, Yenice, to camp out for the There is no evidence for a Roman or medieval road con- night after the first day. On the second day he crossed the necting Satala with Acilisene, but the importance of both Tchak-sou, river Zevker, at Cat and went due south to Kard places suggests that such a road should have existed. The Keu, Kerdagikoyu. On the third day he crossed unspecified likely course for it would seem to be along the line of the desolate country and came down to the Euphrates at Pignan, nineteenth-century military road, along which the Cumonts Pingan, below Zimara, Zinegar.*’° traveled.*°° This ran south up the river valley and crossed the Texier appears to have followed the same route in the Sipikor Daglari at about 2,250 m and dropped into the plain reverse direction from Eruin, Egin, via Kourou tchai,

directly north of Eriza in Acilisene. Kurucay, and Gerdjanis, Basgercenis, to Chairam, Siran; From Cheriana, Ulu Siran, to Eriza, Erzincan in but gives no details of his path.*”! Acilisene or the Euphrates valley. No certain classical or This route from the crossing of the Lykos south to Zimara Byzantine evidence is known for such routes, but a section of and the Euphrates, seems very likely to correspond with the fourth Hadji Khalfa route must travel through this part either or both the Antonine routes, and the Peutinger route, of the country, for it includes Cherdgiamis, Basgercenis, and from their respective stations of Carsagis or Draconis.* ’?

Kemah as stations. These roads are mentioned because a , ;

part appears to coincide with the Antonine and Peutinger From [Souls ourmaina South to Paipertes, Bayburt

We place the site of Sourmaina on the headland to the west

360. See p. 303. Cumonts, SP, II, 321, suggested the region of of the mouth of the river Hyssos, Karadere, not at the Tersun for Haza. Miller suggested Aschuz for Hassis, which might modern Stirmene.*’? The shelter afforded by the headland of

pe [ay Route 9 Ped kos, Kelkit, or Agut south of Kelkit: /R, cols. Araki; Burunu makes the mouth of the river an excellent 361. The Cumonts used it: SP, I], 340—42. They speak of it as the

new military road to Trebizond. 367. See p. 175. 362. Panaretos, ed. Lampsides, 71, 72, 74, 77, 78, 82 and map, 368. See p. 61. AP, 4-5 (1933). 369. Hell (1846), 386-96; see this chapter, p. 50. 363. See p. 308. Chrysanthos, places Solochaina approximately 370. Hell (1846), 396-99. at Kodil. 371. Texier, Asie Mineure (1839), 591. In Tschichatschof (1858),

p. 365.50. See p.C. 170. Ritter. 372. See p. 32.

364. Papadopoulos-Kerameus, FHIT, 86-87, and this chapter, 339, Kiepert refers to particulars of this journey in the hands of

366. Cumonts, SP, II, 340—42, 351-63. This road is marked on 373. This is agreed by Kiepert and by Chrysanthos although

Lynch (1893-98), II, map. neither mentions the ancient site. See p. 323.

54 CHAPTER TWO anchorage, the delta is a wide area of flat land suitable for to the pass across the Soganli Daglar1 at over 2,600 m. No cultivation. The cohort stationed there in the Notitia digni- natural route exists here and the motor road was blasted out tatum, and hagiographical evidence link the place with of the mountain side by the Russians during their occupation

Satala.>’* The Ravenna Geographer has Medocina and _ of this territory in the First World War.

Solodicina connected with Ysilime, Sourmaina, and The southern slopes of the pass are more gentle, and Ofeunte, Ophis, in a list of coastal cities. From Peutinger we typical of this side of the Pontic chain in being barren of know that Medocia and Solenenica are stations on the road vegetation except along the watercourses. At Cuma Venk from Satala to Trebizond, and geographical considerations there was a Georgian monastery which may tentatively be suggest that routes from Sourmaina or Ophis would join the identified as the Rkinis-Palo of Wakhoucht, marking the Satala to Trebizond road in the mountains, where these two frontier between the lands of the Georgian province of stations ought to be found. Geographically, this route makes Dchaneth and Byzantine Greek territories.*°'

good sense since it constitutes; From a direct between the Pharangion, Black , a. Ispir oe, Eskilink Pazar to Syspiritis, Sea and Bayburt, and with the routes south from Pirahmet or ; . . This route across the mountains from Eski Pazar is ;sugZindanlar Arazi in the Kanis, Harsit, valley, to the Lykos , ; 375 ; gested by the ruins of a fort and bridge at the mouth of the

valley and Satala.*’° The Turkish map marks numbers of ; , , Maki Dere, which are possibly of Byzantine date, the situtracks southward over the mountains from ation the Hyssos, ; of the Byzantine monastery of Fetoka, and some re-

Karadere, valley, and there is no obstacle to of travel except thedate. , ; ; 382 . mains at Hayrat which may be medieval height of the watershed at over 2,500 m. This would have ae . ; ; Although it will be seen that much of this route uses the made the winter use of this route hazardous. The existence of ;

; ; or larger more important Kalopotamos valley rather than such a ’route in theand Byzantine period is suggested the Maki, we nevertheless place its start in by the, latter. It

Lazaropoulos’ account of the workwhereas of the Abbot . found no trace of any 376 , / where contains ruins, we have Ephraim, heearly mentions the river Surmena, the crossing mouth of the riveros Kalopotamos, nor was there any Karadere. Chrysanthos says that; atthe from . ; . 377 cL: modern habitation until very recently. It may be that Sourmaina to Paipertes took twoatdays.”’’ Evidence forwhich the , faces ; the . ; , complete lack of shelter the mouth of the river, use of the ,route is provided byworst Fontanier, who observed thatto. ae ge ; ; had directly intoarrived the weather, and its liability disastrous aBayburt, caravan just at Surmene from Baibout, ; flooding, prevented itroute from being useful as aas place of, ;habiand378 by Blau.°’’* The as far south the ; . ; . tationby andKiepert starting as point for a route across the mountains. watershed is .marked having been traveled by ; ; river ;., ; Above Eski Pazar is Halt Tepe and the first village in the Krause. Deyrolle had intended to use this route, but was ; 7 ves valley is called Halt. ‘‘Halt”’ is the name given by the coastal

prevented from doing so by .the kaymakam of Bayburt, people of therobbed eastern Black Sea people because; 379 the French consul had been it region aoccurring fewto the 393 .onfrom across the mountains,’°” and the on name the

months previously.

coast here might suggest settlements of people from across

From Ophis, Of, to Paipertes, Bayburt the mountains in connection with the transit route.

There are no historical remains at present-day Of, or The valley of the river Maki is fertile, with gentle slopes; evidence for the use of this route. It is however given a place the river runs through hilly country as far as Hayrat, and by Janssens*®° and the Kiepert map marks a journey by only low hills separate it from the next valley to the east

Deffner from Of to Bayburt. formed by the river Kalopotamos. A visible sign of the The modern road runs along the river valley, which is steep prosperous countryside is the great three-story konak of the

and narrow, but only rises gently inland as far as the ad- Cakiroglu family who ruled these parts. At Hayrat the ministrative center of Caykara at a height of about 500 m. character of the country changes and the mountains close in From there the valley rises more steeply through forest with on the valley. There are signs that this modern administrative some patches of cultivable land to the last village of Yukari center for the Maki valley is an old site, since storage spaces Ogene and the tea house at Derebas1 at about 2,000 m. The cut into the natural rock in the shape of jars were discovered final climb is up a steep face with over twenty hairpin bends when a track above the village to the south was being wid-

ened. From the region of Hayrat there are three ways of

Winfield, BMGS, 4 (1978), 26. c a joming me yo

374. Notitia dignitatum, ed. Seeck, 85; Brown, Bryer, and rossing the watershed and joining the valley of the

unimportant. th

375. See note 15 above. These stations also appear in a confused Akampsis, Goruh, to reach Pharangion, Ispir. The first Is to land route in Schnetz, Ravennatis Anonymi Cosmographia, 23, sec- take the valley of the river Baltaci from its confluence with

tion 12. Janssens, Trébizonde, 20, notes the route but regards it as the Mak: north of Hayrat, or to join this valley near its 376. FHIT, ed. Papadopoulos-Kerameus, 113: the river tov headwaters at Yente, well south of Hayrat, andt en cross the Svous pass between the Haldizen and Kemer Dagéglari at over PHEVOV.

377. Chrysanthos, AP, 4-5 (1933), 90, following Ritter, 3,000 m. The track forks at the pass and one branch leads

Erdkunde, XVIII, 916-17, who also mentions a route from Yomra south westward to join the motor road from Of to Bayburt,

to pemuatane. 11 10.BI 0). 375 ; , while the other leads more or less due south into the

. Fontanier (182 ), IT, 0. Blau (1860), note I, remarks a Akampsis valley. caravan from Bayburt via Sirmene. More modern evidence is pro- . vided by Stratil-Sauer (1934), 403. He observes that this route is The second way is to make the easy crossing of the hills to 80 km shorter than the Trebizond route. Cf. Stratil-Sauer, 4 P, 29 (1968), 318. 381. Wakhoucht, Géographie, 129. 379. Deyrolle (1869), XXXI, 416. 382. See p. 330.

380. Janssens, Trébizonde, 20. 383. See p. 1.

LAND ROUTES, NORTH-SOUTH 55 the east of the river Mak: at any point up to, or including, his guide had to deliver letters to the Lord of “Aspri,”’ [spir: Hayrat and join the tracks up the Kalopotamos valley. this journey took them one day.*8° Their direct road would The third way is to continue south from Hayrat through presumably have taken them across the Akampsis river and Makitoromanli and then climb steeply up past the monastery over the Tatos and Varos Daglar1.*?°

at Fetoka. The track above the monastery leads south- From Ispir, Clavijo’s party took four days to cross the east at about 2,000 m to join to headwaters of the river mountains and come down to the sea a little way east of Kalopotamos, (Kalopotmos), and eventually the line of the ‘““Xurmenia,” Surmene. He says that the Lord of Ispir gave motor road from Rize to Ispir. The pass at Dagbasi is at them a guide to cross from his frontier into the territory of about 2,750 m, well above the tree line, and leads across into the Empire of Trebizond, and that when he had crossed the the valley of the river Capans, which is a tributary of the mountains he came into the country of ‘““Araquiel,”’ Arhakel, Akampsis. The confluence is about 5 km above Ispir, to the inhabited by Armenians but governed from Ispir. We assume west. Earlier Turkish maps show dans along the track which with Zdanevitch that Clavijo’s route followed more or less preceded the present motor road and it seems likely that in along the lines of the modern motor road up the river this case the motor road more or less follows the course of the Capans, across the pass at Dagbasi, and down into the valley earlier track. This route across the mountains is the hardest of the Kalopotamos. We follow the Zdanevitch map, but not and highest of any so far described, but represents the easiest his text, in assuming that Clavijo left the Kalopotamos valley

crossing from the Black Sea coast to Ispir. in the region of Giineyce and crossed the hills to the west to Clavijo’s Return Route from Allequix or Allexque, come down to the sea in the valley of Maki. Four days is a Aleskirt, in the Autumn of 1405%** (see fig. 3). Clavijo’s route reasonable period for the journey from Ispir to Eski Pazar. from Aleskirt is given here, since we believe that in part it and agrees with the estimate of forty hours which was given

followed paths described above. Clavijo was prevented from to Hamilton more than four centuries later.*?' There is, returning through Erzurum on the direct route to Trebizond furthermore, the anomaly that he says that he came down to because of a Turkmen rebellion. The route is interesting, but the sea at a place on the coast six days journeying to the east there is an immediate difficulty where we take up his journey, of Trebizond. This would place his arrival on the coast since the indications which he gives for his next stopping somewhere in the region of Hopa, which is manifestly implace at ‘“‘Aunique”’ are contradictory. He is specific in saying possible, and we prefer to follow the other indications that he

that he journeyed for four days north from ‘“‘Allequix,” gives; namely that he was a little way east of ‘““Xurmenia,”’ Aleskirt, through barren country to reach Aunique. This Surmene, and that he arrived in Trebizond on 17 September, would theoretically rule out Avnik, which is the identifi- 1405. Working backward, this indicates that Clavijo took cation for ‘“‘Aunique”’ made by Zdanevitch,*®° since it is only one day, or perhaps a little more, for the journey along the a day’s journey at the most from Aleskirt. It would, however, coast, and the six days may be a misreading for six hours’ make good sense for the old Armenian capital of Ani, which sailing which would make better sense. A further point in corresponds with the concentric walls and the importance of favor of placing Clavijo’s coastal destination in the region of the Aunique described by Clavijo, and with the four day Eski Pazar is that he makes no mention of Rhizaion.

journey. However, on the next stage of his trip, Clavijo states From Rhizaion, Rize, to Theodosioupolis, Erzurum.

; ; ; via Syspiritis, Pharangion, Ispir

that he passed by the castle of ““Corcon,”’ Tortum, two days _ , or from Aunique.*8° Avnik and Tortum are indeed two days / , lies in th we f the th apart, but it would be well nigh impossible to get from Ani to The evidence for this route hes in the position © © ares

; . towns theThe needdirect for a road means of communication Tortum in twoand days. runs through Kars, arbetween .i them. The circuit of the walls of Rhizaion, Rize, is large and Bardis, and Oltu or Narman, and crosses more than 200 km P ; ; 392is; . rocopius specifies the importance of the town.”’* There of rough mountainous country. It was only on the third day out from Kars (and, therefore, the fourth day from Ani) that another traveler by horse, Hamilton, came into the valley of 389. R. Fedden, “Four travellers in search of Turkey’s mountain

the river Tortum.387 For the most part we follow the flora, The Times, 18 January, 1969, describes the high pasture

. country of the mountains between Ispir and Tortum.

Zdanevitch interpretation of Clavijo’s route, arrived at in- 390. O. Blau, “Miscellen zur alten Geographie,” Zeitschrift fiir

, , . ; ACTOSS and : i

dependently of his work, but D. C. W. suggests Ani, rather allgemeine Erdkunde, N.S., 12 (1862), 296-99. Blau plausibly sug-

than Avnik as a plausible alternative for ‘“‘Aunique.” gests that Xenophon’s route was down the Tortum valley and then

A.A.M.B. would like to retain Avnik on philological and 1 ih cea. He suggests that Mt. Theches should be Makur Dag

geographical grounds (see fig. 3). . Thus Clavijo would have been following in Xenophon’s footsteps if Two days beyond Corcon, Clavijo reached the castle of he had been able to take the direct crossing rather than go via the “Viger” which Zdanevitch identifies as Kaleifisrik, noted by Capans and Kalopotmos valleys. See p. 54.

Deyrolle.*8® On the next day he left his direct road because 391. Hamilton (1836), I, 220. . . ow,

392. See p. 331. K. Koch traveled from Rize to Ispir in 1843, and

384. Clavijo (1404), trans. Le Strange, 332-36; ed. Estrada, recrossed the mountains further east from Peterek to Athenai,

242-46. Pazar: Koch (1844), II, 11-40, 84—118. Itineraries summarized in J.

385. Ilia Zdanevitch, L‘itinéraire géorgien de Ruy Gonzales de Edmondson and H. W. Lack, *““The Turkish and Caucasian collecClavijo et les églises aux confins de l'atabégat (Paris, 1966), 7, 8, and tions of C. Koch,” Notes from the Royal Botanic Gardens, 35(3)

map. (Edinburgh, 1977), 321-335. This article lists all of Koch’s travels,

386. Clavijo (1404), ed. Estrada, 243. together with a map, and gives a bibliography of his writings, which 387. Hamilton (1836), I, 214-16. seem to be rare. We have been unable to find copies of his maps or

388. T. Deyrolle, ““Rapport sur une mission scientifique en Asie books in British libraries. We are grateful to W. Finlayson for this

Mineure,’” AMSL, Ser. III, (1875), 375. reference.

56 CHAPTER TWO nothing in the hinterland to justify this importance, but a direct. Not far off it are six villages with castle names, so that

route to the interior would do so. the modern road may well mark the general line of the earlier We follow Adontz in identifying Strabo’s Syspiritis, Ispir, route.*?° It comes down into the plain of Erzurum along the

as the later Pharangion, a Persian possession which the river Sergeme and joins the main road from Bayburt, just to Romans took over in the time of Justinian.*?* The produce the west of Ilica.

of its mines would need to be exported, and Rhizaion was its . ;

natural port po From Athenai, Pazar; the Mouth of the River Theodosioupolis, Erzurum, was an important forward tryians: ee Dere; Arcesens OF arrpoula. Byzantine military base which commanded the northern R ates ; ' . d fom the snouths o F the rivers Adienos

access routes from Persia and the Caucasus into Asia Minor. ° inane ne ‘ The supply and communications problem along the length of Pazar, or Prytanis, Furtuna, are suggested by castles at these

,; ll f ; inl fi

Asia Minor to Constantinople was formidable, while the had ae Southward to Svspiritis route to Rhizaion and thence by sea was much shorter. A Pharangion, Ispir A few kilometers inland from Athenai, direct route from Rhizaion ran across the mountains from SiON, SPI. ; ; , ‘ Giindogdu, a few kilometers to the east of the town. This Pazar, the castle of Kise or Cihar dominates a bend in the

followed up the eastern slopes of the river Salor. We have not Athenai eens teat ahetk ty men

explored this route, but it is attested by the han names on the ; ; ‘

Turkish map: Rickmers traveled up it.2°¢ The track joins Sapo to the east, would have run in or along the shoulders of with one running up the Kalopotamos valley on the 3.000 m the Adienos valley south as far as the meeting of tracks at a

. . _° . height of 2 9 in. Here it joined the rout

ridges between Seytan Dagi and Cimil Dagi. A route inland “18 000 m at Asagi Hemsin. Here it join © rome

; ; Kale.*?’

from Bozuk Kale, farther to the east of Rhizaion, would also up the Prytanis valley. Or the traveler may cross at a much

‘ . - , earlier stage intocoast the Prytanis valley, perhaps,passing Kise have started from the at Gundogdu. However, it seems likely that the main former route to a rom the Mouth of the River Prytanis, Furtuna, to

Ispir followed the course of the motor road southward over oo, ; paly ra }

° a . uyuk D motor road, for-

the hills via Karadere and down into the valley of the river Pharangion, Ispir. The river Prytanis, Furtuna, divides into Kalopotamos, below Giineyce.3°5 There are no physical two major branches about 5 km inland from the coast. The obstacles to such a route, which crosses hilly country, rising western branch is the Buyuk Dere, and a motor road, OF toa height of about 700 m. From Giinevce to Ikizdere. where merly a track, runs along the western shoulders of the river

0 a Neg nT OF abou a. ; x ‘ valley and comes down to the river at a second major con-

there is a confluence with a tributary river, the valley only fluence where the modern administrative center for the

rises a little, but from Ikizdere the ascent is through forested or _

Caml: Hemsin is now situated. There is a fine humpgorges to the snow pastures. district, and the pass at Dagbasi Bel at . bout 2.800 m backed bridge across the confluence, carrying tracks up the

a The descent n the south side is through the barren brown eastern river, variously known as the river Hala or Kackar.

° fof6 the . The track Buyuk River landscape valley ofto theIspir rivercontinues Capans toupitsthe confluence ; . or and re-

th the Ak sis at a height of about 1.200 m. and Ispir lies crosses it at Mollaveysi. Up to this point, at a height of about about 5 an dovametream . the east where the valle . ens 500 m, the route 1s an easy one, with well cultivated valley

; y Open slopes, but southward the valley becomes much steeper and out for a short stretch. From the river crossing above Ispir is ; ; . the track climbs past Zil Kale and up to Varos, the highest of

another long ascent up the barren slopes to the pass at about ; 398 There j das f 2,500 m between Cilak and Sulak Tepeleri the Pontic castles.°’® There is now a motor road as far as , South of the pass the land ovens u ‘ato rolling hill Mollaveysi. Beyond Mollaveysi the paved track to Varos is

ae P pe P ; Some now in greatlargely disrepair;ofRickmers recorded avalleys paved track all country which is, , composed pasture inthe thepath the way down the valley. From Varos south and barren or scrub-covered hill slopes. At an averageclimbs height .: of over 2,000 m it differs only from other Anatolian plateau through snow pastures to meet the track from Rize at Asaf)

a, y pia’e Hemsin and from thence crosses the Tatos Bogazi at about

country in being better watered by the many streams which

3,300 m under the eastern flanks of mount Varsambek, now are the tributary headwaters of the Akampsis. The motor ; . . d ther | to th th of Ovacik. which Vercenin, and descends into the valley of the river Salacor.

POAC CTOSSES ANO MET TOW Pass TO Me HOPED OU Nang, Wie This flows into the Akampsis east of Ispir. The difficult strangely understates itself in is thisdescribed land of prominent features, . , ; 399 . a southern descent by Stratil-Sauer.

for it forms the watershed between streams flowing into the

Akampsis and the Black Sea, and those southward flowing | streams which feed the Euphrates and the Persian Gulf. 396. This whole region around the headwaters of the Akampsis There are many possible ways across this high country (worun) is tne area o Tsourmen oF Tehormar given by Basil II to between Isp ir and Erzurum, but the motor road runs fairly mark one actual castle as Harapkale, near Kavaktepe Koyu. This

. : avi € Aouropa ates. see note adove. € 1urkisn maps

may be the same as the Karakale marked on the Tarhan Map, with 393. Adontz, Armenia, 22—23. The identification was made in- two groups of ruins. We have traveled over this road but have not

dependently of Adontz which perhaps gives it more value. explored for ruins.

394. Rickmers (1934), 476-78. 397. Koch (1844), 84-118. Rickmers (1934), 476-78, made this

395. It has been pointed out above that the terminus for this route crossing but does not specify where he crossed. was not at the mouth of the Kalopotamos because of its exposure to 398. For these castles, see p. 341 below. Rickmers (1934), 476-78,

bad weather, whereas Rize has a fairly sheltered harbor. On noted Zil, but not Varos. Giineyce and the Kalopotamos, see Pereira (1969), 34—52. 399. Stratil-Sauer (1934), 406-9.

GEORGIAN ROUTES 57 From the Mouth of the River Prytanis, Furtuna, to Parhal direct means of reaching the Black Sea coast from the well and Yusufeli, or to Hunut, or to Dortkilise. The track up the settled valleys of the rivers Oltu and Tortum.*°? Findikh, Prytanis as far as the fork at Camli Hemsin is the same as for Arhave, and their hinterland as far as the watershed, are now the route described above. Here a track, now a motor road, the heartland of the surviving Laz.

runsPup steep eastern the Inland Kackar to Ilica,Coruh, , biethe steep © Bore may gorge ; Fromof Hopa to River the Akampsis, at above is a second trackmakes up the western arm nd ,place ; . The lie of confluence. the land atThe Hopa it an obvious for formerly Hala, where there are hot springs. A few kilometers Borcka

of the river (Hala Dere, subsequently Kavran Dere), leads up ; ; settlement, but no evidence of classical or Byzantine occu-

into the summerLoe pastures belowfound. mount Kagkar, thelowhighest , pation A route across the mountains of the Pontic peaks. The has lastbeen habitation is the large yayla of , eastward from Hopa cuts out a section of the Akampsis

Kavrun. Thence a rough pass at aboutdangerous 3,400 for mnavigation, leads round oo . . and provided the river could be the southern shoulders of Kackar and down the steep valley oo. . , river ; crossed at Borcka, was a useful of communication of the Hodicor to its this confluence with means the Akampsis a .; . . .;

; with Georgian hinterland. FromanHopa the road climbs little belowthe Hunut. A castle is reported Hunut, whence , patches , ne Leg easily throughatcultivated land ahd of forest to the

easy ascent to Vicer, Kaleifisrik, on Clavijo’s route, and a

. . ; . 400 with Cankurtaran pass at about 1,000 m. The to Borgka junction routes followed by Hamilton anddescent Deyrolle. . . ee _is an easy one, and the river valley at this point is less precipiAn alternative track from Kavrun Yayla leads 404 tous than aboveround or below the the town.*°*.

northern shoulders of Kackar and over an equally rough pass at about 3,250 m to drop into one of the headwater

; ae the GEORGIAN ROUTES streams feeding river Parhal, which is the largest of the . .

northern tributaries of the Akampsis. From the upper Eastern Pontos marches with Georgian and Armenian

reaches, a track crosses a second mountain ridge to the territories. The lines of division fluctuate at different periods east to come down the river Dortkilise, passing its famous and are at the best uncertain.*°° monastery church and chapels, to join the Akampsis at the Routes from Ad confluentes, Cobandede hamlet of Dortkilise, near which a castle stands on a rock These routes are the continuation of the road eastward

One that we kd nade to elim up oe vel from the plain of Theodosioupolis through the region which ‘e vallen . ack escen i , sth foe compara’ y the Byzantines knew as the Phasianes; the crossing of the ae © val cy 0 the r oh ar 4 wit nea d ; + The Erax, Aras, is near the point where the river turns southward. Stopes, alternating wit harch ane cu a , and. T L One route, probably taken by Romanos IV on his march to renecentury Georgian churc ' Parhal, A ‘ Parma Mantzikert, Malasgirt, led southeastward to Chleat, Ahlat, Noy lee i me ma therm ope in valley ie oie 1,500 and the region around Lake Thospitis, Van. ar ast es vb er own t IS vod - cy a - in nae A route more or less due eastward passed under the slopes ° " “th int Byzantine the Ak sma y satel, Her of Mount Ararat; stations along it are listed in the Peutinger he Ake © con no nak ae t © ane ‘table , ‘he Tables, and also by Mustawfi and Pegolotti.*°° ee "bef, Gorul , - cy IS Wide ane ! ospitabdie e the Another Peutinger route continued northeastward to hit crore the me ab ae Pooh acto BOrges, f the Artaxata, and no doubt the tenth-century route given in the wale " only emerges above rtvin. The oe les arm 0 he Armenian source followed much the same course to the later Fiver above Ilica is the Kagkar. A track up it leads past t c capital of Doubios/Dvin, Artasara.*°’ These were invasion poverty-stricken Kackar Yayla to the Six Finger Mountains, routes Alt: Parmak Daglari. A pass at over 3,000 m between these

peaks leads down into the river Parhal and onward to The Main Highway into the Caucasus and through to

Yusufeli as described above.*°? the North

From Marthoula, Findikh, or from Archabis, Arhave, to The natural route runs from Ad confluentes, Cobandede

Parhal and Yusufeli. Neither author has looked at these | Koprt, probably along the line of the modern railway routes but the Turkish map shows a paved track across the through Asagi Pasinler, Horasan, and across the mountains from Findikli and a reported castle near the town | Yagmurludagi mountains to Sarikamis and the plain of may indicate a Byzantine or Georgian settlement. The Kars. From Kars there is a choice of routes to Tiflis. One mountains recede from the coast a little around the Findikl | route would follow the line of the railway eastward through

and Arhave deltas, giving more favorable areas for settle- 403. For the Georgian routes, see p. 59. ment, but there is little flat land. A direct track is marked 404. Koch (1844), II, 144—53. For the Akampsis, Coruh, River

inland from Archabis into the valley of the river Parhal, and —_‘ route, see p. 19. . a it may be remarked that these crossings provide the most 405. Time did not allow us to include a section in the gazetteer devoted to the history and monuments of the region of Tao400. See p. 38 above; Deyrolle, ““Rapport” (see above, note 375); Klardjeti, but D. C. W. hopes to publish a separate account of this

Hamilton (1936), I, 214-20. area at a later date.

401. See Takaishvili, Arkheologicheskaia, 82-87 and pls. 115-30 406. Manandian, Trade and Cities, 190-96. His sketch map on for the church, formerly known as Othta Eklisia; and 87—88 and p. 93 misplaces Mount Ararat, Agri Dag.

pl. 131 for the castle. 407. Manandian, Trade and Cities, 90-100. For the tenth-century

402. For the church at Parhal, see Takaishvili, Arkheologiche- route, see p. 169. For the Peutinger route, see Miller, ZR, cols.

skaia, 93-101, pls. 132-48; D. Hills, “‘Turkey’s richness in old 676—77; J. Schnetz, Ravennatis Anonymi Cosmographia (see above,

churches,” The Times, 20 April 1963. note 15), 4-17, section 2.

58 CHAPTER TWO Leninakan and across the mountains to Karakilise and then Russo-Turkish frontier. From there, a track leads down the

northward to Tiflis. The second would go northward valley of the river of Acaristan to its confluence with the through Ardahan and follow the line of the river Kyros, Kur, Akampsis near the point where it opens out from its gorges to Akhaltzike, and then eastward to Tiflis. From Tiflis north into the flat delta country. From this point, it is an easy ride

there is only one, difficult, route across the Daryal pass. to Batumi. These central routes up the Caucasian peninsula as far as A more difficult route to the sea is to descend into the Tiflis were used in turn by Pompey, Corbulo, Vespasian,*°® Akampsis valley and cross it by boat at one of the wider

Heraclius, and Basil II.+°? reaches of the river in the bend between Artvin and Sirya. The Route from the Black Sea at Phasis, Poti, Inland Thence the track leads south over the mountains to the west

. of theand Akampsis, down toRoute the river Parhal. The route is to=Tiflis, the Peutinger from Sebastopolis

to Artaxata unlikely to have been much used as a means of reaching the

. . , , , sea, but could have served as a means of communication

Versions of the Peutinger route are given on Kiepert’s between Ardanoutzion and the important Parhal valle

map, and also by Miller*'° and Manandian.*!' The routes to anouraton ange Imp y: Tiflis, or Artaxata, or Doubios, Dvin, probably ran along the From Ardanoutzion, Ardanuc, to Ardahan and Tiflis same path up the valley of the river Phasis, Rion. The moun- or to Kars and Ani tain crossing at about 2,200 m is well described by Lynch.*! Ardanoutzion, Ardanug, lies at a height of about 700 m. The roads then separated at Akhaltzike, with the Peutinger From the fertile valleys around the capital it is a long but route continuing to the southeast, while the route to Tiflis easy climb eastward to the inhospitable upland plateaus of followed down the valley of the river Kyros, Kur Cay1. This Ardahan and Kars.*'* The motor road winds around the route inland from the coast is an important one since it hills to find an easy gradient and serve a number of villages, represents the only relatively easy way to cross the coastal but it is possible that the medieval route followed the more

Sea ..

chain of mountains and come into the fertile highland basins direct track up the river Bulanik and passed by the monastery

which alternate with mountain barriers to form the of Rabat. The pass over the Yalniz Cam Silsilesi mountains Caucasian Isthmus. There is a route from Bathys, Batumi, at about 2,500 m was guarded by the now ruined castle of via Ardanoutzion, Ardanug, but it seems most likely that Urumdere, and the traveler then drops easily down to the travelers from both Bathys to the south and Dioskourias to great plain formed by the upper reaches of the river Kyros, the north came along the coast to the mouth of the Phasisin Kur, ata mean height of about 2,000 m. From here one route

order to travel inland. runs northward with the river to Akhaltzike, and thence to

Routes from Ardanoutzion, Ardanuc, to the Black Tiflis, while another route runs southeast through Ardahan and Kars to Ani, keeping always to the easy high plateau

; ; or country. In the vicinity of Kars 1t would have crossed the

There is no natural direct means of communication Pent; . F the plain of E

between Ardanoutzion, Ardanug, and the Black Sea, except cutinger route coming up Irom the plain of Erzurum. by river.*'* However, there was a fairly direct route (above, From Ardanoutzion, Ardanug Southward, into Upper

p. 57) provided that the crossing of the Akampsis was Tao and the Upper Araxes, Erax, Aras, or the Kyros, possible. An alternative landward route to the sea at Bathys Kur, Valleys was rather longer. It led downstream from Ardanoutzion for From Ardanoutzion, Ardanug, to Oltu. The mountains to a few kilometers to the confluence with the river Imerhevi, the south of Ardanoutzion, Ardanug, are high and there are Berta Suyu, and then up that river past the town of the same no easy tracks through them into upper Tao. One medieval name (now Heydancik) to the headwaters of the river. The route probably followed the western slopes of the valley of pass over the Karadagi at about 2,300 m marks the present the river Ucsu up on to a summer ridge road. This stays at around a 2,000-m contour southward until the point where sce Magic. Roman Rule. 356-59. "554 er 1598 Ok AI 34 it could have dropped southeast to the valley in which the 409. Honigmann, Ostgrenze, 160 ff. For a 9th century itinerary of monastery and fortress of Niyakom are situated. From the Caucasus routes by Ibn Khordadbeh, see A. Sprenger, “Die Niyakom a track leads westward past the churches of Taos Post- and Reiserouten des Orients,” pt. 1, Abh. der Deutschen Kari, Tavusker, to the episcopal church at Ishan, or south Morgenlandischen Gesellschaft, III, 3 (1864), section VIII, 57-62, from Taos Kari along the valley of the river Calagutsuyu to

and map no. 8. Idrisi, ed. Jaubert, I], 312-13, gives the distance , . .

between Maldeni, or Melitene, Malatya, and Tiflis as 17 days, but his Its confluence with the river Glaukos, Oltu. From here, routes post stations appear to be confused. On p. 325, he gives a route from lead either south across the mountains to the town of Oltu, or

Tiflis via Qaliqala, Erzurum, to Trebizond but the number of days east and then south to the same town up the valley of the has been transposed; clearly it should read: Tiflis to Trebizond, 12 Glaukos.

days; Tiflis to Erzurum, 4 ; Erzurum to Trebi , , Morkwart "Skizzen aur histrischen Fomcorahie und Gosthihte von A second route probably went up the river Bulanik and

Kaukasien (Vienna, 1928). There is a general description of the crossed the pass between the Capayur and Horosan mouncentral Caucasus region before it had much changed in Klaproth tains at a height of about 2,700 m. Tracks from the pass (1813), and an account of military campaigns in Allen and Muratoff, follow the valley of the river Panazkert down to the impor-

410. Miller, ZR, cols. 649—50, 653-54. ; ; .

Caucasian Battlefields. tant fortress and former town of Panaskert, Panazkert. This 411. Manandian, Trade and Cities, 106-10. stands in fertile country at a little over 1,500 m, with prosper412. Lynch (1893-98), I, 37 ff.

413. See p. 19, for the Akampsis, Coruh, River route. 414. Koch (1844), II, 200-8.

GEORGIAN ROUTES 59 ous villages around it. Tracks lead southwest downstream to to the Tortum castle valley and climbing up barren ravines to the confluence with the river Glaukos, guarded by a castle, the pass at about 2,500 m on the western shoulder of the K1z1l

and thence upstream past other castles and the fortified Dagi. From the top of the pass, which is snow pasture monastery of Pernak to the town of Oltu. Or, it is possible to country, it is a gentle descent through an eroded and treeless cross a range of hills to the south of Panaskert at about 2,300 landscape to the town of Oltu. If the Haho or the Kisha m and come down into the valley of the Komur Cay. This valleys were used, more direct tracks probably led to the valley contains important churches, among them the round Kizil Dagi pass. The southern route may have cut across the church of Bana, Penek, and the fortress of Kahmis, which we hills due west of the castle of Tortum at Oskoy, and left it by identify as Kalmakhi, frequently mentioned in the Georgian Kaledibi castle to cross the pass north of Ziyaret Tepe at annals.*'*> The river KOmir flows into the Penek, thence about 2,500 m. This is the route used by Hamilton, who down a wide valley to join the river Glaukos below the town mentions Id, Narman, and Liesgaff, Liskav, and is now

of Oltu. A track also leads from the Kalmakhi, Kahmis, the motor road from the Tortum valley to Mamrovani, valley and from Panasker up through Dortkilise and Narman.*'? Mamrovani stands in a wide valley at about

Turkesen to Ardahan. 1,700 m. It is an easy road as far as Narman Pitkir; here the

These valleys average about 1,500 m in height and are track leaves the valley to avoid gorges and a long bend in the wide, but with sadly eroded hills to flank them. This was the river and crosses low hills to rejoin it once again as far as the

heartland of upper Tao, and the number of churches and confluence of rivers southwest of Oltu. castles situated in it are evidence that it was once a fertile and The route continues east of the town of Oltu up to the high rich land. The town of Oltu and its fortress were the center plateau of Kars and Ardahan.*?° The track is easy down the from which Basil II directed his reorganization of Tao, since wide and fertile valley of the river Glaukos, Oltu, as far as the

it stands at a crossroads.*!° confluence at Dolgah Koyt; it continues gently up the unFrom the Town of Oltu, Eastward and Westward. From named eastern source of the same river as far as Bana, Penek.

the west comes the continuation of the Lykos, Kelkit— A track for Kars leads from here south up the river Bardiz Bayburt—Ispir route.*!7 It could have come into the Tortum Cay to the town of that name, and thence east over the pass valley by one of several tributary streams. The northern between the Ziyaret and Ortodoruk mountains on to the route came down the river Odiik or Vihik, past the monastery plateau and town of Kars. This was the route traveled westof Haho. A central route came over the high passes and down ward by Hamilton from Kars to Bardiz. But from there he the river GOkdere, past Kisha,*'® or down the river Corman, went down the river Bardiz Cay for four hours, and across passing the monastery of Ekeki, Ekik, and Kisha. A southern the mountains to Narman Pitkir via Tabranek, Tecerek (?). route came down the unnamed river valley, the lower reaches East of Penek the Glaukos enters a short gorge, but it of which are dominated by Tortum castle. From the Tortum seems likely that any medieval route stayed in the river valley, two routes continued eastward to Oltu. The modern valley, which is in general not too steep. The watershed at motor road is the northern route, leaving the valley opposite | about 2,400 m separates the headwaters of the Glaukos, 415. The whereabouts of Kalmakhi was considered unclear by Oltu, flowing into the Black Sea, from the headwaters of the Honigmann, Ostgrenze, 220-21 and by Toumanoff, Studies, 437 ff., Kyros, Kur, flowing into the Caspian. Beyond the watershed 491. We place it at Sogmon Kale at Kahmis, on account of the sizeof _—it is an easy run down a short valley onto the plateau at Gok, the fortress and the fact that its position seems to fit in with what and numbers of tracks cross the hills to reach the route going

little is known of it. Zdanévitch, L ‘itinéraire géorgien (see note 385 south and north through Kars and Ardahan. If Clavijo’s

above), 12, 14, appears to come to the same conclusion without “Auni is Aniand not Avnik. it Id have b b

giving reasons. But while his map suggests Kahmis, his spelling 1s unique Is AN ane not Avni ~ it wou ave been Y one Kamhis, which is a village east of Mamrovani, Narman. When Taiq, of these routes through Kars, Gole, and Oltu, or Bardiz and

the future Samtskhe, became the short-lived early eleventh-century Mamrovani, Narman, that he traveled to Ispir.47' And it Byzantine theme of Iberia, Kalmakhi, Kahmis, has been identified as was by one of these same routes, or by Panaskert, Panazkert, the Kalmouche of the will of Eustathios Boilas of the 1050’s. But we and Kalmakhi, Kahmis, that Basil I] came to Oltu.

cannot otherwise identify any of the names in the will and agree with ,; ; Lemerle that the location of Boilas’ estates should be sought else- From Oltu Southward. A direct route south into the where. See S. Vryonis, ““The Will of a Provincial Magnate, Araxes, Erax, Aras, valley at Hasankale, or Ad Confluentes, Eustathios Boilas (1059),”” DOP, 11 (1957), 266, 165—76; P. Lemerle, Cobandede, went up the river Id as far as its confluence with

Cing études sur le XT¢ siécle byzantin (Paris, 1977), 44-47. Sandwith the Egrek. Here one route turned west to cross into the

(1856), 306—28, went past while Kalmakhi on his journey Kars to . Ardanoutzion. Tortum valley, the southern road wentfrom on up the river 416. Stephen of Taron, Histoire universelle, ed. and trans. F. Egrek and crossed the eastern shoulders of the Kose Dag

Macler (Paris, 1917), 165. mountains to come down to Hasankale. Here the traveler

417. See p. 54. . was on the main route westward into Anatolia via Theo-

418. Identified as Ketzeon by Zdanevitch, but we know of no dogioupolis, Erzurum; he could go east to Tabriz, northeast ruins there except for a small Georgian chapel, and certainly there . are no signs of a walled city. The whereabouts of this town is of into the Caucasus, or southeast to Lake Van. some interest since Constantine Porphyrogenitus notes that the

Byzantines wanted it and that it was a source of supply for Theodosioupolis: DAJ, I, 67 ff. We suggest Tortum Kale or Hinzorik 419. Hamilton (1836), I, 211-16. as other candidates for Ketzeon, on p. 60. We are sure that Ketzeon 420. Koch (1843), 220-48, see note 385 above. must be looked for in these Tortum or Glaukos river valleys since 421. In comparing the Clavijo and Hamilton journeys it 1s interthey are fertile and low-lying in comparison to the high plateau of esting to note that Hamilton took five days to ride from Kars to

Theodosioupolis, and therefore an obvious source of supply. ispir. Ani would have been a further day’s ride from Kars.

60 CHAPTER TWO From Theodosioupolis, Erzurum, Northward into the district of Asisp’eri.*?* From here routes go north to the

Tao Black Sea, crossing the Akampsis and passing Dortkilise, up Such a route would almost certainly have followed the line the river Parhal,*?* or northeast back into the Tortum valley, of the modern motor road across the plain to the north of the and via Ishan and Taos Kari, Tavusker, to Ardanoutzion, city and up the headwaters of the Euphrates, Kara Suyu, to Ardanug. the watershed at the ‘‘Georgian Throat,” Girctt Bogazi.*? A second route from the “Georgian Throat” runs north, This divides the waters running into the Black Sea from those keeping high on the eastern shoulders of the Tortum valley, of the Red Sea. North of the watershed, a direct route leads to Hinzorik, which should be noted as another possible site down past Tortum castle and thence down the valley of the for Ketzeon.*?° From here it was possible to drop into either river Tortum as far as Unguzek castle. Thus far the valley the Tortum valley or the Glaukos valley routes.*7°

affords a natural means of communication. But from

Unguzek northward, the great landslip which formed the

Lake Tortum in the post-medieval period h led th 423. Toumanoll, Studies, 466-67.

ike post-medieval period has concea e 424. See p. 57.

original nature of the valley. North of the lake the river 425. Wakhoucht, Géographie, 125, speaks of the great town of descends steeply, for the most part in gorges. These widen Khendzoreth. M. F. Brosset, Additions et éclaircissements aUhistoire only at the confluence with the river Glaukos, Oltu, where 4% /a Géorgie (St. Petersburg, 1851), 106-7, suggests Ardahan for there is a natural ford and a route from southwest to north- Ketzeon, but this seems most unlikely since it is very distant from

,; : Erzurum and could only grow the same type of crops as that raised

east running past the cathedral of Ishan. It therefore seems on the plain of Erzurum. F. M. Corpi, “The catastrophe of probable that a route from Erzurum northwards left the Kantzorik, Armenia,” Quarterly Journal of the Geological Society,

Tortum valley below Ungiizek and continued northeast past 46 (1890), 32-35, describes a flow of mud about 7 to 8 km long and

the monastery of Ochki, Osk Vank, and across the Harkever 100 to 300 m wide which buried the whole village. Dagi mountains to Ersis, which is a possible candidate for 426. Itis regretted that the most important work of F. Hild, Das

byzantinische Strassensystem in Kappadokien (Vienna, 1977), was

published too late for us to make comparative use of it in this

422. Koch (1844), II, 265—74 for Oltu to Erzurum. chapter.

EXCURSUS ON THE ROUTES TAKEN BY MEHMED II IN 1461 As a demonstration of one of the many ways in which the subject since 1964, but conclusions and errors are his D.C. W.’s Chapter II on Routes may be used, A. A. M.B. own. offers the case of Mehmed II’s movements in 1461. This There were three elements: an Ottoman fleet under Kasim excursus is not intended to provide a history of the fall of | and Ottoman armies under Mehmed and Mahmud. The Trebizond in that year, which has still to be written, for at Sultan did not do things by halves; the fleet was of 100 to 300

least seventy-eight fifteenth-century accounts and contri- ships and carried cannon, and even if the armies did not butory scraps of evidence have yet to be correlated before include the 140,000 cavalry and infantry claimed for them, the equally substantial, but generally unhelpful, secondary they must have been one of the largest forces yet seen in literature is considered. The most important source for Anatolia, posing peculiar logistical problems. As the force Mehmed’s routes is an eyewitness account by Tursun Beg, may have exceeded the entire population then ruled by either

private secretary to Mahmud Pasha, Grand Vizier in of the Sultan’s principal victims of 1461, Ismail Isfen1456-68 and 1472-73. This has not been employed by his- diyaroglu of Sinope and David Komnenos of Trebizond, torians of Trebizond, or even (as Inalcik pointed out) by it is not surprising that there were few casualties and that biographers of Mehmed II.' Nor has the evidence of the no cannon was fired. But, for different reasons, Ismail dating of Ottoman documents been applied to establish the and David were surprised to find their respective capitals Sultan’s movements in 1461, and at least one Greek source attacked. remains to be used. So A.A.M.B. claims the luxury of The Sultan left Edirne (Adrianople) after 23 March 1461, ignoring the secondary literature and reexamines the subject crossing from Gallipoli to Mudanya to enter Anatolia.” He largely in the light of Tursun Beg and D. C. W.’s chapter. He is indebted to Dr V. L. Ménage for making Tursun’s account 2. H. Sabanovic, Krajiste Isa-Bega Ishakovica: zbirni katestarski

available to him, and for guidance and correspondence on Pep 49. gad: Monuments Turck history Savona me Asikpasazade, trans. Kreutel, 219, with additions by Negri (an in-

1. H. Inalcik, ““Mehmed the Conqueror (1432-1481) and his habitant of Bursa), Gihadnniima, Die altosmanische Chronik des

time,”’ Speculum, 35 (1960), 408-9. Mevlana Mehemmed Neschri, ed. F. Taeschner, I (Leipzig, 1951),

ROUTES OF MEHMED II IN 1461 61 reached Bursa after Mahmud, who was there on 21 April.* Palaiologos had ridden in triumph through Constantinople. The Ottoman aplekton of Bursa seems to have been on the After settling things, Mehmed marched back along the coast. plain of Bitinos (Madenos, now Biladiyunus airport), 8 km This was perhaps the most trying part of the campaign. The north-northwest of the town. Mehmed and Mahmud moved coast road proved very difficult and there was no fodder. on to Ankara, where the Sultan held court and documents Kasim had to supply the Ottomans by sea and they lost many

were issued on 12 to 21 May.* Thence they climbed to beasts.© From Canik, Mehmed turned inland to Niksar Kastamonu and marched on Sinope, where they arrived (where he camped between 28 August and 6 September), after Kasim’s fleet from Constantinople had invested the Tokat, and Gerede (formen Krateia Flaviopolis, where he place, and before about 10 July. Ismail of Sinope was ally of was between 17 and 26 September), reaching Edirne, via David of Trebizond, but he was also brother-in-law of the Istanbul before 4 December.’ Niccolo Segundino’s report of Sultan and had reason not to expect Mehmed’s attack. He the fall of Trebizond was posted from Anatolia on 21 surrendered Sinope. Kasim sailed on to Trebizond. Sultan September and reached Venice before 20 October. The news Mehmed and Mahmud Pasha marched south to Tokat and arrived in Rome before 26 October (when it was passed on to Sivas, and east to Koyulhisar, the Akkoyunlu frontier castle, Mantua and Hungary), Volterra before 27 October, and which they took from Uzun Hasan, brother-in-law of David Bologna before 23 March 1462. By the time Amoiroutzes of Trebizond. They then moved east to camp on the Yassi sent his eyewitness account from Edirne to Bessarion on 11 Cimen and north to (or toward) Bayburt. Either on the Yass December 1461, the news was already stale.® Cimen, or on the Barkar, Bulgar, Dagi, Sara hatun, Uzun Mehmed’s route is easily followed as far as Koyulhisar. Hasan’s mother, came to plead with Mehmed and was taken Yassi Cimen presents the first problem, which is curious, for on to Trebizond. On about 28 July, either near Bayburt or on this “broad pasture” was used either as an aplekton, or as a the Barkar Dagi, the armies divided. Mahmud Pasha went battlefield, by a Turkish army on at least four occasions. Yet, first, taking an unusual westerly route, encountering (accord- as Taeschner pointed out, it appears on no map—even ing to Tursun, who was with him) terrible mountains. Sultan ours.” Yassi Cimen was the scene of the great three-day Mehmed took an unusual easterly route, encountering (ac- battle of Ramadan 1230 between Kaykubadh and Celal alcording to Mihailovic, who was with him) terrible moun- Din, which saved the Seljuk state from Khwarazmian domitains. It poured with rain. There was little to choose between nation.'° It may have been used again before the battle of being stuck in the mountains (with the baggage camels), or Kose Dagi, to the north, when the Seljuk state went down to stuck on the coast (where the mud was as sticky as porridge). the Mongols in 1243. Mehmed II used it in 1461. According

But the mountains encountered by Mehmed may have been to Tursun, ““The Sultan ... marched against the land of less terrible than the mountains encountered by Mahmud. Kasim’s fleet had invested Trebizond from after about 13 siege was 32-40 days, and the mountain crossing 18 days): see July. Mahmud arrived there on about 13 August. Mehmed Kritoboulos, Historia, 117-43; Amoiroutzes, PG, 161, cols. 723-78; arrived there on about 14 August. David surrendered hiscity Mihailovic, Memoirs, 119, 224-25 (where the commentator, Svat

on 15 August,® two centuries to the day after Michael VIII Soucek. unaccountably makes David surrender Trebizond to ahmud on 26 October).

TO 6. Tursun, Chronicle, 102.

191; I] (Leipzig, 1955), 274; Kitab-i Cihan-ntima Negri Tarihi, ed. F. 7. Sabanovic, op. cit., fols. 152%, 154’, 157%, 123°; Mihailovic, R. Unat and M. A. Koymen, II (Ankara, 1957), 742. The dating of Memoirs, 121. Mehmed returned to Adrianople by 4 December the letter supposedly written by Mehmed to Skanderbeg from 1461. Giovanni Antonio de’ Colli, copyist of MS Harvard Typ 17, Constantinople on 22 June 1461, hitherto used as a terminus post written at Chios and finished on 13 June 1461 (surely he means 1462),

quem for the start of the Anatolian campaign, must therefore be was an eyewitness of the fleet’s return to Constantinople and Pera rejected: see Barletius, De vita... Georgii Castrioti (Strasburg, 1537), from Trebizond. It then consisted of about 150 ships, with artillery 311-13; Miller, Trebizond, 100. Notes in Sabanovic, op. cit., fols. 93°, freighters and 4,000 cantara (almost 200 tons) of gunpowder which 155", reveal that the Sultan left Constantinople for Adrianople after the Sultan had kept dry. The copyist reckoned the army, which he

3 January 1461 and before 14 March. did not see, at 200,000 men. See E. H. Wilkins, ‘““The Harvard

3. Kulliyyat-i Divan-i Mevlanad Hamidi, ed. 1. H. Ertaylan Manuscript of Petrarch’s Africa,’ Harvard Library Bulletin, \2 (Istanbul, 1949), 283-86, has a panegyric (kKasida) by Hamidi pre- (1958), 320—35.

sented to Mahmud Pasha at Bursa which ends with the date 10 Rajab 8. Lambros, NE, 2 (1905), 331-33; P. D. Mastrodemetres, and chronogram for 865 (21 April 1461). Hamidi, hitherto at NixoAaoc Lexovvitvodc (1402-1464) Bios kai Epyov (Athens, 1970), Ismail’s court at Sinope, was perhaps hastily changing masters; 89: Pastor, op. cit., HI, note at p. 255; Lambros, NE, 2 (1905), Mahmud, as beylerbey of Rumeli, had been mustering the troops of 325-26, cites 26 October 1461, and Babinger, REB, 7 (1949), 205-7,

Rumeli. 27 October 1461, respectively, as being the first mentions of the fall

4. Sabanovic, op. cit., fol. 116". On the plain of Bitinos, where of Trebizond in Italy. See note 5 above. Mahmud Pasha was injured, see the unpublished anonymous chron- 9. Taeschner, Anatolische Wegeneitz, 5. icle in Oxford, Bodleian Library, MS Marsh 313, fol. 139% (and on 10. Ibn Bibi, trans. Duda, 166-71, 328; Juvaini, The History of the manuscript, see the Catalogue of the Persian, Turkish ... manu- the World Conqueror, trans. J. A. Boyle, II (Manchester, 1958), 451; scripts in the Bodleian Library, begun by E. Sachau, completed by The Chronography of Gregory Abu’l Faraj, the son of Aaron, the

H. Ethe (Oxford, 1899-1954), II, 1163-64, no. 2051. Hebrew physician commonly known as Bar Hebraeus, being the first 5. On the date of the fall of Trebizond, see F. Babinger, ‘‘La date part of his political history of the world, trans. E. A. Wallis Budge, I de la prise de Trébizonde par les Turcs,” REB, 7 (1949), 205-7: (Oxford, 1932), 395: ‘““And very many of them fled to Trapizun and which ignores the evidence of S. P. Lambros, ‘H GAmotc ti>¢ to the country of the Iberians, and about fifteen hundred horsemen

TpaneCobvtos kai f Bevetia, NE, 2 (1905), 324-33; which ignores fell from a high rock during the night and died.”; Taeschner, the evidence of L. Pastor, The History of the Popes from the close of Anatolische Wegenetz, 74; H. L. Gottschalk, ‘‘Der Bericht des Ibn the Middle Ages, Il1 (London, 1894), note at p. 255. Other dates are Nazif al-Hamawi tiber die Schlacht von Jasycimen,” WZKM, 56

extracted from Kritoboulos and Amoiroutzes (that the maritime (1960), 55-67.

62 CHAPTER TWO Erzincan.... The army encamped in the yaylak of Yassi east of the Cimen yaylak and five hours west of the Pulur Cimen overlooking the plain of Erzincan and one day’s Dagi. Two stages and eight hours on is Camur (Dyanik; march from it....”'' Following the same Kelkit route in Taeschner’s Tschamur).*" Then comes Tulus (Toloslar; 1473 as he had in 1461, Mehmed may have again used the Taeschner’s Tolos),”? five and a half hours beyond, and Yassi Cimen when he finally came to grips with Uzun Hasan Tercan, Derxene, Mamahatun, two stages and ten hours on the Otlukbeli, Baskent campaign.'? But this time he further east. Murad IV’s route ran from Yar Pinari to Yass1 preferred Tercan, Byzantine Derxene, then Mamahatun, as Cimen, then to a Biryuk Yass: Cimen, and on to Palahor his principal aplekton. Mehmed’s routes were traversed by (now Balahor), each five hours apart. This route does not Selim I, who camped on the Yassi Cimen for a few days from drop down to Erzincan and is therefore not the one used by 18 July 1514, on his way to his great victory over the Safavids travelers, such as Burnaby,?* from Erzincan to Tercan. It at Caldiran on 23 August.'? It may have been used on the provides a Yassi Cimen which lies, as Tursun stated, one day

Georgian campaign of 1578 and was a stage in Murad IV’s from Erzincan but which does not actually overlook the

march against Baghdad in 1638.'4 plain of Erzincan.

Uzungarsili places the Yassi Cimen in the neighborhood of The itineraries lead one to a Yassi Cimen on the north side the Kurutepe yayla, east of Susehri (a vast area);'> Cahen is of the Cimen Dagi, along the Balahu Dere and just west of more precise with “the grassy valley of Yassi-Chimen, to the Balahor and Sipanazat on our Map II. As well as offering a

west of Erzinjan,” but even that does not help.'® Evliya ‘broad pasture,” this aplekton is also something of a stayed at ‘“Jemen, an Armenian village on the plain of ‘“‘Bathys Rhyax”: the river turns water mills and the valley is Erzenjan,”'’ but, while the name may be the same, he was enclosed by steep escarpments (not found on the southern coming southwest from Kemah, Kamacha, rather than from side of the mountain), which, geologically, could (and may the northwest like Mehmed. So commentators have, rea- still) harbor the notorious grotto of petrifaction. sonably enough, seized upon the Cimen Dagi, north- If our placing of the Yassi Cimen is correct, three observanorthwest of Erzincan. But the Cimen Dagi and the Yassi tions, economic, social, and strategic, may be made. Cimen are different, if related, things. One of the features of First, the Yass1 Cimen and the inhabited mountain of the Yassi Cimen was a peculiarly unpleasant spring. In ca. Cimen Dag: formed part of the Turkmen summer pasture 1340 Hamd—Allah Mustawfi cited Qazvini for the report system. Broadly speaking, the related winter pastures lay “that in Armenia in the meadow called Yasi Chaman there is around Diyarbekir to the south. But each area could also a spring where the water gushes forth with such violence that provide unseasonable grazing. The Barriye summer pastures the sound made by it can be heard at a great distance away: are an island within the Diyarbekir winter pastures and, we any animal that falls into it forthwith perishes; and its waters suggest, Yassi Cimen and the plain of Erzincan offered when drunk are violently purgative.” '® This is evidently the winter grazing to complement the Cimen Dagi’s summer phenomenon noted by Hadji Khalfa at Erzincan, “une ville pasture for those who did not make the long trek south and riche en paturages ... On y voit, dans une montagne, une for local transhumants.** At all events, there is strong evigrotte remarquable par ses stalactites: 11 tombe du plafond de dence that the largely Armenian and largely urban economy cette grotte de l’eau qui se petrifie.”'? By contrast, Hadji of Erzincan was being severly eroded by the largely Turkmen Khalfa’s Cimen Dagi lay near Kelkit, “‘une montagne qui and largely pastoral economy which engulfed it in the later sert de campement d’ete ...: elle est appelée ... Tchémen Middle Ages.?° dag ‘y (montagne de la pelouse). Un Ulots turkoman... vient Second, the local economic distinctions of the region are y camper pendant la belle saison.... On y trouve encore strikingly reflected in the social boundaries of the earliest d’autres montagnes habitées.”’ ° But the two real clues to the local defters in 1530. These reveal a swathe of upland settlewhereabouts of the Yass: Cimen are in the Sivas—Erzurum ments which were exclusively Muslim on the Cimen Dag1, itineraries of Hadji Khalfa (cited in p. 34) and Murad IV. Balahu Dere, and mountains north of Erzincan as far as Both place it firmly on the north side of the Cimen Dagi. Tulus, along Hadji Khalfa’s route, while villages on either In Hadji Khalfa, the Yass: Cimen lies four and a half hours side of the mountains and in Tercan were almost exclusively Christian.?° 11. Tursun, Chronicle, 101; abridged trans. V. Ménage. Third, there was historic precedent for camping an ap80 2 Woods, Aqquyunti 131-32; Pitcher, Historical Geography, lekton on the north, rather than the south, side of the Cimen 13. J. yon Hammer-Purgstall, Geschichte des Osmanischen Saeh for ou ‘cventh contury anda bishopric in the thie

, 98, maps XV, XXV, XXXI. = ; “:

Reiches, II (Pest, 1828; reprinted Graz, 1963), 409. on 14. Taeschner, Anatolische Wegenetz, 10. teenth.?” Indeed, Yassi Cimen was in effect the Turkish 15. I. H. Uzungarsili, Osmanh Tarihi, I (Ankara, 1949), 251 note 2. 21. Taeschner, Anatolische Wegenetz, 10. 16. Cahen, P-OT, 129-31. 22. Taeschner, loc. cit.

17. Evliya (1644), II, 202. 23. Burnaby (1876), II, 50-76.

18. The geographical part of the Nuzhat-al-Qulub composed by 24. Woods, Aggoyunlu, 42, map. 2. Hamd-Allah Mustawfi of Qazin in 740 (1340), trans. G. Le Strange 25. Sanjian, Colophons, 179, 206-8; Vryonis, Decline, 259, 271

(Leyden and London, 1919), 276. note 761, 281 note 79, 374; Bryer, BZ, 66 (1973), 334 note 9.

19. Chéref-Nameh ou Fastes de la Nation Kourde par Cheref-ou’ 26. I. Miroglu, XVI. yiizyilda Bayburt Sancagi (Istanbul, 1975),

ddine, Prince de Bidlis, dans I’lidlet d’Arzeroume, trans. F. B. 35-86, esp. map opp. p. 112. Charmoy, I, i (St. Petersburg, 1868; reprinted Farnborough, 1969), 27. C. Foss, “The Persians in Asia Minor and the end of

187. Antiquity,” EHR, 357 (1975),-722; Vazelon Act of 1256; Bryer and 20. Chéref-Nameh, 186. Winfield, BUGS, 4 (1978), 26.

ROUTES OF MEHMED II IN 1461 63 Satala, enjoying the same supply lines to the coast and ad- The tekvur begged for quarter for himself and his household vance post at Theodosioupolis as the Roman strategic base and surrendered his kale and realm... .”’ 34

had done. Only in the last century have the exigencies of Tursun was with Mahmud on the western route and promodern warfare made Erzincan the Turkish successor of vides our only, but certain, evidence that the divided armies

both Satala and Yassi Cimen. took different routes. Asikpasazade speaks for the Sultan’s The next problem is the Barkar Dag, which Tursun makes party on the eastern route, which seems to have been marginthe Fatih climb north of Yass: Cimen, after Sara hatun had ally easier. But all are agreed that it was an exceptionally joined him; it is evidently the Bulgar Dagi on which difficult crossing. Kritoboulos was handicapped by having to Asikpasazade makes Sara hatun plead with the Sultan present his account in terms of classical geography. This (the two versions may not be in conflict).*8 Kreutel, convention allowed no means of describing an aplekton such Asikpasazade’s German translator, identifies the Bulgar as Yassi Cimen, which therefore had to be named after its with the Balabandagi;*? Papazoglou, his Greek translator, nearest town, Erzincan, which in turn had to be called with a Medilis Dagi°® (which we cannot locate); and Tigranokerta (an all-purpose name for an Armenian city Minorsky with a mountain east of Refahiye and between the rather than any actual Tigranokerta).*° Similarly, an omEuphrates and Kelkit.*' None offers an argument for his nibus name for an Anatolian mountain range is the Tauros, identification and Minorsky has least reason to do so. We which for Kritoboulos ran from Mount Mykale to Media, suggest, however, that Barkar/Bulgar is, philologically, more touching most parts of Anatolia except the actual Tauros. likely to refer to the whole range of the Pontic Alps, the But for Kritoboulos this identification at least allowed Parachoathras or Paryadres, rather than to a single sum- Mehmed to climb in the footsteps of Herakles, Dionysos, mit, although in the Bulgar Dag: near Niksar of the Alexander, Pompey, and Timur. Behind the conventions of Melikdanismendname to the west and in Georgia Parhal to Kritoboulos’s account, the horrors of the Ottoman crossing the east, it may have given its name to individual peaks and of the Paryadres are vivid enough.*°® places.*? We cannot propose any particular mountain for the Two factors are clear: that, if they used any recognized Barkar, Bulgar, Dagi, which is unfortunate for it would have routes at all, both Mehmed and Mahmud took unconvenprovided a key to how Mehmed and Mahmud reached tional tracks which may not even appear on our Map II; and

Trebizond. that they cannot have gone on the main highways through From Yassi Cimen, according to Tursun, Mehmed, Mah- Mesochaldia and the defile of Torul Ardasa. David

mud, Sara, and ‘“‘the army moved off towards Tarabzun Komnenos, tekvur of Trebizond, was surprised by the Otto.... When the army, traveling by the Bayburt road, had man armies because he assumed that the central routes over crossed over the mighty and snowy mountain of Barkar the Pontic Alps, described in Chapter II above and Section and come near to the borders of Tarabzun, Mahmud Pasha XXI below, were blocked. They were. They had been in 1456

was sent on ahead by the left-hand way, to prevent the too, when Haydar Pasha of Amasya could only attack emperor [fekvur] from fleeing and blockade him in his fort- Trebizond by sea, demonstrating that a marine siege was not

ress [hisar]. The Sultan with his household troops [Aapu] and enough to take it. But David might have reflected on the Anatolian troops marched by the right-hand way. The Cuneyd’s land invasion of a few years before that, when the Sultan’s plan succeeded: his ‘genius’ [himmet] so favored main passes were also blocked. Cuneyd of Erdebil reached Mahmud Pasha that the pickaxemen [kazmacilar] and Kordyle by the simple process of outflanking the central pioneers [ba/tacilar] hacked a way by that impassable road routes to the west, passing through Turkmen territory. and the voynuks*? were able to descend, between dawn and Doukas observes that it was also Mehmed’s policy to bypass afternoon, from the crest of the mountain to its foot. Mean- resistance in 1461.’ while the ships had come and invested the town, but the infidels, thinking it impossible that the Sultan could approach 34. Tursun, Chronicle, 102, abridged translation kindly supplied

by that [mountain] route, fought daily with the marines by Dr. V. L. Menage. a. a and resisted insoucianly in the fortress (ale, But when 0° ¥APEIZAUE am Kretl 25,26 Keita Hiri

the sound of horses’ hooves came to their ears and the scouts disputed (it could not be far west of Nisibis or at Martyropolis [now appeared, they were unable to flee but barred their gates. But Silvan and formerly Mayafarikin] between Diyarbakir and Bitlis),

when the kale was invested by sea and land and later the but is hardly important because Kritoboulos certainly had little idea Sultan himself came up, the infidels realized that there was of its whereabouts. Cf. C. F. Lehmann-Haupt, s.v. “Tigranokerta,”

; RE; Magie, Roman Rule, I], 1214 note 36; and H. Hubschmann,

no escape. When the guns were set up, the infidels were so ‘Die altarmenische Ortsnamen,” /ndogermanische Forschungen, 16

terrified that there was no need for bombardment or assault. (1904), 473-75. 36. Kritoboulos, Historia, 139—40; ed. Grecu, 277—83. His de28. Tursun, Chronicle, 101; Asikpasazade, trans. Kreutel, 225. scription of the Tauros has the strong appearance of being conven-

29. Asikpasazade, trans. Kreutel, 308. tional, but I cannot trace its source. Perhaps it is from a now lost

30. A. N. Papazoglou, Ma@ayue3 B’ 6 nopSntys Kata tov description in the ps.-Plutarch. Strabo, Geography, XI, 1, 2—4; 1, 15; Tovpkov iotoptkov Aoik Ilaod Zavté, "Ex."Et.BuC.27., 16 (1940), x1, 7, xu, 1, uses the term in a very general sense. On possible

note 4. references to other crossings of the Tauros, see Appian, Mithridatic 31. V. Minorsky, EJ, s.v. ““Uzun Hasan.” Wars, Xx, Xxx; Plutarch, Pompey, xxvii. Neither Tigranokerta

32. Strabo, Geography, XI, xu, 4; xiv, 1; Melikdanigmendname, nor the Tauros figure in A. Diller, “‘Byzantine lists of old and new ed. Mélikoff, I, 159; Honigmann, Ostgrenze, 181 and note 5; Bryer, geographical names,” BZ, 63 (1970), 27-42.

DOP, 29 (1975), 139 note 100; Minorsky, s.v. “Laz,” EI’. 37. Doukas, Bonn ed., 342; ed. Grecu, 428. Not the most reliable 33. Cf.H.A.R. Gibb and H. Bowen, Islamic Society and the West, of witnesses, Doukas takes Mehmed to Trebizond via the Phasis, the I: Islamic Society in the Eighteenth Century, 1 (London, 1950), 54. Caucasus, and Colchis.

64 CHAPTER TWO There is evidence that some of David’s subjects abandoned east of the nexus of routes controlled by Torul and their emperor; nine out of twelve pronoiaroi changed sides Mesochaldia by which Mahmud and Mehmed respectively and kept their lands—but then David himself abandoned his —_ bypassed the Kabazitai.

city and realm and negotiated terms only for his family and The Vizier went “‘by the left-hand way,” west of Torul and household.*® Whether Kabazites, David’s erstwhile suppor- hence started through Cheriana (see Section XIX). Here a ter and pansebastos (military leader), abandoned hisemperor local report may in fact give the clue to where Mahmud

by failing to defend Trebizond is unclear, but he did not climbed. In 1901 the village of Matsera, Macera, now abandon his own ancestral strongholds. Chalkokondyles § Alinyayla boasted the tomb of, and a chapel dedicated to, a states that after the fall of Trebizond, the Sultan sent Haydar martyred bishop Basil of Cheriana. The parish priest of Pasha of Amasya to winkle the pansebastos Kabazites and Matsera then had a manuscript Life of the bishop by the his son out of their lands of Mesochaldia, and Clavijo makes hieromonk Kallinikos Phytianos, which we cannot now it clear that Torul, Ardasa, was one of the strongholds of the trace, but which stated that bishop Basil had been killed Kabazitai.*? This confirms that Mesochaldia and the Torul __ resisting the invasion of the Fatih in 1461.4* This makes defile were still in Trapezuntine hands when Mehmed and sense. Mahmud (rather than the Fatih) may have marched Mahmud reached Trebizond. Nor were the results of Haydar northwest from Yass: Cimen through Kelkit and Cheriana Pasha’s expedition long lasting, for the mountain fief quickly | to Mumya Kale, driving bishop Basil from his see. Thence regained its autonomy astride the main routes south. It was he could not take the obvious route northeast, for it was not until Uzun Hasan’s death in January 1478 that Mehmed blocked at Torul. Instead, we propose that he climbed over was able to send his son Prince Bayezid to suppress the the routeless Cepni pastures out of Cheriana, 3,000 m over principality of Torul, the last vestige of the Empire of | the Balaban Dag: (which Kreutel providentially identifies Trebizond, in 1479-80, and to reopen the main routes which with the Bulgar Dagi) and down to Erikli, Emrek, and nearby had been denied him in 1461.*° The last ruler of Torul was Matsera, where he drove bishop Basil to his death. From called Merne; whether he was a Kabazites or not is unclear, Matsera Mahmud again could not take the obvious route for Kabazitai, father and son, had been deported in 1461.*' northeast, for it was still blocked at Torul, now only 25 km The Pontic ballad which describes how the treacherous away. So he would have skirted the defile by continuing Marthas surrendered the Palaiokastron (presumably that at along Cepni pastures to Kiurtiin and Suma Kale, before Hamsik6éy) to the Fatih in 1461 must therefore be part of _ finally entering Matzouka and Greek territory down the local romance, for Mehmed cannot have gone that way.*? Malaka Dere at Dikaisimon, Macka. For much of this way Instead, we must seek unconventional tracks to the west and he would have been on the lines of Ciineyd’s invasion of a

os ; . 4a. decade or more before, which then veered off further west. Kolonizasyon metodu olarak sirginler,”’ Jstanbul Universitesi 38. O. L. Barkan, “Osmanh Imparatorlugunda bir. Iskan ve Mahmud seems to have kept to Turkmen uplands for as long Iktisat Fakiiltesi Mecmusi, 15 (1953-54), 220. Tursun, Chronicle, as possible before descending to the Trapezuntine valleys. He 102, makes it clear that David Komnenos was interested only in may well have had Cepni guides, picked up in Cheriana.**

1 : Owr e SU. : or)

protection for hse! ane pis family and did not request quarter for In skirting east of Torul and Mesochaldia, “by the 39. Chalkokondyles, Bonn ed., 496: Clavijo (1404), 157.InDop, ‘hthand way,” Mehmed had fewer problems than 29 (1976), 130 note 60, A. A. M.B. noted that ‘“‘there is confusion Mahmud, for there are a number of tracks (see Map II). The over the identity of the last one, or two, Grand Mesazontes [of | question is, how far east he went. Evidently he did not take Trebizond]. N. Iorga, Byzance aprés Byzance (Bucharest, 1971), 57, the favorite Turkmen route over the Larhan yayla.*°> He names a Kabazites as final Grand Mesazon of Trebizond, citing an probably did not take the next route east, down the Panagia

enigmatic ‘““Seconde chronique grecque,”’ which,” he adds, “‘I cannot . ; trace.” A.A.M.B. thinks he can now trace it as the Ecthesis valley either, for it runs beneath the great monastery of Chronica, ed. Lambros, 26, which Iorga appears to have misread: Soumela. There is no archaeological or literary evidence of ‘‘Altamourios” was last Grand mesazon, Amoiroitzes was last proto- Soumela being surprised by a vast Ottoman force, which vestiarios and, as Chalkokondyles makes clear, Kabazites was last surely would have found its riches difficult to leave alone.

pao. See 302 below, and Inalcik, Speculum, 35 (1960), 425: The Short Chronicle of Soumela has twenty-four entries Hammer, Geschichte, Il, 174-75. Tzanicha and ‘‘Mathakhel” between 1253 and 1481, but does not mention what would (Borcka or perhaps Soterioupolis) fell at the same time. have been the most awesome secular sight of any of its 41. Barkan, “Osmanli [mparatorlugunda,” 219 note 88. ‘“Aleksi monks’ lives. The entry for 1461 indeed notes the fall of Istranik” was another defender of Torul. A. E. Vacalopoulos in- Trebizond, but the compiler was under the impression that it corporated some of Barkan’s findings in Origins of the Greek Nation. | .

The Byzantine Period, 1204-1461 (New Brunswick, 1970), 229: they “4S to Uzun Hasan that the capital fell. Even allowing for are not in his original version, ‘Iotopia tod Néov ‘EAAnviopob, I monastic otherwordliness, it is hardly conceivable that he (Thessaloniki, 1961), 304. This is the fullest modern account of the would have made this error if he had actually witnessed the fall of Trebizond, but is so riddled with misunderstandings and passage of the Sultan beneath his monastery.*° inaccuracies that it would be fruitless to attempt to unravel them. Vacalopoulos identified the Merne of Torul both with Kabazites and

the Marthas of Palaiokastro. It is, however, worth pointing out that 43. Kandilaptes, ChP, 2 (21-22) (1946), 505; Bryer, Isaac, and Barkan’s evidence is derived from the earliest Trabzon defter, of Winfield, AP, 32 (1972-73), 227-32. 1487, and that the fall of Torul which it reveals occurred in 1479-80, 44. Clavijo (1404), 120.

rather than 1461, which Vacalopoulos assumes. See Inalcik, 45. Panaretos, ed. Lampsides, 73, 76.

Speculum, 35 (1960), 425; and Lowry, Thesis, 33. 46. This is, of course, an argument against a Soumelan origin for 42. See note 41 and Triantaphyllides, Phygades, 47; and A. its lame chronicle. The chronicle was transcribed by Minoides Vacalopoulos, “Zur Datierung zweier griechischer Volkslieder,”’ Mynas in MS Paris. Suppl. gr. 1248, fols. 35-36; cf. Lambros, Short

Zeitschrift fur Balkanologie, 3 (1965), 4-11. Chronicles, 84-86.

ROUTES OF MEHMED II IN 1461 65 Moving further east, there are two routes which bypass and that Mahmud may have traveled some of his way over Torul and Mesochaldia, as well as much Greek habitation: the twelfth-century supply routes of the same monastery to the Task6prii route which would have brought Mehmed to its lands in Cheriana.*® the mouth of the Pyxites, just east of Trebizond; and the David was perfectly justified in assuming in 1461 that if Hyssos, Kara Dere, route, which would have brought him to Uzun Hasan did not stop Mehmed, the strongholds of the sea at Herakleia in the Sourmaina bandon, 30 km east of Mesochaldia and defile of Torul would, for they were not Trebizond. Both routes offer formidable mountains, of up to brought into Ottoman control until 1479-80. He just over2,500 m. The former is more direct, but that is perhaps looked the possibility that Mehmed and Mahmud might against it, for although Mehmed started his eighteen-day simply outflank Torul and Mesochaldia. Kritoboulos was crossing a day or so after Mahmud, he arrived a day late, perfectly justified in comparing the Ottoman feat in crossing despite the fact that his march seems to have been less his notional ‘“Tauros” mountains with those of heroes of the difficult. This delay might be accounted for by the march past, for the Sultan’s pincer movement on Trebizond exfrom Herakleia to Trebizond. Furthermore, the Hyssos hibits great daring and sound strategy. But Tursun reveals route has a certain historic appropriateness, for it would that it was probably Mahmud, rather than Mehmed, who have revived old military links between Herakleia and had the more formidable task in implementing the plan until Satala, perhaps last used by Heraclius.*’ We propose, there- the Sultan rejoined his vizier at Skylolimne (the Gul Cayir) fore, that Mehmed may have also followed the eleventh- on the eve of the fall of Trebizond.*? century supply routes of the monastery of St. Eugenios in Trebizond, to its lands round Bayburt through Sourmaina; 48. Janin, EMGCB, 266-68.

47. Cf. Brown, Bryer, and Winfield, BMGS, 4 (1978), 22-30. 49. See p. 200.

«

Section I

FROM CAPE KARAMBIS TO SINOPE Although David, brother, “‘herald and forerunner”! of From Cape Karambis to Sinope the Paphlagonian Alps Alexios first Grand Komnenos, reached and held ‘Apaotptc largely dispense with foothills and fall almost directly into (Amasra) and ‘HpaxkdAsia tov Movtov (Eregli) briefly in his the sea. Despite what appears to have been a route along the expedition against the Laskarids in 1204—14, the most west- coast, in use in the twelfth century,’ land communications erly outpost of the Empire of Trebizond proper was, until between the small settlements were still impraticable west of 1214, Cape KapapBic (Kerembe Burunu).* Geographically, "“ABa@vov tetyos ClmvonoAic, Ginopoli, now Inebolu)® in the unassuming headland, on which tunny-fishermen cluster 1972. The coastal villagers watch international shipping hug

with long rods today, is an even more significant limit. the shore but are isolated from the outside world and from Commentators from Strabo to the British Admiralty em- each other by the vagaries of the weather on the cape. phasize that Cape Karambis is the closest Anatolian point There are three indications that Karambis was more than to Cape Sarych, the southernmost tip of the Crimea, the a simple cape. In 1091 (St.) Theodore Gabras, independent Trapezuntine Perateia.* The Black Sea is narrowed here to duke of Chaldia, kidnapped his son Gregory, then a hosonly about 225 km by these two capes, “by which [it] 1s tage of Alexios I Komnenos, from Constantinople. The divided into two seas,” * “which, owing to different winds Emperor’s agents caught up with the Gabrades near what

blowing at the same time in each of them, are often very Anna Komnene describes as a “polis locally called distinct.’ > Cape Karambis is the westernmost extremity of | Karambis.’’? Here Theodore surrendered his son. Anna’s part of the Paphlagonian coast which, stretching almost due reference appears to be the sole literary Byzantine one to a east-west, projects into the the Euxine. The easternmost settlement on, or close to, the cape. But in 1608 Bordier sailed extremity, 140 km away, is the Syrias-Sinope peninsula, past “Carambis, sur la pointe duquel se voient encore les which lies even further north but is slightly more distant from vestiges de quelque ancienne ville, maintenant... déserte.”’ '° the Crimea. Persistent tales that mariners are never out of — Finally Evliya observed enigmatically of the cape that “‘on sight of lands when sailing from Karambis to the Crimea, the rocks are some remarkable inscriptions.” '' However no

are, however, hard to credit.® traveler seems to have landed on, or near the cape.'? 1. Nicetas Acominatus (Choinates), Bonn ed., 828, 842. MONUMENTS

2. George Acropolites, ed. Heisenberg (Leipzig, 1903), 11-12:

Nicetas Acominatus (Choniates) in Sathas, MB, I, 115-26; Nothing appears to survive today on the cape itself. Ephraemius, Bonn ed., 305; Tomaschek, K/einasien, 78, Longnon, — However, about 2 km east of Cape Karambis lies the village

re tn de Constantin eta ping de Moves Par, and kala of Fakasit was in 1972 efetively reached only by pereur de Trebizonde,” JRA/K, 8 (1903), 247-48, pl. xxiv; the same’s ——_—_— To poAvBPdSoBovAAov Tod adtoKpatopos TpaneCovvtoc Aavits, in Sea,” Byzantine Studies — Etudes Byzantines, | (1974), 20-21, gives

DENA, 8 (1905), 237-48, and 10 (1907), 113-56: K. M. Konstant- useful sailing times for the crossing; Eustathios of Thessaloniki opoulos, To Aeyopevov LOALVPSOBOvAAOV Tod adtoKpatopos AaPid, reckoned three days. See also A. Bryer, *“The Latins in the Euxine,”’

in DENA, 8 (1905), 121-30; the same’s To poAvBpdoBovAdrov AaBid XVe Congres International d'Etudes Byzantines (Athens, 1976),

Kopvyvov. ‘Anavtnoic mpoc tov xk. I.I.BeyAepiv, in DENA, 8 Rapports et Co-Rapports, 1, Histoire, 3, p. 4. (1905), 293-322: Schlumberger, Sigillographie, 424-25; V. Laurent, 7. Idrisi, ed. Jaubert, 394; see p. 100 note 55. The coastal route in ‘““Sceau inedit de David Comnéene, libérateur du Pont et cofondateur Miller, /R, col. 634, makes no sense. For a general account, see G. de l’empire de Trébizonde,” AP, 19 (1954), 151-60; Miller, Trebi- Jacopi, Dalla Paflagonia alla Commagene (Rome, 1937). zond, 18; Vasiliev, Speculum, 11 (1936), 25; Perrot and Guillaume 8. Strabo, Geography, XII, m, 10; Arrian, 20; Anonymous peri-

(1861), 18 (for the inscription of David at Pontic Herakleia). plus, 19; Tomaschek, Kleinasien, 78. It appears as I@vOnoAtc in 3. Papadopoulos-Kerameus ed., FHIT, 117; Vasiliev, Goths, Hierokles, Synekdemos, ed. Honigmann, 696.2, the only place listed

187-88. between Amastris and Sinope. No ancient or medieval site, let alone 4. Strabo, Geography, XII, iu, 10; cf. II, v, 22; XI, 11, 12. a ‘Wall of Abonos,”’ is evident there today. Its 19th-century Greek

5. Black Sea Pilot, 396. church is, like that at Ordu, now a prison.

6. Strabo, Geography, VII, iv, 3: Robinson, AJPhA, 27 (1906), 136 9. Anna Comnena, Alexiad, VIII, 1x, 4; ed. Leib, II, 153. It was and note 7; Leaf, JHS, 26 (1916), 4. But reference to the tables in near Aiyiva, which we cannot trace. Cf. Bryer, ““Gabrades,” 176.

Reed’s Nautical Almanach (London, 1971), 320-21, suggests that 10. Bordier (1608), 97. simultaneous sightings of the mountains behind Capes Karambis 11. Evliya (1644), II, 36. and Sarych in mid-Euxine are unlikely. Jonathan Shepard, 12. Clavijo (1404), 106; Tournefort (1701), IT, 150; Hell (1846), “Another New England?—Anglo-Saxon Settlement on the Black I], 339 (‘‘Fakas Skelessi’’).

68 SECTION I sea. About 500 m inland, bearing 100° on Fakas village, is a with a few small pebbles. There are broken ridged tiles in the

rocky hill. A feature projects from an outcrop on its south- area, but no other masonry could be found in the underwest slope. This consists of a brief stretch of banded brick growth. The purpose of this feature is obscure. Locally and stone masonry. The latter consists of three sections of | knownas “‘the Genoese Hamam,” it stands near a spring and regularly-coursed roughly-faced blocks above a footing, di- may have formed part of a fountainhead. It provides, howvided by two bands of brick, the upper of four and the lower ever, evidence of medieval settlement, and perhaps even of of three courses of brick (see Appendix). The bricks are Anna Komnene’s town of Karambis. divided by 7 cm of lime and pulverized earthenware mortar

Section II

DESCRIPTION for it is situated on the neck of a peninsula, pe and has on either Ewan (Sinopolli, Sinub, now Sinop)! stands on the north- side of the isthmus harbors and roadsteads and wonderful east corner of an extensive peninsula, roughly square in tunny-fisheries. ... Futhermore, the peninsula is protected all shape with sides of about 30 km. The northwest corner is round by ridgy shores, which have hollowed-out places in marked by the cape of Zupidsc, Aenth dpa, “the fine cape” them. ... Higher up, however, and above the city (1.e., on Boz (now Ince Burunu). Midway between the twin headlands Tepe), the ground is fertile and adorned with diversified Xenophon’s Ten Thousand were directed, to the anchorage market gardens. ... The city itself is beautifully walled, and is of Apunvn (Armone, Erminio, Mépto Appupd, Eren, now also splendidly adorned with gymnasium and marketplace Ak Liman). Armene itself was of slight account and it was and stoas.” ° In the fourteenth century Al Umar described said that “whoever had no work to do walled Armene.” The Turkish Sinub and Christian Boz Tepe more quaintly. It was bay of Armene is watered by the "Oy 8opavnc (now Karasu).” commonly cas ue ssiane er overs f h yeas8 Mf rar.

° 9 bP) . . a 7 . .

The whole peninsula is made up of low, gently rolling hills. It vn more ‘cin u U an t © ane s of the ful th, he is fertile and heavily cultivated, supplying the immediate . nh and Poot an isthmus he Tbs Bat 0 a ae 4 agricultural demands of the city. The mountains begin to rise sienderest OF loins. CW years carler ton attuta oun to ca. 1,300 m about 35 km inland: they close in on the sea that Boz Tepe then sustained eleven Christian villages and a short of Ayancik to the southwest and at Karousa (Gerze) to hermitage of St. Elias. A lake in the center of the island

the south-east irrigates it; enquiry suggests that no upstanding remains Like Amastris, Sinope straddles a narrow causeway which survive in the area of the military base which now crowns

links what is all-but an island to the mainland. The island, Boz Tepe.

now Boz Tepe (Ilép3anac, M6Céanac—its ancient name is Classical roads ran from Sinope west and east along the not known but later Greeks called it ‘““Karapi”’ or simply Paphlagonian and Pontic Shores, and south Over the vijooc) is roughly triangular, lying west-east, about 4.5 km Paphlagonian Alps to a junction at Thomia (Germanicopolis long and 3 km across at its broadest.* Off its eastern ex- “) pemmaps modern Boyabat.” Robinson noted a number tremity (now Boz Tepe Burunu) is an islet called Sts. Peter ° milestones m the area, but these discoveries can only and Paul or Skopelos (now Gazibey Kayas1): only fishing tantalize the historical geographer, for he failed to state boats can negotiate the channel between it and Boz Tepe. veers he made them; Leaf, however, argued strongly that The isthmus between Boz Tepe and the mainland narrows to these Toutes were not commercially important. Modern a width of less than 400 m. The main harbor of Sinope, the experience of the still abominable roads may support Leaf’s finest on the Pontic coast, is sheltered on its southern side and thesis, which was first implied by Hamilton and Munro.

; ; ; ; 3 e ° . . ° 10

boasts one of the few evident ancient moles on the Euxine. A

i oe ondon, , vol. p. V, ,

lesser cove beaches boats below the walls at their northeast L 5 b Lona 008" ae Ny 60, trans. (adapted) H. L. Jones, corner. By land the two ports are divided only by the walled 6. Al Umari (1342-49), 190 (apparently the only translation of town, but they are separated by over 7 km of Sta." . the Pontic section of this Moroccan geographer’s work, which is here Strabo’s remains the best description of the city: “*Sinope is derived from hearsay). Among more useful genuine travelers’ acbeautifully equipped both by nature and by human foresight, counts not otherwise cited below, see Clavijo (1404), 107; Tafur

(1438), 130: Tournefort (1701), II, 153-60 and the engraving (view

from the east) opposite p. 153; Evliya (1644). II, 38; Beauchamp 1. Skylax, 89, 102; Anonymous periplus, 21-23; Arrian, 21; (1796), 145-54: Lechevalier (1800), 377-81; Teule (1842), I, 433-36; Kretschmer, Portolane, 648; Delatte, Portulans, 1, 238-39; II, 30. and Hell (1846), II, 345-52: IV, 238—40. 2. Xenophon, Anabasis, V1, 1, 15; Strabo, Geography, XII, m1, 10; 7. Ibn Battutah (1332), I, 465-68.

Skylax, 89; Arrian, 20-21; Anon. periplus, 40; Tomaschek, 8. Miller, 7/R, cols. 643—44, 670-71; Bekir Basoglu, Boyabat ve Kleinasien, 78; Miller, [R, col. 643; Kretschmer, Portolane, 648; Cevresi Tarihi (Ankara, 1972), 21. The present Boyabat-Sinop road

Delatte, Portulans, I, 239. is a recent construction and cannot follow the lines of the ancient

3. Tomaschek, Kleinasien, 78: Bzhshkean (1819), 41—45, trans. one. Andreasyan, 30; British Admiralty Chart 2237, inset plan. See also 9. Robinson, AJPA, 27 (1906), 139; Leaf, JHS, 37 (1916), 1-10.

fig. IE. 10. Hamilton (1836), I, 313; J. A. Munro, “Roads in Pontus, 4. Black Sea Pilot, 398: Itinéraire de | Orient (1861), 519. Royal and Roman,” JHS, 21 (1901), 53-55.

70 SECTION II Sinope, then, lies at the head of no major route. Yet it flourished not as a producer or exporter, but as an was a major port—sometimes the major Euxine port— entrepot—and, sometimes, as a pirate center. The most throughout the classical and much of the medieval period. striking evidence of the poverty of the city’s own resources Why? The explanation is surely that while the Sinop-Ince was revealed when, 1n the mid-nineteenth century, direct Burunu peninsula satisfied the immediate and mundane steamship services were introduced from Constantinople to needs of the city, the real hinterland of Sinope was not the the Crimean ports and to the caravan heads at Amisos and inhospitable Paphlagonian interior, but the other Greek col- Trebizond. They naturally bypassed Sinope which, deprived onies of the Black Sea, especially in the Crimea. In this of trade, swiftly entered a decline from which it has not

respect, Sinope was the Venice of the Euxine. recovered. Modern hopes for the revival of Sinopitan comIn emphasizing this aspect of Sinope, one should not, merce, after an unsuccessful attempt to reduce the timber of however, ignore the political significance of its own Pontic Paphlagonia in a now abandoned match factory, have holdings or the economic significance of their produce. On turned to an emulation of Strabo’s tunny fishers, but the the map, the three strips of coastland which obeyed Sinope pattern of modern shipping is unlikely to restore Sinope to its from the seventh century B.c. to the early fourth century A.D. ancient wealth. are insignificant enough.'' But they presage the later sep-

aratism of the littoral and, for over a millennium, constituted HISTORY

what amounted to the earliest Pontic empire—punctuated Robinson’s exhaustive but unreflective monograph on anpolitically by invaders from the south and east and geograp- cient Sinope makes it unnecessary to do more than outline its

hically by the holdings of the upstart cities of Amisos and early history.'* Modern opinion ascribes the foundation of a Polemonion. But in the late twelfth and early thirteenth Gree, colony in the place to about 700 B.c.; in fact the earliest centuries Andronikos Komnenos and his grandsons David material evidence (from a cemetery on the mainland) dates

and Alexios were doing no more than unite a former to around 600 B.c.'* The subsidiary colonies of Kotyora, Sinopitan empire which had long set the distinction of these Kerasous, and Trebizond followed. Pericles inspected the coastlands. The Sinopitan empire consisted first of the colony and Diogenes housed himself there. By then it held a Sinop-Ince Burunu peninsula and the coastland stretching to near-monopoly of the Euxine carrying trade and was its the Halys in the east, where it met the lands of Amisos. The principal emporium. In 183 B.c. Sinope fell to the Pontic second section was a strip from Cape Jason (just east of the kings; already their window on the west, they soon made it lands of Polemonion) to the Sinopitan colonies of Kotyora their capital. Mithridates VII, who was born there in 135 B.c., (now Ordu) and Kerasous (now Giresun). The third enclave was responsible for its first major defense, embellishment, stretched from Koralla (now Gorele Burunu) to the east of and port facilittes—much of the material in the walls of the final Sinopitan colony of Trebizond. The exclusiveness of Sinope described below probably belongs to this period. these Greek settlements, “hems of a barbarian cloth,” | yculius took it in 69 B.C., but it entered the Roman Empire which, as Xenophon found, turned their backs upon their as a free and autonomous city in 63 B.C., a position it retained immediate hinterland and looked to a wider world, set the (arty under the guise of the Colonia Iulia Felix Sinope) for

pattern for all subsequent Pontic development. another three and a half centuries.'° Sinope lies midway between Constantinople and the Pliny obtained Trajan’s permission to investigate the Phasis, yet is substantially closer to the Crimea and its col- possibilities of building a 16-mile aqueduct to bring water to onies than to either. It 1s at the hub of the ancient Greek and the city; it was to be financed by the Sinopitan citizens but, if

medieval Italian trading stations of the Euxine, none of

which (save ancient Tanais, Venetian Tana on the Don) lies —_ at a distance of more than 600 km. d'Asie, IV, 568; Sinop Il Yilligi 1967 (Sinop, 1967), 151-61. The Sinope had, it is true, its own not inconsiderable exports: classical coins of Sinope depict a ship’s prow. In 1461 Ismail oil from the olives which resume their growth eastward from Isfendiyaroglu of Sinope had in. his yards a ship of 900 pithoi, probably the largest then in the Black Sea: Chalkokondyles Bonn this point; the famous “‘Sinoper” or miltos earth from the ed., 489; cf. Bryer, “Shipping,” 7. Probably mistakenly, Nusret southwest; timber (boxwood and wood for the masts of Kuruoglu (in Nazmi Sevgen, Anadolu Kaleleri [Ankara, 1959], Paphlagonia); salted mullet and tunny (which grow large *“Sinop,”’ 280-91), ascribes the building of part of the walls of Sinope enough to be caught profitably here on their gyration round not to “Sebastos” (see below) but to the architect Abu Ali Ibn Abir-

, ; Rakka el Kettani of Aleppo, who was in 1227 responsible for

the Euxine); and lesser items. But such products are hardly Alaeddin Kaykubad’s great naval arsenal at Alanya, which still sufficient explanation for the abundant evidence of Sinope’s _ stands by the sea. Alaeddin (1219-36), however, entitled himself ancient and (to a lesser extent) medieval prosperity, or for the ‘Sultan of the Two Seas” on the grounds that he held both Sinope ship-building industry which, from the fifteenth century (and and Alanya, and a Seljuk shipyard, on the lines of the Alanyan, certainly earlier), was associated with this city.'2 Sinope might be expected. Perhaps it is represented by the great blocked sea gate in the south walls, but there is no other physical evidence for it.

11. The map attached to Magie, Roman Rule, II, best displays the In the seventeenth and eighteenth centuries, the shipbuilding yards

Sinopitan empire. a . were on are Foreshore i ne southeas of an rye Boner v 608),

12. Robinson, AJPh, 27 (1906), 140—44 (excluding listings of iron 101-5; National Bank of Greece, The Gree erchant Marine and livestock for which there ° sneuffcient evidence that Sinope (1453-1850) (Athens, 1972), pl. 36. was the actual exporter); Leaf, JHS, 26 (1916), 1-15 (especially on 13. Robinson, A/PA, 27 (1906), 125-53, 245-79.

miltos; cf. Mary P. Merrifield, Original Treatises on the Arts of 14. Boardman, Greeks Overseas, 250, 266; J. M. Cook, The Painting [London, 1849, reprinted New York, 1967], I, 246); Strabo, Greeks in Ionia and the East (London, 1965), 52-53. Geography, loc. cit., Magie, Roman Rule, 1, 183-35; Cuinet, Turquie 15. Magie, Roman Rule, 1, 183-85; Strabo, Geography, loc. cit.

SINOPE 71 it was in fact begun, it has left no evident remains. '® Pliny did betrayed Karatekin to Alexios I Komnenos and handed the not mention that evidence for Christian life in the city begins port over to Constantine Dalassenos. Karatekin’s lure seems traditionally also under Trajan, with the martyrdom of St. to have been a large quantity of gold and coin of the imperial

Phokas of Sinope. treasury which lay in Sinope.** The place was never a

The evangelization of Sinope by St. Andrew, and his meet- Byzantine mint or theme capital, but maintained a flourishing of the gruesome anthropophagoi there, belongs to later ing customs station, which may account for the treasure— legend—although there may have been a marble effigy of the which was left untouched when Sinope was restored to the Apostle near the city which iconoclasts attempted to destroy Byzantines. Karatekin desecrated, however, the church of in the years 741—75.'’ But the real patron of the mariners the Panagia.?° A century later Sinope became one of the and merchants of the city was Phokas, the gardener and first chief strongholds of the future Emperor Andronikos | bishop of Sinop (the two figures are seemingly identical).'® Komnenos during his Pontic career.?° Whatever the actual origins of Christianity at Sinope (and The history of thirteenth- and fourteenth-century Sinope Trajan’s period is a reasonable supposition), the cult of St. is important, complex, and partly unresolved. But the basic Phokas soon became one of the most popular in the Euxine; feature that emerges is that the port managed to evade conSinopitans dedicated a share of wheat to it and there was an trol by any major Anatolian or Italian power for much, if not annual panegyris in the patron’s honor.'? St. Phokas’ sée most, of the period and that its heterogeneous population

was a suffragan of Amaseia. It appears in lists until the made a modest revival of Sinope’s ancient role as the thirteenth century; the last bishop of Sinope is mentioned in emporium of the Euxine. In 1204/5 the city passed into 1315. Counting St. Phokas himself, literary or sigillographic the hands of Alexios and David Komnenos of Trebizond evidence exists of at least seventeen medieval bishops.?° who, like their grandfather Andronikos, reunited the old Byzantine Sinope received a few attentions of Justinian, to Sinopitan empire as a Pontic entity. But in 1214 the Seljuk whose reign boundary stones are attributed, and in 580 Sultan Izzeddin Kaykavus (1210-19), frustrated of an outlet Tiberios II sent an expedition into southern Russia from the on the Black Sea at Aminsos to match the Seljuk port of port;*' further evidence for the port’s continuing association Antalya, ensnared Alexios, first Grand Komnenos, on an with the northern Euxine is demonstrated by Inscription 5, incautious hunting trip outside Sinope. There was a siege and published below. Sinope became a stronghold of the the Sultan and his hostage negotiated. David Komnenos— if Armeniak theme, the revolt of which in 793 ended with the he had not already died as a monk on Mt. Athos—may have execution of its bishop Gregory.*? The port lay on the outer been killed at this stage. Eventually, the Sultan recognized edge of Arab raids, but in 834 Nasir, called Theophobos by Alexios as vassal of the Pontic territory east of Sinope which the Byzantines, a rebellious Kurdish chieftain of Caliph al- was called “Canik,” like *‘Rum,” after its real or supposed Ma’mun, was proclaimed king by a **Persian”’ garrison at inhabitants—the Laz (Tzannol). The treaty, drawn up by a Sinope, variously estimated at between 7,000 and 30,000 Seljuk notaran, fixed the annual tribute of the tekfur (submen. Nasir-Theophobos hastened to ally himself with king) of Trebizond at 12,000 gold pieces, 500 horses, 2,000 Theophilos (829—42) and, himself betrayed, took refuge with cattle, 10,000 sheep, and 50 bales of precious goods; it was the Emperor in Constantinople in 838. Twenty years later the not as yet a military vassalage and one must suspect exagger-

Arabs again reached Sinope.?* ation in the great scale of this tribute. Alexios and the Sultan The city enjoyed over two centuries of peace thereafter, rode together into Sinope on Sunday 1 November 1214, but the second Muslim conquest of Sinope came in 1081 a date given precisely by both an inscription on its walls when Karatekin, one of Meliksah’s emirs, set up a tiny and and Ibn Bibi. Izzeddin showered honors on the Grand ephemeral Turkish state there. But Cavus, like Nasir- Komnenos and forthwith invited him to remove himself and Theophobos before him, a Christian convert and renegade, the Sinopitan archontes by ship to Trebizond.?’ The walls of Sinope were rebuilt then, they themselves

na ogee Tagg teo A Schutet and R; provide out first evidence for them, other than structural (1906), 125, 245: Magie, Roman Rule, 1, 590, 597. since the works of Mithridates VII. A splendid bilingual 17. Ps.-Epiphanius, PG, CXX, col. 220B; Gedeon, PP, 87: F. Arabic and Greek inscription on tower 38 (see fig. 4) of the Dvornik, The Idea of Apostolicity in Byzantium and the Legend of the citadel records the fact that Bedreddin, son of Abu Bakir,

Apostle Andrew, DOS, IV (Cambridge, Mass.. 1958), 225. See p. 218. completed the works as Izzeddin’s vassal in April to 18. Ch. van de Vorst, “Saint Phocas,” Ana/Boll, 30 (1911), 252-95.

19. N. A. Oikonomides, “Aytog ®mKGc 6 Livonevd The blocks are distinctly long for their width. towers | tooes 4 and “‘cardo” of Sinope, conveniently , , ; 4 .;the Embossed masonry isthe found reused at Amastris and wedged between “‘citadel’’ and classical harbor (see ; ;in; an er : ; ; arch of the causeway to the west gate in Trebizond. In Sinope fig. 4), may indicate the site of the agora. What is clear is that ; ; it appears the lower courses type at least towers and 30, the first ,acropolis ofinSinope (masonry A)ofwas the 29 north,; and 37 to 43—a rectangular block on the only eminence in ern part of the citadel—towers 30 and 37 to 43—-and that the city which we theref, the oricinal i

the citadel itself followed (towers 30 to 43). It may be that © CILY WIMCY WE MAETETOTE PROPOSE aS MAG ONEINAT ACTO Polls

(pls. 2a, 7a, 8b, 11a). Roman to walloftowns abandonwith suchfacing acro- , stones ;; ; ys ; practice Type Bwas consists ashlarbut masonry

poleis. The position of what appears to be the Roman reli- ; ; arranged inofalternate coursesSerapeum of headers stretchers. gious center the city—the (inand the museum .The , ; headers are set regularly, in contrast to the more irregular grounds north and of tower 17) 6 a” behind . headerthe andBelediye stretcherSaray: work atand Amasra the different

and the “palace” even further east—suggests that in thein : . system at Amasya, where headers and stretchers alternate Roman period the center of the city moved east. It seems h Tvpe B is found chiefly in the citadel

highly likely that the east wall (towers 12 to 20) was not on its the same course. TYPE @ Is found chieily in the citadel area

; ; ;site. ; andInalong the main harborthe front, from towerssuburb 29 to 36 (?), to present Amastris (Amasra) main classical :mole ; , “4: : towers | to 6. It could perhaps be associated with the of with the larger public buildings lay on the mainland; at .

; .; great blocks which extends from between towers 6 and 7 at

Sinope, Boz Tepe seems to have been used for this purpose.

moor ; ; about 260°, meet tower and can bemay seenhave today. ;below In Early Byzantine times theto notion ofthe an31modern acropolis . . io water, its blocks reused on mole which extends returned, if only as the microcosm of the familiar larger scale , ; from towers 7 and 12 (pls. 3a, 8b). market on C a plain which moved to a nearby gar- .and ; ; stone ; 54 .; town ; Type consists of had banded walling of brick

Seljuk periods. ) .;

rison fortress.°* At any rate, the old acropolis and citadel of , , . S; ; courses, lying out placeain the inope returned to these functions in theofByzantine and seaofaround ; ;towers 28 and 29particularly and associated with stretch walls between towers | to 6. It appears to have five or more brick

2. The Walls (fig. 4, 2a—11b) courses alternating with eight or more stone courses; the lime The wails form an irregular rectangle in plan, the longer — mortar included pulverized earthenware. A subsection, type

a *C, designated by form rather than masonry and consisting 535 R. E. Wycherley, How the Greeks built cities (London, 1949), of the V-shaped towers 41 to 43, is discussed by James Crow

54. This thesis is difficult to characterize: despite the fact that in a note on p. 78 below. Procopius states that Justinian surrounded the entire polichnion of Type D, represented only in the northeast corner from Sura (now Suriya) on the Euphrates (Buildings, I, 1x, 1), he in fact towers 18 to 21, (see pl. 3b), is now only a rubble core of seems to have walled the acropolis only. This feature of sixth-century abundant pulverized earthenware and brick flecks in the lime

town building is not noted in D. Claude, Die byzantinische Stadt im

6. Jahrhundert (Munich, 1969). 55. Cf. F. G. Winter, Greek fortifications (London, 1971), 80-81.

SINOPE: THE WALLS 77 CONCORDANCE OF MASONRY TYPES AND TOWERS IN SINOPE

Type Letter: 27}/A/B }C {DIE} F/G/H|J;K;LIM

Tower No. 1 B}| C G

23 B |C CG G B 45 B GK B |CCG

67BH CG | JJ

8 F H 9 10 |?

. 12 J L 13 F 14 F\G L 11 |?

15 G L 16 18 D 19 D 20 D M 21 22 M 23 24 M 25 26 M 27 F M 29 A|B C E G L|M 17 |?

28 |?

30 A|B C EG G 3] B C 33 B G

34 B G 35 B G 36 B?; ©G G 37 A 38 A G 39 A 40 A G 41 A #C 42 A *C 43 A *C E

SUGGESTED PERIODS OF TYPES

A Pre-Mithridatic or Mithridatic D,E,F Early Byzantine to thirteenth century?

B- Mithridatic or early Roman G_ Reconstruction of 1215; Seljuk

C Roman: *C fifth century?—-see p. 76. H,J,K Isfendiyaroglu to 1461 L,M_ Ottoman after 1461

NOTE: A tower number includes the walling to the right of the tower proper, which may be of a different build from it. Not all masonry types are

included, and all assessments should be regarded as approximate.

78 SECTION II mortar, which is well packed with few air holes and without of 1215. It is probable that type B represents Mithridatic beams. There are a few heads of reused columns. work. It is difficult to pick out much damage by the bombardTypes E, F, and G employ material from types A and B. ment of 1853, but the sea-gate between towers 33 and 34 They are distinctive, but difficult to place in order of building. (one, at least, was built after the gate was made) appears to Type E consists of a facing of reused blocks from A to B, have been blocked after the Russian bombardment. with a mortared rubble core to the wall. The header and Mr. James Crow kindly contributes the following note on stretcher system is still employed, with use of columns as the V-shaped towers 41 to 43:

incorporates pulverive’ ear thenware The tpe i voond On a section of the north wall of the citadel and immediately

; , east of the massive northwest corner tower 30 are four small,

chiefly in the middle-lower courses of towers 29 to 30 and 43. closely spaced, V-shaped towers (41 to 43—one is unnumbered

It is associated with Inscription 5 (pls. 8b, 46). on the plan; pls. 7a, 9a). This form of tower is unusual anyType F, found chiefly in towers 8 and 9, 13 and 14, and 27 where before the invention of gunpowder and the Sinope group includes elements from A and B, with some brick and spe- is probably unique on the Pontic coast. The towers and curtain cially cut conglomerate stone in sections where the reused are similar in construction to the work of the north face of

material does not fit. tower 30, with coursed tooled ashlar, incorporating a number

Type G, acommon one, is a variant of F, using no headers of reused architectural fragments. About 2 meast of tower 30 is or stretchers, but some brick and very large blocks from A a clear straight joint, but this represents a distinction of struc-

and B. It is found chiefly in towers | to 6 (see pl. 3a), 14 to 16, den rather an phase. on the ne iene a es Lk, 30 to 38, and 40, often in conjunction with standing A or B Mithridatic construction of the tower nee A and 3 This and in association with the group of inscriptions (including defined by the use of isodomic, quarry-faced, ashlar, taken in

that of 1215) in towers 37 and 38 (pls. 8b, 9a, 1 1b). conjunction with the chamfering of the corners of the tower, Type H, found only in towers 7 to 9, employs rough blocks strongly suggest a Hellenistic date.*® The masonry of the curwith a soft white lime-and-pebble mortar and a little brick or tain with V-shaped towers is clearly later than this and may be

tile infill. assigned to a late Roman or post-Roman date.

Type J consists of regular courses of square stones with Both the form and construction of the towers provide terlittle mortar—something on the lines of the fourteenth- mini post quos for the curtain with V-shaped towers, and this century masonry at Kordyle (q.v.). It lies below type L in chronology can be narrowed further by reference to parallels tower 12, and above types G and H in towers 6 and 7. from northern Greece and Bulgaria. In form, the V-shaped Type K consists of strikingly black basalt blocks; it is tower Is closely linked to the more common pentagonal Prowe

shaped tower, and both seem to have been introduced in the

found only on the added prow-tower 4. mid-fifth century A.p. in the Balkans and on the eastern frontier Type L is faced with courses of fairly small irregular of the Empire.5’ The closest parallel to Sinope is found in the blocks, with only a small proportion of reused ashlar; there is west walls of Thessalonike, dated by Vickers to ca. 450, a substantial mortar facing, beam holes, and brick flecks in amongst the earliest examples known.°® These match the the mortar. It is found in the original northwest gate in tower close spacing and small dimensions seen at Sinope; and the 29, with its high rounded arch, in upper stories of gate towers at both are closer in form to buttresses than to the large towers, and in arched windows, especially over type G, in flanking towers commonly found in late Roman fortifications.

towers 12 to 15. Other examples of V-shaped towers are considerably larger Type M is chiefly represented along the north sea wall by a and are restricted in distribution to the eastern Balkans: at

. ; Sofia, Plovdiv, Varna, and Kjustendil.*? None of these facing of small rough-cut in regular courses, .than ; four 3 . ; can be more firmlydark datedgrey by stones archaeological evidence the

with interstices filled with smaller stones; the mortar contains fifth-sixth centuries, and have consequently been assigned to

much pebble and a trace of brick. The core is strengthened Justinian. Apart from Varna, the only other example known with stringers and tie-beams in the form of crib-work. A date from the Black Sea is a Kaliakra, where, although the apex of contemporaneous with type L 1s suggested by plates 4a and the V has been squared off, the scale of the towers is closer to

lla, where the lower courses of the wall have a facing of that at Sinope.©° A date similar to that of the other Bulgarian reused ashlar blocks, while the upper section is faced with examples has been suggested. On the basis of this comparative small rough-cut stones—but the core of the two sections evidence, the examples from Sinope would seem to fit into the

appears to be of one build. The towers in this north wall (20 same context. to 27, 29) protrude only slightly on the interior and a gate 56. Winter, Greek fortifications, 196. gives access to their second storey from a catwalk (pl. 6a). 57. S. Bobchev, “‘Krepostnite kuli s izdaden oster reb i znacheThe only decorative feature here is a blind arch with stone nieto im za ukrepjavaneto na antichnite gradove, ‘/zvArhInst, BAN

voussoirs (pls 5b). 88 Vickers “The late Roman walls of Thessalonica,” Rom

This typological program IS very tentative, but we can at Frontier Studies ‘1969. Eighth International Congress of Limes.

least suggest that the earliest acropolis(type A)isrepresented —_forschung, ed. E. Birley, B. Dobson, and M. Jarrett (Cardiff, 1974), by towers, 30, 37 to 43 (although the V-shaped towers 41 to 251; O. Tafrali, Topographie de Thessalonique (Paris, 1913), plan.

43 as they stand may be fifth-century work—see the note 59. D. Ovcharov, “‘La trés ancienne edification des forteresses below); that the classical citadel attached to it is represented a4 aa dans nos terres (Ve-VIle s.)," JzvistDr, 24 (1974), towers 30 to 36; that the east wall is, as it stands, compara- 60. I. Zachariev and VI. Vladimirov, ‘‘Parva ukrepitelna linia na tively late; that the northern defenses, as they stand, are even nos Kaliakra,” /zvestija na Nazionalna Voennoistoricheski muzei, |

later; and that type G probably accords with the rebuilding (1973), 180.

SINOPE: THE “PALACE” 79 The absence of a number of Black Sea towns, including Area XV is a further rectangular space, of which nothing Sinope and Mesembria, from Procopius’ Buildings may be can be determined without excavation. Areas II and III significant in providing a more secure date for the examples correspond to areas IV and VII on the north-south axis. It from Sinope. At Mesem or ananene has shown, - re may be, therefore, that areas IV and V were continued to the

bas! s of archaeological evidence, that the construction ot the west, forming a second cruciform hall in this area, with the

main gate was in the later fifth century, rather than during a IV] b f f 5 that th

Justinian’s reign.°! Epigraphic and other evidence from the Crossing mn area 1¥. It can be seen Irom Bure that the Dobrudja and Cherson® indicate that Mesembria was not masonry piers at the corners of area IV were thick enough to

alone in receiving imperial attention in this period. The close have supported a groin vault. | parallel between Sinope and Saloniki makes it quite possible The ruins are of one build of banded brick and stone that this building activity was not restricted to the north and courses, with a mortared rubble core. Four courses of brick west coasts of the Black Sea, and that V-shaped towers 41 to 43 alternate with four courses of stone; the walls average about at Sinope belong, as they stand, to the late fifth century rather 1.58 m in thickness. The facing stones are limestone blocks than to a later period. Without more specific archaeological or averaging 10-15 cm in width and height and 23 to 41 cm in

Oe evidence this eating d Ana a hypothetical partof —_jength. The bricks are 40 to 44 cm square and 3 to 4cm thick;

the Black Sea policy oF Leno and Anastasios. they are well made of a light red clay. The brick courses run

3.sae The “Palace” right bricks throughisthe the wall (pl. the mortar eee ; ; between of thickness about theofthickness of a16b); brick—3.5 cm. About 300 m southeast of the city walls are a series of ruins A f brick 30 to 35cm in height and

locally known as Balat, or Mithridates’ Palace.°* They stand set of Drick courses measures 9\ [0 > em in height anc’ a set

ythe , yfinal of stone courses 60land to 65 cm, the including the top layer of on stretch of level before ground rises to ; ; ; ar ; Boz T Th hich th ; df mortar in each case. The mortar is of lime and sand; bricks nde i sth resent oneroune ie - 2 “l wy ve enone tha 1 5 i and stones are wellofbedded with few air spacescourses in the core. Bareens an © A change in the rhythm brick above the level of the ,original floor—as can betheseen inand thestone ; ne . yen . 1s

hurch in th ter of the ruins. The rui tend evident in the higher register above the blind arches in the

cnr’ rab ‘ 10.000 « “ Qe © Fuins exXtenc over ai walls of areas VIII and X, where a brick band is omitted, so

: The anon t , iv me F thi kabl that there is a set of eight stone courses in area VIII before the Ane true ex ent and original purpose of this remarkable next brick courses, which mark the springing of the vault (pl. ruin can only be determined by excavation; it is not now clear 12b), and a set of nine stone courses in the wall of area X (pl

how far the surviving walls represent the original scope of the 13 , A similar j ath ber of building complex. See figure 5 and plates 12-17 ). A similar increase in the number of stone courses apTh 6 -" t , P xX XII XII 40 pears wherever the walls stand to a sufficient height; this may ore le th 4 np angular th, A xX d XII. or therefore have been part of the regular system of building. sich b vig ane dk we 13, 14, iN SCT . “ib ' 0 re There are only two exceptions to the rhythmic alternation of

cacy abour 8S NWI e (p s. 13, 14a). Area isabouw im brick and stone throughout the ruins. At point Q (pl. 16a), a wide and terminated in an eastern apse or exedra, probably set of brick courses ceases for no apparent reason; stone

with a masonry semidome. seems unlikely that areas ; PP , the ; ——Itcourses replace them, amounting to aX setand of ten; and at

XII were ever entirely roofed over, but pilaster strips the ; ; wall I to J large rectangular blocks are used, in averaging north wall of area X (pl. 14a) suggest that it may have been a 80 x 30 This wall alland k

courtyard with a peristyle. Area XIV has part of a semi- nS om Ais ee ne acts as a terrace wah Ee METAS @ circular wall with niches and could have formed an exedra rise in the ground level of about 4 m. Wall K is also a terrace

th d f XII Area XIII have b wall, with a ground level about 2 m higher to the east of it.

™ f dow ome - area | re ‘d ire bie , th 4 There is no trace of any regular lacing of the structure with rooted over, Since It was only ty Mm wide, probaory Wii a beams. The only signs of the original woodwork are four timber roof, for the wall is too thin to carry a masonry vault. b hol aging of th It ofarea VIII (ol

Area VIII (pl. 12b) has a cross-shaped plan with the south arge beam holes at the springing of the vault of area VIII (pl.

/a; squat 12b). These may have heldAdjacent tie beams to the arm elongated to form cruciform space. 1help y : maintain pms they ; a equilibrium of the walls and vaulting, or )(more likely)

chambers act as buttresses for its walls, and it is clear thatstage this .and were used only to erect . belong to the constructional great hall was vaulted. Apart from their structural purpose as the centering of the vaults

buttresses for area VIII, the functional purpose of the com- In the oath wall of area X there are a number of beam plex of smaller chambers is not clear. The only area where the holes. but their positions are irreeular and mav date from a

roofing survives in its entirely is area I, later used as a church or P d: 6 , 7 ,

“ the original walls.

ofia, , 155-56. . .

and described below (p. 81 ff.) later period, when secondary structures were built against The walls are characterized by a marked and pleasing

Gof . 969). 155. 56 Nessebre, 1, BAN Arkheologicheski Institut surface articulation of blind arches. Where there was a high 62. I. Barnea, ‘Contributions to Dobrudja history under Anas- wall surface, as m areas VIII and probably X, there is a

: tasius I,” Dacia, 4 (1960), 363-74; CIG, IV, no. 8621. second storey of blind arches. At point A on the plan there is 63. W.E. Curtis, Around the Black Sea (London, 1911), was told a single blind arch, and at AB there is a second blind arch : that Mithridates here killed his wives and sisters to prevent them above the first. At points D and G there are blind arches at

at E Akurgal an sew eS gerichs ther ate Auserabungen on second floor level and there was probably a similar series Sinope (Ankara, Turk Tarih Kurumu Yayinlarindan, V. [14], 1956), 39 below, but the wall surfaces are so damaged or covered by ivy

and pls. xx, xxi; and the sames’ “Sinope,” TiirkArk Derg, 6 [1] that it is difficult to determine their aspect. The reveals of the

(1956), 47-61; 6 [2], 5—10, esp. 8. blind arches are straight and not broken by any receding arch

80 SECTION II to frame the opening. The blind arches in the north wall of Among secondary constructions, modifications in area I area X were about 2.20 m wide and they have nine or ten are discussed below. There are three large semicircular stone, and no brick, courses in the recessed wall. The func- arched cuttings in the long north wall of area X (pls. 13, 14a). tion of the blind arcading is difficult to determine. Forming a The regularity of the cuttings and the fact that the easternvery pleasant visual pattern in themselves, they could con- most corresponds to a similar one in the south wall (making

ceivably have held statues. The voussoirs are all of brick. an arch into area XJ) and the extraordinary labor which they The interior surfaces of the arcading retain fragments of | would have required suggest an adaptation of the complex to two layers of plaster; the lower of lime and pulverized earthen- a new use at a fairly early date. one semicircular cutting was ware, the upper of very white lime with a small quantity lined with a plaster of lime and pulverized earthenware. The of chaff as a binder. The walls of area I (the church) have an other hatched openings on the plan do not appear so regular, original plaster base of lime and pulverized earthenware but they may all belong to the same period of reuse. The only covering the whole surface where visible. It seems likely that functional feature of these archways is that they open up a all the wall surfaces of the “‘palace” were plastered in this number of new means of communication between the areas. way, for the banded brick and stone has a very rough finish. On the west wall of area II are traces of two later roof levels The lower parts of the walling may have been enriched with of gables. marble revetment panels since fragments of thin slabs of At point N is a careful cutting of a semicircular niche with verde antico with a high polish on one side are found in large semidome roof. This has been partly walled in front with a

numbers in the area. sheet of fine black and white marble; the local explanation

The only sign of an upper-storey window is in the west wall that it was a water stoup is reasonable. of area II (pl. 14b). This would have stood about 6 m above Plate 15b shows a repair high in the wall where a fragment, floor level. The opening is large enough to be a doorway but probably from a templon screen, has been emplyed as a sort it is better regarded as a window as there are no signs of an of quoin. The position of the fragment suggests that it was

associated stair or floor. intended to be decorative and it may therefore have been

At point H there are three round-arched niches with brick placed there after the collapse of the original vaulting. voussoirs. They are semicircular in plan and are part of an Perhaps this, and the water stoup, are connected with the use apse-shaped wall which could have been a raised exedra to of area I as a church.

the east of area XII. The “‘cisterns”’ lie about 100 m south of wall K. They are

Round-arched openings with brick voussoirs are found at four rectangular stone-built structures which were originally points E and F (fig. 5, under the springing of the vault). These covered with stone barrel vaults. Their walls contain stone of are archways between the chambers: points E are high arches all shapes and sizes, including reused blocks; the voussoirs of

and points F are lower arches which probably formed the the semicircular arches are thin slabs of stone employed as if

frames for doorways. they were bricks. The vaulting appears to have been made in

Both open archways and blind arches are semicircular the same way, with stones used like brick. The walls bear a with a single row of brick voussoirs and no recessed molding. very hard plaster to about 4 m above the present ground Exceptionally, at points E, S, and P, a double row of brick level-which may be about 2 m above the original floor. The

voussoirs form the arch. wall is recessed about 20 cm at the level where the plaster

Arch IV F was provided with shuttering and the marks of stops, and here are traces of openings which could have been

the wooden frame can be seen in its mortar joints. doors or windows. Point T marks the archways connecting The south arm of area VIII retains the springing of a vault the rectangular chambers; points U mark late openings cut in stone and it may be conjectured that the four arms of the into the outer walls to give access to the interior, which are

cruciform space were roofed with concrete and masonry now used as cowsheds and as a garden. A house has been barrel vaults which formed a groin vault over the crossing built against the wall of the southern rectangle.

(pl. 12b). The heavy plastering, the lack of access into these chamAt the same high level, area V M (pl. 17a, b) was roofed bers and the thickness of their walls suggest that they were with a concrete and stone barrel vault; the broken edge of a built as cisterns. Like the “‘palace,”’ they lie well outside the piece of cross-vaulting can be clearly seen jutting out into present town walls, suggesting a period of security for their area IV. Perhaps area IV was, therefore, also roofed with a building. There is a lake, but no springs, on Boz Tepe, which groined vault formed of intersecting barrel vaults. However, could have fed them; rainwater could easily have been chanthe springing of the vaulting over areas II, III, and VIIisof — neled into them.°* They stand well above the town and could brick; these areas may have been entirely vaulted in brick. have conveyed water to it by simple gravity, without compliThe only vault which survives intact is the brick one over area cation of pumps. I (see below); the vaults over areas III and VII were probably The complete absence of brickwork suggests that the ‘‘cisat the same height. Parts of brick barrel vault survive at a terns” may be of different date from the “palace” complex, lower level at points L; they may have acted as relieving with which they appear, however, to be aligned. But there are arches for the higher vaults (pl. 15a). They carry traces of the same two layers of plastering in the blind arches of area VIII 65. cf. T.S. R. Broughton, An economic Survey of Ancient Rome,

which have been mentioned above. IV (4) (Baltimore, 1938), 778. This ts not, of course, Pliny’s proposed

vault of brick. built.

The smaller area VI is also roofed with a low-level barrel scheme of waterworks, and may well confirm that they were never

SINOPE: THE “PALACE” 81 no reliable features to indicate a date for their construction. palace it is difficult to ascribe a governor or official to Sinope The church in area I of the “‘palace” lies in what was in Roman or Early Byzantine times who would warrant such

originally a rectangle with arched entrances in the north, a building, although it must be remembered that there apsouth, and west walls. The arches are semicircular and have pears to have been a branch of the imperial treasury in double rows of brick voussoirs. The church is roofed by a Sinope in the late eleventh century. The analogy of Miletos, brick barrel vault about 6.90 m high at the crown. It had a called Palatia by Byzantines and Balat by Turks, is not floor of limestone blocks about 1.07 m square and 18 cm encouraging, for it seems to have derived its name from the thick. Three of these were still in place. The walls of this massive classical theater (later transformed into a Byzantine original rectangular chamber were plastered with lime con- castle) which dominates the city, and not from a palace

taining pulverized earthenware. proper.

The conversion into a church is marked by the cutting of A second explanation is that the complex represents a the shallow apse S, and it may be that the low central niches gymnasium and baths complex.°® The cruciform hall apof this apse were cut out at the same time, together with the pears in the baths of Titus (if Palladio’s plan is correct), prothesis niche in the north wall (fig. 10). It is possible that roofed on three sides by barrel vaults which meet in a groined the arch in the west wall was also modified at the same time vault at the crossing.©’ This was the frigidarium; the tepi(fig. 7) by hacking out the lower 3 m of walling and shaping darium, a T-shaped building, lay to the south of it. The great the upper part of the cutting as if to form part of a low wagon Hadrianic baths at Lepcis Magna have the same features. In or barrel vault. A cutting (at a level and of a shape which the Sinope “palace” area VIII 1s a cruciform hall which was would correspond to the modifications of the west arch) has probably roofed with intersecting barrel vaults; the smaller been made in part of the north wall. It is about 30 cm deep area II to V, and VII, to the west of it, is also cruciform. These and runs for 3.30 m to the east of the archway. It looks as ifa areas could have corresponded to the frigidarium and tepisecond and lower internal roof, with a barrel vault of ma- darium, while the adjoining large rectangular areas X, XII, sonary, projected here, but the absence of any equivalent and XIII would have been either a xystus or gymnasium cutting on the south wall suggests that the project was never rooms. The date of the banded masonry might be of the third

completed. or fourth century A.D. The Byzantine baths of the sixth A second modification to the north wall was the blocking century at Ephesos, with their numerous apses, have someof the archway P R with a thin filling, leaving only a small thing of the same layout but are more modest in scale and do rectangular window in the wall. A rectangular window is also not employ so much stone.®8

cut in the west wall, just below the vault. It may be that these If the ruins are those of the baths and gymnasium of window modifications are contemporaneous. The rectan- Sinope, placed, like the enormous “Bedesten”’ or gymnasium gular window of the apse belongs to a second modification; of Amastris,°’ outside the walled city, it seems likely that the the apse was reshaped when part of the original cutting was “cisterns” are associated with them, for they are ideally blocked up and further cuttings were made in the semidome placed as reservoirs for baths. The banded brick-and-stone

around it. masonry is hardly likely to belong to the gymnasium menThe untidy fill in the second modification of the apse looks tioned by Strabo, and it may be that the “‘cisterns’’ survive very similar to that in the north and south archways. If this is from a both complex earlier than that now standing. so, we have at least one point of reference for modifications. However, the masonry of the “‘cisterns” points to a much The original south archway was filled in to leave only a small later, rather than earlier, date. rectangular doorway and a small rectangular window. The The secondary use of the ruins can only be determinded by painting over this filling fits in with its shape (fig. 12)— excavation, but the massive scale and regularity of the cutnotably the figure of St. Marina and the inscription. It seems tings suggest an early date, Justinianic or Middle Byzantine. likely, therefore, that the modification of the arch and paint- There is one possible explanation for the modifications. ings are of the same date; the inscription states that the Until the early seventh century the supply of grain to the

church was repainted in 1640. populace of Constantinople was an imperial obligation— In the late eighteenth or early nineteenth century, the and an imperial concern thereafter. With the loss of Egypt,

church seen by Hamilton must have been an enlargement of the Empire had to look elsewhere for the bulk of its grain, in

the original one in area I. The extension probably covered particular to the Euxine. Unlike the situation in antiquity areas II and III and the traces of two different gabled roofs and later in the fourteenth century, when the Crimean ports described above on the west wall or area II may be connected became the granary of Genoa (as Crete was to become the with this extension. So also may be the insertion of the granary of Venice), in the Middle Byzantine period they fragment of templon screen (pl. 15b) into the external ma- imported corn from the southern Black Sea ports (such as sonry and the cutting of the holy water stoup 9 (?) at point N.

Dating is difficult. The type of banded brick and stone 66. Cf. Strabo, Geography, XII, m, 11. masonry (something akin to type C of the city walls), of 67. A. Boethius and J. B. Ward Perkins, Etruscan and Roman which the bulk of the “‘palace’ complex is constructed, = architecture (London, 1970), 225 and fig. 94. cannot be of Mithridatic date, so the popular association of (Istanbul. 197 ar Ancient civilizations and ruins of Turkey the ruins with the Pontic kings may be ruled out. However, 69. S. ‘Eyice, Kuguk Amasra Tarihi ve eski eserleri kilavuzu they have been described as a “palace’””—Balat—at least (Ankara, 1965), 64—68, pl. 9; Necdet Sakaoglu, Cesmi Cihan Amasra since Makarios’ day in the seventeenth century. If this were a (Istanbul, 1966), 226-32, 276 (where it is identified as a basilica).

82 SECTION II Aminsos), rather than export it. There was a grain surplus at by the Evangelist symbols of the Lion and the Man. The Lion the eastern end of the Euxine and along the Balkan coast. is red; in effect only its paws and head are shown. The Man is

Sinope was ideally placed for the trade of Pontic and Laz outlined in red on a yellow ground color, and, as usual, is grain to the Crimean ports and to Constantinople during this represented by a head and wings. The Lion and the Man period. If the “palace” complex was adapted as a granary for seem to belong to a late repaint and are out of proportion Constantinople as part of an imperial policy in the early with the remains of four small winged creatures, two on the seventh century, safe from the thieving fingers of the mob in south side and two on the north side of the circle. One of the the capital, it could have continued as a more commercial creatures on the south side shows vestiges of a yellow Gospel granary for the Black Sea area after its more limited function Book with white pearl decoration. would have ceased in 618 and 626.’° The suggestion is more Next to this composition were two narrow panels framed conjectural than our proposition that the complex was orig- by red borders. In the southern panel was a standing figure, inally built as a gymnasium-bath structure, but should be about 2.50 m high, wearing a blue tunic and a red cloak.

considered. Traces of jewels at the collar and a crown on the head suggest 4. “Balat Kilise” (fig. 5) that the two figures represent David and Solomon. Nothing / —_ « - , ae is visible in the western part of the vault; an Ascension or This church lies in area I of the “‘palace.”’ An inscription of Pentecost would be appropriate

1640 dedicates it to the Theotokos, and a painting of the ppropniate. Koimesis on the exterior of the lunette above the south door The East Wall (fig. 8) would suggest that it was more particularly dedicated to the On each of the two faces of the eastern reducing arch which

Dormition. However, in 1658 Makarios stated that the links the naos and apse were four standing figures, the shape church was dedicated to “The Divine Ascension.” ’' Mostof of three of which can just be distinguished where whitewash the interior and part of the exterior of the church are painted. has flaked off. The backgrounds were grey and the inscripThe original mortar of the walls was of lime and sand, tions in neat white lettering. One of the lower figures on the devoid of pebbles, well bedded down as elsewhere in the south side of the west face was a bishop wearing a red and ““palace.”’ A rough-cast plaster layer of lime and pulverized white polystavrion.

earthenware over the masonry face covers the walls and Of four roundels in the arch soffit the upper figure on the vaults of the church; a similar layer is found over some of the south side was bearded. The background of the roundel was

later cuttings in the walls. The surface layer of plaster on yellow; the figure had a yellow halo outlined in red, and which the paintings were made is of lime without any ap- wears a red robe and holds a book. parent binding material. The plaster joins follow the red In the lunette of the east arch is a seated Christ, but little borders. The seventeenth-century plastering for the south can be discerned beyond the red outlines of both the throne doorway paintings is composed of two layers of lime plaster, and the figure. The flowing curly hair of the figure on the

each with a tow binding. south side of Christ almost certainly identifies it as an Two compass incisions describe the halo and medallion archangel.

oeuines: Red poraer lines and some es lines are inened The South Wall (fig. 9)

th iced li t in ee an fenndan he Sar - The lower 1.50 m of the wall was probably occupied by a

with incised lines, but no such lines are found in the sacrifice patterned dado; the lower horizontal border of the lower

with incisions. , ;

of Isaac. Theaincisions were withis a rather blunt instru; rae ; aor ; . register of made painting scored with an incised line.

-Vision ; The of fragments ofAlexanderia figures to the east the south doorway the St. Peter of were notofmarked out -with . HLL. earaeeer appear to represent; in the east spandrel, an angel hand —_ , , a. held up toward another figure; only a few red outlines Preliminary drawing was executed in red ochre which, in ; . ment. The later paintings in the south doorway and that of

a , ; earlier remain, together with a yellow ground color for the flesh of; many of. the paintings, all that survives; may . ; the outstretched arm. The is figures at the east endit of thebe wall

that, originally, was very little to them. . the ; oowesternmost oo.there had yellow haloes withmore red outlines and Only partialaedescriptions of the paintings can be made for ;;; ar figure had a remains bejeweled collar to his tunic. much has gone and much of what is covered by . ; ; hit h ath Its) b id The middle register consisted of five or six panels sep-

white-wash or (in the vaults) by green mold. arated by wide red borders. The Sacrifice of Isaac (pls. 19,

The Vault (fig. 6 and pl. 18) 20b), has a yellow foreground and a grey black upper At the east end in the center is a circle about 1.5 m in background. Abraham’s cloak is green, with dark green fold diameter. It has a wide red border in which there are traces of lines and highlights in a lighter shade of the same color. The white lettering. The subject of the painting within is obscured —_ faces of both Abraham and Isaac are yellow with red feature

by whitewash and mold but may be assumed to represent a lines and outlines; Abraham has white hair and a yellow halo full-length Christ in Majesty or a bust of the Pantokrator. To outlined with white. The rock background is red, with dark the south of the circle is a yellow triangle which may rep- red shadow lines and white highlights. The ram caught in the resent a ray emanating from the Glory. The circle ts flanked thicket is white. At the top west corner of the panel is a yellow triangle outlined with red. The divine figure appearing in it is

70. See J. L. Teall, “The Grain Supply of the Byzantine Empire, €ardless and has a yellow halo. |

330-1025,” DOP, 13 (1959), esp. 117-32 and 136. The Annunciation (pl. 20b) has a plain green foreground

71. Makarios (1658), 429. and a grey black upper background. The architecture is light

SINOPE: ‘“‘BALAT KILISE” 83 grey, with red outlines and is topped on the left by a red The South Archway canopy supported by four columns. In the top center is a The soffit (fig. 11) contains St. Kosmas in its west half and segment of Heaven in yellow; three white rays descend from St. Damianos in its east. Both were full-length figures but St. it toward the Virgin. The haloes of the two figures are yellow Kosmas has been destroyed from the thighs down, where the with a red inner and white outer outline. The angel wears a arch was at some later point cut away. The lower backwhite robe and red cloak with an agitated tail of drapery, but ground color for the figures was yellow and the upper all the clothing has been overpainted. The wings now exhibit background grey. The flesh color is yellow with red feature a yellow ground color with red outlines, but upper layers of lines and the hair is red. The haloes are yellow with a red

paint may have fallen away from them. inner and a white outer outline, but traces of an earlier thick The Nativity survives only in a fragment at the bottom white outline for the haloes and of a larger outline for the left-hand corner, where there is a seated figure resting his heads, also earlier, can be detected. head on his left arm. He wears a red garment and has a yellow St. Kosmas wears a grey robe and red cloak, which have

halo with red inner and white outer outlines. been blurred by overpainting. St. Damianos wears a white The Vistitation (pl. 22a) has a yellow ground and grey robe and a red cloak with a jeweled hem. The inscriptions are black upper background. The architecture is grey, with red in the same large white letters as those of the donor inscripwindows and a red roof; yellow drapery is suspended across tion of 1640; St. Kosmas, however, has traces of smaller and the back of it. On the left is Elizabeth wearing a white robe earlier lettering in red, which has now turned black.’> Figure and a pink cloak with red outlines. Mary wears a grey robe 11 shows that the form of these letters is regular and traced

with blackish fold lines (there may have been light blue with less of a flourish than the work of the later painter. highlights); her cloak is red with dark red fold lines. Both The North Wall (fig. 10)

figures yellow haloes with red inner than and white outer vanwall, 1B hbut b outlineshad Less of the painting remains on the south Th dels in th st I. 20a, b) tin: enough to make clear that the arrangement was the same. In

© "dee . van iptand on “t iP “s b | Pa, . the upper register there are traces of three roundels and in the

narrow decorative Strip and’ caciy encloses a dust cans © middle register of three Feast scenes (which can be partly

eight roundels remain; there may have been a total of eleven. re , The seven busts whose heads have survived all represent made out from incision marks where the paint has all flaked

P off). The surviving scene is certainly the Entry into beardless. ,young men whose faces have a yellow color ; be ; Jerusalem, and the panel adjacent to it toaground the west might and red feature lines; the haloes are yellow with thick white , — a: the ae the Raising of color Lazarus would putinscriptions the scenes in; .,the outline; background is (which light grey with in white. The roundels have white inner outlines and a wide wrong order). The easternmost scene contains the remains of

red outer border. The decorative motif framing the roundels vee nas: necting enter maa Traneh enthroned

is yellow and grey with red outlines. The outlines of the In the | ‘ster at th 7 , d. ab he th ,

haloes and roundels, the shape of the decorative motifs, and , n Ie Owe re Bister at the cast ene, 4 Ove . pro esis the upper and lower horizontal borders were all marked out niche, is the Vision of St. Peter of Alexandria. It 1s largely with incised lines. There is also a horizontal incised line whitewashed, what is still visible is now blackened by smoke,

, , ; and only a few outlines can be distinguished. The flesh color-

running through the center of the heads which may either be . f both St. Pet d the Christ Child js vell th red

a guide line for the centering of the roundels or mark a border Ing OF DS Teter ange Sas We TS YOROW WHEN TS

which was later abandoned outlines and feature lines. St. Peter wears a white omophorion The small white let terins ‘5 of a type found elsewhere with red crosses. Christ stands on the altar table clothed in a

YP be . a, discussed chiton, which he holds his left hand where it was which. gwill below. Theincharacteristics which . torn. All

appear here are the B, with well separated curving members other details are obscured save for the sigla IC XC and three

an almost cursive d and the flat based on ligature, X , lines of slanting lettering of medium size and neatness, of

which recurs in the donor inscription oo which the word yeitov(a) can be made out, identifying it as

The first two busts represent two of the seven Maccabees, Pa en standard text: tic Gov tov xitwva, LotnNp,

Abelbous and Ourias. They are listed in the Painter’s Guide tee , ; ,

of Dionysios of Fourna as Abeib and Gourias.”? In In the register below is part of the halo of a figure which Supplementary MS 3 they are named Abid and Gourias 73 was otherwise destroyed when the prothesis niche was cut: and in Suoplementarv MS 5 their names are spelt as at the letters 8nAoo1 can be distinguished. It seems likely that Sinope.”4 The third figure of whose name only io N sur- the Vision of St. Peter also antedates the cutting of the niche, vives, could be Antonios, who is placed third in the lists. The which would have truncated the main figure. But the absence

seven Maccabees are always listed in Dionysios with the of incised guidelines suggests that t he scene may not have Seven Children of Ephesos. If we assume them to have been formed part of the original decoration. all represented at Sinope, they would account for fourteen out of the twenty-two roundels (on both sides of the church), 75. Perhaps vermilion, which changes easily to black sulphide of

leaving eight roundels filled, perhaps, with the Anargyrot. mercury.

76. Dionysios of Fourna, “Eppnveia, 154, 155, 219, 268; MS 4 72. Dionysios of Fourna, “Epynveia, ed. A. Papadopoulos- (p. 279) places the scene in the prothesis, as at Sinope;.at Manasija it

Kerameus (Petrograd, 1909), 161. is on the southeast column of the naos. See also G. Millet, ‘‘La vision

73. Dionysios of Fourna, ‘Eppnveia, 272. de Saint Pierre d’Alexandrie,” Mélanges Charles Diehl (Paris, 1930),

74. Dioysios of Fourna, ‘Eppnveia, 297. II, 99-117.

84 SECTION II The West Wall (fig. 7) The division of the tunic by a horizontal line belongs to a Little remains on the north side. On the south side there later repaint. The hem of the tunic is yellow with red outlines. are two registers, each with a full-length standing figure. The The boots and buskins are similar but have been overpainted lower figure wears a white two-part tunic, with fold lines and with further red lines. The flesh is yellow with red feature outlines marked out in red and yellow. The tunic has a lines and outlines; the hair is red. The halo is yellow with red jeweled yellow hem. The trace of a foot belonging to an inner and white outer outlines. Some of the fold lines of the

earlier figure underneath is visible. cloak fall in a broad and graceful pattern; however, they In the upper register is the figure of a bishop in a white belong to a repaint. Above St. Panteleimon is the fragment of cloak with yellow fold lines and thick red outlines. Thetipsof _a third figure.

grey shoes appear from under the robe: the omophorion has The figure in the lowest register of the north reveal has red crosses. The flesh was yellow with red feature lines; the been destroyed; that in the middle register, balancing St. face was bearded and had large staring eyes. The halo is Panteleimon, wears a white tunic with red fold lines, and a yellow with a wide white outline marked by two incised light red cloak with dark red fold lines. This figure 1s beardcircles; the lettering of the inscription is white and of medium less and its flesh is yellow with red feature lines and outlines; size. The robes have incised main outlines; despite damage, the hair is red. The halo is yellow with a red inner and a white

they give the impression of falling in graceful folds of | outer outline. The lower background is red, the middle secdrapery; thus, this figure is in strong contrast to the squat and tion yellow, and the upper grey. Much of this figure, includdumpy figures in the Visitation scene which adjoins it on the ing its identificatory inscription, is obscured by whitewash.

south wall. The figure in the upper register is also largely covered by

The backgrounds of both figures are in three sections. The whitewash and mold. lower is red, followed by a thin yellow middle section and a On the reveals and soffits of the narrower west arch there

grey black upper section. were two standing figures and seven or eight roundels conThe lunette figures were too remote and indistinct to allow taining busts, of which six survive.

of a detailed description. In the spandrels the symbols of the The roundels are set on a grey background, with the sun and moon are white with outlines in red; they make little exception of that at the top center, which has been oversense on their own and are divided from the figure above bya painted in blue with a wide red frame, a thin white outline, red border line. It seems likely that there was originally a and, finally, a thin black outline (most of which has now solid west wall, or that there was a filling to the arch bearing a flaked off). They are linked together by smaller circles conCrucifixion scene, of which the sun and moon symbols are all taining equal-armed crosses and the sigla IC XC N K in red.

that survives. The rectangle at the top center has a grey The spaces between the circles are filled with a stylized leaf ground and a wide red outline; the diamond has a yellow _ design in yellow. There are no signs of incised guidelines and ground with a red outline which seems to have been over- the roundels and circles are very crudely executed. painted in grey. The star within the rectangle 1s yellow. The From south to north, the roundels contain figure busts as

winged creature has yellow wings with red outlines. The follows: figure within the diamond could represent the Ancient of 1. Robe white with yellow fold lines; cloak, repaint, green Days, or, possibly, the Majesty of the Last Judgment: it has (pl. 25a). Flesh, yellow with red feature lines. Hair, red. Halo,

red outlines but whitewash obscures most of it and the yellow. Untidy red inscription reads “Ismail.” cutting of the rectangular window above has destroyed most 2. Robe, repaint, blue. Book white. Flesh, same as 1. Halo,

of its head and shoulders. yellow with a red inner and a white outer outline. Inscription

The West Arches (fig. 13 and pl. 21a, red b) obscured bycloak, w hitewash ; , green fold lines; repaint, standing figures—originally there probably six.3.The 4. Same , only oo 5. Robe andwere cloak, same as Flesh, as |. Itasis3 the

; ; arch 3. Robe, white with On the reveals of the wider are the remains of five

lowest figure the south 1s clearly as St. -_ _. a ; oe ; ,on surviving bustside to have a beard,identified red. Inscription, originally

Barbara by an inscription in white lettering to theonly leftMofisitclear ; , red, nowthe black; (pl. 21b). The lower background is red, middle ;; ; 6. Outline of roundelsection only remains.

yellow, the upper grey. St. Barbara a white robe ;, f ; ; grey Of and the lower standing figures onlywears the head of St. Eirene with fold lines, and a red cloak and cowl with dark red . ; hashas survived (pl.overpainted. 22b). The The large white letters identifyingathe fold lines;which been much flesh is . , a feature ; ; Saint lines; have mostly gone. Theyellow two forms A noteworthy: yellow with the halo isother also withofacross redare . stroke. :inner ; . ; :and oneared has a stepped, and the a slanting white outer outline. The Saint holds a martyr’s ; , -_The ain ; , face is yellow with red feature lines; the hair is red, falling crosslet in her right hand. On her robe is a red K, for which , .; ;explanation, curls around theunless neck. The face was later built up with we have no. colors, a memory of survive. the en- thick . The ;;— flesh of which onlyitabe few fragments igmatic ellow has a white outline bordered each ye side. os by y a Abovegammadia. St. Barbara, St.halo Panteleimon is identified byon white

,; , thin red line. The topaof the Saint’s costume wasofyellow with ettering (pl. 21a). He wears white tunic, the lower part , ; which has red fold lines and the upper part yellow fold lines red outlines; it was doubtless an imperial garment. The Pper pany background is yellow. The crown is outlined in red, with

77. See the RBK, II, 617-19; and W. Oakeshott, The mosaics of | White pearls. Beneath it, and under the yellow background

Rome (London, 1967), 378-79. color, are traces of red outlines of an earlier crown of a

SINOPE: ‘“BALAT KILISE” 85 simpler Byzantine design. There are also traces of an earlier Tokat had a substantial Christian population.®° The bishop and larger head. The more noticeable features of St. Eirene Ezekiel is probably the metropolitan of Amaseia who in the

are the long elegant nose and nostril lines, and the largeeyes = seventeenth century resided at Sinope.?' The archon with upward gazing pupils. The roundel busts, both on this Kyriakos is perhaps the local Greek primate. His title brings arch and on the south wall, as well as the figure of Isaac on to mind the other great bilingual inscription of Sinope, of the south wall, share this characteristic upwards gaze. 1214-15, and the Armenian Reis. °?

South Doorway, Interior Paintings (figs. 12, 14, and Exterior

pl. 23) The exterior of ‘*Balat Kilise”’ is also decorated with paintOn the lunette above the south doorway a donor’s inscrip- ings. On the lunette over the south door is a representation of tion is associated with a pattern and a painting representing the Koimesis (fig. 15 and pl. 24a). Formerly there were paintSt. Marina and the Demon. St. Marina is a badly damaged ings accompanying it on the soffits of the south arch but of half-length figure with red outlines, holding the demon by the these only a few tiny fragments remain, and are now covered hairs of his chin’® while she raises her other arm to belabor with whitewash.

him. Apart from the architectures, which are rational in their

The inscription is composed of five lines of Greek and a arrangement, the scene reproduced in figure 15 corresponds sixth of Arabic in red letters on plain white plaster, and is toa standard Late Byzantine Koimesis. There do not appear framed with a red outline and a wide red border (fig. 14, to be any mourning women (although they could be included pl. 23). By 1973 much of the last two lines had become among the background figures) and the episode of the Jewish

illegible. It reads: Prince of Priests, Jephonias, is not depicted. The composition fits fairly well around the rectangular window and was

AVLOTOPIOON 6 TAVOENTOS KAL probably painted after the window was opened. Q10G vaos THC O(EoTto)Kov d1a GuydpOoUNs Kal EFOdo0v The only haloed figures are the two bishops, who hold

TOD EVTLLOTATOV APYOVTOG KUPOV KN- Gospels. Their haloes are yellow with a red inner outline and

piax[oc] 6 petzi[s] apyiepatevovtos Kup |ek3€- a thick white outer outline. The background is green; the 5 Kn K(ai) epoupyovvtos Kvp otaPpiavoc etouv AX M upper portion of it—the sky—1is reddish black. The lettering now shows black, probably as trace marking of white paint.

al-hagir Grigorios min beled al-Togat fi al-musavur fi The architectures are rendered in grey and light red.

sene 1(0)50.’? Garments are in red, green, two shades of black (one of

which would originally have had a blue overpaint and the ‘The most sacred and holy church of the Theotokos was other a red overpaint to represent purple), yellow, and white. repainted by the aid and contribution of the most honorable Christ and the attendant angels wear white robes with red archon kyr Kyriak(os) the Reis, in the episcopacy of kyr fold lines. The painting is so severely damaged that any (1)E(k)zekiel and in the priesthood of kyr Stavrianos in the judgment on its style is made impossible. The faces are of the year 1640. The humble Grigorios of the land of Tokat in the Late Byzantine type, with small features and curly hair and enkleistra? in the year 1(0)50.” (A.D. 23 April 1640—a.p. 11 beards—quite unlike those of the interior of the church.

April 1641) Some of them have been repainted and the face of one of the

bishops on the west side has crude white white paint similar The church was therefore redecorated in the period 23 to that used on St. Eirene in the interior (pl. 22b). The robe of

April-31 December 1640. St. Peter, near the head of the Mother of God, has a red

There are peculiarities of letter forms in the Greek: the ground color, dark red shadow areas, light red highlight stepped N and the three forms of A employing stepped, areas, and may have had white highlights—but the glues slanting, and horizontal cross-members. Typical of Late used by the painter were poor and the colors have run. It ts Byzantine forms are the high horizontal stroke of the A and the most elaborate surviving system of coloring of a garment low central member of the M. The middle member of the E in the entire decoration of the church and is a good example has a high upward turn and the ligature OY hasa flat bottom of the conventional Byzantine system. But the workmanship

member. is poor and a seventeenth- or eighteenth-century date is In the Arabic the epithet haqgir may suggest tapeinos; it is possible; it may even be a consciously revivalist painting of also reminiscent of fagir and is here a synonym. Fagir can the second half of the nineteenth century. The similarity indicate connections with a dervish order; hagir may here between some of the colors of this painting and those appearindicate a priest or monk. There is some uncertainty about ing in the retouching of some of the paintings of the interior the reading ‘“‘a/-masavur,” but, if correct, it may indicate an suggest the work of the same artist. The original Koimesis enkleistra. One may envisage a monastic cell at Balat Kilise painting was painted with yellow, red, green, white, and to which Gregory of Tokat (perhaps the painter) retired. black (and possibly blue) colors; it was later touched up with 78. Dionysios of Fourna, ‘Epynveia, MS 3 (p. 273), gives instruc- a garish blue and violet red. tions to paint St. Marina thus.

79. We are most grateful to what amounts to a committee which 80. M.A. Cook, Population pressure inrural Anatolia, 1450-1600 has assisted in the discussion and interpretation of this inscription. (London, 1972), 63. They include Professors Cyril Mango, Rudi Lindner, Heath Lowry, 81. Makarios (1658), 429.

and Speros Vryonis. 82. See p. 72.

86 SECTION II On the west face of the niche to the south of the west arch is The rendering of the flesh is all done with a yellow ground

the figure of an archangel (pl. 25c). Of the background, the color and red feature lines—the simplest Byzantine lower portion is yellow, the upper is blue painted directly system—and there is no evidence of the more sophisticated upon the white plaster surface. The archangel wears imperial build up of color from a green ground. The basic colors seem garments of a debased form and holds an orb and staff. The to be white, black, yellow, green, and red; the adherence of robe is decorated with flower patterns in red and green, also the yellows and reds at least suggests that they were painted painted on the white plaster ground. The loros has a yellow on the fresh plaster. Evidently, the painters of Sinope did not ground with red and green motifs and a white pearl border. use the famous “‘sinoper’”’ much. Blue and certain other The flesh color is yellow, the hair red with curls in light red varieties of red belong to later repaints.

lines. The halo is yellow, outlined in red. From a stylistic point of view the most striking features are The figure is too damaged to allow comment on its style, the squat, heavy figures, their limited number in the scenes, but seems to be eighteenth- or nineteenth-century in date. It and the lack of complicated architecture or iconography. was made by a painter unfamiliar with Byzantine technique, Such characteristics point, again, to an early date, parallel as may be seen from the lack of overall ground colors. The with the Cappadocian, Cypriot, and Kastorian examples. It alarming blue of the background appears again in the over- could be argued that the crude quality of the paintings is

painting of the Visitation in the interior. consonant with the work of a provincial painter in the postA few graffiti are visible on the lower parts of the walls, Byzantine period; but this is unlikely, for there is virtually no including two of simple sailing ships—one on the figure of St. post-Byzantine painting in the Pontos before the nineteenth Barbara and the other on a figure in the lower register of the century. Post-Byzantine painting normally exhibits some of

south wall.8? the complexity of Late Byzantine iconography, whereas at

The technique, style, and dates of ‘‘Balat Kilise” are a Balat Kilise we have a very formal example of Middle matter of discussion. The surface plaster of the paintings of Byzantine iconography. the interior is of a single period, save for that of the tym- A second stylistic characteristic, which is paralleled in

panum over the south doorway. The latter is dated by the early Cappadocian painting, is the division of the inscription to the year 1640 and this date concords with the background into three bands of color: yellow, green, and Late Byzantine use of tow as a binder. The remainder of the grey black (instead of the formal green for the lower and blue plaster surfaces apparently lack any organic binder and con- for the upper foreground, both painted over an initial grey or form to a system of plastering which was employed from late black in wall paintings from the eleventh century onward. Roman times up to about the tenth century and was not However, the three-colored backgrounds appear as late as

common thereafter.** Parallel examples are found in the fifteenth century in the tower of the Hagia Sophia at

Georgia and Cappadocia. Trebizond.

The use of incised guidelines is irregular: the Annunciation A third characteristic which is unconventional is the single scene has fairly extensive incisions and the Sacrifice of Isaac wide white outline for some of the haloes. This is more likely next to it has not. The roundels on the south wall are marked to be early rather than late in date, since a double outline for

out with incisions; those on the west arch are not. If our haloes becomes normal from about the eleventh century hypothesis that there was originally a Crucifixion scene on onward; at‘*Balat Kilise” the wide white outlines have in the west wall is correct, the roundels on the west arch would some places been overpainted with the conventional red be later than the other paintings in the church interior. inner and white outer outlines. The bulk of the painting now visible is of the simple type, The faces of Isaac (pl. 20b) and Abelbous (pl. 20a) bear a aptly described by Vasari as being just ‘outlines in a col- strong resemblance and suggest the same painter. They share ored field.’ The system for clothing consisted of an overall the upward gazing eyes, whose pupils are joined to the upper ground color, fold lines and, sometimes, highlights. This eyelids, with the figures of St. Eirene (pl. 22b) and Ismail (pl. simple technique is found in the archaic decorations of 25a). In other respects Ismail and Eirene look rather different; Cappadocia, in Cyprus at St. Salamoni (Rizokarpaso) and moreover, the system of roundels on the west arch is unlike St. Mavra (Kyrenia), and in Greece in the earliest work in St. that on the south wall for there are no incised guidelines and

Stephen and the Sts. Anargyroi (Kastoria). None of these the pattern in the spaces between the roundels is not the paintings is securely dated, but a consensus of opinion puts same. the early Cappadocian and Cypriote work, as well as the A different and more sophisticated drawing can be seen in early painting at Kastoria, in the ninth and tenth centuries.°° 83. NotinO. F.A. Meinardus, ‘“‘Medieval Navigation According group of wall-paintings,” JBAA, 30 (1967), 19-36; S. Peleto the Akidographemata in Byzantine Churches and Monasteries, kanides, Kastoria, I (Thessaloniki, 1953), pls. 38, 41, 87, 88,

AeAt.Xpiot. Apy.Et., 6 (1972), 29-52. 97, 100. The long-awaited text to this volume is still lacking and it 84. See Restle, Wall painting, 1, 224-34, 236-37; and D. Winfield, is regrettable that, in the quarter century that has elapsed since

“Middle and Later Byzantine Wall Painting Methods,” DOP, 22 publication of vol. I, Byzantinists have been actively impeded (1968), 64—79 and tables 1—vi. It must be emphasized that too little from study of these important paintings and that no cleaning or

attention has been paid to plastering practices and that too few conservation work has been undertaken. The Cypriote examples Byzantine examples have yet been described to provide sufficient are unpublished, but a ninth- or tenth-century date would be

evidence for dating by composition of plaster alone. stylistically acceptable (although a date after the reconquest of 85. Cf. R. Cormack, ‘“‘Byzantine Cappadocia: the archaic Cyprus in 965 is more probable).

SINOPE: COLUMN INSCRIPTION; CHRIST RELIEF 87 the face of St. Barbara (pl. 21b), and in the proportions of her white lettering and of the red lettering of the 1640 inscription figure and that of St. Panteleimon above her (pl. 21a) which suggests that a redecoration of the rest of the church was are much more elongated than those of the figures of the carried out in 1640, as the inscription states.°°

scenes. All else is speculation. We may tentatively conclude that

On the exterior, the Koimesis provides the only example in the decorative program, the simple technique of painting, the church of a fairly convential Late Byzantine iconography and the division of the backgrounds into three differently (pl. 24a), but it might have been painted at any time between colored sections point to an early date for the original decorthe seventeenth and nineteenth centuries. The only charac- ation, perhaps somewhere in the period between the ninth teristic of a very late painting is usually a westernizing in- and the eleventh centuries. fluence, and this is entirely lacking here, unless it be found in The paintings were retouched more than once; the church the relatively correct perspectives of the architecture. By was altered by the cutting of niches and windows, and the contrast, the flower patterns on the robes of the archangel in filling and remodeling of the south door and apse. One the niche in the west wall are certainly very late and show a period of redecoration 1s attested by the inscription of 1640. western influence; for this figure an eighteenth- or Travelers’ accounts make it clear that the building was in use

nineteenth-century date is likely. as late as Hamilton’s visit in 1836. The dilapidated condition of the paintings makes it im- 5. Inscription on Column

possible to offer a more detailed stylistic appraisal of them rE dati ¢ what ared to be a church :

without a serious cleaning and conservation program—and OUDEAOMS Of Maal appeared tO be & emu’ were re this is no more than a hope, for they have greatly deteriorated vealed in 1963 in the course of excavations for a Bas pump in the last fifteen years. But what could be seen on the surface, not far west of the walls. They also brought to light an altar together with our observations on differences of style and made ofa stumpy fluted Doric column (pl. 25b). A clean-cut

; ; . inscription carved on two successive flutings reads: of painting before the final repainting. AEAQINIOX technique, makes it clear that there was more than one period

The styles of lettering confirm this conclusion. There are OPIIAAEOX

three types of white lettering which are distinguished by size and form. The large white lettering 10 to 12 cm high identifies The splayed ‘‘sigmas”’ suggest a late classical date. The in-

Sts. Kosmas, Damianos, and Eirene; the forms of K, M, and scription is not otherwise published and the whereabouts of A concur with those of the red letter inscription of 1640. The the altar is now unknown.

medium-size white lettering, averaging 5 cm in height, ident- , aa

ifies Sts. Barbara and Panteleimon and is also found in the St. 6. Relief of Christ with Angels (pl. 246)

Peter of Alexandria inscription. The form of the M in this In the Sinope Museum there is 2 block of stone about medium-size while lettering is similar to that of the inscrip- 0 cm high with a representation of Christ with angels carved tion of 1640 which is in red letters. The small white lettering 17 !0w relief. It ts reported to have come from a village in the

averages about 3 cm in height and appears in the medallions neighborhood. ,

of the south wall. The oy ligature is close to that of the 1640 The carving represents a full-length seated figure of Christ inscription. The B with separated lobes is similar to that of within a mandorla held by four angels. He carries the Book in the medium-size lettering accompanying St. Barbara. The His left hand and blesses with His right. The peculiar features almost cursive form of A is peculiar to the small white of the figure are the discrepant SIZES of the two hands, and the

lettering. mae rectangles inscribed with the sigla IC XC on each side The red lettering is of two types. The larger is that of the of the halo. oo, , 1640 inscription (fig. 14 and pl. 23). The smaller is clear on The mandorla outline isa flattened oval and is treated as a the labels for St. Marina and St. Kosmas. A characteristic of tangle aised shape grasped Py the lingers one angels me tis the M, the central member of which depends from the standing rather than flying. Their bodies are somewhat awk-

- angels themselves are arranged in pairs and appear to be

tops ofwardly the uprights rather from halfway down .with ar them. arranged andthan onlystarting one wing apiece 1s shown, Thus, there are five types of lettering. Once again the sone attempt to balance its shape with a hanging end of

evidence is incomplete through damage and the whitewash drapery falling trom the shoulder on the side opposite to that

which covers the paintings; these types probably indicate the of the wing. ; ; ,

work of different scribes, yet they need not be far separated in The narrow face of the Stone Is decorated with an interdate for they often have one or more characteristics in laced rope P attern which is neatly executed with COMPASS:

common. drawn circles. It seems fairly clear from the manner in which We have evidence of different sorts of plastering and of the interlace continues to the left, beyond the figural panel, several different periods of painting. The only secure date is 1640 for the lettering of the donor inscription, the painting of 86. For unrecognized examples of the use of the word ‘avio-

St. Marina, and the tow-bound plaster below them. Not topiodn eee eae. ane Oso aA dg gn wal enough, however, survives of the painting of St. Marina to Panagin. Arakiotissa, Cyprus, date i 1192): de dA. un d ) enable us to relate its date to that of any other painting in the Stylianou, “Donors and Dedicatory Inscriptions of Cyprus.” church. The similarity of the K, M, and A forms of the large = JOBG.9(1960). 101.

88 SECTION II that this block was a lintel of either a doorway to a church or churches of the city of Trebizond. The figural carving is so an iconostasis screen. The carving of Christ would thus have worn that no stylistic evaluation of it can be made beyond decorated the flat underside of a doorway, or perhaps the observing that it is a piece of provincial workmanship. The

central entrance of the iconostasis. only objection to a Byzantine date for it is the unusual raised The subject matter of this carving seems to be that of a rectangles with the lettering. Therefore, it could be a carving Christ in Majesty, although it lacks the attendant evangelist of the Byzantine period and its worn appearance could insymbols that might be expected in such a scene. Or it may be dicate an early date; or it may be a piece of Byzantine revival regarded as the central section of a representation of the carving of the nineteenth century, several examples of which

Ascension. Both interpretations are possible. are in the Sinope Museum and one, perhaps the best, is near Its date is difficult to judge. The rope interlace is found in Sebinkarahisar.®’ reused medieval blocks of stone in the church on Cape Jason

and there are various fragments of interlace pattern in the 87. Bryer and Winfield, AP, 32 (1974), 247-50.

. gn %

A ®\ i; LIN QTTIH, SINOP ” / 23 ZF ine | Spo * Plan of Walled City ao “‘ A @y, 21 Cs

GG

EXTANT STRUCTURE a Ny \ ______ RUINOUS OR OBLITERATED My WN r \ \ Hy @ BN

““"""=" STRUCTURE 4 NY \ —-— SUGGESTED CLASSICAL GRID ry: | — \

* A ae S © a Ny 27 ff’ + \ NJ B YK \ Cl a Y ‘\ a \C.gee 4ipe Z~ (_\ ——"* \ c =

—— = x’ Xa ss a 29D 28 { ale p > 4 ee < Y “UY “re. === = at 4 Cp FEA —_ [ aZi 42Acropolis eA heIM LK o> (= et 7 43 Dt” H i 7] 4 ‘\ ¢ Offa 1 G aN = 360% \ — ° q ./| Ss ./ %, *, 2 ——— ae % * J DNS : Ny on EX: 4 ©, Q, LS tM 4 4 “ -e 7" a “ot X\ e, & A LB, Y= mae fis (@) INSCRIPTION NUMBER 26 \ \\ 40 TOWER NUMBER ye AN Serapeun

Cp = x ' Ua 1 a \

AN 2 4, r/ °“en XA ia \G: 7 Yj Dy KS 34 Mole ~,,.33 A Off 32 2:

é % nN Citadel X, \ Z 2 Uy ee 4 _— S —e - S UW,

4. |

, Js ™ 35 Submerged Ris BLACK SE 12 YU

0 iy 1G? 1/4 ey) 1 KM. APPRO KIMATE , mounds, we cannot locate, and have no reports of, even the but it must have fallen to Izz al-Din Kay-kavus II, with castle of Paurae. There are, however, reports of castles on the Sinope, in 1214 and was not recovered by the Trapezuntines Balik Goli, at Akalan, 21 km west of Aminsos, and of two castles on either side of the Halys, about 24 km south of 21. Idrisi, ed. Jaubert, 393; ed. Nedkov, 96—97 and note 297 on Paurae. The latter are Asar Kale, an apparently substantial Pe The 12th-century geographer still referred to the “Aly” or fortification, on the west bank, and the significantly named 22. Albert of Aix, Historiae, in Recueil des historiens des croisades, Kostantinusagi nearby, on the east bank." We have not Historiens occidentaux, 1V (Paris, 1879), 570: ad castellum visited them. Imperatoris Pulveral. Anna Komnene’s account makes it clear that

*‘Pulveral” is Paurae. 26. Ibn Bibi, trans. Duda, 318, 346, and note 457; Al Umari 23. Anna Comnena, Alexiad, ed. Leib, III, 38: ti¢ Mavpane (or (1342-49), 341.

Tavpaxys). 27. Ritter, Erdkunde, XVII, 440; Cumonts, SP, IH, 117-21; 24. John Cinnamus, Bonn ed., 176: Mavpanv. Oikonomides, Pontos, 49-51: Bryer and Winfield, AP, 30 (1970), 25. G. L. F. Tafel and G. M. Thomas, Urkunden zur alteren 251-53.

Handels- und Staatsgeschichte der Republik Venedig (Vienna, 1856), 28. Tarhan, Map. We understand that the late Mr. Robin Fedden

476 and note: “Pabrei,” Pabrii,” ““Babriti,” or “*Pauriti.” canoed down the Halys in 1970, passing Asar Kale.

Section IV

AMISOS, AMINSOS, SIMISSO, AND SAMSUN DESCRIPTION to the east of classical Amisos, and discharge cargoes and From earliest times a settlement at the mouth of the Lykastos passengers by lighters. A tonnage of between five hundred (Merdirmak) in the bay between the Halys and Iris deltas thousand and one million annually was thus serviced in the

has provided central Anatolia with a major outlet to the late nineteenth and early twentieth centuries. Euxine. Its two great rivals have had different functions.

Sinope, to the west, has more difficult access to a hinterland HISTORY

and is in fact a Crimean entrepot. Trebizond, to the east, is Evidence for the early prosperity of Amisos, as a Greek essentially the gateway to Persia. Amisos has both a fertile colony and as one of the major cities of Mithridates’ kinghinterland (the Phazimonites, Chiliokomon, and the Lykos dom, is abundant.® Significantly, Strabo speaks of the cultivalley) and a major caravan route. During several periods it vation of the olive in the region.’ The fact that the olive is not

has been the port for Constantinople of the most direct route grown further west, along the Paphlagonian shore, may to Sebasteia, Aleppo, and Baghdad. The three gentle passes account for the comparative scarcity of early Greek settlesouth of Amisos (the Mahmurdag1 at 840 m, the Hacilardagi ments on that coast. The acropolis of Amisos was severely at 820 m, and the Karadagi at 900 m) offer the easiest route sacked by the army of Lucullus, but Roman Missos reover the whole stretch of the Pontic Alps. A late fifteenth- covered and appears to have extended its vici to the site of century traveler found that Tokat lay six to seven days south modern Samsun and, perhaps, to the south.® There was a and Aleppo a further fifteen days.' For eastern Anatolia, late considerable Jewish colony.’ Large-scale Armenian settlemedieval merchants, when considering Amisos as a port, had ment appears to have come later—perhaps after the fall of to balance the cheapness of the sea route to Trebizond with Bagratid Amaseia in the mid-1070s, or after Timur’s sack of the comparatively high kommerkion charged by the Grand Sebasteia in 1400. At all events the Armenians of Amisos Komnenoi there. However, despite the difficulties of landing, spoke the dialect of Tokat.'° Amisos has always been the natural port of central Anatolia. The earliest reference to a bishop of Amisos, suffragan of The port certainly presents problems. At Amisos a large Amaseia, is first made at the Council of Chalcedon in 451. flat-topped natural acropolis, called Kara or Eski Samsun, Nine bishops are known, the last in the mid-twelfth century. overlooks the sea. It is about 2.7 km long north to south and The see is not mentioned in lists after the thirteenth cen1.5 km wide, rising to 159 m. The site 1s easily defendable but, tury.'' But Amisos certainly had Christians from an early like most Pontic coastal settlements, can hardly be called a period. It was one of the places supposedly evangelized by St. port. The acropolis reaches the sea at a steep promontory Andrew; St. Phokas of Sinope was brought up there under (called Hagia Anna in sixteenth-century Greek portulans”) and there is some shelter on the east side, where there were clear remains of a massive classical mole.* But although 5. Bryer and Winfield, AP, 30 (1970), 253—55. personal experience

Abul Fida describes the place as a “famous harbor,”* there {HE Fen amd FanN EO nae is no evidence that it was still serviceable in the Middle Ages. 157-65: John Boardman, The Greeks Overseas (London, 1964), In the tideless Euxine, caiques of up to 40 tons are still 266-67. The coins of classical Amisos are encountered more abunwinched up the beach and chocked up; they can be launched —_ dantly than any other in the Pontos.

on rollers with the manual aid of uP to seventy assistants. i Miller, TR, cole 645~46 Cuinet. Turquie d’Asie, 1, 101-2. Anything larger (and most Italian shipping in the Euxine in 9. Anderson, Cumont, and Grégoire, SP, III, 26—27 (inscription the late Middle Ages was larger) would have to anchor. Until from Carsamba). However, Sharf and Starr do not record Jews in

very recently ships had to stand in the open roads of Samsun, Amisos thereafter.

10. Macler (1909), 105. An average 19th-century proportion of

Armenians at Samsun was 10.8 percent: see Bryer and Winfield, A P,

1. Rieter (1479), 61-62: **Zschomschon;”’ Oikonomides, Pontos, 30 (1970), 252. 39. 11. Parthey, Notitiae, 64 no. 236, 108 no. 173, 179 no. 288, 187

2. Delatte, Portulans, 1, 238; 1], 31-32. no. 196, 207 no. 292, 249 no. 151; Le Quien, OC, I, 533-36; Laurent,

3. Hamilton (1836), 290; Admiralty, Black Sea Pilot, 400. CS, V (1), 305-7 nos. 420-22; DHGE, II (1914), cols. 1289-90; 4. Abul Fida, ed. Reinard, II, 39; Le Strange, Lands, 147. Hierokles, Synekdemos, ed. Honigmann, p. 37, no. 702 (1).

AMISOS, AMINSOS, SIMISSO, SAMSUN 93 Trajan; a little army of virgins—Alexandria, Claudia, establishment of a rival port of Samsun, side by side with Euphrasia, Matrona, Juliana, Euphemia, and Theodosia— Aminsos. Samsun was under Seljuk rule during the decade were martyred under Galerius, and St. Charitina under ca. 1194—1204, but it is quite likely that Aminsos remained

Diocletian. '? Greek and that there was no fight but a local accommodation Amisos retained its importance under Justinian’? and of interests.

throughout the Byzantine period. It was a stronghold of the By the fourteenth century, Turkish Samsun and Genoese theme of Armeniakon. But its second sack came at the hands Simisso were distinct settlements which had superseded of an emir of Melitene in 860 or 863—probably the latter Byzantine Aminsos and coexisted beside the ruins of classical date.'* He despoiled the town and complained that the sea Amisos and Roman Misso. Despite popular Turkish explashould be whipped, as it prevented him invading further; but nations for the distinctions of name, they are all, of course, Amisos recovered. Seals of kommerkiarioi confirm the com- variants of the same name. The fourteenth-century situation mercial importance of the place and, in the final words of the can be compared with that in Smyrna, where the castles of De administrando imperio, Constantine Porphyrogenitus the emirs of Aydin and of Genoa faced each other, and with noted that, “if grain does not pass from Aminsos ... the that in Pontic Oinaion, where Greek and Turkmen villages

Chersonites cannot live.” !° existed side by side.

This statement is interesting because, by the fourteenth All sources emphasize the proximity of Simisso and century, the position was nearly reversed and it was the Samsun. Three derive from the period 1400—4. Arabshah Crimea that exported corn, and also because this is one of the states that Samsun, ‘“‘a fort on the shore of the sea of the earliest spelling of Ami(n)sos with an intrusive “‘n.”’ Various Mussulmans, [was] set opposite a like fort of the wicked theories have been brought forward to account for the new, Christians, which two are less than a stone’s throw apart and and increasingly popular, form of the name, but it seems to each fears the other.” 2° Schiltberger explains that ‘“Samson be no more than the common “rational ‘n’”’ before a sibilant consists of two cities opposite each other, and their walls are

of speech, which began to penetrate written forms.'® distant, one from the other, an arrow’s flight. In one of these Like the Arabs, the Tuirkmens and Seljuks sought to reach cities there are Christians, and at that time the Italians of the sea through Aminsos. It passed into Turkish hands in Genoa possessed it. In the other are Infidels, to whom the about 1194, becoming part of the lands of Rukn al-Din, ally country belongs.”’?' Clavijo found, two years later, that of Alexios III who, however, put pressure on the Turkish ‘this city possesses two castles: one of these belongs to the merchants of the place in 1200.'’ Four years later the city Genoese, while the other with the adjacent harbor and town-

passed to Alexios and David Komnenos. ship is in the hands of [the Turks], for which reason we dared What was the situation during the first Turkish occupation not go into port here but kept well out at sea.” *? of ca. 1194-1204? To begin with, Sabbas, Greek dynast of The problem is to separate the sites, for neither castle now “Sampson” of this period, has long been relegated to his survives. It will be demonstrated that classical Amisos was rightful fief of Priene.'® But Aminsos seems to have passed to probably abandoned by the twelfth century; it is rather more the Turks without a fight and to have been recaptured distant that ‘‘a stone’s throw” or “‘an arrow’s flight” from equally casually—despite the fact that the recapture caused Samsun proper and it 1s hardly likely that the Turks would considerable disruption to Seljuk commercial and political have countenanced a Genoese castle on the acropolis where aspirations. Ibn al-Athir states that the Komnenoi ‘‘closed it would have overawed their own fortress on the shore. Both the sea”’ to the Turks and that the loss of Aminsos caused a castles, therefore, were probably on the beach. If the acrocrisis in the great Seljuk emporium of Sivas (Sebasteia).’° polis was abandoned by the twelfth century, the site taken by Thwarted on the Euxine, Kaykhusraw sought an outlet at the Turks in 1214 probably also represents Late Byzantine Antalya (taken in 1207) on the Mediterranean coast, with Aminsos; the Genoese castle followed later. widespread consequences to the future of the Seljuk state. The modern town of Samsun stretches about 3 km along What seems to have happened at Aminsos was Turkmen the coast southeast of the acropolis of Amisos. It has four infiltration and settlement in the decades before 1194 and the distinct quarters. Two (Kadik6y, slightly inland on the acropolis slopes to the south, and the Ciftlik Caddesi quarter 12. Charles van de Vorst, “Saint Phokas,” Ana/Boll, 30 (1911), inland to the west) are Greek and Armenian creations of the

507-109 Pyne . u “Cumon 1 and Gregoire, SP ed 115, cols. nineteenth century. Closer to the sea are two Turkish quar-

13. Procopius, Wars, VIII, 11, 2. ters, divided by the massive wall of the old bazaar: one to the

14. Constantine Porphyrogenitus, De Thematibus, ed. Pertusi, 21; northwest and the other to the southeast. The northwest Theophanes Continuatus, Bonn ed., 179; A. A. Vasiliev, Byzance et quarter 1s clearly the older one and incorporates the old les Arabes, 1 (H. Gregoire and M. Canard, La dynastie d’Amorium bazaar and the oldest Turkish building in the town, the [820-867] [Brussels, 1959], 250—51).

15. Constantine Porphrogenitus, DAJ, I, 286; Anderson,

Cumont, and Grégoire, SP, III, 4. 20. Arabshah, trans. Sanders, 190.

16. Stamatios B. Psaltes, Grammatik der byzantinischen 21. Schiltberger (1402), 12. It 1s at Samsun that Schiltberger

Chroniken (Gottingen, 1913), 81; Tomaschek, Kleinasien, 79. placed his tale of the battle between the sea serpents and land vipers, 17. Nicetas Acominatus (Choniates), Bonn ed., 689, 699; Cahen, which the vipers won. The sultan is said to have taken it as a P-OT, 117; Brand, Byzantium confronts the West, 138-39. heartening omen, but the story could conceivably be an allegory of 18. Jerphanion, OCP, 1 (1935), 257f.; Orgels, Byzantion, 10 the situation there—the sea serpents representing the Genoese, and

(1935), 67 f. the land vipers the Turks. 19. Cahen, P-OT, 119, 164. 22. Clavijo (1404), 108.

94 SECTION IV thirteenth-century Pazar Camii, built under the Iikhanate During Timur’s incursion and after the Ottoman reoccuclose to the beach and about 1.2 km east of the acropolis. We pation of 1419, trading conditions became less profitable, propose this as the site of Late Byzantine Aminsos and of _ although it is not clear whether Simisso did not simply join

early Turkish Samsun. Hence, the neighboring southeast the general decline of Italian Euxine trade. The Venetians quarter is likely to have been the site of Genoese Simisso. The sent their three ships for a two-day trading period in 1421,

two towns will be discussed separately. but in the same year an Armenian colophon records a deThe town and castle of Samsun (the northwest quarter) vastating fire in the town. At the incanti of 1426 one of the

was probably established as a Turkish settlement after the three (now prudently armed) ships auctioned on the Seljuk “recapture” of the place in 1214. It is likely that it Trebizond and Simisso route went for a derisory 1d.—the passed into the hands of the pervane, and by the end of the normal figure had been about £100. The Genoese colony 1s century it was a possession of his grandson Masud Beg. After last mentioned in Italian sources in 1424 and largely left the Mongol withdrawal, it was ceded to the Isfendiyaroglu shortly after—perhaps before the Venetian incanti of 1426.

dynasty of Sinope. Bayezid I captured it from Ciineyd in The Genoese set fire to their base before leaving. 1392 or 1394 but the Mongols blocked all trade through the Nevertheless Ottoman defters of 1481-1512 show that six place in 1401.?° By 1404 it was in the hands of Bayezid’s son, Frankish households still survived in Samsun then.??

Mir Siileyman Celebi.?* The Isfendiyarogullari retook it in Sphrantzes was wrecked at Samsun in 1449. It was one of 1419, but it returned to the Ottomans and Mehmet I shortly the ports denied access to Constantinople during the siege of afterwards. Except for a Seljuk and Mongol mint in the 1452.°° Thereafter the, now presumably single, town fell into period 1233-48, Samsun does not seem to have been notably a decline until its astonishing resurgence as the port of the important although it offered the Ottomans, as it had the great Constantinople-Baghdad highway in the nineteenth Seljuks, access to the Euxine at a commercially important century. As late as the 1860s there was only a small Turkish place. After the Pazar Camu, the earliest Muslim monu- village on the shore and a smaller Greek suburb inland at ments in Samsun are an inscription of 1323 anda mosque of | Kadik6éy; their combined populations did not reach 5,000.

1503.75 By 1910 Samsun numbered over 40,000 souls, and Greeks,

Simisso (the southeast quarter) was relatively important, Armenians, or Franks controlled no fewer than 142 of its 156 however, and was not greatly troubled by events in neighbor- businesses and 85 percent of the shares of the Bafran tobacco

ing Samsun, only a “‘stone’s throw” or “‘arrow’s flight” away, market; there was a slight Christian majority in the until the early fifteenth century. Until then, the two places population.*! were an effective partnership of Italian capital and naval

expertise with Turkish merchandise and supply routes.?° MONUMENTS?” The earliest Italian—a Venetian—known to have visited the ; ,

; ; ; . 1. Amisos and Missos

place came, in 1212, when it was still Trapezuntine Aminsos.site ; ; on oo. the acroLo.station . , The main classical and early established Byzantine The Genoese of Simisso was certainly by ; . me 4:inac; .Genoese polis nowmaintained lies within a amilitary zone and is generally 1285. Here the consul and a fortified ; ae ; ; ; . cessible. The house long enceinte of in Hellenistic walls survived comptoir. A Franciscan reported 1320, 1334, and ; ; were sub.; stantially untilisthe late nineteenth century; there no1390.. It. ,may be an accident that the nineteenth-century — ; . not table remains in 1935 but few traces

Frankish quarter and Latin church were situated the siteor 33are, as —_ now.°* Thereinwere,

we have proposed for Simisso. Simisso

of them can be made out

of of rock-cut of (ora number variants the : tombs, ;

; ; as which twonow of the largest were later converted into on churches. name), it, ;was called Italian documents 6 was 1 The ‘“‘Tomb of St.inPeter” to the south of theand acropolis

portulans, offered a rival outlet to Trebizond, exempt from ; , im “395 . 97tariffs still aof cult 1905, as was the larger **Manastir1” on the the .high thecenter Grandin Komnenoi.”’ To the local ; ., ; ; on the strength a west side, which Cumont identified, of an

Greeks and Armenians, still probably the majority of the population, Simisso provided protection. George Scholaris, 29. Sanjian, Colophons, 150; Thiriet, Régestes, nos. 1811, 2021: Grand Logothete of Tr ebizond, who had Genoese connec- Mordtmann, s.v. “Samsun,” E//; N. Beldiceanu, ““En marge d’un

tions, took refuge there in 1363.78 livre sur la Mer Noire,” REI, 39 (1971), 392 and note 7.

30. Sphrantzes, ed. Grecu, 74; Ducas, ed. Grecu, 209, 307.

23. Schiltberger (1402), 12; M&M, A&D, II, 547; Wachter, 31. Bryer and Winfield, AP, 30 (1970), 252, 254. Verfall, 20; Evliya (1644), II, 39; Mordtmann, s.v. “Samsun,” E/!; 32. Beside travelers’ reports cited elsewhere in this section, the

Darkot, s.v. ““Samsun,” JA; Vryonis, Decline, 139. following have been consulted: Ritter, Erdkunde, XVIII, 796-806;

24. Clavijo (1404), 108. Beauchamop (1796), 144; Kinneir (1813), 308; Rottiers (1829), 247;

25. Mordtmann, s.v. “Samsun,” E//; Asikpasazade, trans. Aucher-Eloy (1834), I, 759; Stuart (1835), 346; Bore (1836), I, Kreutel, 127 f.; Wittek, Byzantion, 10 (1935), 41; Vadala, Samsoun, 292-95; Moltke (1838), 207, 212; Zacharia (1838), 310; Suter (1838),

13-18; Le Strange, JRAS, 12 (1902), 260. 434: Teule (1842), I, 437-38; Badger (1842), 14-16; Wagner (1844), 26. Finlay, History, IV, 323, who describes the situation some- 247; Hell (1846), If, 355-62 and plan in IV, 392-93, pl. xx; Finlay

what imaginatively. (MS, 1850), fol. 528; Van Lennep (1864), I, 40-51; Cunnynghame 27. Iorga, N&E, I, 359; Golubovich, BBB, Il, 265; Bratianu, (1871), 346-47; Tozer (1879), 5-10; Lynch (1890), I, 6; Warkworth Geénois, 156-59; Heyd, Commerce, I, 92-107; Pegolotti, ed. Evans, (1898), 54—55; Papamichalopoulos (1902), 311-29; Hawley (1910), 28, 63, 65, 91, 229 (Sivas); Cahen, Mélanges Halphen, 92; Bryer, AP, 303-5; and Childs (1910), 10-17. The MS in PRO/FO 526/14, 26 (1964), 296, 301; Cahen, P-OT, 166; Kretschmer, Portolane, 649; encouragingly entitled History of Trebizond, Samsun, ete., is in fact Bratianu, Actes, 80, 215—16, 226; Balard, Sambuceto, nos. 107, 171, an account of the commerce of Samsun from ca. 1866 to 1904, by the

174, 213, 236, 480, 676, 740, 769, 903. French consul there, H. Cartanze.

28. Panaretos, ed. Lampsides, 75. 33. Vadala, Samsoun, 13-18.

AMISOS, AMINSOS, SIMISSO, SAMSUN 95 epitaph, with a church of St. John Prodromos.** Traces of tary. Hamilton observed that the castle stood by the sea, wall painting were still visible in the larger cave. On the which came up to its northeastern corner; built at different acropolis were a number of cisterns, one 12 m wide with four periods, the lower part was ‘“‘composed of large square marble columns and tile and mortar vaults, which reminded blocks, while the upper part has been repaired with small Cumont of those in Constantinople. In 1959 D.C. W. saw stones. I do not believe that the former is Hellenic, but rather this cistern when bulldozing operations uncovered it and Byzantine, though constructed with materials derived from confirms that it appears to have been Early Byzantine work; the ruins of Amisus; the upper part is merely a Turkish in 1970 A. A. M.B. observed the uncovering and destruction restoration or addition.” *’ Smyth had the same impression: of substantial brick and mortar walling (apparently Early “The Turkish castle of Samsun has been built at two very Byzantine), and also a large vaulted building of well-faced different epochs; for the lower part is constructed of large stone (probably classical) set into the northeastern slopes of well-hewn stones, while the upper has more recently been the acropolis. In this area Hamilton noted a cave called 4 added, in a very inferior style of workmanship. But this mnyn and a stuccoed cistern and, a little to the south, ‘“‘re- building ... would have formed a very insufficient protection mains of a square building with a round tower at one corner, to the town in case of attack.” °? One must therefore specuapparently of Byzantine construction, with Roman tiles late that the lower courses of the castle represented the late mixed up with it.”"*> At the southern end of the acropolis Byzantine fortress of before 1194 and the remainder the Hamilton noted a ruined church of St. Theodore, which had Turkish additions of after 1214. been converted into a mosque, and Schmidt saw semicircular But what of the Genoese castle of Simisso? Van Lennep towers, marble slabs, and the remains of a temple with and Bryce thought they had seen it. In 1864 Van Lennep

columns and bas-reliefs.*° observed that a lighthouse had been built upon Samsun A number of classical floor mosaics have been found; none castle which was “‘as usual’’ attributed to the Genoese, that have been published and several are said to have been de- the surviving walls and towers of the structure were decaying

stroyed when a large part of the northern acropolis was and that the interior contained houses and shops.*? In 1869 leveled for a military installation and in street repairs in the the old Turkish quarter in the northwest of the town was northwest quarter of Samsun proper. In 1959 D.C. W. sawa razed in an attempt to eradicate malaria and other infecmosaic with a geometric border near Hamilton’s cistern; in tions.*° But in 1876 Bryce could find ‘‘no sight whatever 1967 similar mosaics and simple granite columns were except the remains of a fine old Genoese castle with mouldercleared away when bunkers were excavated for an American ing yellow walls, dating from the fourteenth century,” witha golf course on the acropolis. A surviving mosaic in a bunker, galleried courtyard.*’ Little reliance can be placed on the last

seen by A.A.M.B. in 1969, showed figures of the Four two reports. Pontic Turks habitually inform visitors that all Seasons within a geometric border (pl. 26a). A Greco- old castles are Genoese; Van Lennep and Bryce do not Roman mosaic, part of which depicts Thetis and Achilles, appear to have seen any more than Hamilton’s and Smyth’s was found before 1961 and is preserved in the Erkek Sanat part-Byzantine and part-Turkish castle. The fortress of Enstitusu (pl. 26b). Photographs of other classical mosaics, Genoese Simisso was probably destroyed with the Italian but no indication of their original or present whereabouts, departure from the place after 1424.

; 3. Other Sites

are available in the Samsun Museum. ; All our, andcoastal previous, observations suggest that there wasseems ; ad The settlement, rather than site the acropolis, noaesignificant occupation of the acropolis after the Early ,;to ; ; —_ be the more appropriate place for a kaunoc, of which there. is Byzantine period. This is contrary to the usual Middle i, ; 42

. ; mention in thewhen Life of St. market Clement of Ankyra.** monastery Byzantine experience, lower towns tended to beALoa ; oe tns Tévvac, to which John Mauropous refers in his Life of superseded by upper garrison fortresses, but the absence of 43 , . the ; as —- Dorotheos, cannot be located.*” The Life of St. Nikon later Byzantine building, of inscriptions, and of sherds we , : . in . . the . Penitentconfirm namesthat ) ypvon nétpa and have noted on the, acropolis the convenience of its a .; monastery 44 ; . settlement on the beach near the broken classical mole over-

hinterland.** As good a site as any for this would be Cakalh,

rode the defensive advantages of the acropolis 25 km southwest of Samsun on the old road to Kavak. Here

e pow: in 1836 Boré found a “‘vieille église de style byzantin,” *° of

2. Samsun which we can trace nothing today. The evidence we have, therefore, suggests that the later

Byzantine city moved down from the acropolis to the shore . a

at Aminsos before 1194, that there was a period of Greek- 37. Hamilton (1836), 289; Evliya (1644), II, 39, indicates a much

Ti , b h his site b kish larger castle. urkmen coexistence, but that t is site ecame Tur 1S 38. Smyth (1850), 150; Rottiers (1820), 250. Samsun after 1214 and that Genoese Simisso was established 39. Van Lennep (1864), I, 49. beside it later in the century. Hamilton, Smyth, Van Lennep, 40. Mordtmann, s.v. “Samsun,” E//.

and Bryce noted a castle in Samsun. Hamilton’s and Smyth’s OD bet C 70). wo accounts, in 1836 and ca. 1850 respectively, are complemen- 43. J. Bollig and P. de Lagarde, Johannes Euchaitarum Metropolitae quae in codice Vaticano Graeco 676 supersunt, Gesell. der

34. Cumonts, SP, Il, 111-17; Grégoire, BCH (1909), 4-6; Wiss. Gottingen, Abh. XXXII (n.d.), 210.

Schultze, Kleinasien, 157-65. 44. Lambros, NE, 3 (1906), 135. The place stood on the borders of

35. Hamilton (1836), 290-91. Paphlagonia and Pontos. | 36. Ritter, Erdkunde, XVIII, 805. 45. Boré (1836), I, 298.

Section V

THE IRIS DELTA, LIMNIA, AND THE PROBLEM OF KINTE The bay of Amisos is flanked by the deltas of the Halys and modate a “fine fleet.” > Lazaropoulos also suggests that it lay the Iris. The Iris delta is the more substantial of the two, for east of Chalybia and Oinaion.® So, on the basis of the ruins of the river has wandered over a wide alluvium before reaching a large monastery of St. Barbara, complete with mosaics,

the sea. Until the 1950s it was malarial but, unlike the Halys near Phadisane (Fatsa), Ioannides places Limnia there,’ delta is now quite heavily populated. The alluvium is rich but while Chrysanthos boldly indicates on his historical map an sandy; there is no eminence from which to view any part of | area marked “‘Fortresses of Limnia” between Phadisane and the delta. The fenced holdings of fifteen villages are inter- Oinaion, where he scattered for good measure no fewer than spersed with grazing lands for sheep, cattle, and water buf- fifty-two castles, all of which seem to be notional.®

falo. The villages are divided by stagnant watercourses and The only answer to the problem has, in fact, always been modern cuttings which have drained the four main lagoons, available. Limnia appears under various guises (Laliminia, where maize and sunflower are now grown. For 34 km the Liminia, Limonia, Lomona, Limonia) on most Italian and

beach 1s lined by high dunes where tortoises lurk. Greek portulan maps from 1318 until the sixteenth century, We propose that the great Trapezuntine stronghold and and on early printed maps (such as Ortelius’ map of 1580) administrative district of Limnia lay in the delta and, less and nineteenth-century historical atlases which followed the conclusively, that its known career had begun as Late portulans.’? They agree in placing Limnia at some point on

Byzantine Kinte. the coast of the Tptc (Yesil Irmak) delta.

The most detailed portulan, a sixteenth-century Greek A. THE STRONGHOLD AND DISTRICT OF description which apparently follows an Italian prototype,

LIMNIA places Limnia the most precisely.'° It states that, working

east from Amisos, the mouth of the Iris was 18 miles; 12 miles

SITUATION AND IDENTIFICATION further on was a cape called Gorgotzas or Simonites, and During the fourteenth century ta Aipvia was a. anim- Limnia lay another 15 miles on. Moreover, Limnia could be perial stronghold of the Grand Komnenoi, and trading port identified by a mountain called Kessarion, which stood controlling _b. a district of the same name with, according inland from it and afforded good pasture. Limnia itself lay on to Lazaropoulos, thirteen forts or fortified places,'and ec. flat, even, wooded land (still a precise description of the

a suffragan bishopric of Amaseia. modern delta). Then, moving southeast round the Iris alThe whereabouts of such an important place has vexed luvium, Lamperon or Thermodon lay 20 miles from Limnia. historians of Trebizond since Fallmerayer; Uspenskij de- Near Lamp cron Was a castle, anc inland a mountain called

voted a whole chapter to the problem.? Miller, but not Mazos, which was cut on the northern side. Finlay, accepted Gregoras’ statement that Limnia lay 200 The measurements in this portulan are usually no more stadia from Trebizond.* This distance of about 38.4 km east than approximate, but the relative distances ar e usually acindeed brings one to the conveniently named Biyiéik Liman curate. In fact, from Amisos to Thermodon, it is closer to 65 of Vakfikebir at the mouth of the Fol Dere. But Panaretos’ km than the 65 mi of the portulan. But the relative spacing many references to Limnia make it clear that it lay much

further west, beyond Oinaion,* and that it could accom- 5. Panaretos, ed. Lampsides, 77.

6. Lazaropoulos, ed. Papadopoulos-Kerameus, FHIT, 61. 7. Ioannides, Historia, 207; cf. G. Th. Kandilaptes, Ta Aipvia,

1. Lazaropoulos, ed. Papadopoulos-Kerameus, FHIT, 61; but PPh, 2 (1937), 60-61; and see the strictures of S. Kokkinides, Ta one reading could, however, attribute the thirteen strong places to Aipvia, PE, 3 (1952), 1581; and Sec. VIII, p. 113, on the monastery Oinaion. Balard, Sambuceto, nos. 767, 768; in 1290 Limnia exported of St. Barbara.

Cuman slave. . Phygades, 38.

muslin and wines to Caffa, and a certain Kale of Limnia bought a 8. Chrysanthos, AP, 4-5 (1933), map. Cf. Triantaphyllides, 2. Fallmerayer, Trapezunt, 303; Uspenskij, Ocherki, 90-99. 9. Kretschmer, Portolane, 649; Thomas, Periplus, 271; Delatte, 3. Gregoras, Bonn ed., II, 680; Finlay, History, IV, 379 note 2; Portulans, 1, 289; 11, 31-32; Spruner-Menke, Hand-Atlas fur die Miller, Trebizond, 48 (who presumably follows Gregoras in stating Geschichte des Mittelaters (Gotha, 1880), maps 85, 88, 89; that Limnia was ‘“‘some twenty miles west of the capital.’’) Tomaschek, K/einasien, 80; Bryer, AP, 24 (1961), 101-2.

4. E.g., Panaretos, ed. Lampsides, 66. 10. Delatte, Portulans, I, 31-32.

IRIS DELTA, LIMNIA, KINTE 97 can be interpreted clearly enough. The portulan’s measure- with our site.'® Theoretically, the only objection is that the ments and early printed maps strongly suggest that one of the mountains to the south do not rise from the plain for 30 km mouths of the Iris was then at the skala of Karabahce, where inland and none of their present names can be related to the swamps and streams today still indicate its former course. unique mention of Mount Kessarion in the portulan.

The cape of Topyot¢a-Lipovitov is almost certainly the Practically, there are more serious objections. A. A.M. B., Calti Burun, which stands by the present main mouth of the who first sought Limnia in the Iris delta in 1962, visited

Iris. Taslikkoy and the other fourteen villages of the area in

The Aapnnpov of the portulan is what a classical geograp- August 1971 and took aerial photographs of the delta a her described as “‘a great port called Aauupov, giving shelter month later. Enquiry and observation revealed nothing for ships and supplied with water.” '' It is associated by other which could be construed to represent Limnia. However, he authors of peripli with the sanctuary and cape of “HpakAsgtov, was unable to examine the now heavily cultivated lake beds the Heracleon of the /tineraria, plausibly identified by Miller in the wet summer. It is reliably reported that a number of with a Heracleum Burun.'* Although Greek names have antiquities had been found in the Tashikkoy area and that, lingered long on the coast (local Turks still called the Iris before the Dumanl Gol was blocked and turned over to ‘‘Lirios” in the seventeenth century),'* the name of this cape maize and sunflower, foundations of well-dressed masonry now seems to have been forgotten, although its whereabouts, were revealed above the surface of the swamp during dry projecting from the Semenik Gol into the sea, is clear summers when the level of the lake was low.'? One of our enough. Lamyron survived as Lamiro on the Italian por- reports arouses the strong suspicion that the masonry was, in tulans,'* but it is not mentioned in Byzantine sources. It is part, classical or Early Byzantine. the anchorage at the mouth of the Oepu@dav, which has It is unfortunate that recent physical changes in the delta given its name to the dreary township of modern Terme. since the eradication of malaria (most of the villages are The ancient Ospioxvpa lay, according to Strabo, further newly built) has probably now made it impossible to locate inland—it has been suggested at Cerkezkoy, which we have Limnia precisely; but reports and the portulans place it not visited.'°> But the fabulous Amazon capital was destroyed firmly in the Tashkkoy region. by Lucullus; it was still regarded as a town in Justinian’s time The administrative district of Limnia in the Empire of but is not mentioned thereafter.'° Mount Mazos survives, Trebizond can probably therefore be identified with a geohowever, as the Mason Dag; Hamilton had the temerity to graphically very distinct area: the extraordinarily fertile Irissuggest that the name reflects the Amazonia mountains of Thermodon delta, about 50 km long and 30 km wide at its

the Argonautica."’ greatest extent. Famous for its agriculture since classical Between the (Caltt Burun and the anchorage of time, in the eighteenth and nineteenth centuries it was the

Thermodon-Lamyron, where, according the our portulan, prosperous fief of the great Hazinedaroglu family, whose Limnia must lie (slightly closer to the former than the latter), mausoleum is at Carsamba, the delta’s modern capital. there is a single ska/a, Kuraba. It satisfies portulan measure- The attention which the Grand Komnenoi devoted to it in ments. The coast is low, flat and wooded and the only feature the fourteenth century suggests that it was then an equally is four small lagoons, close to the beach. Kuraba stands desirable province. Strabo speaks of the fine grazing on the between two of the former lakes, the Dumanlh and Kargal1. plain,*? which made it a natural attraction to the Turkmen The Dumanll is the larger, about 5 km long and, before it was enemies of the Trapezuntines. Among its thirteen fortresses

recently dammed, had access to the sea through a narrow mentioned by Lazaropoulos (unless they be simply thirteen channel. At one time it could have been an excellent natural towers encircling Limnia itself) may have been counted the

harbor. The Turkish 1:200,000 map is misleading at this castles at Lamyron and on Mount Mazos, but, apart from point. The skala of Kuraba lies 2.5 km north of Kuraba local reports of sites south of Carsamba, which we have not proper; it consists of one tea-house and a lighthouse on an visited, none can be made out today. Strabo pointed out that empty shore; the place marked Kuraba Iskelesi on the map is the whole delta has been created by the alluvium of its broad in fact the western end of the straggling village of Tashikk6y. rivers,”' and the meanders and abandoned mouths of the Iris

The map marks Taslikkoy, “Village of Stone,” as being a and the oxbow lakes of the Thermodon suggest that these ruined site. Using the map alone, it is the obvious site for the rivers have long obliterated any fortresses that stood by their

imperial stronghold of Limnia, which would have taken its fords. name from the small lakes which surround it. Kiepert placed

the classical Ayk@voc Aipnv of the White Syrians further to HISTORY the west on the Calt: Burun, but it might also be identified The earliest mention of Limnia as such comes with

11. Anonymous periplus, 29. Panaretos’ record of the death of the Grand Komnenos John 13. Bordier (1609), 112. 18. Arrian, 22; Anonymous periplus, 28; Ptolemy, Geography, ed.

12. Miller, 7R, col. 646; Arrian, 22.

14. Kretschmer, Portolane, 649. Muller, 866; Miller, 7R, col. 646; Kiepert, |: 400,000 map; Rottiers 15. Strabo, Geography, XII, 11, 15: Tarhan, Map. There are re- (1820), 252: ‘*We went down first to the ruins of Ancona on the river ports of a castle called Karpi Kale, near Terme. We have not visited Ekil” (Yesil).

it.16.19. Reports from Bay Emin Tuksal of Samsun, and Bay Ihsan Procopius, Wars, VIII, 1, 2. Nemlioglu of Trabzon (who used to shoot in the delta). 17. Apollonius Rhodius, Argonautica, Book II, line 977; 20. Strabo, Geography, XII, my, 15.

Hamilton (1836), I, 283. 21. Strabo, Geography, 1, i, 7.

98 SECTION V II there, on 17 August 1297.?? It was then evidently of some end of January to end of May 1369°° strategic importance for, during John’s reign since 1280, October 1379, when he “‘took control’ of Limnia.°° Chalybia, to the east, had been lost. Limnia therefore became

a Greek enclave between Turkish Samsun and Tiirkmen The timing and length of Alexios III’s visits strongly sugChalybia, a place of refuge for Greeks fleeing from the south, gest that the Turkmens were seeking winter quarters on the where the Church of Amaseia was, by 1315, in disorder. The fertile coast, as shepherds today come down from the mounbishop of Zalekon-Leontopolis had fled and the Christians tain pastures to the Semenlik Golu in the delta, and that the

there transferred their allegiance to Sinope. Amaseia itself Grand Komnenos was opposing them. Further west his was without a metropolitan, and in the same year its strategy was to attack the Turkmens when they first reached Christians asked for the bishop of Zela (apparently the only their summer pastures in May. one of its traditional suffragan sees to have survived) to fill But Alexios III could not shadow Turkmen transhumance the office.?* But instead, Kallistos, bishop of Limnia, was everywhere and by October 1379 seems to have been forced elected metropolitan of Amaseia in 1317. Prudently, how- to bargain for security in Limnia by marrying his daughter, ever, he stayed in Limnia and its surrounding villages, lands Eudokia, to the local emir, Taceddin; the final admission of of the Grand Komnenos.”* This is the first mention of | defeat came by 1386, when Panaretos describes Taceddin as Limnia as a see, when the most junior suffragan of Amaseia “emir of Limnia.’’*’ Between those two dates, in 1384 there became the effective metropolis. Limnia lay within the is a second and final reference to a bishop of Limnia. He Empire, but not the Church of Trebizond; it was therefore was called Joseph and had been consecrated by the “‘irrevcomparatively safe. A parallel case is found when the metro- erend”’ “‘pseudo-monk” Paul Tagaris, successive charlatan politan of Adrianople moved to the suffragan see of | Orthodox patriarch of Jerusalem and more-or-less genuine Agathopolis, which was in Byzantine territory, until such Latin patriarch of Constantinople. In 1375, when Tagaris

time as the Turks would yield the metropolis.” was passing through the Pontos and probably consecrated The isolation of Limnia, which could now only be reached Joseph, this presumptuous prelate styled himself bishop of by sea, made it an obvious place of exile. Tzanichites and a Tabriz (to which he had been appointed by the patriarch of group of nobles were banished there in 1340 and executed the Antioch).78 But Tabriz lay within the patriarchate of following year; the Grand Komnenos Michael was impri- Antioch and Limnia in that of Constantinople, so it appears soned there from 1341 to 1344.7° But its isolation meant that that Tagaris was characteristically exceeding his canonical it could also be detached by a rebel leader or local governor. rights. However, Joseph of Limnia’s position was officially Constantine Doranites was certainly kephale of Limnia in confirmed in October 1384, when he was also given, like his 1351, when Alexios III mounted an expedition against predecessor, the administration of his near-defunct metrohim,”’ and other probable rebel governors were Grand Duke polis of Amaseia.*? John the Eunuch, gaoler of Michael Komnenos, who raised a When did Limnia pass into Turkmen hands? The existence “large army” there in 1340,*8 and Basil Choupakas, who of an important bishop of Limnia in 1384 cannot be used as brought “‘followers” from Limnia in 1355.7? Panaretos de- evidence because the emir of Limnia was a client and ally of scribes Doranites as “exercising the headship” (kegadat- Trebizond; his wife would be expected to protect local iKEv@v), the same office (kemadatixiov) which he ascribes to Christians. Nor is there any evidence that Joseph was actuKabazites in Chaldia,*° which suggests that Limnia was not ally resident in Limnia; it is within the bounds of possibility a simple bandon but may have been regarded as one of the that, in its final throes, the once great metropolis of Amaseia

diminutive Trapezuntine themes. had been moved from the swamps of the Iris to Trebizond From the 1350s, the Grand Komnenos Alexios III seems itself. Perhaps Limnia had been in Turkmen hands before to have determined to keep Limnia under direct imperial Alexios III was obliged to “‘take control” in 1379; possibly control. He made six recorded visits there, the timing and the district was Eudokia’s dowry for Taceddin.*° Taceddin length of which are consistent and signficant. The visits were Celebi died in 1386 and was succeeded by his son Artamir (I

on: 2).41 Clavijo noted an emir of the same name in 1404.*7 An 22 September 1351 to late January 13523! Artamir (II ?) held one of the most important Trapezuntine 19 December to soon after 25 December 1356°? 35. Panaretos, ed. Lampsides, 77. soon after 6 January to mid-March 1357°° 36. Panaretos, ed. Lampsides, 79; Bryer, DOP, 29 (1975), 129. 6 December 1360 to about 20 March 136134 37. Panaretos, ed. Lampsides, 80. 38. M&M, A&D, II, 228; cf. Bryer, BZ, 66 (1973), 336-37 and

22. Panaretos, ed. Lampsides, 63. note 20. The Limnian adventure is not recorded in D. M. Nicol,

23. M&M, AD, I, 34-37; Wachter, Verfall, 17-20. ‘The confessions of a bogus patriarch: Paul Tagaris Palaiologos, 24. M&M, AD, I, 69-71; Vryonis, Decline, 291, 318. Orthodox patriarch of Jerusalem and Catholic patriarch of

25. Vryonis, Decline, 325. Constantinople in the fourteenth century,” JEH, 21 (1970), 289-99.

26. Panaretos, ed. Lampsides, 65-67. 39. M&M, A&D, II, 64-66; Vryonis, Decline, 291, 335. 27. Panaretos, ed. Lampsides, 70. 40. Perhaps indicated in Panaretos’ sour comment on the original 28. Panaretos, ed. Lampsides, 65. marriage negotiations between the Grand Komnenos and the Celebi 29. Panaretos, ed. Lampsides, 71. in 1362: “But nothing came of this for the emperor”’: Panaretos, ed.

30. Panaretos, ed. Lampsides, 70, 73. Lampsides, 74. 31. Panaretos, ed. Lampsides, 70. 41. Panaretos, ed. Lampsides, 80.

32. Panaretos, ed. Lampsides, 72. 42. Clavijo (1404), 109. But see now Elizabeth A. Zachariadou, 33. Panaretos, ed. Lampsides, 72. “Trebizond and the Turks (1352—1402),” AP, 35 (1978), 344 and

34. Panaretos, ed. Lampsides, 73. note 4.

IRIS DELTA, LIMNIA, KINTE 99 offices, that of Grand Mesazon, at the surrender of the city in convenient confirmation of the existence of the place in the 1461.*° The victorious Ottoman army took the coast road twelfth century was Idrisi’s mention of a “‘small town’”’ called

west from Trebizond thereafter, and the minuscule emirate (in Jaubert’s translation) ‘““Kendia,” seven days west of of Limnia was perhaps swept away then. Limnia itself lin- Trebizond.°? gered on the portulan maps, but its port, cathedral, thirteen Grégoire’s Kundu was noticed by Hogarth and Munro?®! fortresses and stronghold may be presumed to have been lost and is evidently the modern Kundir Kopru, a bridge over

in the malarial swamps of the wandering Iris. the Lykos (Kelkit), 25 km southwest of Mesudiye (formerly Hamidiye), which takes (or gives) its name to a village 3.5 km

B. THE PROBLEM OF KINTE south of the river. By road it is 63 km east of Niksar and That Limnia should have sprung fully armed into promi- 29 km west of Koyulhisar. This identification, upon which nence, ex nihilo, in the late thirteenth century is improbable, Chalandon built his account of John 8 movements in otherespecially as we have reports that the remains near Taslikkoy WIse unjustifiable detail, IS SO attractive that we are loth to

included walls of well-dressed masonry, suggesting a much disturb It. However, it presents Insup erable problems. earlier site. Similarly Kinte appears as an important place in Kinte was evidently a place of some importance, a Pontic the twelfth century, and, inexplicably, is not heard of as such polis. Yet, apart from Idrisi, it seems to have escaped menagain. What we know of twelfth-century Kinte fits in with the tion in any other source. Late Medieval Pontic p oleis (or even

known site and purpose of fourteenth-century Limnia. Very sm all towns ) do not go unno ticed. The possibility that tentatively and, if only for want of a better site, we propose Kinte (and Its district of Kintike) is another name for a

that Limnia began its career as Kinte. known town is ther efor ca veryt eal one. ,

Kindiur and its bridge (which, strictly speaking do not lie

SUGGESTED IDENTIFICATION in the Pontos) stand on a major classical and medieval road. Nevertheless, the place does not correspond with any station

The expedition of John Komnenos from Constantinople _ in the Jtineraria. If Kiindiir is derived from a Greek name, it

against the Danismendids of Neocaesarea (Niksar) in would have been something like KovvSodpoc, rather than 1139-40 is described by Niketas Choniates, by Michael the Kivtfj or its district of Kivtici). More important, if John Syrian, and in Prodromic verse. Choniates states that the Komnenos was worried by lack of provisions and Tiirkmen Byzantine army followed the Pontic coast to take advantage attack, he would have wintered in Byzantine coastal territory of local supplies and to avoid encounters with the enemy.** rather than in the heart of the Danismendid lands on the He does not state how far John followed the coast or highway between the Tiirkmen strongholds of Niksar and where the expedition turned south over the mountains to K oyulhisar, an indefensible position in a gorge overlooked Neocaesarea. On the winter solstice (21 December 1139), the by mountains rising to 1,500 m and with precarious supply army went into winter quarters in the moAet TlovtiKt ti) _jines (as he later found when investing Niksar) to the north. Kivtqj (described in another text as moAet tov movtov tii The Prodromic verse indicates that the Byzantines faced KivtiKT] Aeyopevn).*° The Prodromic verse describes the the rigors of winter in the mountains during February 1140, rigors of February 1140, when the army, having evidently left evidently after the Byzantine army had wintered at Kinte, Kinte, reached the Lykos and then began its abortive siegeof = and on their march to the Lykos. Combining the verse and

Neocaesarea.*®° Michael the Syrian states that the Choniates’ description, it seems most probable that the Byzantines and Danigmendids faced each other for six wintry mountains lay between Kinte and Niksar. The route months,*’ presumably the spring and summer of 1140. But fom Kiindiir to Niksar follows the Lykos valley and prethe Greek sources reveal that the Byzantine troops at sents no problems; the route from Niksar to the coast (either Neocaesarea were ill supplied, lacking equipment and sump- tg Fatsa, via Bartae, or to Unye, both a little east of the Iris ter beasts. John’s nephew and namesake, son of the sebas- delta) is very mountainous and is usually under snow in tokrator Isaac Komnenos, defected to the Turks. The — February. Niketas Choniates indeed makes no suggestion Byzantine army disintegrated and straggled into Constan- that Kinte was located elsewhere than on the Pontic coast,

tinople on 15 January 1141.*° along which John had been marching to afford security and Chalandon argued that John must have turned south from provisions. The most convenient place for him to winter

the coast near modern Ordu (Kotyora) and taken the = would have been the nearest stronghold to the Niksar Melanthios (Melet Irmak) River to the Lykos, for at the — (Negcaesarea) route in Byzantine lowland territory which point where the Melanthios and Lykos almost meet Gregoire could provide a port to Constantinople and winter grazing noted a place called Kundu. Here there wasa Late Byzantine for the horses, which he was later to lose in the mountains. bridge, on the piers of a Roman predecessor; the site was The Jake-harbor of Limnia was a port, while Unye (Oinaion) deserted in Gregoire’s day but was believed by him to have and Fatsa (Phadisane), the outlets to the Niksar route, a little been of some importance in the past.*” For Chalandon, a to the east, were not. The Iris delta, as Strabo observed, was 43. Ecthesis Chron., ed. Lambros, 26: Hist. Pol., Bonn ed., 37; famous for its horse grazing, while the Pontic Alps come too

Hist. Pat., Bonn ed., 96; Synopsis Chron., 579. close to Unye and Fatsa to provide equivalent pasture.

44. Nicetas Acominatus (Choniates), Bonn ed., 45. It is here that Idrisi’s reference to what appears to be Kinte

te nenas inane PG 133 cole 1340. i and note 8. is revealing. Written just before 1154, the Sicilian Arab 47. Michael the Syrian, ed. Chabot, III, 249.

48. Nicetas Acominatus (Choniates), Bonn ed., 47-49. 50. Idrisi, ed. Jaubert, 394. 49. Chalandon, Les Comnene, II, 177, 178 note 1. 51. Hogarth and Munro (1891), 730.

100 SECTION V geographer’s land itinerary from Trebizond to Constan- Following Idrisi, Kendia or Kandia should be found in, or

tinople may be summarized thus:> near, the coast, five-eighths of the way between Kerasous and Aminsos—or, at any rate, nearer to Aminsos than Kerasous.

Trebizond—Hirsunda (Kepaoovdvta, Giresun), two This brings us to the east side of the Iris delta and to the

days only major settlement in the 92 km between Aminsos and

Hirsunda—Kandia (in Nedkov) or Kendia (in Jaubert), Oinaion: Limnia. This long and otherwise featureless stretch

‘‘a small town,”’ five days happily makes it unnecessary to place too much reliance on Kandia (Kendia)— Ania (‘Aptvooc, Samsun),°** “‘a very Idrisi’s spacing. Further east, Oinaion, Boon, and Phadisane

small town,”’ three days were known to Idrisi, and it is unlikely that he would have

Ania— Sinybyli (Xtv@mn, Sinope), two days confused any of them with Kendia or Kandia.°® Similarly, Sinybyli—Samastry (Apaotpic, Amasra), five days Aminsos and Oinaion were known (rather better) to Samastry—Araklays or Haraklia (HpdakdAéeta, Ere- Choniates and it is highly improbably that he could have

gli), three days confused either with Kinte.>’

Araklays (Haraklia)—-Constantinople, eight days. The great Trapezuntine stronghold of Limnia sprang into prominence so suddenly, with its first mention by that name For Kandia, Nedkov proposes Neokaisareia, arguing that in 1297, that it is hard to believe that the place had not existed

it could be a misreading in the Arabic for Kasra or before. A personal possession of the Grand Komnenoi, they Kaisareia.°* As Niksar, however, Neokaisareia retained its doubtless cherished and refortified the site; did they rename prefix in the twelfth century. In any case, the identification is it also? highly unlikely, for the most striking aspect of Idrisi’s route 1s We tentatively suggest, therefore, that Idrisi’s “‘small that it otherwise follows the coast the whole way. To make a town” of Kandia or Kendia became Trapezuntine Limnia difficult detour to Neokaisareia and back would be quite and is also identical with Choniates’ Kinte and its district of

unjustified. Kandia or Kendia must surely be sought on or Kintike, where John Komnenos wintered between 21 near the coast, with all the other places listed. The itinerary 1s December 1139 and February 1140. Our argument rests on

rather erratically spaced,°° but all the place names are in too may suppositions to be accepted without reserve, but the

order. identification with Limnia is at least more plausible than that with Kundur Kopru. 52. Idrisi, ed. Jaubert, 394; but the transliterations in Idrisi, ed. Nedkov, 98—99, have been preferred here.

53. Idrisi, ed. Nedkov, 99, and note 309 on p. 148; ed. Jaubert, Constantinople in 1139 may be accounted for by the fact that it took

394, has the same transliteration. Nedkov argues convincingly for a land route. There is a major gap in the /tineraria along the Aminsos; it cannot be the Halys mouth, for Idrisi refers to that more Paphlagonian shore, suggesting that they took to sea. Is Idrisi’s correctly elsewhere and here calls Ania a town—albeit ‘‘very small.” route west of Sinope a maritime one too? There are two arguments

54. Idrisi, ed. Nedkov, 99, and note 308 on pp. 147-48. against the notion. First, he also states that the journey from 55. The day’s journey ranges from 28 km between Kerasous and Trebizond to Constantinople was 94 days by sea. (This is a fair Aminsos, and 94 km between Aminsos and Sinope, which would estimate; medieval ships normally took 10 to 14 days. The slowest suggest that Ania is too far to the east, were it not for the fact that recorded journey was 25 days and the medieval record was 44 days:

Idrisi’s distances can never be relied upon precisely, and have not northwesterlies sent shipping faster out to Trebizond than the been here. But the total journey of 28 days over 1,154 km may be return. Second, comparison of Idrisi’s times west of Sinope with compared with the 32 days taken by normally slower caravans over actual sailing times show that his route was twice as slow. In June the 954 km between Trebizond and Tabriz. It might be suggested 1389 and in March 1404, Ignatius of Smolensk and Clavyo, respecthat Kandia or Kendia is Kotyora (Ordu) and that Ania is the tively, each took 9 sailing days between Sinope and Constantinople, Anniaca (Koyulhisar) of the Itineraria, were it not for the fact that while Idrisi records 18 days for the stretch. Sea travel was not, both Kotyora and Anniaca lost their names before the Middle Ages, however, necessarily twice as fast as the land route, for it must be and that this excursus inland would make nonsense of the next remembered that Clavijo had spent 5 days in abortively sailing to section to Sinope, which would be at least five days by land from Trebizond in November 1403, before being driven back until naviKoyulhisar. More puzzling is that, while the whole coast was still in gation from Constantinople resumed the following spring, while Byzantine hands in ca. 1154, coastal communications by land west Ignatius was held up by contrary weather at Herakleia for 9 days. of Sinope are difficult. In 1972 A.A.M.B. found that they were See Miller, JR, col. 643; Clavye (1404), 101-10, 338: Ignatius of

impossible for modern transport and had to take caiques from Smolensk (1389), 82-87. Inebolu to Cide. On foot and on horse it is possible, though labo- 56. Idrisi, ed. Jaubert, 395; ed. Nedkov, 96—97. rious, for the traveler would be obliged to double up and down 57. Nicetas Acominatus (Choniates), Bonn ed., 296, 463, 689, valleys. The slowness of advance of John Komnenos’ force from 699, 842.

Section VI

OINAION AND THE DISTRICT OF CHALYBIA DESCRIPTION wide and 30 km deep. The lower slopes are heavily wooded The site of the classical and medieval Oinaion is modern 294 more densely populated to the east. To the south, forests Unye, a coastal town lying midway on the broad bay be- and permanent villages give way, as the Pontic Alps rise from tween the Iris delta and Cape Jason, 27kmeast of the mouth _—_W€St to east, to upland summer pastures.

of the Thermodon and 26 km west of Polemonion. To the

immediate west, Oinaion is sheltered by cape dytoc HISTORY

Nixodaos (Ainikola), and, 9 km to the east, by the Gxkpa Medieval Oinaion became prominent in the twelfth cenMntpoxodic (Metrepol Burunu)— perhaps a reminder that tury with the Seljuk and Turkmen aim to reach the Euxine Oinaion (never itself'a see) was the seat of the metropolitanof there and at Aminsos; it may indeed have been in Turkish

Neokaisareia from the twelfth to the nineteenth century, hands in the period 1157-75.’ But it was recaptured, when he removed to Ordu.' Cape Metropolis appears to supplied troops to Manuel’s expedition of 1175 against the correspond with the AunAntos of the peripli and with the Danismendids, and was regarded as the Pontic capital when Camila of the /tineraria.* It is now bare of any upstanding Andronikos Komnenos, future emperor, held it as a sort of

monuments. sovereign fief (having earlier maintained a fortress for a

The classical name of the place (shared with that of two Saltukid emir near Koloneia), for a brief period before Attic demes, a Corinthian stronghold, an Ikarian town, and 1182.'° It was the most easterly possession named in the a place in Elis)? was Oiovn. In Byzantine times it was more Partitio Romaniae of 1204 as part of the Latin Empire,'' but commonly called Oivaiov (“Yveov by 1605);* it appears on was taken, almost simultaneously, by Alexios and David the portulans as Omnio, Honio, Onio, Homo, Homorio, and Komnenos.'? But the Tiirkmens were already in the district tO Oiveov.* Choniates notes the (probably spurious) vinous and had controlled Neokaisareia, to the south, for almost a

; . 7 .;

connections of the name.°® century. Panaretos reports that they seized Chalybia, ‘‘so Oinaion owes its importance to its shipbuilding industry that all those places became uninhabited” '? during the reign

(first mentioned in the twelfth century),’ to its route to of John II (1280-97). This probably explains the almost Neokaisareia about 70 km over the mountain to the south, complete absence of medieval Christian monuments in the and to its hinterland of ) XaAvBia, famous from classical area. A dynasty of Chalybian emirs emerges by the early times until the nineteenth century for its ironworkers, who fourteenth century. Bayram Beg (fl. 1313-32) harrassed probably gave the area its name.® Chalybia may broadly be Trebizond.'* In 1341 the Grand Komnenos was exiled to described as the northern slopes of the Pontic Alps between Oinaion.'° The nadir of Trapezuntine fortunes in the area the river Thermodon and Cape Jason—an area about 70 km came in 1347, when Oinaion itself and St. Andreas (pre-

iy = Decline. 204: Ki 1813). 318 sumably Cape Jason) were lost.'®° In November 1357 Hac 2, Aten. 23. Anonumoney periplus, 3. Miller, IR, col. 646. Omar, son of Bayram, invaded Matzouka,"’ but the turning loannides, Historia, 207, notes old tombs and a number of churches

on the cape, where the Karaklar Dere debouches, but A. A. M. B. 9. Cahen, P-OT, 100, 117.

could find nothing of them. About 1.5 to 2 km south of the cape is a 10. Cinnamus, Bonn ed., 176; Nicetas Acominatus (Choniates), tall tepe, behind which there are said to be two chapels and, at a Bonn ed., 295-99, 462-63, 842. Choniates quotes an oracle, sup-

village called Asarma, about 3 km inland, a large, and therefore posedly referring to Andronikos I, for the subject arose “from a

probably nineteenth-century, church. place full of wine”’—1i. e., Oinaion: see C. Mango, “‘The legend of 3. Strabo, Geography, VIII, 1, 5; vi, 16; vi, 22; vir, 1; IX, m1, 25. Leo the Wise,” ZVI, 6 (1960), 63-64.

4. Tomaschek, Kleinasien, 80. 11. Tafel-Thomas, Urkunden, 476. 5. Kretschmer, Portolane, 649; Thomas, Periplus, 251-52, 271; 12. Lazaropoulos, ed. Papadopoulos-Kerameus, FHIT, 61, 118:

Delatte, Portulans, II, 32. Vasiliev, Speculum, 11 (1936), 6.

6. Nicetas Acominatus (Choniates), Bonn ed., 463. 13. Panaretos, ed. Lampsides, 63: Mote doikoucg yevéoSat tac 7. Idrisi, ed. Jaubert, 393; ed. Nedkov, 96—97, and note 298 on yopac OAac.

p. 146. Cf. Makarios (1658), II, 435. 14. Panaretos, ed. Lampsides, 63-4. 8. Apollonius Rhodius, Argonautica, Book II, lines 1001—8: and 15. Panaretos, ed. Lampsides, 66. Theophilus, ed. Dodwell, pp. xm and 162, s.v. ‘“‘Calibs.”” For a 16. Panaretos, ed. Lampsides, 68. discussion of the name, see Magie, Roman Rule, 1, 179; II, 1068-69. 17. Panaretos, ed. Lampsides, 72.

102 SECTION VI point in Greek fortunes came only a few months later, in a castle by the port and an inland fortress (presumably August 1358, with the diplomatic marriage of Theodora Caleoglu Kale) “‘on a neighboring hill summit.” *° Komnene, daughter of Basil Komnenos, to the Turkmen In 1445 a Burgundian crusade ran amok in the Euxine, and emir.'® Oinaion was restored then, if not earlier. By Geoffroi de Thoisy and a small fleet “‘coururent toute la Mer

December 1361 the Turkmens were well in hand, for Maior et prirent sur lesdits Turs pluseurs navires et ung Panaretos reports that “we went to to 6omitOKaotpov of — chateau nommé Onyo, lequel ilz brulérent. Et de la alerent

Hac: Omar, son of Bayram—or rather he joined us at veoir l’empereur de Trapezonde .... ”’*© The Burgundian Kerasous. We went to Kerasous from Chalybia by land, and sack of the castle in Oinaion (apparently without Trapethe emir Haci Omar and the Turks followed us in an almost Zuntine protest) and subsequent visit to the court of the servile manner.” '? It is difficult to know what Hac: Omar’s Grand Komnenos John IV suggests that the town had passed OONITOKAGTPOV was— possibly a fortified house in the later finally out of Greek and into Turkish hands between 1404 Derebey style of the coast, or merely the castle he lived in. and 1445, although the Burgundians were remarkably casual Caleoglu Kale, just south of Oinaion, is a possibility, for in identifying which of their targets was Greek, Italian, or Greek and Turkmen communities existed side by side there. Turkish, as subsequent complaints revealed.*’ Another possibility is Kekirkalesi (or Mahalle Kalesi), Evliya described Oinaion castle in the 1640s as ‘“‘a square about 13 km west of the small mountain and communi- stone building in the seashore” and recorded a tradition that cations center of Akkus (formerly Karakus). This formid- it had been built by the Grand Komnenoi.”® In about 1806

able castle, whose foundations are classical, would have the castle passed into the hands of Suleyman Zade dominated the medieval (but not the modern) route from Hazinedaroglu, pasha of Trebizond and Camik who transNeokaisareia to Oinaion and the mountain grazing lands formed it into a splendid palace that remained one of the above the rhododendron scrub which are particularly fine at chief architectural curiosities of the Pontic coast until it was this point. Jerphanion has proposed the castle as the site of | burnt down in about 1900.7? Oinaion itself flourished as the Mithridates’ great Kainochorion (to Kalvo y@piov) and port of Neokaisareia, particularly in the early nineteenth describes it in some detail (pl. 28). He comments: “‘Le site century. But a disastrous fire in 1839 gave Samsun preemiest un des plus sauvages que nous avons rencontrés en Asie nence and Unye relapsed into a backwater thereafter.*° Mineure.”’ 7° The present authors have glimpsed it only from

afar. There are also said to be ruins at Ahretkoy, 3 km to the MONUMENTS south of Kekirkalesi.*' About 9 km northeast of Kekirkalesi 1. Oinaion Castle is a village called Bayramlh, whose name may reflect that of The foundations of Siileyman Zade’s palace were deHaci Omar’s father. Alternatively the word OOMITOKAOTPOV scribed by Hamilton as being “apparently of great antimay suggest a more domestic fortress, of which the obvious quity.”°! We propose that they represent the remains of the

example is that at Boloman Kale (p. 114). castle described by Evliya, mentioned in the episode of 1445, On 8 October 1379 Alexios III betrothed his daughter and by Clavijo, and that they survive in the wall, still pointed Eudokia to Taceddin Celebi, emir of Limnia, in Oinaion.?? out as belonging to Siileyman Zade’s palace, on the With the loss of Limnia in 13 80-87, Oinaion became the northwest side of the old square of the town (where an old most westerly outpost of the Empire; Chalkokondyles con- plane tree is traditionally that beneath which the firms the fact in the 13 90s.*° But Oinaion could only have Hazinedarogullari dispensed justice), and near the sea (fig. been a Greek enclave in Turkmen territory and the emirs of 16). The site overlooks the elbow of the bay, where the Limnia and Chalybia commanded much greater forces than foreshore is best protected from stormy weather. The survivthe Trapezuntines ever aspired to. In October 1386 Alexios ing northern sea wall runs about 105 m, with five buttresses,

III's two sons-in-law, Taceddin of Limnia and the new and 1s thirty courses, or 11 m high at its highest point. It 1s Stileyman Beg of Chalybia, fell out. Taceddin invaded ade of regularly shaped basalt stones, about 0.27 m square, Chalybia with (according to Panaretos) as many as 12,000 With very thin wedges of stone, 0.05m thick, in the lime and men. The Limnian Turkmens failed, losing their emir, 3,000 sand mortar between the courses. There are no tiles, bricks,

men, 7,000 horses and many arms.?* or pounded brick.

In 1404 Clavijo noted that, apart from a suburban settle- Between the two most southerly surviving buttresses there ment of about 300 Turks (probably Gepni, nineteenth- is a blocked door, topped by a semicircular arch with neat century travelers also mention the village), the population of voussoirs. Only the upper 0.50 m is now standing above Oinaion “for the most part were Greeks.” It was then ruled gg ound level; the present ground level outside the castle wall by an archon of the famous Melissenos family who held both must be at least 1.50 m above the medieval level (pl. 27 a).

18. Panaretos, ed. Lampsides, 72. 25. Clavijo (1404), 108. 19. Panaretos, ed. Lampsides, 73. 26. lorga, Sarrazines, 33. 20. Jerphanton, Me/USJ, 5 (1912), 135* — 141*; 13 (1928), 39-40. 27. Bryer, BK, 19-20 (1965), 183 and note 32, 191 and note 61.

Cf. Strabo, Geography, XII, m, 31. See now Elizabeth A. Zachari- 28. Evlhiya (1644), I, 40. adou, “Trebizond and the Turks. (1352~—1402),” AP, 35 (1978), 344. _ 29. Bryer and Winfield, AP, 30 (1970), 243-48; O. Bora, Yeyil

21. Tarhan, Map; local information at Akkus. Unye Rehberi(Ankara, 1969), 69; S. Eyice, ““X. Hommaire de Hell ve

22. Panaretos, ed. Lampsides, 79. Jules Laurens,” B7TTK, 27 (1963), 98, 104, pls. 17, 30. 23. Chalkokondyles, Bonn ed., 64—65. 30. PRO FO 195/101; despatch of 9 October 1839.

24. Panaretos, ed. Lampsides, 80. 31. Hamilton (1836), I, 270-78.

OINAION AND DISTRICT OF CHALYBIA 103 The impressive medieval walling is the facing to a platform “magnificently built in stone.’’ °° In 1836 Hamilton found it upon which Suleyman Zade’s palace stood—now occupied __in ruins,*® but it seems to have been repaired shortly afterby houses and gardens. Despite Evliya’s statement that the ward, for Hell remarked that it was much frequented on feast

castle was square, Lauren’s engraving of 1846 shows that it days.*’ During the nineteenth century it received pilgrim

was not and the modern street plan and the shape of the gifts from Greece and Russia. It is last mentioned by mound behind the medieval walling suggest that the castle Cartanze in 1904 and still figures on British Admiralty in-

was originally triangular (fig. 16). structions to mariners.*®

The masonry of the medieval walling is comparable to that The structures on the mainland perhaps confirm that the employed in fourteenth-century Trapezuntine imperial for- islet was a pilgrim center of some importance, but the church

tifications at Kordyle and Rhizaion (qq.v.), but could be itself must have been exceedingly modest in size. By 1969 earlier. If Oinaion became an important Byzantine strong enough of the topsoil had eroded to make it clear that it was point aganist the Turkmens in the twelfth century, there is a basically a single-apsed chapel about 2.50 m long (3.40 m, distinct possibility that the origins of the castle must be including the apse), and was evidently entered from the west. sought then, and that it was here that the wayward Only the first course of part of the apse was then visible (for Andronikos Komnenos dallied with his most famous mis- treasure seekers had all but destroyed the remains by 1970), tress, Theodora Komnene, ex-queen of Jerusalem. but there was enough to show that it was a regular semicircle

2. The Church of the Theotokos with an inner radius of 0.74—0.76 m and that the walls at the

a . ; ; - base were 0.16III m thick. The foundation stones were Patriarch Makarios Zaim of rough Antioch, who visited ; oo. , ; ; bonded with a mortar of lime, pebble, sand, and pulverized Oinaion in 1658, was the first to note this church, which was ; , ere

32 brick.and There is everyaindication that the chapel then surrounded by, ; stone walls had garden.*? The(and ; probably Le the causeway and its adjoining cave) is medieval and that. it;

more modern church of, the Panagia, mentioned by Ritter, . plan. ; was not (Kinneir circular but of conventional seems to be another- .monument mentioned ; red earthenware 33 . In 1963 the islet was strewn withtwo coarse

3. Th Ni ; , .

Greek churches and one Armenian).*°* Theresherds. is no One trace of ; sherd with lead green glaze could have been

any church in the town today. ,

either Byzantine or Ottoman.

e Church of st Nicholas _. ; ; 4. There is a local report of a castle in the hills to the west of

The islet of St. Nicholas, still called Ainikola, lies opposite Caleoglu Kale, probably above the Zindan Dere. We have

Ainikola lighthouse, about | km west of Oinaion. The islet is not vite dit >P y connected with the mainland by the vestiges of a causeway of ,

rough stone and mortar, 47 paces long. Near the mainland 5. A castle called Ginca Kale or Gencgaga Kale is said to end of the causeway there are the remains of a structure stand about 6 km west of Ainikola Burun.*? We have not about | m square. Its walls are of uncoursed stonework and visited it.

. _pulverized ; 6. Caleoglu Kale The has earthenware in it. Possibly ; . ; lower ; Thislayer castle (named after an eighteenth-century its interior surfaces are rendered with two layers of plaster. :

Derebey) this was a smallatcistern. The5 flat plateau above the cliff has the . -. ; ; a: stands Kalekoy, km southwest of Unye and overlooking foundations of a rectangular building, about 12 x 26 paces ~ (Unye n: , . size. _ . Ta the caravan route southnorthwest and the is Piyapots River in Under the cliff and facing an artificial 40 . ; , 41 Dere).*° It may perhaps be identified with classical Caena. cave.. The The floor has been leveled with pebble and mortar and ; i is on a greatfor rock, of |striking aspect. The south and is carried castle over the seashore about msteep beyond the cave . east sides of the rock are and sustain such structures as entrance. The present beach level is more or less thewest same asfall , ; sheer for . remain; the north and sides that of the cave floor.sloping At the down back of the river. cave The is a whole small rock rock-isover ; densely ; ;50 m ; before to the

cut step, about 0.20 m high and 0.20 m deep. Both the step ;

; covered with an undergrowth of bramble, daphne, and the roof of the cave are rendered withinathe plaster made of , and , . thorn. It seems to have been barer early nineteenth aieislet , . ; century, the features of the castle could picked The itself is when no than 37 paces sizebeand its ;out ; . ; more clearly, formore a plan of 32 thexruins todayinwould require a

lime and pulverized earthenware. .

perimeter was entirely walled. The walls were about | m thick ,

and liberally pointed on the exterior to offer a smooth surface major clearing operation.

; ; a Near base of center the southeast side of its thehighest rock, .and apagainst stormythe seas. In the ofoutside the islet and , parently any of theat defensive walling, ;is.a large

. ° . 34

point (about 8 m above sea level) are the foundations the , ; | . classical rock-cut tomb in the form ofofa tetrastyle temple This church, described by Bzhshkean as circular and said Basd S Pe

church of St. Nicholas (pl. 27 b). having a square door and reliefs of eagles above the pediment by Ritter to be Byzantine and restored in 1629," was also 35. Makarios (1658), II, 435. visited by Makarios III in 1658; he describes it as being 36. Hamilton (1836), 279. 37. Hell (1838), EH, 369.

38. PRO FO 526/14: Black Sea Pilot, 401.

32. Makarios (1658), H, 435. Cf. Laura Ridding, The Travels of 39. Tarhan, Map; unsigned article “‘Ordu,” Turkiye Turizm, 5

Macarius (London, 1936), 110. (27) (August 1965), 52. . 33. Ritter, Erdkunde, XVIII, 847; Kinneir (1813), 318. 40. Hell (1838), II, 368; Arrian, 23; Bore, Unye (see note 29), 34. Bzhshkean (1819), 55; Lebeau, Bas Empire, XX, 486 note 3; 71-72.

Ritter, Erdkunde, XVIII, 847; Boré, Unve (see note 29), 73. 41. Miller, /R, col. 646.

104 SECTION VI (pl. 29); the type, so familiar in Paphlagonia, is here the rock-cut entablatures, and a rock-scooped water pan, which farthest east example on the Pontic coast. The tomb iscut ina is plastered. The entrance and first few meters of the tunnel rough cliff, about 14 m above its present footing, a maize cistern are also plastered. The tunnel cistern enters the rock garden. Below and to the right of it are two, or perhaps three, at an angle of 45°; it is about 3 m high and 2.75 m wide. rock-cut recesses, one or more of which may have formed Today only forty-five steps are visible, each about 0.25 m subsidiary tombs whose facing has now fallen. Hamilton was deep, the total depth being 11.25 m; D. C. W. estimated that told that within the main tomb there were paintings on either the total distance from the entrance to the level of the acside of the interior, “apparently of Greek saints.” +” To reach cumulated debris at the bottom would be at least 30 m. the tomb, would require a ladder longer than that at present About halfway down there are remains of an artificial blockpossessed by the Unye Fire Brigade, and D. C. W. and A. A. ing of the tunnel, creating an upper plastered section which M. B. have been frustrated on their several visits to it. may have been used as a storeroom, for space on the castle However, a certain amount can be made out from below. rock must have been limited and the upper section of the There appears to be only one layer of painting over a white cistern can never have been filled with water. plaster, now very battered. The outer bay of the tomb is To the south of the platform, before the cistern and below narrowed into a central one by two angled jambs, of which it, are some walls, which recent excavation by the local the right one appears to have a painting of a standing saint authorities have revealed to have belonged to a chapel. (The (or, possibly, the Panagia). The central bay, in turn, leads area 1s now completely overgrown again). The plan was a into a smaller inner bay which has a semicircular vault and is simple rectangle with rounded apse, most of which had fallen made narrower by two straight jambs. On the right hand away, and a transverse narthex running from north to south jamb is a scene which appears to depict the Koimesis. On the the west wall of which is formed by the solid rock. The ceiling of the central bay are paintings divided into four masonry 1s of ashlar blocks of the local yellow limestone, sections by white lines; but the vault is so blackened with soot fitted without mortar at the external face, and, as in the that it is impossible to make out the subjects. In general the Hagia Sophia at Trebizond, there is an occasional decorative paintings have a blue background, red outlines, and figures course of red limestone. In the west wall of the narthex there distinguishable by their lighter haloes. The tomb was evi- is a rock-cut niche, from which bones and other objects are dently used as a chapel or hermitage; there is no reason to said to have been retrieved. A round-arched doorway led

doubt that the paintings are not Late Byzantine. from the narthex into the naos, and on the north side of the Bzhshkean claims that the castle proper had four en- naos at the west end there was a small chamber which might ceintes; each of the outer three had one gate, the innermost have served as a tomb or possibly a baptistery; it is now so two gates.*° Today only three enceintes can be made out, the filled with debris that only the rough shape of it is clear. Both gate to the outer lying a few meters east of the rock-cut tomb. the naos and narthex were barrel vaulted with ribs at interIf Bzhshkean’s statement can be relied upon, an outer ward vals; it would appear that the external roof level corre(no trace of which can now be found among the hazelnut sponded with that of the cistern platform. groves) may have existed on the comparatively level ground Above the cistern to the south are rock-cut steps leading to the southeast of the castle. Practically, this would make through what must have been a third gateway into the citadel sense, for there is little room on the rock proper for housing a at the top of the rock, but no masonry now survives at this

garrison or beasts. point. Above and to the south of this conjectured gate is a

The present gateway is framed in the lower courses with second cistern tunnel of even more impressive dimensions well-cut blocks of the local yellow limestone laid 1n heavily than the first one. This is cut into the rock at an angle of 65° to sanded lime mortar. The blocks are rectangular and there is 70° and is 3.05 m wide at the entrance. It is entirely stepped,

no sign of long-and-short work such as can be seen in the like that of the lower cistern. The bottom cannot be disHellenistic work in Amaseia. At some later period the whole tinguished. In 1963 it took over fifty seconds before a stone gateway was reinforced with extra masonry and buttressed falling down apparently reached the debris below, and in

on the east side with a rounded bastion. The reinforcing 1971 forty-four seconds. masonry is of smaller blocks of waterworn stone laid in The heavily overgrown peak of the rock has stone-cut regular courses with a lime-and-pebble mortar. Its core isof | steps here and there and presents abundant evidence of an mortared rubble and has stringer beams across the wall, but upper citadel, now almost entirely collapsed. There is a ver-

there are no signs of cross tie beams. tical excavation into the rock which may have served as a About 15 m higher is a second gateway framed by well-cut third cistern. The masonry on the south side of the upper yellow limestone blocks, most of which have now gone. citadel is of roughly-shaped stones in regular courses, comWithin this enceinte, and about 15 m higher than the second _ parable to that of the reinforcing in the lower gateway. gateway, there is a platform cut in the rock, forming the Classical Pontic masonry so far east does not exhibit the entrance to a tunnel cistern; to the right of the entrance are same finesse as that found so obviously in sites such as Amaseia. The agglomeration of defenses on Caleoglu Kale 42. Hamilton (1836), 274, 277-78; Winfield and Wainwright, clearly belong to many different periods; the place, after all, AnatSt, 12 (1962), 134; information from local children who have has a history of two millennia. There is no masonry that ray Botahkean (1819). 55: Lebeau, Bas Empire, XX, 486 note 3: looks specifically classical Pontic (perhaps this was one of the Jerphanion, Mé/USJ, 13 (1928), 16, 22-23, 40; Texier, Asie Mineure, Strongpoints razed by Pompey), but the masonry of the

619-20. chapel and that of the lower gate are similar and are

OINAION AND DISTRICT OF CHALYBIA 105 medieval. What is clear, however, is that the rock-cut tomb of the fall of Amorion in 838,°° celebrated in hagiography; and cisterns are classical Pontic, that the castle was used in but it is probably unnecessary to seek a specific event. the Late Byzantine period, and that the chapel in the tomb Triantaphyllides and Ioannides identify the Kaotpo tic

can be expected to be of that time also. ‘Opyiac of Pontic ballads with Caleoglu Kale.°' The imposCaleoglu Kale, its impressive rock cloaked in undergrowth ing and inaccessible forest castle with its rock-cut tomb and and saplings, is one of many Pontic sites which will excite stairways and apparently fathomless cistern-tunnels would speculation. Perhaps its importance is not in the little it can naturally arouse popular imagination. As such, Caleoglu tell the field worker, but rather in its function in myth. Kale is no more remarkable than any other castle around Caleoglu Kale in Myth. Like a number of Pontic castles, which local tales gathered (like, for example, Zil Kale, q.v.). Caleoglu Kale has been popularly identified with to kaotpo But in this case its fame spread further. tTHS Qpyidc, the fortress of a beautiful princess. But it is In the West the tale of the ““Castle of the Sparrowhawk” is unique in having probably also been identified in European apparently first found in literature in Jean d’Arras’ romance romance with the legendary “‘Castle of the Sparrowhawk.”’ of Melusine (1382—94),°? but part of the theme must have Significantly, the tales have points of similarity, and at been known in 1366, if not as early as 1312. Melusine (Meli-

Caleoglu Kale, even actual contact, which deserve sande), supposed wife of Raymond count of Lusignan and

examination. ancestress of the Houses of Rohan and of Luxembourg, is Although the most famous example of Pontic akritic one of the most celebrated fées of French medieval romance. poetry, the so-called Ballad of Gabras, has been shown to be It must be remembered that the Lusignan family ruled largely the composition of a nineteenth-century Gabras,** Cyprus from 1192 and were titular rulers of Cilician Armenia the wealth of modern Pontic Greek and Turkish popular epic from 1396: Merlier, sister of Melusine, was condemned to and folk-tales clearly goes back to ancient tradition. imprisonment in the Castle of the Sparrowhawk in Cilicia Spyridakes has drawn attention to several references to until the Day of Judgement. She was guarded by a sparballad-makers in the Trapezuntine horoscope of 1336.*° to rowhawk, but noble knights were enjoined to keep the vigil them may be added a reference occurring in the earliest Life of St. John the Baptist outside the castle. As a reward they of St. Athanasios the Athonite (born ca. 920) that the family were entitled to the grant of any wish by the imprisoned of his “‘Colchian’” mother was celebrated in songs— princess, so long as their wish was “‘of earthly things.” >? presumably popular songs, for the author states that he does They could not ‘demand [her] body nor [her] love by mar-

not wish to discuss the matter further.*° riage nor other wise.” °* Knights who persisted in unsuitable Among the oldest and most widespread themes of Greek requests suffered ill luck to their ninth generation. In ca. 1366 popular poetry is that of a castle defended by a beautiful Mandeville embellished the tale and placed the castle not in princess against the Infidel for a long period (usually twelve Cilicia but between Trebizond and Erzurum. But he admityears). It would eventually be captured by treachery (by a ted (with curious honesty, for this armchair traveler) that disguised renegade who makes an entry into the castle, as in “this is not the right way to go to the parts I have named the story of the apostate Amazon Efromiya who plays this before, but to see the marvel I have spoken of.’ °* According

role in the capture of the monastery of St. Gregory to a more common tradition, the castle lay beyond Layays Thaumatourgos in the Melikdanismendname,*’ or by the (Lajazzo, now Ceyhan), near Pharsipee (now Persembe ”) diversion of a river, as in the ballad of Palaiomatzoukan and in Cruk (Korigos, now Gavurkoy).°® Here knights were Palaiokastro, g.v.**). The princess, to save herself from the required to wake the sparrowhawk and to remain outside the Infidel, leaps from the topmost tower castle to her death (as castle for three (sometimes seven) days and nights without in the popular tradition of the fall of Trebizond itself*”). sleep, sustenance, or company before asking the princess for Spyridakes finds origins for the theme in the Arab tradition any worldly wish. After one such vigil, a king of Armenia

- oo, asked for the princess herself, but he and his descendants

“ won a rot TeBoe Pc 958) 159.3 ie ‘the same. To were given eternal war. A poor man wished for wealth, which aKPITLKOV Enos Kal TO “Gopa tov Pabpa,” AP, 23 (1959), 33-38: he was granted in plenty, but a Templar who asked for a Bryer, Gabrades, 168 and note 22; and the same in A P, 30 (1970), 248 neverfailing purse of gold was destroyed with his Order.*’

note I. . _ . (The Templars were indeed suppressed, partly for tales of 45. G. K. Spyridakes, Hontat ONHOOMV GOLATWV EIS their wealth, in 1312). TpaneCovvta kata tov IA’ aidva, AP. 16 (1951), 263-66. , ; 46. Zitie prepodobnago Afanasija Afonskage, ed. 1. Pomialovskij In 1402 Johannes Schiltberger, a credulous Bavarian mer-

(St. Petersburg, 1895), 3: not in L. Petit, “Vie de Saint Athanase cenary in Bayezid’s army, was captured by Timur at the

l’Athonite,” Ana/Boll, 25 (1906), 13. battle of Ankara. When he finally retired home in 1427 he 47. Melikdanismendname, ed. Meélikoff, 1, 434-35.

48. Triantaphyllides, Phygades, 18-22, 29-35, 47; Legrand, 50. G. K. Spyridakes, To dnp@des dopa tod Kaotpov thc Chansons, 76-78; Marcellus, Chants, 1, 94-97; A. Papadopoulos, ‘Opiac, Ex. Aaoyp. Apy., 13-14 (1960-61), 3-34. ATP, 2 (1886), 391-92; the anonymous T’OQpyidc 16 kaotpov, PPh, 51. Triantaphyllides, Phygades, 32: loannides, Historia, 206-7. | (March 1936), 26-27; and the modern Pontic play, To Kaotpo tij¢ 52. Jean d’Arras, ed. A. K. Donald, Conte de Melusine (London, ‘Opidc, by K. Kallides, serialized in PPh, | (1936). There are similar 1895), 1, 15-16.

tales about the castles of Tarsos, Kordyle, Zara, and of Anna, 53. Jean d’Arras, ed. Donald, I, 15-16. daughter of the last Grand Komnenos. In Turkish, such castles are 54. Jean d’Arras, ed. Donald, I, 15—16.

often called **K1z Kale.” 55. Mandeville, ed. Hamelius, I, 97-98; II, 89. 49. The palace of the Grand Komnenoi in Trebizond is called 56. Mandeville, ed. Hamelius, I, 97-99: II, 89. “Kara Kizin Kale” (“Black Girl Castle’): Miller, Trebizond, 106-7. 57. Jean d’Arras, ed. Donald, 15-16.

106 SECTION VI described his captivity and subsequent Pontic travels. The relates how “‘we asked a man to take us to the castle and gave sources of Schiltberger’s stories have yet to be analyzed but it him money; and when we got to the place, one of my comappears that, like many western travelers who brought tales panions wanted to remain and keep watch. He who brought back from the East, he was not an inventive man but simply us advised against it; and said that if he did not carry out the put together first-hand experiences, garbled versions of local watch, he would be lost; and nobody would know where he stories, and tales taken from earlier travelers—or, as seems went; the castle is also hidden by trees, so that nobody knows to be the case here, all three elements. Schiltberger certainly the way to it. It is also forbidden by Greek priests, and they enlivened his account with lavish, but unacknowledged, ex- say that the devil has to do with it and not God. So we went tracts from Mandeville, the bogus “‘traveler’’ who had him- on toa city called Kerason.” °? There are hints of an element self transferred the Castle of the Sparrowhawk from Cilicia of truth in this account. The priests of Oinaion would have to near Trebizond. Schiltberger moved the castle still further, been naturally suspicious of the castle concealed by trees, to between Aminsos and Kerasous, and suggested that 1t was with its rock-cut tomb (albeit, possibly a chapel or herin Greek hands. When he was in the area,°® Oinaion was the mitage), which the Melissenoi held against Turkmens from most westerly town in Greek hands between Aminsos and the south. Could it not be possible that Schiltberger was also Kerasous, and Caleoglu Kale is the most striking and ap- told of Caleoglu Kale as being the Kaotpo tg ‘Qpytiac, and propriate candidate for the famous castle, which Schiltberger subsequently, after reading Mandeville, “identified” it with actually seems to have visited. After repeating the story of the the “Castle of the Sparrowhawk”’ of Western romance? Armenian king, the poor man, and the Templar knight, he

58. Clavijo (1404), 108. 59. Schiltberger (1402), 41-41.

Section VII

SITUATION quiet and prosperous backwater and Neokaisareia itself lies Neokaisareia, now Niksar, was successively Mithridates’ awkwardly distant from both the Pontic coastal communiKaPetpa, Pompey’s Atoonodtc, Augustus’ EeBaoth cations and the main Anatolian trunk roads to the south. Hadrian’s ‘Aépiavn, and the Byzantine Neoxatodpena. It Hence it has been forced into strategic prominence only lies at the eastern end of the broad valleys and alluvial plains twice: when first the omen s and ven the ae ks found i a which the Lykos creates in its search westward to join the Iris temporary key to Inner Pontos. But expeditions against at Mayvonodtc (Evnatopia, now Tasova). This region is Neokaisareia have always been hampered by troublesome Mavapo.a, the heart of Inner Pontos and of Mithridates’ supply lines and so have usually come to grief. Lucullus state, comprising more properly two valleys which are joined pound it eeu enough in 0 Bf and only took the place

by a narrow neck east of Erbaa. The eastern section is about Cause Mit m ates panicked. , ;

15 km long from east to west and up to 5 km wide: the Neokaisareia finally passed to Rome in A.D. 64, becoming

western section is rather larger a civil metropolis in the late second century. The fame of St. The great wealth of the Phanaroia gave rise to the towns Gregory the Thaumatourgos, evangelist of the Phanaroia, and cities of Neokaisareia, Erbaa, and Magnopolis. It is a patron and (from ca. 240) first bishop of Neokaisareia ene green island between the Pontic Alps and the Anatolian sured that the city also became the ecclesiastical metropolis uplands which has always produced a surplus of foodstuffs. ponte Polemonion. r gregory 's ane only Pontic saint Strabo notes that the foreland of Neokaisareia is the richest WHOSE stature the roae ; Ni at Co, the h ee ootiny in the Pontos, yielding abundant grain, oil, and wine; here Cappadocian Fatt ers, and NEw Kaesarea, the byzantine were Mithridates’ mines, palace, watermills, zoological gar- name for the city, 1s appropriate. His cult assumed the local dens, and hunting preserves.' Neokaisareia itself is placed on importance or at of " e old Pontic “erty Men. rom St a spur of the Paryadres? at the point where the steep northern Sregory i Ne kai . at cast ‘recente S 540 th etre. slopes of the valley break up into gently rising hills, making it po he a ° a foun caffras sted n 4 to th © me fey the obvious point of departure for the road northward over pols claimed four sullragans, recuced to mires aist the mountains to the sea. A fair road runs south to Comana Trebizond became a metropolis. But in the late eleventh Pontica and the Dazimonitis plain; at Neokaisareia this century a final military effort against the Turks iS marked by north-south route meets the military highway which runs an ephemeral expansion to ten suffragans. Like other inland eastward up the river valley to Satala.> An Arab itinerary sees, such as Amaseia, the Church of Neokaisareia had its finds Neokaisareia four days from Koloneia: in 1658 share of troubles: the metropolis was moved to Oinaion in Makarios reached it from coastal Oinaion in about the same the mid-twelfth century and had a series of difficulties from

time.4 1318. Thereafter, Neokaisareia itself lay on the outer fringe of Pontic Hellenism. By 1658 Makarios found Greekless

HIsTORY priests mindlessly mouthing the liturgy in the church of St. Nicholas in nearby “‘Argosti’”’ (Akkus ?), but in Neokaisareia

The economic and geographical factors which placed ‘‘we durst not appear ... in our character of Christians.” ° Neokaisareia where it stands are plain enough, but in an Anatolian context the Phanaroia is no more than an enviably 5. Magie, Rome Rule, 1, 333-37, has a good description of the campaign.

1. Strabo, Geography, XII, m, 30. 6. Gregory of Nyssa’s Life of Gregory the Thaumatourgos, in 2. Called “‘Bulgar Dag” in the Melikdanismendname, ed. Mélik- PG. 46, cols. 897 and 905 ff.; Schultze, Kleinasien, I, 165-71: off, I, 157-59, probably the Balahandag, part of the Greek Par- Laurent, CS, V (1), 356-63; Darrouzés, Epistoliers, 182-83, 229; yadres. Cf. Asikpasazade, trans. Kreutel, 225; Bryer, Gabrades, 179 Pachymeres, CSHB, I, 286; Dolger, Kaiserregesten, nos. 1590, 1610;

note 52. Chalandon, Les Comnene, I, 111; II, 638 note 7, 641 note 3, 650, 651 3. Miller, /R, cols. 669, 679. The roads run southwest to Seramisa, note 6; Gelzer, Texte, 539, nos. 241-44; 554, nos. 299-301; Parthey,

west to Pidis, and north to Bartae, Polemonium, Camilla, and Notitiae, 110, nos. 218-27; S. Petridés, ““Documents sur la rupture

Caena. de l’Union de Florence,” EO, 14(1911), 206; Mélanges Mercati, III,

4. Honigmann, A/PHO, 4 (1936), 263, 266; Makarios (1658), 215; Le Quien, OC, I, cols. 499-508; REB, 14 (1956), 103, no. 22; 437~—38. Grumel, Régestes, nos. 1041-43; Cumonts, SP, II, 259-73;

108 SECTION VII Despite severe earthquakes in 344 and 499, the city and its break-up of the Danismendid state, but it was the Seljuks Roman bridge survived the Early Byzantine period well. who took the pickings. Neokaisareia fell into Seljuk hands in Mustawfi described it as medium-sized with many fruit gar- 1175. Manuel promptly sent Andronikos Vatatzes and Dul dens. It appears in Hierokles, Justinian, and Constantine Nun to capture it—which they failed to do. Once again the Porphyrogenitus; it was represented at the Council of Byzantine expedition was ill-timed and ill-supplied and this Ephesos and stood fairly high (eighteenth in Leo VI’s time the disaster at Myriokephalon, in 1176, supervened. Notitia) in the metropolitan lists. It lay off the main routes of The Seljuks went from strength to strength in the area, Persian and Arab invasion and raiding. Comana moved to exterminating the last Danigsmendids in 1178. Kilig Arslan the safety of Tokat but there was no need for Neokaisareia to left Neokaisareia to one of his ten sons; Rukn al-Din took it move, for it was already a fortress town. But the area was far in 1197. There was no hope of its falling to the Grand from being safely Orthodox and Hellenized: there were local Komnenoi thereafter, for it had almost completely ceased to Jews, Paulicians, and Armenians—the latter had their own be a Greek city. The earliest dated mosque is of 1180, but

bishop in the thirteenth century.’ there is an earlier Danismendid medrese. The Goregi Buyuk In the late eleventh century Neokaisareia found itself in Tekkesi is thirteenth-century. Neokaisareia finally joined the the front line of Pontic resistance to outside invasion for the Ottoman Empire in 1397 and has subsided into a quiet second time. For over a century the place was a bone of backwater since.” contention between the Seljuks and Danismendids on the

one hand and the Gabrades (local Pontic leaders) and the MONUMENTS

Komneno! on the other: the struggle passed into epic in the It is probable that the whole Phanaroia would repay more Melikdanigmendname, where the Christian hero, Sah-i pattat intensive investigation than we have given it. There are, for represents, among others, St. Theodore Gabras. In the epic, example, suffragan sees such as Eunikos!° and Kokkos!! the Christian hero s Amazon daughter turns Tur kmen and, which we have not been able to locate; and the great pilgrim attractively disguised as a monk, lures “‘Gavras”’ and man- church and monastery of St. Gregory the Thaumatourgos, ages to seize the key monastery of St. Gregory the which was doubtless the most considerable monument of the Thaumatourgos at Neokaisareia, with which the place falls. area, has also eluded identification. It was still standing in In fact, Neokaisareia seems first to have been sacked by 1658 when Makarios “sighed to visit the ruined church Afgsin in about 1068; Roussel de Bailleul and Romanos IV which we saw at a distance, of the grandest architecture, and fought the Seljuks for the castle, but it was not back in Greek with the cupolas still existing .... The belief is spread among hands again until St. Theodore Gabras defended it in the the people, that a hot spring here, called Eboas, now at a 1080s. The Pontic martyr died in 1098, by which ume distance of several miles from the town, was formerly in the Neokaisareia had probably fallen to his great rival, the emir very center of the city. This place contains a stupendously Danigmend, who made it his capital. Here Bohemund of large church, of the most magnificent architecture, called Antioch was imprisoned from 1100 to 1103—1in 1101 a @avpatovpyoc, with many monuments still remaining.” ! Lombard crusade failed to reach the place. But control of This had gone by the time of the Cumonts: they noted only Neokaisareia was shadowy: Gr BOTY and Constantine the rude churches of St. Nicholas and of the Transfiguration Gabras at least claimed it before it firmly returned to at Niksar!3 which seem in turn to have gone.

— ghed to visit the ruined churches,

Danismendid hands am the I 120s. In 1139/40 John i If Makarios is right in placing the church of the Komnenos made It his objective in his abortive expedition Thaumatourgos outside modern Niksar, a possible canagainst the Danismendids; he transfer red a number of didate for the site is the substantial and certainly medieval Greek refugees from the area into Byzantine territory—a — masonry which lies by the Erbaa road about one kilometer Neokaisareites family appears in the thirteenth century—

and probably moved the see to the coast at the same time. . . ; 9; Chalandon, Les Comnéne, I, 11-12; H, 176-80, 506—7; Cuinet,

John’s expedition would have had better luck after 1142, 149. Cahen, P-OT, 27, 111, 115, 221, 240, 263; Polemis, Doukai, when Melik Mohammed Danismend’s death led to the Turquie d’Asie, 1, 735; Nicetas Acominatus (Choniates), Bonn ed., 47,61; Vryonis, Decline, 95, 107,115, 119, 122-23, 162, 221, 405, 441;

———__———_— Michael Attaliates, Bonn ed., 105, M&M, A&D, I, 551.

Theophanes Continuatus, Bonn ed., 72; M&M, A&D, I, 83; 10. Edvixoc appears as a suffragan bishopric from the late elevSchlumberger, Sigillographie, 291 (rejected by Laurent, q.v.); enth until the thirteenth century, in Parthey, Notitiae, 110, no. 224; Polemis, Doukai, 149 note 8; C. Astruc, W. Conus-Wolska, J. 209, no. 345; 251, no. 205. The name of its only known bishop, Gouillard, P. Lemerle, P. Papachryssanthou, and J. Paramelle, *“Les Anthimos, is found on an eleventh-century seal, in Laurent, CS, V sources grecques pour I’histoire des Pauliciens d’Asie Mineur,” 7M, (1), 265-66, no. 498. It is not in DHGE, or in Le Quien, OC. 4 (1970), 65, 167: BZ, 1 (1894), 253-54: Vryonis, Decline, 204—5, 304, 11. Koxkoc appears as a suffragan bishopric from the late elev-

307 note, 318, 320; Makarios (1658), 437-38; Gedeon, PP, 415-16, enth until the thirteenth century, variously as 6 KoxKov and 6

434, 449, 636. KyKkov, in Parthey, Notitiae, 110, no. 223; 209, no. 344; 251, no.

7. Cuinet, Turquie d'Asie, 1,735; Cumonts, SP, IH, 261; W. Ruge, 204. The name of its only known bishop, Nikephoros, is found on an s.v. ‘“Neokaisareia,” RE; CIG, no. 4186; Theophanes, Bonn ed., I, eleventh-century seal, in Laurent, CS, V (1), 264-65, no. 497. See 37; Justinian, CIC, Nov 28, preface; Hierokles, Synekdemos, ed. also R. Janin, s.v. ““Coccos” in DHGE; not in Le Quien, OC. It

Honigmann, 37, no. 702; Vita Eutychii in PG, 86, col. 2344; cannot be a misapprehension for KovKovods (Gokstn), suffragan Vryonis, Decline, 49, 52, 304; Lestrange, Lands, 147; Constantine bishopric of Cappadocian Kaisareia, for the two appear on the same Porphyrogenitus, De Thematibus, ed. Pertusi, 64; Anderson, SP, I, lists: see Parthey, Notitiae, 108, no. 182; 110, no. 233. But see p. 310.

56-59; Cahen, P-OT, 212. 12. Makarios (1658), 439. 8. Melikdanismendname, ed. Mélikoff; Bryer, Gabrades, 178-89. 13. Cumonts, SP, II, 262.

NEOKAISAREIA (NEOCAESAREA) 109 southwest of the town, but excavation would be needed to Basil (Ayvasil). Confirmation is provided by a small frag-

determine the identity of the building (pl. 30a). ment of wall painting in one of the western chambers. This Some scant remains of walling indicate that the Roman consists of a plaster ground of lime, with a straw or chaff town lay south of the present site, which is in foothills (pl. binding, and the shape of a halo of a standing figure painted 30b). A Roman town on the flatter terraces of the river valley upon it. Consequently, the plan in figure 17 is labelled itself agrees with the usual Roman practice. The fortified site “Church” and the side chambers *“‘Chapels.”’ lies on a spur above it; the flat top of the spur, which forms The second certainty is that this building was not designed

the citadel area, is protected on three sides by the natural fall as a church, as a glance at its plan must show. Cumont of the land. Its only vulnerable side is the north-northeast, suggested a bath. But there are no signs of piping or of the where the spur joins the main shoulder of hills. The western tenacious hydraulic plaster normally used for bath houses. side is the steepest; the medieval town seems to have stood — However, there is a third possibility. The plan of this building

largely on the east and south sides. closely resembles that of the tenth-century Georgian palace

The Fortifications and domed audience chamber at Geguti.'> The type of the Neither author has found any masonry on the citadel spur masonry (small stone courses, good vor’, simple molding, which can be ascribed to a Mithridatic or Roman date sparse brickwork) would be compatible with our proposition Cumont is not specific as to early remains. but there are that the building represents the audience hall of the twelfthreports of nolygonal masonry!4 a dof a ‘cistern tunnel century Danigmendid emirs of Niksar. The site 1S certainly

which were not located. Our own account of the fortifi- appropriate for such a structure. If so, this is where cations is by no means exhaustive and any proper study of it Bohemund of Antioch and the heroes of the Melikdanis‘n the future should be made in the context of the fortifi- mendname met in legend as well as in fact; its brief career

cations of Tokat and Amaseia would have ended with the Seljuk conquest. This would also

The main citadel enclosure on the flat top of the spur explain why a Turkish building in Anatolia subsequently comprises an area of more than | km from north to south by became, perhaps uniquely, 7 church. ; less than 100 m at its broadest section from east to west (pl. The other building in the south bailey is represented by the 31a, b). At the south end is the south bailey with remains of bon a : nineteen vantain : 35h) One ets waren ones walling round it. An enigmatic building stands in the south- Oasted 4 Pond ane SOUDEAIT APE 2207 nemne on

west corner (pls. 32a—34a) of the south bailey wall on the west side (pl. 35a) has a facing

Beyond the ruins of a nineteenth-century police station, oF roughly Sduare mone aie An Tegular courses. the south

and toward the south end of the castle hill, is the ruin of a ,; oo, e ; a:

domed cruciform building, the arms of which are divided bailey and the inner citadel, is now, like all structures within into rectangular barrel-vaul ted chambers (fig. 17). The ma- the inner citadel, a confused mass of fallen masonry (pl. 36).

; Visa: A wide archroughly at the east sideand of the wall; may sonry is of small stones, squared laidcross in regular ; ., represent

courses. The core is of mortared rubble, the stones of which the gateway to the inner citadel. The arch is rounded and the are well tamped in, with few gaps The flanking chambers masonry 1s of solid mortared rubble on the inside of the arch:

were entered by doors from the central area, but their form 1s the exterior face "oma aoe a ones oO.

not now clear since they have been robbed of all their facing 6 a qu .; stones. The chambers were lit by windows over the doors the western side the cross wall juts out to the northwest and a ; 4. has similar exterior facing, but thisonappears to be a repair. looking into:athe central area. masonry Two windows the south side . .; for a different of rough-cut stones in random were round arched, with stone voussoirs. One, on the north; courses can be observed below the laterworked work (pl.; ,34b). east corner, has brick voussoirs. The only carefully ; stone features that survive are corbels high up in the north- In the inner citadel there is a rock-cut cistern with masonry east and northwest walls of the central area. These corbels ae whch nas a hore in ue OP he rai or crawing hema have the form of a cornice with a simple receding profile, and b ok b die “hich tor “| : 8 t ‘rab th a sma ° Th may have carried supporting beams across the corners of the One ae ne PM the dent parr © d ‘ a aie de wi ‘h

central area (pl. 33b). Above the corbels, in the center, is a Face Nie ot b Le oe 7 tl " ‘d thick qnoct . Th triangular projecting block. If the central area was originally ragmen's OF broxen ores Orage maren onan als covered by a masonry dome (which should have incorpo- wall on the southeast side (pl. 37a) is of rough stonework laid rated some windows, or the flanking chambers would have in fairly regular courses, with series of tie-beams and string-

lain in forbidding darkness), these triangular stones may “rs h reeular Meee one ose a aes ner. represent the base of pendentives or squinches. To the south hi he, pare the vast wall of the town side of the northeast corbel, an earthenware jar is embedded w Th orth in lb . a the inner cit: del and the inner

in the masonry (pl. 34a), presumably as a means of lightening ee Wa OWE ; at ; the mass of the corner masonry citadel bailey stands to a considerable height; its construcWhat was this singular building? There are two certainties. 15. R. Mepisaschwili and W. Zinzadze, Die Kunst des alten

the first is that, at some stage of its career, it served as a Georgien (Leipzig, 1977), 49-50.

church. D.C. W. was informed that it was the church of St. 16. E.g., in the castles of Zil, Varos, and Koloneia (qq.vv.). Outside Anatolia, it appears in the Byzantine walling at Saone, ina 14. H. H. van der Osten, ‘Explorations in Hittite Asia Minor,” thirteenth-century Muslim tomb in Syria, and in the palace platform

AJSLL, 43 (1927), 135. wall of Tblisi, Georgia.

110 SECTION VII tion is of four types (pl. 38a, b). First, the masonry in the The Town northwest corner (pl. 37c) is mostly of small stones roughly Little survives of Byzantine Neokaisareia. In the upper squared and laid in regular courses, without any sign of town at the northeast side on the edge of the cemetery are the reinforcing beams. A peculiarity is the occasional use of ruins of three barrel-vaulted buildings faced with small much larger rectangular ashlar blocks. The second type of stones laid in regular courses. Local tradition ascribes these masonry (pl. 38a, right) consists of roughly-squared stone buildings to the Christians, but there is little to distinguish blocks laid in courses less regular than those in the first type their functions: they could even be nineteenth-century wareand brought up to an even surface with small flat stones. houses. In the gardens to the south of the town there was in Four square holes in the base of the wall probably indicate 1962 a short stretch of bonded brick and stone walling which drains rather than tie-beams. The quoins are of ashlar may be ascribed to a Roman or Early Byzantine period. blocks; nearly all have been robbed. The third type of ma- Nearby, on the west side of the main road, a marble sonry (pl. 38b) employs regular horizontal courses of tie- Hellenistic sarcophagus in 1968 was serving as a fountain. Its beams which penetrate to the external face of the masonry. ridged lid imitates round-ended roof tiles; the akroteria bear The stonework is much the same as that of the second type rustic scenes. There is no inscription. except that it lacks ashlar quoins; much of the heavy lime In the little garden cemetery surrounding the tomb of the mortar pointing remains in good condition. The fourth type Melik Ghazi, there is a fifth- or sixth-century Byzantine of masonry is similar to the third but has quoins like the impost capital (pl. 40b). This is the only piece of Byzantine second type. One column ts used as a header. The wooden ornament which we have found in Niksar. Its date probably stringers and tie-beams were used so that stringers, concealed coincides with that of the pilgrim church of the Thaumatby lime-mortar pointing, line the outer face of the wall. The ourgos; one may only speculate that it came from it. mortar has mostly fallen away to expose the beam ends (pl. The richly decorated turbes of the Danigsmendid and Seljuk 38b). This was an inefficient method of wall construction: periods and the size of the Ulu Cami indicate that Niksar lime mortar does not adhere well to wood, and once it has enjoyed considerable prosperity in the twelfth and thirteenth fallen the exposed beam soon rots or can easily be hacked or centuries. The turbe in the north bailey is a good example of fired by besiegers. This crib-work of stringers and tie-beams this kind of monument. The turbe of the great Melik Ghazi is repeated, without the herringbone pattern in the masonry, Danismend, captor of Bohemund, has been largely rebuilt. in the wall along the southeast side of the inner citadel. In the garden around it ts a fine collection of early Turkish The most impressive surviving stretch of wall, and prob- headstones. ably the greatest work in the fortress, is the cross wall between the inner citadel bailey and the north bailey (pls. 37b Dates: The citadel would need excavation to determine and 39b). This wall is up to 5 m thick; its core is of mortared with any certainty the sequence of building. The polygonal rubble well laid in and tamped down without gaps. Afewof masonry reported there should be Hellenistic, and Cumont’s the facing stones have not been robbed. The setting bed brick vaulting may be Roman—or perhaps he may be referretains the shapes of large ashlar blocks of the local grey ring to the ruins of the small baths described above.'® Mr. stone, laid without mortar at the joints on the external sur- David Wilson kindly reported banded masonry of brick and face. The stones were of regular size but laid in the long and stone (with five brick courses grouped between the stone) short pattern, so that one course is a facing and the next is which must be late Roman or Byzantine work. The cross wall bonded flat into the wall, narrow side outward, as a line of — and ditch which separating the north from the inner citadel headers. The bastions are rectangular. The gate is at the east bailey are similar to some sections of the walls of Antioch— end, lying some meters back along the east wall in the recom- also undated. This could be Justinianic work; the masonry mended manner, so that an attacker had to approach it with technique of facing and header stones is paralleled in his right side exposed to fire from the battlements.’ ’ In front Justinianic walls in Africa.'? But it could also be Middle of this cross wall is a massive ditch faced with masonry. It is Byzantine or Danismendid walling and the latter possibility now largely filled in. At a later period the upper sections of _ has the supporting evidence of ashlar facing of a similar type the cross wall were repaired with small rough-cut stones laid found in the turbes of the town. The smaller stone work (with

in random courses (pl. 40a). or without reinforcing beams) may represent the theme castle The west wall of the north bailey projects as far as the of the Middle Byzantine period and has parallels in Amaseia. modern town cemetery. It is made of a mortared rubble core The prow-shaped towers of the town wall may also be faced with small rectangular blocks in regular courses. It is Middle Byzantine, since this type of. fortification (first

about 1.75 m thick. recommended by Philo)?° recurs in other Byzantine The general circuit of the walls of the north bailey and of fortifications. the town is still fairly clear. Within it internal walls running

up the slopes of the eastern side of the citadel may have divided the wards of the town. There are rectangular, poly-

: “din: . Cumonts, SP, II, ,

gonal, and prow-shaped towers (pl. 39a). The last are faced IBC is. SP. 11.250

we roughly-shaped stones laid in irregular courses. Ashlar 19. C. Diehl, LAfrigue byzantine (Paris, 1896), 149, fig. 1 and

ocks (almost certainly reused) serve as quoins. note 4.

20. A. de Rochas and C. Graux, “‘Philon de Byzance, Fortifi-

17. Vitruvius, On Architecture, I, v, 2. cations,” RPh, 3 1872), 156 (for the Anonymous of Byzantium).

Section VIII

PHADISANE-POLEMONION AND THE DISTRICT OF SIDENE DESCRIPTION (the last specified as a river).’ Although the name of Between Chalybia and the substantial promontory of Cape Polemonion is remembered today in Boloman Irmak and Jason lies ) X1t6nvH, whose classical name did not survive in Boloman Kale, the Kale received Its name very recently and

the Middle Ages. The Sidenos or Polemonion River does not represent the ancient site. Phadisane is, however, (Boloman Irmak) flows through the district, correctly de- more or less represented by modern Fatsa, an unlovely stragscribed by Strabo as fertile and largely low-lying,' to the sea gling town with a comparatively good anchorage; there 1s

where it emerges in the center of a shallow bay flanked by slight shelter also at Boloman Kale. modern Fatsa, 2 km to the west of its mouth, and Boloman _The district of the Sidene 1S heavily wooded, with conical

Kale, 5 km to the east. Ancient Sidene had three coastal hills and hillocks typical of this part of the coast, but the strongholds: 4 Zi8n, XaBaxa, and @aBsa.2 Side was named foothills of the Pontic Alps do not rise for more than 20 km TloAgumviov (or ToAguwvn), either by Polemo I or, more inland. probably, Polemo II, who relinquished his lands to Rome in

A.D. 64.*° The name of Phabda is perhaps remembered in HISTORY

modern Fatsa. Chabaka cannot be identified. Polemonion, capital of the classical Pontic Polemonion, Polemonion is one of the handful of eastern Pontic was clearly a major settlement in antiquity—at any rate by classical cities which would have boasted the appurtenances Pontic standards—but its long decline may already have and public buildings of the substantial sites of western begun when in the third century it was Neokaisareia that Anatolia. As has often been observed, it is represented today became the ecclesiastical metropolis of the province to which by a series of mounds which run for 2 km along the coast Polemonion had given its name, and Polemonion itself a from Fatsa to Boloman Irmak. Although it is probably the suffragan; the final indignity came when the metropolitans of most important classical site of the coast, it has never been Neokaisareia removed to the comparatively secure Greek investigated. No inscriptions and, apart from a rockcut tomb coast not to Polemonion, but to Oinaion, which had no noted by Hamilton,* no finds have been reported. We record sufragan even, in the twelfth century. The see of Polemonion below only one inscription and two churches (the latter of — 1s found in lists until the thirteenth century. Seven bishops some significance) from in or near the site which will be lost are known by name, the last, Andrew, in the tenth or eleventh in the creeping suburbs of modern Fatsa in the next few century.® The see probably did not survive in fact much

decades. longer, not so much through Turkish pressure on the coast or Classical geographers place the “‘fortress”” of Dadioavn, the collapse of the metropolis of Neokaisareia and its as@Maddio0a,° Pytane, or Fitane,° and the “‘city”’ of Polemonion sumption by Oinaion, as because of the probable desertion of

so close together to the west of the mouth of the Sidenos the site of Polemonion itself. The place and its bishops do not River (corresponding to modern Fatsa and the ruins in the figure in Trapezuntine sources. eastern suburb) that it seems likely that one was the scala of By the twelfth century Phadisane had certainly eclipsed its the other. Nevertheless the twin sites retained distinct names twin settlement of Polemonion.? The inland route, which led

on the portulans, where they are called Uatiza, Vathiza, eventually to the great Seljuk emporium of Sivas, gave Vatiza, Vatisa, Uatica, Fatiza, Fadida, Fadnica, Fadissa, 6 Phadisane a brief prosperity upon which the activities of a Matiodc, } Padioa, or h Badioa (the last specified as a port); Genoese notary, Federico di Piazzalungo, shed unexpected and Pornom, Pormom, Pormoni, Porman, and 6 Ileppop light in 1274. In the early decades of Genoese trading in central and eastern Anatolia there was evidently work for 1. Strabo, Geography, 1, m1, 7; IH, v, 25; XII, m, 14-16. 7. Kretschmer, Portolane, 649; Delatte, Portulans, 1, 238; II, 32:

2. Strabo, Geography, XII, m, 14-16. Tomaschek, Kleinasien, 80. 3. Magie, Roman Rule, 486, 561-62. 8. Ptolemy, ed. Muller, 873; CJL, III, no. 6818; Hierokles, 4. Hamilton (1836), I, 270. Cf. Schultze, Kleinasien, II, 183; Synekdemos, ed. Honigmann, 37, no. 702; Gelzer, Texte, 539, no.

Ritter, Erdkunde, XVIII, 843—44; Kinneir (1813), 321. 243; Le Quien, OC, I, 515; Laurent, CS, V (1), 366, no. 499.

5. Arrian, 25; Anonymous periplus, 30. 9. Idrisi, ed. Nedkov, note 301 on p. 147, mentions ‘‘Fatisa,”’ but 6. Miller, 7R, cols. 646-47. not Polemonion.

112 SECTION VIII itinerant notaries traveling between isolated groups of merchants were looking further east, to Tabriz, for their Italian merchants, before resident professional notaries commerce and to Trebizond for their port. By then, too, the became available. Federico di Piazzalungo is first found in Turkmens may have made the Sivas-Phadisane route too Lajazzo in early 1274 and at Sivas on 19 July of the same dangerous. year. Less than a week later—and this suggests that he must What appears to be a loose enceinte of fortresses round have traveled very fast on a direct summer route (perhaps Phadisane and Boloman Kale, with its apparently thirteenthSivas-Hafik-Ipsile-Resadiye-Aybasti-Fatsa), he reached century church, may have been established at this period by Phadisane, setting up some sort of record. But the fact is that Trapezuntines against Turkmens from the mountains. he drew up four documents for the Genoese of Phadisane on Phadisane retained some importance as an anchorage 24 and 25 July. By 21 August he had reached Sougdaia beside the overgrown wreckage of Polemonion. A sixteenth-

(Sudak) in the Crimea, presumably by ship from century portulan mentions two churches there, dedicated to Phadisane.'° The documents he drew up in Phadisane large- — the Theotokos and to the Prodromos respectively,'* which ly concern loans, but also indicate a fairly substantial trade were replaced 1n the nineteenth century by the church of St. (in unnamed commodities) carried by Genoese ships from George, which still stands.'® Evliya found three hundred there to Amastris (Amasra), Constantinople, and the houses at ““Fatsha,’”’ mostly Greek,’ ’ and Hell a small ruined Crimea. A total of twelve different Genoese are named as fort, of which there is now no trace.'8 parties or witnesses to contracts, indicating a comparatively

large colony for this date, but the merchants had not reached MONUMENTS the stage of local organization of having a factory or loggia 1. Bucera Kale (fig. 18) of their own and were doing business in the house of one of This building lies 7 km east of Fatsa, 500 m south of their number In P hadisane, Guglielmo Mastraccio. In Baskoy. It is reached from the coast at Kizilcerkes (Kizilot). Sougdaia Federico di Piazzalungo drew upa document for The Kale is a more or less rectangular construction, about another Genoese resident of Phadisane, Nicoleta 59 y 38 m, in a maize field and on no particular eminence.

d’Albenga."’ . The walls average a thickness of 1.10 m and stand up to a

The chance survival of Federico di Piazzalungo’s docu- eight of 7m; there are few remains of them on the south side ments reveals that Genoese merchants were active in 1274 and only traces on the north. They are built of rough stone

both in Phadisane and in the emporium of Sivas. At with lavish use of mortar. The exterior is whitewashed in Phadisane one Genoese had a ship, the San Giovanni. In yarts. In an angle of the east side are a series of windows 1267 a Genoese ship, described as “de Savasto” (1.e., Of Which, although they are of faced stone, opening inward in Sivas), was pillaged, and it was probably from Sivas that a the style of nineteenth-century Pontic churches, appear to be cargo of ginger was exported to the Champagne a little later. :ore domestic than defensive. There are beam holes and the There is further mention of Genoese activity in Sivas IN traces of a door on the west side. Despite local opinion that 1280. o Through which Euxine port was the merchandise of — Bucera Kale is a very old castle, it appears to be no more than

Sivas carried? The earliest definite mention of a Genoese _a substantial fortified farmstead and yard, perhaps built by station in Trebizond does not come until 1285, and in an eighteenth-century aga. However the site of a kule (fortiAminsos until 1289. It seems likely, therefore, that for a brief fied tower) was pointed out ona steep eminence about 750 m period in the 1270s and 1280s Phadisane was used by the _ to the south-southeast of Bucera Kale, and another supposed Genoese as their principal Euxine outlet for the commerce of castle, called Cingutkaya or Cingirtepe, on a lower hilltop Sivas. Phadisane was then part of the Trapezuntine Empire about | km to the north-northwest.'? Both sites were heavily and Genoese activities there do not appear to have been wooded, difficult of access and were not visited. authorized by the Grand Komnenoi, but the Genoese may have chosen the place because imperial control was slight in 2. Evkaf Kéyti Kale*® (pl. 41a, b) the area for the Tiirkmens were overrunning the Chalybian The castle overlooks the route south and the Fatsa Dere, 5 interior in the 1270s and 1280s.'3 Possibly Phadisane and the | km south of Fatsa. On the east side of the river, about 500 m Sidene was a no-man’s-land; certainly Genoese documents of 1284 and 1290 indicate that its population was mixed. The ° peatte, vor rulans, M503 307: th ‘ened article “H documents mention, respectively, “Echisene,” daughter of gio “be 4 (1950), 07. Brver and Winfield. 4P. 30 (1970). Corcha,” and “‘Yerena,” daughter of ‘“‘Murit,” as inhabi- 14243. tants of Phadisane.'* It is impossible to say whether these 17. Evliya (1644), I, 40. names are Greek (‘‘Yerena” could be Eirene or kyr’ Anna), 18. Hell (1846), II, 369-70: also in Texier, Asie Mineure, 619. The Armenian (‘“‘Corcha” could be Krikor), or Turkish (as fort may be represented by a section of the east wall of Polemonion, “Echisene” sounds). After 1290 there is no further mention whieh The vnsien wavicke “Orde.” Tiirkiye Turizm, 5 (27) (August of Phadisane (Vatiza) in Genoese sources, but by then Italian 1965), 61; Tarhan, Map. 20. This is the site apparently described as Goregi Manastiri in

10. Bratianu, Recherches, 158—59, 172-73, 302-8. Turkiye Turizm (see note 19), but there is no monastery there, and

11. Bratianu, Recherches, 205, 308. extensive equiries in Fatsa and on the site do not evince any recog-

12. Bratianu, Recherches, 166-67. nition of the name, or of those of other sites listed: Kiz Kulesi 13. Panaretos, ed. Lampsides, 63. (supposedly 15 km from Fatsa, near Yalikoy), and Akkaya Kalesi 14. Bratianu, Actes, 172; the same, Recherches, 172—73; Balard, (supposedly 20 km south of Fatsa, and illustrated as a fortress upon Sambuceto, nos. 416, 424, 434, 55S. a steep rock, with curtain wall and bastions).

PHADISANE-POLEMONION, SIDENE DISTRICT 113 south of Evkaf Koyu, a track branches right; the castle 1s we have.?* Hell writes: about 350 m further on. It stands on a heavily wooded hill and is approached from the east. The western and southern ... une église qui présente une construction tout a fait étrange sides rise steeply from the river and the castle commands the pour moi: elle est de forme octagonale; chaque face se compose surrounding countryside as far as the sea and for about 5 km d’un arceau a volute cylindrique; les archivoltes de ces arceaux

to the west and south. Traces of up to four enceintes of walls reposent sur des piliers orn€s d’une petite corniche ayant la

survive, which are built of rough stone with much lime forme de la figure 6, et percés en A d’une ouverture formant mortar and, unusually, little pebble. The two outer enceintes passage et faisant comamaniqner sans doute a une petite galerie , - ; régnant autour d’une partie de l’édifice. Au-dessus des arceaux, have small semicircular bastions and enclose a roughly les murs, tout en conservant la figure octagone, s’elévent vertioblong area of about 200 x 50 m. The single, well- calement. Ils sont en ruines et ne permettent pas d’apprecier la preserved, gate is in the second enceinte on the east side; it 1s naissance de la coupole qui recouvrait évidemment l’édifice. flanked by bastions of which the northern (a later addition) Cette construction, comme celles du Bas-Empire, est irréhas a straight join with the wall. There are door-bar holes in guliére. L’arceau qui donne entrée dans l’abside, et celui qui est the jambs of the gate and remains of a parapet running over immediatement a gauche, sont plus grands que les autres. it. A substantial cave runs into the hillside from the south, L’abside est de forme elliptique et percée de trois fenétres reputedly for 150 m. The supposed site of a church (ac- ornees d’un chapiteau sculpte. Point de briques dans la concompanied by the usual tale of the discovery of gold) was struction: moellons grossiers. Les arceaux paraissent donner

; , passage a des chapelles laterales, tellement deétruites qu’on ne

pointed out on the west side, opposite the outer gale, but the saurait s’en faire une idée. Restes de peintures byzantines:

scattered stones among the trees gave little idea of its form, if figures. Le dessous des archivoltes orné de dessins. Diametre de

indeed it had been a church. The summit, to the north, 1s réglise, 18 pas. Profondeur de l’abside, 14 pas.2° reached through another buttressed gate, less well preserved,

in the third enceinte. There are remains of a small keep, or An approximate date for the building would be in the fourth enceinte, at the top. Here was some loose tile (not geyenth to tenth centuries: comparisons for the masonry may found in the walls) and three loose well-cut limestone blocks, be made with Dere Agzi, the city walls of Ankara, about 1.00 x 0.30 x 0.80 m in size, two of which were Mesembria, and Bodrum Camii: for the cornice profiles with simply molded on the edge (including guttae), as foracor- — Dere Agzi; and for the recessed apse with St. Nicholas at nice. It is possible that they came from the site of Pole- Myra, Dere Agzi: and for the recessed apse with St. Nicholas monion. Byzantine sherds included a good example of white at Myra, Dere Agzi, and the Fenari Isa Camii.?° glaze sgraffito ware.

Evkaf Kéyii Kale is a typical example of the local for- >. Funerary Relief (pl. 42, from photograph by D. C. W.) tresses of the Empire of Trebizond which become increas- Professor J. J. Wilkes kindly comments that it is a

ingly numerous as one proceeds eastward. tombstone with relief of the deceased and Latin epitaph ona The site, or area, of Polemonion itself yields two churches, raised panel, measuring approximately 1.80 x 0.60 m. The St. Barbara and St. Constantine, a Roman funerary relief | ‘P'S gabled with a plain roundel in relief in the center. The

with inscription, and a classical epitaph. relief figure shows the deceased on horseback, facing right.

He is wearing a short tunic, loose leggings, and short cloak

3. Monastery (?) of St. Barbara . . fastened at the shoulders: on his right side is a short sword or A large stone-vaulted construction 32 m long, withamain —_—_ dagger. The horse has full trappings. The text beneath reads:

apse, pentagonal on the exterior in the thirteenth-century

Trapezuntine style, was described by D. C. W. in 1962.?! C-NVMERIO’ There are traces of associated buildings. It probably cor- MAIORI-DEC: responds to Ioannides’ monastery of St. Barbara, said to VETERANO: have had mosaics,”* although only traces of wall paintings TOSSIDENE PRO

survive today. CLA-VIRO-SVO-ET-

4. Church of St. Constantine C-NVUMERIVS MAIOR This remarkable building was first noted by Hamilton: FILIVS PATRIH C

“The ruins of an octagon church, dedicated to St. Constantine, and the remains of a massive wall to the south © © (4/0) Numerio Maiori dec(urioni) veterano Tossidene of it, are probably the only evidence of ... the former site [of Procla viro suo et C(aius) Numerius Maior filius patri

Polemonion] ... about a mile and a half to the east of — 7(eredes) c(uraverunt).

Fatsah.” 23 Hommaire de Hell reported of Fatsa: “Tout “To Gaius Numerius Mator, decurion and_ veteran, prés, a lest, est une église assez remarquable, non loin d'une vallée et d'une riviere, qui portent, comme elle, le nom de 24. Hell (1846), Il, 369, and the unclear sketch in pl. xx, fig. 6, Polemona ....”’ Laurens’ lithograph for Hell, reproduced in which is lettered, referred to in the passage here quoted. plate 44, and Hell’s commentary provide the best evidence 25. Hell (1846), IV, 393 and pl. xxx1 (6).

26. Cf. Krautheimer, Architecture, pls. 111B, 112A, 112B, 113A, 113B,129B, 137A, 141A. We are most grateful to Dr. Beat Brenk for

21. Winfield and Wainwright, AnatSz, 12 (1962), 155-57. discussion of the building and for views on its date, based on the

22. Ioannides, Historia, 207. lithograph reproduced in plate 44, but not on the commentary by 23. Hamilton (1836), I, 270; cf. PE, 1 (1950) (see note 16), 207. Hommaire de Hell.

114 SECTION VIII Tossidene Procla, for her husband, and the son Gaius ground. Let into the exterior, 0.81 m above this, is a relieving Numerius Maior, for his father, saw to this being set up as arch of five voussoirs and two more blocks flanking it on

heirs.” either side. These stones have moldings of four rows of beading, recalling the decoration of the exterior of the apses The deceased had served in, and been properly discharged of the Hagia Sophia, Trebizond. The ironstone step has a from, a Roman cavalry unit in which he had held the rank of neat molding (pls. 43b, c). troop commander. It is most unusual to find a military The stonework of the interior is less well cut and is set in tombstone that omits any reference to the unit (presumably a pebble mortar, except for the four cut stone arches which cavalry ala) in which the deceased served, but this may be due carry the dome rising from simple cornices and the two to its being erected by his widow and son, instead of by fellow receding arches in the eastern bay. The interior has been soldiers from his unit. Also noteworthy is the omission of any plastered (for its conversion into a mosque, according to its

details of the age of the deceased. owners, although no sign of this use could be seen), but

The name Numerius is of Latin origin, as is also the beneath the plaster no evidence of wall painting was found. cognomen Maior. The wife’s gentilicium, Tossidene, is un- The floor was once paved but is now of concrete. In the usual and probably of local origin, although one hesitates to eastern, and only, apse there are two small windows, each

suggest any connection with Ptolemy’s Sidene.?’ about 1.77 m from the raised bema and about 1.40 m high, Judging by the lack of abbreviations in the inscription, the one roughly central and the other facing northwest. High tombstone appears to be early, and is unlikely to have been above the west and north doors are small windows (the set up after the middle of the second century A.D. Since the western of which is now blocked). The four pendentives are deceased was a veteran, it is not evidence of a Roman cavalry geometrically near-perfect, but the dome (which rests entirely

station in the area, of which there is no record. upon them) is very shallow (pl. 43c). The exterior of the dome The tombstone is now in the Samsun Museum. can be seen in the attic of the house now adjoining and

6. Boloman Kale ; a

surrounding the church. It is a rough mass of mortar and has

; ;fortified . beenstructure entirely stripped oforiginally any facing it may once have had. This stands on what was an ;

The castle is surmounted, between the church and southislet, aboutwalls 7 km east of the mouth of the Dere.by Its the finest early , , ern upon which it Boloman perches,

picturesque site and appearance hasnineteenth-century always been noted by ; , ; Ottoman house in the region (pls. 43a,

nineteenth-century and modern travelers, none of whom . ; 45a). This house is built entirely of wood and its saloon and appear to have penetrated the castle to observe that the ae ; ee miniature kiosk are decorated with charming paintings of Ottoman house perched above it partly encloses a cruciform ; ; seascapes and with good wood carvings. In 1836 Hamilton

domed church (pl. 43a).

reported that the castle wedge was called Hayar (i.e., Kaya) Kale The castle ;is,Co. built on a belonged roughly triangular of rock , “38 . and to an Ali Bey; ten years later Hell noted about 65 m long,. .projecting into thebelonged sea, which has traces of athat ,; of it and the little palace to Ahmed Bey, a relative complete defensive walling on its perimeter. On the east side 39-1

. . athe pasha Carsamba.*”’ pasha was the famous Osman is tiny but of sheltered harbor.This The made of of large .-, 4 . Hazinedaroglu, who walls, was also pasha Trebizond

roughly-shaped stones laid in regular courses 1n which the 30 , (1829—42).°° Today the castle belongs to the family of Argun mortar has now been largely eroded by wind and sea, great-great-great-grandson stand ; Kademoglu Bey,

up height of m on the side for to the south. There . of Osman . a:to; .aPasha, to 6whom we landward are indebted permission to survey the

structure rises. . .

is some evidence the ground levelAccording within this enceinte ;; _ Ls church and that for other kindnesses. to family tradi-

was originally up to 4 m above the external level, for it is from ; he selamlik ab h | built bv O

this point that the church (fig. 19) enclosed in the castle ton, the selami above the castle was built by Osman

, Pasha’s father, Suleyman Zade Hazinedaroglu of Carsamba, ;of pasha of Trebizond 1811, asthat a guest house. main The only; parts thewas exterior of from the church arecastle nowin.His _ ve 31 . residence the palace he built above the Unye.

exposed are a section of a well-faced polygonal apse which Th ;

e house of Boloman Kale consists basically of only one emerges through the; east wall of the andthe thepresent northern A . ground ; oor, standing 5 to castle, 6 m above exterior wall and door of the structure which evel opens on to a small | ; on what must be about 4 m of solid fill above the rock.

courtyard (pls. 43b, 45b). This andapeak the apse are made ; . fill ,; beneath the ; ewall owners of certain findsofinTh this

large localtoyellow limestone thethis ;, a: well-faced ae , floor, butblocks it wouldofbethe unwise place much reliance of upon

Chalybia-Sidene region which is also found in the church of hearsay evidence.>2

the Hagia Sophia, Trebizond. In places in the footing of the

north door the stone 1s inter-leaved with long tiles (including 28. Hamilton (1836), I, 270.

some roof tiles), but judging by the areas from which a 29. Hell (1846), II, 369-70. modern facing of plaster and cement has fallen away, it seems 30. Cf. Bryer, BK, 26 (1969), 191-210.

that tile is rarely employed and that the blocks were laid with 31. See page 102. _ | | | .

little or no mortar. A west door is blocked and the north . 32. Reported finds included two coins which, from their descrip-

d the threshold of which is t t bove the ground tion, sounded as if they could be eleventh-century anonymous oor, te Fesnold Of WHICH Is TWO steps a 8 ; bronze; a Christian “token”; and skeletons lying beneath simple,

and floor, is now the sole means of access. The present door is apparently gabled, brick canopies, which were inscribed in

2.10 m high but the original had an arch 2.76 m above “European,” not “Roman” (i.e., Latin, not Greek) characters. Further questioning revealed, however, that they were not in the ‘“‘Lombardic” epigraphy employed by the Genoese, and that the

27. Ptolemy, ed. Miller, 873. only letter recalled was a “‘reversed L”’ or Greek I.

PHADISANE-POLEMONION, SIDENE DISTRICT 115 The chronology of the buildings on Boloman Kale is There are caves to the east and west of the castle. difficult to determine. The selamhk belongs to the first dec- 7. About 1 km west of Boloman Kale, just inland of the delta ades of the; nineteenth therises church, with itsis ;a ; of the Calislarcentury Cayi, whereand a bluff on the west bank,

pentagonal apse, local limestone, and molded beading, may 33 . vor

cave noted by Hamilton.°” In its walls there are artificial be; ,placed more or less securely among a group of largelyrock-cut ;. ; niches; it may conceivably have been a Hellenistic thirteenth- and fourteenth-century and largely imperial tomb

foundations in Trebizond, although its cruciform plan is

unique in this context and area. But does the church antedate 8. Caves the castle walls? This must be the case in the section where the Two caves, locally described as having been “’Christian apse emerges through the walls and faced stone runs into the churches,” lie about 1.5 km east of Boloman Kale in the cliff joint. However, the form of the church may have been dic- face beneath the present coastal road and just above the sea. tated by that of buildings already existing in the south side, The larger cave, to the west, is a natural cleft very carefully the foundations of which may lie in the fill beneath the floor worked and chiseled, about 18 m deep. It is decorated with a of the Ottoman house which in turn stands 1.50 to 2.00 m single band of painting, about 1.25 m high, which runs right above that of the adjoining and encompassed church. Except round the cave. The subjects have been damaged by fire or for the dome (which may have been stripped of its facing erased beyond recognition, although it was reported that when the house was built, or, more probably, considering its pictures of ‘‘a woman and a fish” were visible until about shallow shape and absence of drum, entirely reconstructed in 1965. There is a single layer of painting on prepared ground Turkish times), it is impossible to reach any wall of the and the treatment of the surviving purple red and blue borchurch which is not exposed on the exterior without excavat- ders suggest that it could be medieval work subsequently ing beneath the house. A blocked west door and window varnished with an oil wash. A second smaller cave to the east show that it was once exposed, but there is no sign that there and below the first is perilous of access and devoid of interest. was a corresponding south door or window in what Is now a 9. There are reports of a castle south of the gendarmerie

decidedly; dark building. It seemsWe possible, therefore, that . . it. , , station at Boloman. have not investigated there already existed a structure and, perhaps, defensive walls on that side. Further conclusions than that would be unjustified.

The church and castle are not mentioned in any medieval

source. 33. Hamilton (1836), I, 271.

Section IX

DESCRIPTION context, at least) must indeed be Koyulhisar but it 1s Two classical roads ran south from Polemonion into the Mesudiye that is Matuasco and Sauronisena falls at GolkGy. Lykos valley. One went southwest to Neocaesarea. Its only the Roman mileage is in fact very reasonable. staging post was Bartae, 30 milia passuum from Neocaesarea The later Greek name for Mesudiye (or Hamidiye) was and 14 m.p. from Polemonion. For Bartae, Kiepert suggested MeAettoc, Meder, or Mma,’ probably derived from the Serkis, now Nefsiserkes, near Kumru, 21.5 km southwest of | MeAav@toc (Melet Irmak)” which runs through it and down Polemonion and 44 km northeast of Neocaesarea.! Apart to Ordu. Perhaps Matuasco is also a garbled form of the

from the fact that no modern track takes this route (which same name. . ok:

can never have been important), Nefsiserkes is in the right The later Greek name for Golkoy was Xayapava.'” Itisa area. Tarhan, however, reports a site at Kizelma Harap, just. | CO™MMunications center and market for summer grazers of south of Kumru, which may in fact represent Bartae,2 while | S©™Me importance, 37km from Mesudiye, about 100 km from D. C. W. notes the claims of Aybasti, which lies onaknown Ordu and standing at about 1,000 m above sea level. —

route to Neocaesarea.? Sauronisena is Ptolemy’s Lavpavia.'' Ramsay pointed

The second road ran southeast to Nikopolis. Its staging Ut that it probably corresponds with Strabo’s Divopta. ' posts from Polemonion were: Sauronisena (ca. 33 m.p., in A.A.M.B. is prepared to accept this identification, dispute); Matuasco (16 m.p.); Anniaca (18 m.p.): and D.C. W., however, prefers Sebinkarahisar for Sinoria.'* Nikopolis (18 m.p.).* The equivalent road today begins not _Golkoy lies on the northern side of a wooded pass which at Polemonion but at Ordu, but was diverted only with the divides the valley of the Melet Irmak from the tributary rise of nineteenth-century Ordu and the building of thisnew, Valleys of the Boloman Irmak and, eventually, the Pontic military road which seems to have given Ordu its name. The coastlands from the Lykos. the town itself is surrounded by old road would have run up the Boloman Dere, past Evkaf gentle slopes where the Bigincik Dere valley widens to offer K6yii Kale (q.v.), and then crossed over to the present Ordu- cultivable land. Above the town is a fepe (hill) on which, until Koyulhisar road at the most convenient point. A track still 1957, stood a great konak (government house). The castle leads up a tributary of the Boloman Dere, past Hisarkaya stands on the Aybasti track about 2.4 km beyond the remains (which suggests a local rock castle) and Akmescit (signifi- °F the konak, west-northwest at Kale Koy Pinar Mahalle, cantly the former Ak Kilise),> below the massive Gélkéy overlooking a tributary of the Boloman to the south. In Kilise Kale, and reaches the road at Gélkéy itself. This isthe terms of the modern road system, Golkoy and its castle are of only practical method of crossing over to the Nikopolisroad. | "© Importance, but they would have effectively controlled the Kiepert and Miller proposed Melet Hamidiye (now ancient road from Polemonion to Nikopolis at this point. We Mesudiye) for Sauronisena (although it entailed greatly in- Propose them as the site of Sauronisena. creasing the Roman mileage); a junction of the Melet Irmak

for Matuasco; and Koyulhisar for Anniaca.® But these MONUMENT

identifications must be modified with the comparison of Golkdy Kale stands on a massive granite outcrop which relative Roman distances on the map.’ Anniaca (in this projects east and at right angles to the valley for about 550 m.

1. Miller, 7R, col. 669. 8. The unsigned Turkiye Turizm, 5 (27) (August 1965), 54 (with 2. Tarhan, Map. illustrations). 3. Tarhan, Map; local reports. 9. Arrian, 23; Miller, 7R, col. 647.

° Pp . .

4. Miller, /R, col. 679. 10. Saltses, Kotyora, 32; Tiirkiyve Turizm, 5 (27) August 1965), 55. 5. Saltses, Kotyora, map. 11. Ptolemy, Geography, ed. Miller, 874. 6. Miller, 7R, col. 679. 12. Ramsay, Asia Minor, 56; Strabo, Geography, XII, 11, 28.

7. From Polemonion to Mesudiye is about 84 km; Mesudiye to 13. See page 35. Sinoria was one of the most notable of the crossing of the Melet Irmak about 13 km; the Melet Irmak to Mithridates’ supposed seventy-five strongholds, “close to the borKoyulhisar about 12 km; Koyulhisar to Pirk (Nikopolis) about ders of Greater Armenia, and this is why Theophanes changed its 35 km. But from Polemonion to Golk6y is about 47 km; Gélkéy to spelling to Lvvopia” (i.e., “border land”’): Strabo, Geography, XII. Mesudiye about 37 km; and Mesudiye to Koyulhisar about 25 km. I, 28.

SAURONISENA AND POLEMONION HINTERLAND 117 At its widest it is about 180 m across. The interior is roughly cut cistern with masonry vault, through which there is an rectangular, divided by levels (the only determinants) into opening. Despite the name Golkoy Kilise Kale, sometimes what may have been inner, middle, and outer baileys. The attributed to the site, there is no trace of a church, nor is any only entrance (to the outer bailey) is at the narrow isthmus at to be found in the vicinity, although the area was inhabited the west end where the castle meets the valley side. Most of by Greeks until recently.'* the rock sides, particulary on the south and southwest, stand The eastern curtain wall runs almost straight across the sheer for up to 150 m, requiring no more than curtain walls. rock promontory and is punctuated by three solid semiThe rock itself is roughly flat-topped, with a stone outcrop circular bastions and a larger hollow one. The stonework is running along its spine; the castle encompassed the entire very similar to that found elsewhere, but the two or three bluff of about five acres. The principal surviving walls are at courses of header holes, set here | to 1.5 m apart, run right the three most vulnerable areas where the rock is not sheer: through the walls, some containing clay pipes. It 1s possible the isthmus about 80 m wide to the west; an area to the that they were built for musket or rifle fire—the wall, which southwest of the isthmus; and the eastern end where, al- has no surviving gate, overlooks the valley road—and that though steep, the lower projection of about 100 m 1s walled this structure is later in date than the rest. The walls here

off, enclosing the inner bailey. stand to a height of 6 m.

The gate at the west-northwest corner is reached by rock- It is very difficult to hazard a date for the rest of the cut steps rising at right angles to it and below a semicircular surviving walls. On independent visits, neither A. A. M. B. bastion. The gate is about 3.5 x 5.5 m on the exterior and nor D.C. W. noticed Ottoman or Byzantine sherds, but fragopens into a short tunnel with a roughly vaulted archway and ments of ridge tiles and the high quality of the mortared a large arched niche on the west side. The voussoirs of arch rubble core of the walls suggest to D. C. W. an Early By-

and vault have disappeared; they seem to have formed a zantine or earlier date. He ventures that the hill at Golkoy semicircle. The arch of the niche 1s formed of flat stones proper, where the konak stood, might represent Roman employed in the manner of bricks. a hole in the reveal on Sauronisena; that the occupation, like in many other cases, either side of the gateway must have carried a stone lintel or removed thereafter to a fortified site on Golkoy Kale, which massive beam. There are holes for smaller beam emplace- represents Byzantine Sauronisena; and that the town rements across the inside of the door space. On the interior and turned to its original site in Ottoman times. exterior of the wall, at the point of the gateway, are traces of While not disagreeing with the general lines of this hypoup to five rows of beam holes, for headers, which do not run thesis, A.A. M.B. formed the impression that some of the right through the structure. The headers are at intervals of | defenses, particularly on the eastern and southern sides, about 1.50 m vertically, and of about 0.50 m horizontally. could be quite late. It is possible that the castle was used until There were no signs of stretchers within the thickness of the the early nineteenth century by local derebeys'* to command wall (2 to 3 m), but they may have been present. At their the road to the sea and that it was abandoned only with the highest point, near the gate, the walls stand uptoaheight of last feudal risings of the 1830s—40s and the building of the

10 m. new road to Ordu which bypasses it. This does not contradict

In all parts of the castle the walls are of regular courses of the fact that some of the defenses of the castle, particularly roughly-faced, nearly square blocks, alternating with smaller around the gate, are probably Byzantine or earlier. wedges of stone with substantial pebble-and-lime (but no The site is an object lesson in the difficulties of assigning a tile) mortar and a core of mortared rubble well laid in and date without the more usual constructional indications and

tamped down with few air spaces, now as hard as rock. in the absence of literary sources. Bastions of roughly semicircular shape are at the east-

' ' ' rl Hires.

southeast and west-northwest corners, and others punctuate

the curtain walls. They are hollow and their internal struc- 14. Saltses, Kotyora, 32, lists twelve local Greek villages with a ture seems to have been entirely of wood. The bastions in the Ons Peoamichalope ° a 1902). 3 ; a See now Elizabeth A. southwest section enclose what was (probably correctly) Zachariadou, “Trebizond and the Turks (1352-1402),” AP, 35

pointed out as a hamam. (1978), 344 and note 2, who proposes Golkéy as the ospitokastron of In the middle bailey, near the southern side, is a large rock- the Haci emir in Panaretos, ed. Lampsides, 73.

Section X

THE CENTRAL LYKOS The lower Lykos (Kelkit), before it joins the Iris (Yesil MONUMENT (pls. 46—48a-—c) Irmak), is touched on in the section on Neokaisareia. The The spur upon which Yukari Kale is built starts as a thin upper Lykos is described in the section on Cheriana and the neck which rises from the river valley and broadens into a problem of Arauraka. The central Lykos never lay withinthe wide shoulder, topped by a plateau. This shoulder narrows lands of the Empire of Trebizond, but its great and enigmatic into a neck once again, and then widens out into a second castle at Koyulhisar, and its surrounds, cannot escape notice shoulder, with gentle slopes, about 50 m higher than the first.

here. The second shoulder has the remains of a curtain wall around it and must have formed the southern, outer, bailey of the

SITUATION castle. On the south side of this shoulder, the ground again The Lykos valley narrows eastward and above rises steeply to a citadel. Little of the citadel is left standing, Neokaisareia into impressive gorges, which never allow it to save for the lower part of a tower which enclosed a barrel-

open out again to the extent of the wide plains of Phaneroia. vaulted chamber. The masonry of the tower consists of The next sizeable broadening of the rift is at Resadiye. rough stone, laid in random courses with the flat face outFurther east, the valley breaks up again into the gentle slopes wards. The outer surface is heavily pointed with lime morar. and alluvial stretches of the Koyulhisar region. The west end On the east side was a rectangular room, with a doorway of the Koyulhisar valley is marked by a large rock, called which may have served as a postern. Asagi Kale. Despite its name, there 1s no trace of walling on North of the citadel the ground drops to a northern bailey. it, although it is strewn with sherds. There 1s a bridge and a It was roughly triangular in plan, with the apex to the north,

ruined han below Asagi Kale. where the ridge of the spur was closed off by a main gate. This At the east end of the valley the river makes a bend to the was the most substantial remaining part of the fortifications, south to flow round a spur of the mountain chain. The spur is and was no doubt always the strongest defense work, for it is crowned by a castle called Yukari Kale. The modern bridge the only part which is easily open to attack. The masonry is

over the Lykos lies below it. There are ruins of piers of two strengthened throughout with stringers and tie-beams, laid earlier bridges, of which too little masonry survives to hazard at close intervals through the walls.

dates. The modern town of Koyulhisar lies to the west of the There are ruins of two barrel-vaulted rooms, built up castle, but appears to have no antiquities. Nevertheless the against the west wall of the north bailey. A semicircular reasons for a settlement in the area are long-standing and projection from the west wall must have been a kind of obvious. Like Neokaisareia, its surroundings are fertile, and bastion or buttress. it stands on an important road junction. In this case the There is no sign of Roman, or earlier, occupation of the Polemonion and Kotyora routes meet the Lykos road from site. The castle may have received attention in the sixth

Sebasteia to Nikopolis here. century, or have been one of the unnamed strong points of There was in the district a Roman station, Anniaca, which Armeniakon or Koloneia. The present ruins would not conought to be sought somewhere on the edge of the alluvial tradict a Byzantine date. But they were no doubt added to in plain. But there is no doubt that the important medieval site later times: by the Akkoyunlular from whom Mehmed II

was the castle above Yukari Kale Koyii.' took Koyulhisar on his way to Trebizond in 1461, and by the Ottomans themselves before the castle was destroyed in the late eighteenth century.”

1. Miller, ZR, col. 679; Cumonts, SP, I, 290-95. 2. Cumonts, SP, II, 290-95.

Section XI

DESCRIPTION Arrian’s possible error, there is nothing against Pastiades’ Cape Jason is the most substantial promontory east of identification and Fernek may well repay field investigation,

, Lge : 9? .;

Sinope, projecting 14 km into the Euxine and culminating in wen we have not mace. The Grand Komnenos Bona Portus. ’® In Idrisi it is ““Byna’’; in Clavijo ‘“‘Leona” (as Alexios III celebrated Epiphany there in 1357.'© Clearly with Ba@tc-“*Lovati,” the Italians gave it a definite article); there were a number of churches in the area, of which the on the portolans ‘“‘Lauona” and Asova;?’ and Evliya calls it

Theotokos or Panagia (probably a monastery) and St. ‘““Wuna,” “where the largest ships can anchor at any Andrew are most often mentioned.'’ The latter may, how- time.” *® Today it is called Persembe, but the local names of ever, be the present church on the headland, built in 1868 and Vona and Vonalimani are still commonly used.

published elsewhere. '® Kotumpa is the next great classical site on the coast. There

Between Cape Jason and Boon, Arrian names an Island of is no doubt that modern Ordu stands on, or close to, the the Cilicians (KtAikawv vijooc),'? corresponding exactly with ancient site, but it is equally certain that there is little or no the island of Hoynat Kale, 4.5 km east of Cape Jason, lying continuity of settlement with it. Kotyora probably did not

off the point where the modern road enters a tunnel (pl. 5Ob). survive the classical period. Originally founded by The eastern cape of the promontory was called Tevytne, Sinopitans, it had already greatly declined by Arrian’s day, is after a local people, and, according to Apollonius of Rhodes, misspelt by Strabo, and is unknown to the compilers of the

a temple to Zeus stood there.*° There was also a river Itineraria—although it lies on an important classical road.?° lévyntoc,?' which may correspond to the modern Caka Kotyora follows the pattern of other Greek colonial settleDere. The classical name did not survive; in one portulan the ments. It stands at the head of a route inland; two wide deltas cape is called kaBo Tévec, but more often it has been called to the east provide ample food supplies; and its sheltered Cape Boon (to &kpwtnpt tod} Bovva).?* Today it is named beaches are overlooked by an acropolis, Boz Tepe. That the Cam Burunu, perhaps after a tree which once marked it. route inland is an ancient one is suggested by the fact that not The southern section of the east side of the promontory far south of Ordu are the ruins of a castle with a cistern tunnel affords, as many classical and modern commentators have which, like so many other examples, is probably Pontic pointed out, what is probably the best anchorage on the work. Neither author has visited it, but a party of British entire Pontic coast. Hamilton observed that “‘it is considered engineers at Ordu reported that the tunnel is several hundred the best winter harbor on this side of Constantinople, pref- feet deep and has a T-junction at the bottom with horizontal erable even to that of Sinope, on account of the greater _ cuttings. depth of water.’’*° The cliff falls nearly sheer into the sea, Nor have the authors investigated the Boz Tepe of Ordu, which is 10 fathoms deep within 300 m of the exiguous beach. for it is now a military base. But nineteenth-century comThere is no real scala but a small settlement has always mentators are no doubt right in locating an acropolis there; provided and watered sheltering shipping. In the twelfth in Bzhshkean’s day it was crowned by a church of St. George. century the place had some importance, both commercially As for the harbor below, Hamilton observed that ‘“‘some and as an embarkation point against the Turkmens.?* The remains of an ancient port, cut out of the solid rock, are still paternal family of St. John, iconophile bishop of Gotthia visible.” >° They are not visible today. (ca. 755—ca. 791) came €« tov B@voc (or Bovootobd) tov Kata With less justification, other commentators have located tO TloAguoviov Kemévov Atpevos EK TH OEpaTtt T@v ancient Kotyora at Bozuk Kale (‘‘Ruined Castle’”’) ona small Appeviakov.*° In the peripli it is called Bowv, latinized as headland 4 km north of Ordu, but the remains on this minuscule site are entirely medieval.*!

15. Lebeau, Bas Empire, XX, 494 note; Black Sea Pilot, 402. It seems clear that the histories of Kotyora and Vona are

. Bznsnkean ; ; trans. Andreasyan, ; trianta- :

8 Aiibaen T1819). 58: thane And Sampsides, ie . linked and that the focus of settlement in the area moved,

I, 269: ‘‘one large building was pointed out, which the sailors called a " Bes, ©

monastir, but it was evidently the ruin of a Greek church, to which Kotyoran area was represented only by a small castle on the

some adjoining buildings had once been attached.” sea, Vona would have been much more secure from Tiirkmen

18. Bryer and Winfield, AP, 30 (1971), 237-42, with plan. The raids. With the return of security in the nineteenth century, ie: Phendate tudor 27on aHaton 8,gonefMG aeunderthejy,vandalism clonareaaed so ary was inscribed stone, which had by 1970. We Kotyoran was resettled Ordu,to whichYona rivaled Vonaand the stand that Turkish newspapers reported on the assite in

June 1977. for over a century before becoming the major town.

19. Arrian, 23. But the origins of Ordu are an enigma. It appears in no

20. Apollonius Rhodius, Argonautica, Book IH, lines 378, 1009.

21. Anonymous periplus, 32. 26. Arrian, 23; Anonymous periplus, 32; Miller, IR, col. 647;

22. Delatte, Portulans, I, 238; II, 33. Rottiers (1820), 241.

23. Hamilton (1836), I, 269; Black Sea Pilot, 403: ‘‘Vona limani 27. Idrisi, ed. Nedkov, 96—97, and note 302 on p. 147; Clavijo affords the best anchorage on this coast; though exposed to winds (1404), 109; Kretschmer, Portolane, 648—49; Delatte, Portulans, 1, from between north and east-southeast, it is stated that such winds 238; I, 33. Mistakenly following Le Strange, in Clavijo (1404), 350 rarely blow home. Violent squalls, however, may be experienced (note 5 p. 109), Bryer identified Leona with Polemonion in AP, 24

during offshore winds, and these should be guarded against. (1961), 104. Numerous sailing craft, which cannot be hauled up on the various 28. Evliya (1644), II, 40.

beaches, winter in this bay.” 29. As Kvtwpoc: Strabo, Geography, XII, m, 17. Cuinet, Turquie

24. Vryonis, Decline, 161 note 169. d’Asie, 1,84, mentions an inscribed brick and a sizeable ‘““Byzantine”’

25. The version in the Ist edition of the ActaSS Iunii, V, 190 (26 fort on the hill behind Ordu, but it is impossible to verify medieval June), cited in Tomaschek, Kleinasien, 80. However, the 3rd edition continuity of settlement because the site is now in military hands. gives a different wording and the alternative version (Bonostos) of 30. Saltses, Kotyora, 8; Bzhshkean (1819), trans. Andreasyan, 37; the name: ActaSS Lunii, VII, 168; cf. 163: urbs ... ignota geographis. Hamilton (1836), I, 267. See also Vasiliev, Goths, 89—96; and, for another Pontic connection 31. E.g., Pastiades, PPh, 1 (7-8) (September—October 1936), 7;

with the Church of Gotthia, the Sinopitan inscription on p. 75. and in the modern Turkish tradition.

CAPE JASON 121 medieval, or even portulan, source; the earliest reference to the monastery of the Theotokos. The walls of one almost the name, which means “army,” which we can find is in 1813, square building stands in parts up to the beam-holes and when it was already a large village. Hence, suggestions that it windows of a third story. There is abundant tile in the is named either after the Fatih’s army or after the road which masonry, but the tiles are too small and the walls too thin to the Turkish army built south to Sivas after 1861, must justify a Byzantine date. Local sherds are post-Byzantine.

be ruled out.°? The road, coupled with the current On the east side of the peninsula is a well-preserved brick‘‘re-Hellenization’’ of the commercial towns of the coast, is and-tile kiln. Itis a domed circular structure, with an internal responsible, however, for the remarkable late-nineteenth- diameter of 3.34 m. The interior is faced with thirteen courses

century expansion of Ordu, if it does not account for its of exceptionally well-cut stone; the exterior has a rougher name. Ordu became the final residence of the wandering facing. metropolitans of Neokaisareia and local antiquarians (who Until the 1860s the most conspicuous monument was a founded such schools as the famous Psomiadeion of Ordu) Byzantine church,*” later replaced by an equally conspicurevived the name of Kotyora, almost two millennia after the ous church, dated 1868, which stands almost in the center of site and name of the old Greek colony had been forgotten. the peninsula. Inserted in its walls are two blocks of yellow But it was an artificial name, and ordinary Greeks still called limestone carved with differing guilloche patterns on a dia-

the place Ordu.*? mond net (pl. 49a, b). These blocks obviously came from an

A modest element of continuity between ancient Kotyora earlier building and their pattern is comparable to those and modern Ordu is provided by Bozuk Kale, 1.e., ““Ruined depicted by Hell in the castle of Tripolis*® and to other Castle.”’ It may well be the “small castle built on a height examples in Trebizond. The ground around 1s rich in sherds beside the sea, and the name of this is Santo Nicio,”’ ** noted of Byzantine graffito and plain earthenware. by Clavijo soon after he saw the (now lost) castle of Boon. North of the church of 1868 is an isthmus which is 35 paces But later portulans add two more dedications: S. Tomao, wide from east to west; it is scarred by considerable rock Santhomas, “Aytoc Owpdc, and “Ayiog O@Eddwpoc.*° cutting. One cutting is roughly in the shape of a rectangle and Whatever the name of the place, it seems to have been no might be a salt pan.*' A depression in the soil across the

more than a late medieval Trapezuntine stronghold. isthmus suggests that the rock might have been cut right across it to separate the northern end of the peninsula into a

HISTORY more easily defensible islet.

The Jasonian promontory must have passed to the To the north of the isthmus the cape is about 170 paces Turkmens of Limnia between 1357 and 1404—-when Clavijo north-south by 80 paces at its widest point east-west. The found it under the control of an emir Artamir (I] ?).°® But ground is more or less a level plateau from 4 to 5 m above sea Boon remained an important anchorage, which probably level and is enclosed by a boundary wall, the foundations of explains why Clavijo was told that the Genoese had raided its which can be seen in the turf. On the east, and sheltered, side castle in 1400, and the area remained essentially Greek. (the prevailing winds are from the northwest) is a small bay Evliya remarks that “‘the mountains are interspersed with where skiffs might have moored, but no larger boat could well-cultivated Greek villages. ... The inhabitants are known have approached the peninsula since there is a flat rock shelf by the name of Wuna Greeks and Turks.’’*’ There were a extending for about 50 meters into the sea on all sides and number of villages, such as Depvéx, ToAatAn, Texxta, and only about 50 cm below the water. Boon itself on the promontory, which remained Greek until It is impossible to date the arrangements on the isthmus;

modern times.*° there is no reason why they should not be classical.

About 70 paces south of the tips of the peninsula are the

MONUMENTS foundations of a church. Three apses can be distinguished.

1. Cape Jason The scale is about 12 x 8 paces. Hell, who visited the site

Close to the modern road, on the seaward side, are exten- twenty-two years before the nineteenth-century church was sive remains of domestic buildings, perhaps associated with built, describes ‘‘les débris d’une église grecque, dont la 32. Kinneir (1813), 323-24: Saltses, Kotyora, 14: Joanne and forme est rectangulaire avec trois absides, vers la pointe Isambert (1861), 520; local tradition reported to D.C. W. extreme du cap’’;*’ this is probably the same building, its 33. Oikonomides, Pontos, 36-37; Papamichalopoulos (1902), ruins demolished to incorporate the stonework in the new 283-92; Bryer and Winfield, 4P. 30 (1970), 236-37. The Psomia- church.

deion has now been destroyed. its adjoining church ofistheliberally The soil ofstrewn the ‘asulawith is liberally st thf t Hypapante survives as abutprison. peninsula fragments

34. Clavijo (1404), 109. Cf. Texier, Asie Mineure, 619. of brick and ridge tiles and with earthenware sherds. Among 35. Tomaschek, Kleinasien, 80; Delatte, Portulans, 1, 238; II, 33; sherds found near the church were those of a fine dish with

Bryer, AP, 24 (1961), 105. . blue glaze and gold patterning, akin to Syrian thirteenth-

36. Clavijo (1404), 109; Panaretos, ed. Lampsides, 72. There were already hostile Turks in the region in 1357, for fourteen of them were

killed by the then Grand Komnenos: cf. Bryer, DOP, 29 (1975), 130. 39. Joanne and Isambert (1861), 520.

37. Evliya (1644), II, 40. 40. See plate 73.

38. Named as Greek villages in Salses, Kotvora, 31—32 and map: 41. There are rock-cut oil and wine presses below Koralla (see there were 320 Greek families on the Cape itself. But the villager who plate 87a), and on Kilise Burunu, near Tripolis (illustrated in took D.C. W. to Kiliseyani stated that the only Greeks in the region Ballance, Bryer, and Winfield, AP, 28 [1966], 255 and plate 12). in recent times had been the monks of Cape Jason. That there had Pontic oil presses are discussed, and seven weight stones illustrated, been a monastery on Cape Jason is certain, but it was already in ruins in Anderson, SP, I, 14—16.

in 1836, and the church of 1868 is not a monastic one. 42. Hell (1846), I, 370.

122 SECTION XI century work. But the salt air and spray has disintegrated broken ridge tiles and bricks, arranged in such a way as to

most glazes. suggest that there may have been a brick arch for a door or

This site is no doubt medieval, although it may not neces- window. The mortar beds are roughly of the same thickness

° 44

sarily be the conventual church of the Theotokos. Clearly _as the bricks and tiles. | there were several churches on the cape. It was variously The church probably had three apses, but dense underreported to us on different visits that there were four, twelve, growth make it impossible to tell, or to photograph the strucand even thousands more churches. We can vouch for two ture. The dimensions are roughly 10 x 20-25 m. A large oak more Byzantine sites (described below); an exact description growing inside the ruins cannot be less than a hundred years of the whereabouts of a third was given us, and a fourth, ona old. The church is probably Byzantine or Trapezuntine. hill several kilometers to the south, may be seen through Although the name of the site is locally explained as having

binoculars. the Armenian suffix “‘-ian,”’ it almost certainly indicates that 2. Bayadi Kéyii the church is Greek and dedicated to ‘Aytavvi, the Pontic St. The scattered village of Bayadi lies over an area of a few yon £100 m bel dtoth theast of the church i

kilometers along the western peninsula of the cape. The road ou TH DETOW ANNO EAE SOUNNEAST OL INE CONTE IS 8

runs by a small bay west of the first group of houses and water source, surrounded by paving and still in use. Probably Koca Burunu, a small peninsula. About 15m above the road the facing of the church consisted of ashlar blocks similar to to the south, where the road skirts the bay, are three sections those of the paving. The rounded arch over th e pool is ; aced

p Pp

of brick and stone masonry. with thin blocks of stone, set in a manner which recurs in the

Most of the facing stones have been robbed, but the thick- Nakip Camil, a the Hagia Sophia, and in a building in the ness of the wall can be judged to have been about 1.20 m. The citadel in Trebizond. The last vestiges of a well-paved track masonry is banded with three courses of ridged tile alternat- and steps between the water source and church are rapidly ing with four or five of stone.*? The remaining facing stones disappearing. The masonry of the church and water sources

are roughly-squared rectangular blocks and narrow headers may well be contemporary. which band into the core. The ridged tiles run right through 4. The Island of the Cilicians

the thickness of the wall, but broken fragments are used in About 5 km west of Cam Burunu lighthouse by the the mortared core and they are not so evenly placed as on the modern road lies a rocky island, Hoynat Kale, separated by exterior. The proportion of mortar to tile varies from parity about 20 m of water from a steep promontory, through to about two of mortar to one of tile. The mortar is made of which the modern road tunnel runs (pl. 50b). The island has lime, sand, and small pebbles; the rubble core is well laid in sheer sides, apart from the southern, or landward, side.

with few gaps. Except for this side there is little need of walls and little trace Not enough of the three pieces of walling is left to de- of them. Such walling that survives consists of rough stones termine the nature of the structure. By local tradition it had set in random courses with a smooth outward surface. Near

been a church (pl. 49c). the southern end of the island is a small barrel-vaulted buildThe banded brick and stone courses are uncommon in the ing with masonry in regular courses and a stone or rubble Pontos and are unknown to us in any building dating from vault—except that, viewed through binoculars, the halfthe time of the Empire of Trebizond. The good quality of the dozen courses at the crown of the vault appear to be brick. In

masonry suggests a Byzantine date. the west wall are three recesses which might by beam holes.

3. Kiliseyam , No other structures are visible on the island.

, A part of Bayadh village called Kavrayalisi lies about 2 km The purpose of these walls is problematical. It 1s unlikely west of the western peninsula of Cape Jason along the coast to be an island fort because the habitable slope of the land is road. Twenty minutes’ walk uphill south of the road are the tilted toward, and entirely exposed to, the higher mainland ruins of a church in the middle of a hazelnut grove. The nut promontory, from which it would not be safe. The island—

_ ,; to th t. It

.undergrowth rocky and exposed to the prevailing northeast winds—is a. groves here have dense and the site would be ; . difficult to find without a guide highly unsatisfactory anchorage. It is too distant from Gaka Only the west and north walls stand, to a height of about 4 bay to the west to serve as a guardian fort. Its use as a signal m. It is difficult to judge the original ‘thickness of the walls station 1s strictly limited by Yasun Burunu to the east and since all the facing stone has gone, but they were more than a Capraz Burunu (which has no trace of ruins) to the west. Its meter thick. The setting bed indicates that the facing was small size, difficulty of access, and lack of command over the

4: surrounding area make it an unlikely candidate for an armade up of large neat rectangular blocks laid in regular chontic castle. courses. The core of the walls is mortared rubble, with a Perhaps the most reasonable explanation of the ruins on

mortar of lime and a great quantity of pebbles. It is well laid P eos P

in but there are a few gaps between the stones. There are the Island of the Cilicians is that they represent one of the some brick and tile fragments among the mortared rubble monastic foundations which, by tradition, are supposed to

The tiles are: 2—2.5 cm thick site and the cm thick. s; h Tirebolu is called Ki . f . nother onbricks the3—4 coast east44. o Another Irevdoiu IS Caite Kiseyan, At one point in the north wall there are a number o which may also be a Kiliseyani: Winfield and Wainwright, AnatSt, 12 (1962), 153, suggest that it might represent a monastery of “‘St. Anne or of the Virgin.”’ But St. John is more likely—cf. the “Ayana”

110. Jasonian site.

43. For other examples of banded masonry, see pp. 73 note 39, 79, of St. John, Vazelon. Pastiades and Saltses do not identify the

CAPE JASON 123

5. Boon castle ; ; ; , é

have existed on Cape Jason. The site is eminently suitable for sea (fig. 20, pl. 50a). The columnar basalt rock*’ upon which an eremitic life of fishing and mortification, but for little else. it is built is no more than 50 paces wide and is connected to

the mainland by an isthmus no more than 12 paces wide,

; ;which ; which little shelter for skiffs The castle, weaffords have nota examined, is noted by i. on , either side. The

kg entire rock was originally encircled by walls. The outer walls ;now can now only be traced on west, and ningeoa 9945 rocky .promontory, but abandoned with nothe one ; southwest,Here 7 northwest, but stand up to 3.5 m in the northwest. inhabiting it. Evliya observes thatfill“the islime, of asand, . ; and pebble . . masonry is a rough withcastle abundant round shape and stands on a hill by the seashore, but it is not ; h strongly garrisoned: the gate looks to the east”4° (ie mortar. The walls, which are up to 1.5 m thick, stand on the Claviyo: “‘Here was a castle that stood by the seashore, crow-

toward the sea). ;

, 7“ bare rock. At the point where they reach the isthmus, to the , southwest, there is evidence (in the form of footings and

castle cannot be seen fromground) the modern of road, , ;The . depressions in the anawhich entrance to the west; and climbs above Boon. It; ; was perhaps small Trapezuntine .aboveofthea circular tower or bastion to the east. Immediately

fortress or watchtower, erected onhighest classical foundations, and ; , as entrance is a grassy knoll, the point of the castle, with possibly the residence of a kommerkiarios or similar officer oo, a the footings of an almost square keep. The masonry here 1s; who would regulate merchant shipping sheltering in Boon , bay not so rough, but the facing either never existed or has been robbed. There are traces of tiles in the mortar. The walls are

6. Bozuk Kale about 0.94 m thick.

The castle of St. Nikias (?), St. Thomas (?), or St. Theodore The castle and keep are probably Trapezuntine.

(?), is in the shape of a three-fingered hand stretching into the Papamichalopoulos states that in his day there were also ruins of a church on the adjoining Kilise Burunu.*® 45. Clavijo (1404), 109. It no doubt stood on the site of the castle

mentioned at Boon in the Anonymous periplus, 32. 47. Cf. Hamilton (1836), I, 267. 46. Evliya (1644), II, 40. Cf. Triantaphyllides, Pontika, 139. 48. Papamichalopoulos (1902), 285.

LD \ 1 HAMAM = ay Zz a ; \ \ \ H

Approximate

SS A TS A Sa cS

0 10 20 30 ao ©5500 METERS 100 B UJ C F R A K A [ F

;BYZANTINE a \ 4 \ \ .\,\ ££ (site of 2 _ and R > |i { Door 7 ‘ao\s2‘A A4csa©, wy 7 \ KL v\ near PAAIZANH, FATSA

en ae ae

‘\ OTTOMAN Castle) co [Blocked ° pf

« ™,

1— “\\ Eero i!

OINAION s “ \ — UNYE NI Approximate

CASTLESQUARE 0 10(SEA) 20———t 30 Meters ot

16. Plan of Site 18. Plan

a , --

“CHURCH’in CITADEL

rta\i iI

| “Internal doorways. fat / had windows over \

iii them lighting the

! “chapels” H

ee

7a "1 Probable side for van ‘ ~~entrance~

graphy, XII, m, 17; Ptolemy, Geography, ed. Miller, 867; Ritter, 11. Triantaphyllides, Pontika, 82-83; Fallmerayer (Fragmente,

Erdkunde, XVIII, 833-38: and here, p. 152. 1840), 138-39; Cuinet, Turquie d’Asie, 1, 75; Ritter, Erdkunde, 3. Triantaphyllides, Pontika, 174, was among the first to dismiss XVIII, 834. the notion of a wandering Kerasous. Cf. Janssens, Trébizonde, 35. 12. Tomaschek, Kleinasien, 80; Kretschmer, Portolane, 648.

We have not seen B. A. Mystakides, Mia Kepaootc 7 S00; Kat 13. Delatte, Portulans, 1, 237-38. TOLTMV TY MpOG A. mMpdc A. OTO TOD EEvog@™vtocg LvNPOvE- 14. Ioannides, Historia, 218; Cuinet, Turquie d’Asie, 1, 67-68.

vopevn, EvEEtvoc Movtoc, 2 (Trebizond, 1893), 290-92. The Greek for Keci (Geci) Burunu was “Axpov Aiyoc, but both this 4. See note 2, and Diller, Minor Greek Geographers, 160. and Kassiope (for Kesap) seem to be 19th Hellenizations of Turkish

5. Miller, 7R, col. 647. names, rather than survivals of older ones. 6. See p. 119. 15. Bryer, AP, 24 (1961), 106.

CITY AND DISTRICT OF KERASOUS 127 dicates a iepa mpivoc (i.e., ““Holy Oak’’)'® is more, but not The remains of the chapel at Gedik Kaya Kalesi probably

entirely, satisfactory—it is surely more likely to be a yepo correspond to it. The second monastery was in Kerasous Mpivos (i.e., “Old Oak’’). The question is complicated by the itself and was dedicated to St. Epiphanios. In 998 Patriarch

fact that Constantine Porphyrogenitus states that 500 Sisinios II (996-98) granted Nicholas, metropolitan of Armeniak soldiers, respectively from THAataviov and Alania, a typikon for the monastery of St. Epiphanios, to IIpivn, took part in the Cretan expedition of 911.'’ If which a memorandum was added in May 1024.7’ The Alan Platanion is Platana (now Akcaabat), the port just west of mission had begun in the early years of the tenth century but

Trebizond, the possibility that I1pivn (albeit in the fem- had not been initially successful. The typikon and its mem- : inine) is another coastal station in the Pontos, our Girapno orandum is our only evidence for the existence of

or Giraprino, is therefore strong. Metropolitan Nicholas and the monastery, but the Church

‘O nptvoc is not an ordinary oak, but the small bushy of Alania became inextricably entwined with those of Pontic oak associated with an important export of the Trebizond and Soterioupolis, long after the Alans had region, cochineal. In fact we now know that the Pontic — gone.?!4 substance was neither cochineal proper nor a herb. It was Before examining what archaeological evidence there is kermes, a brilliant red dye for silk, which was an expensive for the whereabouts of the monastery of St. Epiphanios, it and much sought-after alternative to ““grana”’ (another coch- would be as well to record the literary evidence for another ineal substitute). Kermes is made from the dried bodies of the religious site in Kerasous, a fourteenth-century hermitage on female Coccus ilicis which clings to the twigs of the oak. The its ““Heights.”’ Andreas Libadenos (who incorporated in

substance was well known to Pegolotti in the 1340s, and in another work two brief ekphraseis of Kerasous, neither of 1434 there were specific instructions to Venetian shipowners which are very informative)?? addressed two letters to the

to buy it in Trebizond.*® anchorite and hieromonk Gerasimos, év “YwtA@ tic We suggest that Geraprino (and variants) may be as- Kepaoovvtoc, probably before and after 1341.°° The tempsociated with a local export of kermes dye, which may also tation to identify this hermit with the monk Gerasimos of the have been used in preparing the stuffs which the Book of the supposed Trapezuntine monastery of St. Euthymios the Eparch names as a product of Kerasous.'” If it is also iden- Great in Jerusalem, who may (if it 1s dated correctly and is tical with the Prine of Constantine Porphyrogenitus, it 1s not a forgery) have drawn up his will on 18 November 1344, unlikely to have been the island of Ares. There is indeed fresh must be resisted.?* All that can be gleaned from Libadenos 1s

water on the island but it is not a particularly convenient han} . lace to station five hundred Armeniak troops. One must 21. G. Ficker, “Das Epiphanios-Kloster in Kerasus und der

th fore look fora. probably fortified. sit pe dto the Metropolit Alaniens,” BNJbb, 3 (1922), 92-101: N. A. Bees, “H év erefore look for a, probably fortified, site near and to the Kepacotvtt povy tod ‘Ayilov “Exigaviov Kai 6 pntponoditnys east of Kerasous. The only site between Kerasous and Cape ‘Adaviac NikoAaoc, AP, 16 (1951), 255-62. On the subsequent fate

Zephyrios which is appropriate is Gedik Kaya Kalesi. of the metropolis, see p. 244.

Although this identification is not entirely satisfactory, 21a. See pp. 348-50. . . Gedik Kaya Kalesi has a sheltered bay and anchorage below 22. Libadenos. Periegesis, ed. Paranikas, 36, 42: ed. Lampsides,

. ; og , 71-75, 101-4, O. Lampsides, LupPodAat civ Biov Kat ta Epya

it and is, rather than Kerasous, the coastal terminus of the “A vépedv AiBadnvoo, AP, 29 (1968), 213. route to Koloneia. We tentatively propose it as the site of a 23. N. Banescu, “‘Quelques morceaux inédits d’Andreas Libaplace that seems to have been known as ‘“‘The Old Pontic déenus,” Budavtic, 2 (1912), 362-63, 380-84; Lampsides, AP, 29

Oak.” (1968). 187-88. 242 and note 2. to "YynAov. or 6 “YwnAoc, is one Theret were leastattleast th asteries in K of the places which have been associated with the famous Pontic. two Other monas eries in crasous or and later Phanariot, family of Hypsilantes (Xiphilinos ?), but see S.

its district. The monastery of St. George, in ruins by the Skopoteas, Oi “YytAdvtar. “H TpaneCovvtiaky Kataywyn tous, nineteenth century, lay beneath the Gedik Kaya summits.*° AP, 20 (1955), 154. 24. The will looks genuine enough and there is indeed another indication of a Trapezuntine monk in Jerusalem in 1391, but it raises

16. Tomaschek, Kleinasien, 81. too many problems to be accepted without doubt for what it pur-

17. Constantine Porphyrogenitus, De Cerimoniis, Bonn ed., I, ports to be: 1. It has been variously dated to 1144, 1344 and 656 (not in ed. Vogt); cf. A. A. Vasiliev, Byzance et les Arabes, 11 (1) 1444. 2. The refounder of the monastery is described as tic de(M. Canard, La dynastie macédonienne) (Brussels, 1968), 201, 204: ONOIVYS EKEIVNS TpanECODVTOS KUpias “Avv7ys Ts NopPvpoyEvand Ahrweiler, Byzance et Mer, 107 ff. The comparatively low pay of vntov, assuming a porphyrogennete epithet never otherwise claimed

these troops suggests that, despite Constantine Porphyrogenitus’ by a Grand Komnene or Komnenos, and not described with the

heading, they were unlikely to have been cavalry. usual Trapezuntine titles. 3. If the document is dated to 1144 or 18. See Pliny, Natural History, VJ, 1v. 11; Magie, Roman Rule, 1, 1444 (which is, in any case, improbable), no empress Anna cor182-83; II, 1074; The Book of the Eparch, ed. 1. Dujéev (London, responds with this patron. If it dates to 1344, the patron could be 1970), 39, 166, 247, 273, 289: A. Lewis, Naval Power and Trade in the Anna Anachoutlou, who had been a nun before she briefly occupied Mediterranean, A.D. 500-1100 (Princeton, 1951), 93; Pegolotti, ed. the Trapezuntine throne in 1341-42. She was, however, strangled in Evans, 119, 123, 144 f., 416, 420; W. Gell, Narrative of a journey in 1342 on the orders of one of her successors, John II], whose father the Morea (London, 1823), 183; Thiriet, Régestes, no. 2349. Signi- (Michael) was reigning in 1344. The will is testified by John Doukas

ficantly, it was prepared with alum. On the confusion between Trichas, imperial apokrisarios and logothete of the household— kermes and “‘grana,” see D. V. Thompson, The Materials and presumably of the Emperor Michael. Anna’s deeds are unlikely to Techniques of Medieval Painting (London, 1956), 111-14; and R. J. have been approved by a member of Michael’s government (who 1s, Gettens and G. L. Stout, Painting Materials (New York, 1966). 123. in any case, otherwise unknown to Trapezuntine sources). 4. The 19. Book of the Eparch, ed. Dujéev, 39, 166, 247, 273, 289. will is also testified by an otherwise unknown Arsenios, patriarch of 20. Triantaphyllides, Phy gades, 45; Cuinet, Turquie d’Asie, 1, 74. Jerusalem, and by three other bishops whose names and titles pre-

It is identified on R. Kiepert, Karte, AV. sent similar problems—for example, there is a bishop of Bethlehem

128 SECTION XIII that he held the old hermit in high esteem, regarding him asa HISTORY suitable confidant for his grouses about the Trapezuntine Like all Greek colonies on the Pontic coast (except for political situation, and that Gerasimos also wrote letters to Amisos, which claimed Ionian, probably Miletan, settle-

Libadenos. . . ment) Kerasous was established by Sinope and was part of its

John Eugenikos eulogized the conveniences which the — empire. Pharnakes took Kerasous after capturing Sinope in Pontos offers hermits?> and it would normally be futile to 183 B.c. It fell to Pompey in ca. 64 B.c. but was not inidentify a particular hermitage. But the chapel and excava- — ornorated into the Roman Empire proper for another centions, described below, on the striking twin rock of Gedik tury. Between ca. 64 B.c. and ca. A.D. 64 it was successively Kaya point to a hermitage and even “The Heights.” Inthe part of the fiefs of Deiotarus of the Tolistobogii, of Darius nineteenth century the chapel was regarded as Byzantine and son of Pharnakes, and of Pythodoris widow of Polemo.*? retained a little sculpture and wall painting. Alternatively The massive lower courses of the acropolis may perhaps be The Heights” may indicate the summit of the acropolis of ated to this, or the immediately preceding, period. Kerasous itself. Bees stated that it is said that the ruins of Roman imperial Kerasous still struck its own coins. One the monastery of St. Epiphanios are on the acropolis of type of Commodus bears the image of a galley, from which Kerasous” *’—apparently without authority, although KX jenast has suggested that it was a station of the classis there are reports of more than one religious site on the — pon rica, but it is not mentioned as a base of any Byzantine

. . . ° 26 e . > : °

927 , ,; ;

acropolis. fleet.>2

One chapel, on the north slope, is post-medieval and has The oft-told story that Kerasous gave its name to the

been published elsewhere.*” But Bees may be referring to a cherry, which Lucullus supposedly introduced from the city,

cave chapel, reportedly with wall paintings, below the has the appearance of an etiological myth.** Kerasous was summit with its rock-cut wells and “amphitheater and and is the major Pontic center of a highly important hazelnut above another deep well, noted by Schultze and Cuinet. trade.*> Otherwise, its hinterland is agriculturally less proWe were at first unable to find it,"” but in 1964 D. ae ductive than other parts of the coast: the mountains come located the site, which is described below (p. 132). It cou exceptionally close to the sea at this point and grain has had mark either the monastery of St. Epiphanios or the her- 4 he imported both from Koloneia and the Crimea.*° In the mitage of Gerasimos, or both; there is no evidence to make Middle Ages Kerasous also exported cloths and, possibly,

the proposal more than tentative. Koloneian alum. At the turn of the seventh and eighth

Finally, soon after leaving the island of Ares, the centuries there was an imperial office of commerce in Argonauts passed the PtAvpnida vioov. There is indeed an —_—k erasous which was associated, on one seal at least, with

unnamed rock lying about 3 km offshore between the island Trebizond and Lazia.2” of Ares and Cape Zephyrios, but Hamilton surely credits Kerasous was a suffragan bishopric of Neokaisareia from Apollonius Rhodius with an improbably detailed knowledge the fourth century; at the end of the eleventh it became an of local geography by identifying this rock with the independent metropolis. Toward the end of the seventeenth

Argonauts’ island. century pressure on local Greeks, widespread at the time, led

to the extinction of the see, and in 1698 the exarchy of nearly half a century before that Orthodox see is otherwise known to Kerasous was transferred to the eparchy of - rebizone In have existed. 5. No monastery of St. Euthymios seems to be other- 1920 the metropolitan of Chaldia, in a fit of untimely exwise attested in Jerusalem. The famous monastery of St. Euthymios pansionism, claimed Kerasous briefly, seeking a corridor to at Khan el Ahmar, 13 km west of Jerusalem, was devastated in the

12th century and is last mentioned in 1177. See PapadopoulosKerameus, A/S, I, 245; H, 254-57; the same in /B, II, 368; V. 32. Magie, Roman Rule, 1, 374, 433, 386, 561; II, 1237-38. Grumel, “Titulature des métropolites byzantins. II. Métropolites 33. D. Kienast, Untersuchungen zu den Kriegsflotten der Rémishypertimes,”’ AOC, I (Mémorial Louis Petit) (Bucharest, 1948), 174; chen Kaiserzeit (Bonn, 1966), 117 and note 142.

Polemis, Doukai, 185; Grumel, Chronologie, 452; P. E. D. Riant, 34. Ammianus Marcellinus, History, XXII, 1m, 16. There are Etudes sur I ’histoire de l’Eglise de Bethléem, II (Paris, 1896), 80 and indeed wild cherries in the Pontic forests, but Don and Patricia note 3; S. Vailhe, “Saint Euthyme le Grand,” ROChr, 12 (1908), Brothwell, Food in Antiquity (London, 1969), suggest that the cherry

183—88; Chrysanthos, AP, 4—5 (1933), 514—45. was already known in Italy when Lucullus may have introduced the 25. Lampsides, ‘Imavvov Edyevixot “Exgpaoic TpaneCotvtoc. Pontic variety. Xpovoaoynotcs kai Exdoo1c, AP, 20 (1955), 30-31 (stanzas 9—10), 35. Movvtov«K1, findik, kapva Tovtika, appear in inventories written probably between 1444 and 1449. Cf. Bordier (1609), 127. of consignments of merchandise shipped to Alexandria as early as 26. Cuinet, Turquie d’Asie, 1, 74; Schultze, Kleinasien, II, 184. 259 B.c., belying de Planhol’s opinion that it is a comparatively

27. Bees, AP, 16 (1951), 262. modern trade. See Magie, Roman Rule, Il, 1073-74; de Planhol, s.v.

28. Bryer and Winfield, AP, 30 (1970), 233-35. The 19th-century “Giresun,” EI’.

church of the Metamorphosis stood on the west of the peninsula. 36. Kinneir (1813), 328. E. C. Colwell, The Four Gospels of The Armenian church was dedicated to Surb Sarkis (St. Sergios). A Karahissar, 1 (Chicago, 1936), 12.

chapel stood on the coastal road, probably near the eastern harbor; 37. On the Koloneia-Kerasous alum road, see p. 149. To the itis illustrated in M. Blanche, ‘““The lame mayor of Kerasund,” Asia, commercial seal of Kerasous cited in Antoniadis-Bibicou, Douanes,

22 (1922), 285—90, 328. may be added two in the Dumbarton Oaks Collections: of the

29. Cuinet, Turquie d'Asie, 1, 74; Schultze, Kleinasien, II, 185. patrikios George Theophylaktos, kommerkiarios of Lazia, Trebi-

30. Winfield and Wainwright, AnatSt, 12 (1962), 133. zond, and Kerasous, in the period 685—95 (no. 55.1.4373); and of 31. Apollonius Rhodius, Argonautica, Book II, line 1232; the imperial office of commerce of Kerasous, in the period 705-11

Hamilton (1836), I, 261. (no. 55.1.4397).

CITY AND DISTRICT OF KERASOUS 129 the sea. The names of sixteen bishops, from ca. 431 to 1673, tory, for its effects were permanent. If Kerasous had fallen in

are known; no episcopal seals have been noted.°8 1301, the Turkmens would have obtained a major access to The Kerasous Gospels (obtained from a church in the the sea and the days of the Trapezuntine Empire would have town in 1906) have been dated to the mid-eleventh century; been numbered. As it was, the place remained a Greek Sirarpie Der Nersessian notes some Armenian characteris- stronghold, frequently visited by the Grand Komnenoi— tics in their somewhat “provincial” illuminations.*? The particularly on their way to state visits to the Turkmens or to place of origin of the Gospels is a matter of discussion, but | Oinaion.*° eleventh-century Kerasous, with its bishopric, monastery of Sgouropoulos describes the building of the fortress of St. Epiphanios, trading port, and Armenians, seems a Kerasous by Alexios II after 1301 in terms so poetic that no possible and appropriate milieu for them. Byzantine factual information (apart from the obvious indication that Kerasous was evidently a modest place—it may not be an it overlooked the sea) can be gleaned. But the keep on the accident that it does not figure in any work ascribed to summit of the acropolis is almost certainly the fortress which Constantine Porphyrogenitus—but was perhaps just large Alexios built, or rebuilt. In January 1348 Panaretos states and cultivated enough to support a scriptorium. It came into that the Genoese made a reprisal raid on Kerasous, ransackits own In the thirteenth century, when it became the second ing and burning the town—there 1s no question of there ever city of the Empire of Trebizond; always on the edge of a being an Italian station in the place.** The fortress was Turkmen “‘border”’ to the west, it was the most westerly of all perhaps unscathed in the 1348 incident, for Libadenos refer-

the possessions under the direct control of the Grand red to its ‘bronze walls” in about 1355.*° Komnenoi by the fifteenth century.*° It is possible that, like The first medieval reference to the island of Ares Trebizond itself, it had a demarch and some form of muni- (Apavimtat in Panaretos) comes when it was raided by cipal status, for it was able to pay formal homage asa city to Ottoman pirates in July 1368—the first appearance of the the Grand Komnenos during the civil war of 1355.*! There is Ottomans in Trapezuntine history.*® This precipitated an no evidence for an imperial kephale, as was true of Limnia, embassy by Panaretos, it seems, to Constantinople, perhaps Trebizond, and Rhizaion; on the other hand, such an official to obtain help. The pirates are not mentioned again. seems likely—in other words, Kerasous probably was the Immediately before his entry on the island of Ares, Panaretos

capital of a bandon. states that Metropolitan Joseph (John) Lazaropoulos resign-

It was at Kerasous that the Grand Komnenoi were able to ed the see of Trebizond on 12 November 1367 and retired to hold, for the first time, the Turkmen advance from the west the monastery of the Panagia “EAsotoa.*’ Fallmerayer was and retain not only the city but its surrounding villages. The victory of Alexios II over the Turkmen “‘Koustouganes”’ at LyovporovaAov, AP, 19 (1954), 262-82. Panaretos, ed. Lampsides,

Kerasous in September 1301 was eulogized by Sgouro- 63, giving the date as September a.m. 6810 = a.p. 1301, not 1302, as poulos, described by Panaretos, referred to by Chioniades, in Miller, Trebizond, 33, followed by Janssens, Ti rébizonde, 93. Letter

Loukites, and Lazaropoulos, and remembered later by °F Chiomtades, in oA P apac poms. rpniyoptoy Xtoviddov Tow

Bessarion.** It was correctly regarded as an important vic- GoTpovovoy ematodat (Salonike. 1929). dated, — this reason, °° 1301 by N. A. Oikonomides, XnpEetwopa nEpt TOV Ex1OTOAdV [pnyopiov tod Xtoviddov, AP, 20 (1955), 40-44; Constantine 38. Parthey, Notitiae, 37, no. 702.7; 65, no. 269; 98, no. 67; 110, Loukites, in Papadopoulos-Kerameus, A/S, I, 421-30; Lazarono. 218; 136, no. 59; 173, no. 321; 189, no. 230; 208, no. 339; 231, no. poulos, FHIT, 61; and U. Lampsides, “‘Zu Bessarions Lobrede

77; 241, no. 78; 250, no. 199; Gelzer, Texte, 599, no. 77; 608, no. 87: auf Trapezunt,” BZ, 35 (1935), 16-17. See now Elizabeth 629, no. 46; Le Quien, OC, I, cols. 513-16; Janin, s.v. ‘‘Cerasus,”’ A. Zachariadou, ‘““Trebizond and the Turks (1352—1402),”” AP, 35 DHGE,; Chrysanthos, AP, 4—5 (1933), 579-83; the unsigned article, (1978), 342-43.

‘“L’affaire de Kerassonde,” EO, 14 (1920), 459-60; later diocesan 43. Panaretos, ed. Lampsides, 73, 79.

maps in G. K. Skalieres, ‘H Avttoxpatopia tic Tpanezobvtosc, 44. Panaretos, ed. Lampsides, 68. (Athens, n.d. [1921]) (appended, showing Kerasous in Chaldia); and 45. Libadenos, Periegesis, ed. Paranikas, 36, 42. Bryer, Isaac, and Winfield, AP, 32 (1973), 129 (showing Bulancak in 46. Panaretos, ed. Lampsides, 76. Grégoire’s earlier reference Chaldia). The first known bishop of Kerasous attended the Council must, sadly, be discounted. By conflating information in Panaretos of Chalcedon (E. Honigmann, “The original list of Chalcedon,” with the famous epitaph of a Komnenos Mavrozomes in Konya, he Byzantion, 16 [1942-3], 54). Kallistos, metropolitan of Kerasous, built a brilliant analysis: that the Michael of the epitaph was son of

was a Signatory to the Council of Florence in 1438. Ioannikios, son of John Axouchos, who later became a monk and 39. MS Morgan 748, and one sheet in Princeton. See Sirarpie Der was hence also the Papadopoulos who rebelled in 1281 (Panaretos,

Nersessian in Belle da Costa Greene and Meta P. Harrsen, ed. Lampsides, 62). This Michael, alias Papadopoulos, happened “Catalogue of Manuscripts in The Pierpont Morgan Library,” also to be emir of Arane, and Arane was none other than the island Exhibition of Illuminated Manuscripts Held at the New York Public of Ares—surely the most picayune emirate ever proposed, even in an Library (New York, 1933), 15, no. 26; and C. Nordenfalk, ‘““The age of fragmentary Anatolian states. Michael—or a Michael—died apostolic canon tables,” GBA, 62 (1963), 17-34. We are grateful to in Konya in 1297. But Grégoire’s analysis was partly based on a

Professor Bob Bergman for drawing our attention to this misreading of the epitaph, and partly on the misconception that it

manuscript. has anything to do with the Grand Komnenoi. The notion and

40. Clavijo (1404), 109; Sphrantzes, ed. Grecu, 128. Balard, argument remain one of Gregoire’s most ingenious flights of scholSambuceto, nos. 409, 768, 843, shows that in 1290 Kerasous was arly fancy. See H. Grégoire, ““Notes epigraphiques; X.— Michel importing salt from the Crimean region (probably to salt fish), and Comnéne, emir d’Arane,” Revue de I’Instruction Publique en

was perhaps exporting wine. Belgique, 52 (1909), 12-17; F. Cumont,”’ Note sur une inscription

41. Panaretos, ed. Lampsides, 70. dIconium,” Byzantion, 10 (1935), 505-15; and 12 (1937), 206—11;

42. Sgouropoulos in Papadopoulos-Kerameus, A/S, I, 431-37; Cahen, P-OT, 210.

cf. T. Papatheodorides, ‘Avéxdoto. otiyor Ltepavov tov 47. Panaretos, ed. Lampsides, 76.

130 SECTION XIII told that the monastery on the island was dedicated to the the information. Later, when Evliya reported on the island in Eleousa,*® and it has therefore sometimes been understood 1644, he noted only that Cossacks had used it as a base for that it was Lazaropoulos, rather than Panaretos, who went attacking Kerasous.°* It seems unlikely that the monastery on embassy to Constantinople after the Ottoman raid.*? would have survived such an experience. There is, however, no doubt about the interpretation of the Today, none of the remains on the island can be definitely text and it also seems most likely that Lazaropoulos retired identified as monastic. Probably, apart from the tradition of not to the island of Ares (which he mentions nowhere in his the 20 May panayir, the only memory that it had been a holy writings), but to another monastery of the Eleousa—the place is a long-standing custom of placing the rags of the Virgin of Compassion—which stood near the Genoese ar- sick on a rock below a tower at the southeast corner of the

senal on the Daphnous beach, east of St. Sabbas in island. Trebizond.°® Furthermore, against Fallmerayer’s identifi- Clavyo describes Kerasous in 1404: “It stands on the cation of the Eleousa with the island of Ares can be cited the shore, with its houses built all up a height that overlooks the

more substantial belief of two local Greek scholars and sea. A strong city wall encircled the whole of this height, Cuinet, that the island monastery was in fact dedicated to enclosing within its limit many orchards and fine fruit St. Phokas. To complicate matters, it was perhaps one of __ trees.”’”° these scholars, Triantaphyllides (then a schoolmaster of The castle was garrisoned with Cepni troops after the Kerasous), who was Fallmerayer’s informant, or misinfor- Ottoman conquest. In 1525 the place included 31 Muslim mant, of the dedication; and to further confound the issue, and 221 Christian households,°’® probably a fair indication modern Turks report a panavir (fair) on the island on 20 of its size-—-perhaps about 1,250 souls—in the later Middle May—principally the feast of the singularly obscure martyr |§ Ages. Bordier noted that the houses were scattered along St. Thalelaios, who has nothing to do with the Pontos.*' We tortuous streets, but that the walls were still impressive.*’

are inclined, however, to follow Ioannides and Trianta- Evliya saw it after the Cossack raids and commented that phyllides in ascribing the dedication of the monastery to St. the castle did not in fact defend the town (suggesting that Phokas, a local martyr, widely cultivated in the Pontos, most inhabitants were then living on the unwalled eastern whose maritime connections are most suited to the island. side of the peninsula), and although it had evidently been a By the time that Fallmerayer, Ioannides, and Trianta- large place, it had shrunk. The town gardens still yielded phyllides knew the island, the monastery had been long plentiful fruit and the anchorage, although good, offered no destroyed. In fact, our only clear evidence that it existed protection against contrary winds.°8 otherwise comes in 1609, when Bordier noted it.°* It might A final blow came in 1764, during the Derebey wars. The be argued that the monastery was founded after 1468 and fortress had been refurbished by Tistaroglu, the Giresun was abandoned before 1644. Bordier states that he was told bey,°’ who lost it to a rival after a long siege which ended that the islanders of Ares held out against the Ottomans for with the devastation of the town and the destruction of its seven years after 1461. This would have been perfectly then surviving Greek church.°° Its bishopric had already feasible: the island has its own water supply and is well gone, and it was only in the later nineteenth century that defended. That there was some local resistance is suggested Giresun saw a quiet revival. by a nineteenth-century Greek tradition that Kerasous itself

held out for many months against the Turks and only sur- MONUMENTS rendered on condition that the Christians remain (which 1. Giresun Kale (fig. 21, pls. 55a, b—60a, b) independent evidence shows that they did) and bear arms 1 The site, a volcanic spur projecting about 1,000 m into the eturn for providing ferry service Over a local river.°° But sea, 1s joined to the mainland by a low shoulder or isthmus to Bordier was not specifically told that the inhabitants of the the south. The northeastern and eastern sides of the spur island included monks in 1461-68, where one might expect were steep and protected by cliffs and a rocky foreshore, but the original form of these has been obscured by the construc-

48. Fallmerayer (Fragmente, 1840), 138-39. tion of the shore road at the base of the cliff. The southern 49. E.g., Miller, Trebizond, 66: Lazaropoulos “fled in 1368 to and western sides slope steeply, but not precipitously, down Lompaides tP 1111956) sland monastery of Kerasunt.” Ct. to the isthmus and the port respectively. The original circuit

50. Finlay (MS, 1850). fol. 43"; p. 350. of walls ran along the seashore and cliffs on the western, 51. Triantaphyllides, Pontika, 82—83; Ioannides, Historia, 218: northern and northeastern sides. On the eastern side the wall Cuinet, Turquie d'Asie I, 75. Fallmerayer states that he was told of turns roughly westward to leave the sea and climb to the the dedication by ‘“‘einem kerasuntische Didascalos (/oc. cit. in note

48), in 1844. Perikles Triantaphyllides became a schoolmaster in

Kerasous in 1842: see D. N. Oikonomides, Lvvontiky iotopia tov 54. Evliya (1644), IT, 41. NEepiovopov “EAAnviKkod Ppovtiotnpiov TpaneCovvtoc, PPh, | (6) 55. Clavijo (1404), 109. (August 1936), 4. On the memory of a festival on 20 May, see the 56. Gokbilgin, BT TK, 26 (1962), 330-33.

Giresun Il Yilligi 1967, 198. It is, however, quite possible that the 57. Bordier (1609), 115. Turks have inherited the Old Calendar with the memory of the 20 58. Evliya (1644), II, 40-41. May feast, in which case it is more likely that St. John the Theologian 59. The keep, called the Ug Kale in Hamilton’s day (1836), I, 263,

(8 May) is being celebrated: see Synaxarium CP, cols. 663, 697. then had ‘‘on one of the angles of the wall ... a small wooden fort

52. Bordier (1609), 115. with loopholes for musketry, said to have been erected by a powerful phyllides, Phygades, 32. 60. Triantaphyllides, Phy gades, 94; Bryer, BK, 26 (1969), 196--97. 53. Miller, Trebizond, 107, presumably following Trianta- Dere Bey, who held possession of the place.”

CITY AND DISTRICT OF KERASOUS 131 summit, then follows the ridge westward and descends to the giving some relief to the external pattern of the masonry. The sea again on the west side. The wall along the ridge is pierced only surviving decorative features are a string course running by an upper gateway and a short stretch forms the outer wall about five courses below the top of the wall and consisting of of the keep (fig. 21, pl. 55a, b). Although there 1s now no trace a single projecting band of blocks, and a cornice of projecting of it, it would seem logical that the wall of the outer ward, or blocks (pl. 57a). One rectangular tower survives in the ridge

town wall, should have continued up onto the isthmus and wall, southeast from the upper gate. The tower projects possibly down to the sea on the other side—thus providing outward and does not break the internal line of the wall. Its the normal first line of defense before an enemy could reach ground (or first) floor entrance is rectangular and there must the gateway to the inner fortress. The spur has two summits. have been side entrances at the second floor level to allow a The slightly higher western summit of 129 m is crowned by catwalk to pass through the tower walls. The form of the the tomb of Topali Osman Aga, who played a notable, if classical tower was followed in the medieval reconstruction idiosyncratic, part in the early years of Mustafa Kemal’s rule of the walls. Leveled seatings for them were cut into the rock, as commander of the Atattrrk’s ‘“‘Laz’” bodyguard.®' The the outer face of which has frequently been cut away to give a

eastern summit, about 100 m away, is occupied by the vertical face flush with the masonry above, adding height to medieval keep. There is now no trace of the amphitheater—if — the wall. There are also rock-cut steps on the side of the such it was—referred to by early travelers, nor any sign of | western summit and two large excavations which may repthe great classical temple or palace whose ruins Bordier saw resent cisterns. One is in the keep (see below) and the other is in 1609.°* This might have stood on the massive rock plat- on a side of the western summit. Both are roofed and water

form near the southwest corner of the town area within the was extracted through a hole in the side. fortress. Elsewhere within the site the natural rock has been The medieval walls appear to be of two periods. The earlier leveled in various places to provide wall seatings, but excava- and major part consists of roughly-squared blocks of stone

tion would be necessary to determine their form. laid in regular courses, with gaps evened up by fragments of The original town evidently lay along the west and brick and small stones. The mortar is of white lime with a northwest sides of the peninsula on comparatively gentle pebble filler. The facing stones average about 20 x 30cm. slopes enclosed by walls, but, as we have suggested, could The ridge wall is between 0.50 and 2.00 m thick and is built have extended across the isthmus and down to the eastern over the classical wall, leaving the remaining thickness of the shore; between it and the clearly delineated upper quarter are earlier structure to form a parapet which served as a catwalk. traces of a curtain wall running north-south, without evident The tower quoins are of larger ashlar blocks (pl. 56b). The

classical foundations and apparently of medieval or towers are rectangular, round, and pentagonal—the latter at Ottoman build. At no point is the wall more than a few the western end of the ridge wall. This regularly-coursed distinct stones in the walls of houses or gardens, but its masonry of small blocks probably represents the work of course can be traced fairly clearly by a drop of about 3 m in Alexios II after 1301—although it could be thirteenth-

the profile of the quarter. century or even Byzantine. It is very similar to the lower city

There are two surviving gateways to the fortress, of which walls of Trebizond and to those of the castle at Oinaion. the lower one by the sea to the southwest must have led to the A later period of masonry is of rough stones laid in walled town (pl. 57). The upper gate on the southeast side random courses and heavily pointed up with lime mortar at must have led into the inner fortress—the modern road the exterior to give a flat, weather-resistant, surface (pl. 58b). leading to the park on the summit runs through what isleftof | This could be later Trapezuntine work, Ottoman, or even the

this gateway (pl. 58a). final defenses of the Tistaroglu Derebeys of Giresun.

The lower gateway, now heavily overgrown, provides a The walled area, which we have called the keep, 1s an good example of the rectangular masonry of the classical, or irregular enclosure with its longest straight side flush with the ‘‘Pontic,” period. The gate itself has now gone, but the finely ridge wall. The other seven sides vary in length, the whole beveled square bastions, stepped to receive it, are particularly enclosing the eastern summit of the acropolis—a maximum

impressive. It is probably this gate which is referred to by extent of 25 x 35 paces. This is a large area to have been Bordier: ‘Les portails de la ville, qui sont du costé de la completely roofed over and it seems more likely that it marine, sont de tres bel aspec et admirable architecture, les comprised a courtyard with wooden buildings against its cintres desquels sont enjolivez, ou ornez de cordons et en- walls. One indication of this is a small rectangular structure, trelacs de tres gentille artifice.’°* There is no sign of this about 5 x 4 m, roofed with a barrel vault and built up decoration today. To the north of the gate is a much later against the southwest wall of the keep. The fact that this

circular tower. structure has a masonry barrel vault suggests that it stood in

The walls exhibit a variety of masonry. The classical work the open air. Its interior walls are faced with ashlar blocks of is of rectangular blocks of green breccia of different sizes. Oinaion limestone, alternately placed as stretchers and headThey are well squared off and laid in regular courses, but the ers in the manner of other Pontic buildings which can be size of the blocks varies within the same stretches of wall dated to the thirteenth century. It has a rectangular window and there may be two periods of work (pls. 57b, 58a,b). Both or gun port facing southwest toward the upper gateway. types of classical masonry have stone headers set at intervals, In the center of the keep is a rock-cut excavation which

61. Giresun Il Yilligi 1961, 59-61. must represent a well or cistern. It has been left with a natural

62. Bordier (1609), 115. rock ceiling and the opening to it is from the southwest; rock63. Bordier (1609), 115. cut steps lead to the opening (pl. 59a). The vertical cavity is

132 SECTION XIII now about three meters deep, where it is blocked with debris, square stones”’ hint at a Byzantine date. Hamilton’s second

but was probably much deeper. church has left no trace either and may have been replaced by The surviving entrance to the keep is through a narrow the large nineteenth-century church of the Metamorphosis passage next to the outer wall on the south side. The facing (now destroyed).°® Nor can the rock-cut steps be found, but stones of the doorway have gone, but a pointed relieving arch they may be associated with the massive rock-cut platforms

of stone voussoirs flush with the masonry face of the wall at the southwest corner of the walled town. Rock-cut

survives (pl. 60b). “troughs” survive along the northeast shore; they were perBeam-holes for the joists of a second-floor structure are haps also used as salt pans. seen on the interior face of the wall on the north side; a niche > Mosaic Floor

or fireplace lined with Oinaion limestone ts set into it (pl. 60a). ; ; ,

To the west of the niche the wall has broken away, leaving In 1958 a mosale floor was un covered during excavations part of a second-floor opening (pl. 59b). This has a gabled for the foundations of a new building in the lower town, near top and may have led to a garderobe—1in which case it would the port. The mosale then lay rather more than two meters be the only such Trapezuntine convenience known to us. The below ground level in a garden; the owner reported that it original keep wall was about 1.20 m thick on the northwest stretched Peyond the hiner end of the garden, some ten side, but appears to have been strengthened twice by ad- hefowe away, DECause is ather had dug there a few years

ditions which are respectively 0.60 and 0.90 m thick. The ore and oun : continued under the garden aL IL masonry of these additions is similar to that of the main re corner © the mosaic Hoor whe was exposed (pl. 61)

eg consists of a chevron framed by a conventional wave pattern.

hen of the keep and it is difficult to suggest and date for The outer border contains white lettering, each letter The features of Giresun Kale which are now destroyed or pot Antes high) framed : " a da rk border, reading: severely ruined are best described by Hamilton. From the The letter forms are compatible with a fifth- or sixth-

; . —perhaps [+ 7 Glyia dvavew([On]. ...

keep (pl. 56a). century date; the inscription suggests that the church, of the ancient walls may be traced almost the whole way ... to the which it must form the floor, was restored then.

with masonry of the some stvle ne oral which ese high 3. About 1,000 meters west of the old town a rock, removed tower overgrown with ivy. Having reached the shore, I re- from the path of the new coast road, crowned with medieval turned by the beach, where the walls were entirely Byzantine | Masonry, stood a few meters from the sea shore. The form of and where are the ruins of a small Byzantine church, built of the masonry was not clear. It could have been a watch tower well-hewn square stones, cemented together with mortar, with or small chapel.

considerable remains of painting on the inside. These walls ; “4:

were very perfect on the west side, and passing through them by 4. Gedik Kaya Kilise a postern gate I descended to the ruins of another church near About 2 km due southeast of Giresun Kale is a steep hill

the beach, where is a small harbour, fit only for very small topped by two rocky teeth (hence Gedik Kaya, “‘split vessels. Here was a double line of walls, the defences having | rock”’)—-the place is distinctive enough in the engravings of been made stronger on this side, partly because, from the depth Tournefort and Hommaire de Hell (pls. 62a—c,-63). The of water, it was the only spot where an enemy’s vessel could _ height of the summit is about 250 m. In the cleft between the

approach the shore in safety. Between these walls we entered a rock, which can be reached from the coast road or, more large and dark apartment: from whence, after procuring a easily, from the hospital on the Sebinkarahisar road to the light, we descended by secret steps to the beach. Here the rock south, are the foundations of a church (pl. 62b). had been cut away, presenting a perpendicular face, up which The foundations reveal that the church had been a long

another flight of steps led to the Agha’s konak. In walking ; , 5 . round the town I had observed many large square troughs cut rectangle with a rounded apse, oriented at 80 - Possibly part

in the solid rock upon the sea shore: they appeared to be the of the length was occupied by a narthex, as in one of the spots from whence the stones used in the old walls had been chapels at Bibat.°’

quarried... At present they are only used by the women as The mortar is of lime and pebbles. None of the walling

washing troughs.°* remains above foundation level. The foundations are of Hamilton seems to have missed the cave church noted by rough uncoursed stones with brick and tile fragments. A Schultze and Cuinet, of which D.C. W. reported no sign in small section of the northeast side of the apse is exposed (pl.

1962. In 1964, however, D.C. W. located it, well concealed 62c). The walls are about 65 cm thick and the external by surrounding scrub, on the eastern slopes of the prom- dimensions of the building are 14 x 67 Paces. Ly ing around

° , . . the site are quantities of fragments of brick and ridged-edged

ontory and below the Keep. Of modest dimensions, it may tiles. The bricks average 4 cm in thickness and the flat base of originally have been a classical rock-cut tomb; very little now the tiles 2. cm

remains of the wall paintings. Hamilton’s first church may There is a rock-cut tomb (pl. 63) on an outcrop im-

have been replaced by northeast the curiously formed nineteenth_- 65 7. mediately of the apse and about 3 m above ;it.; It

century one, which is published elsewhere.®* The “‘well-hewn was cut in such a way as to narrow at the neck and widen at

64. Hamilton (1836), I, 264-65. the head and it is oriented at 80°, with feet to the east. The 65. Schultze, Kleinasien, II, 185; Cuinet, Turquie d’Asie, 1, 74;

Winfield and Wainwright, AnatSt, 12 (1962), 133; Bryer and 66. Papamichalopoulos (1902), 265.

Winfield, AP, 30 (1970), 233-35. 67. See p. 270.

CITY AND DISTRICT OF KERASOUS 133 dimensions are 1.90 min length by 0.25 m in width at the feet chapel within a fortified enclosure (figs. 22, 23; pl. 64 a, b).

and 0.45 m at the shoulders. Only the foundations of the chapel survive in what is nowa

The western summit has two holes, presumably to hold hazelnut grove. The mortar was of lime and sand devoid of water, excavated in the rock. Other depressions may rep- pebbles. The ground plan (fig. 23) suggests that it may have resent excavations for water or for tombs. One measures 2 m been roofed with a dome. in length by 0.35 m in width at the feet, and, perhaps, 0.50 at The steep sides of the spur provide a natural defense on the the shoulders, where the rock has been broken away. A east, south, and north sides, which has been im proved by groove has been cut round the top of it, presumably as a walling. On the western side, where the spur joins the main

seating for a wood or stone lid. body of the hill, there is no natural defense. Here fortifiThe eastern summit has a tomb oriented at 100°. It is larger cations consist of an outer wall with two round towers which than the one near the church and the rock around it has been must have defended a gate.

leveled, with steps cut into it. Below the summit, on the The mortar of the walls and towers is of lime and small southeast side, is a smaller cutting about 1 m long and pebbles. The surface masonry is made up of roughly squared oriented on a northeast-southwest axis. It is now about | m blocks of stone laid in regular courses. The upper part of the deep but its earth fill may conceal greater depth. It could walling on the north and east side looks like a later repair.

represent a child’s tomb or a water stoup. The masonry, while not dissimilar to Alexios IT’s work on A substantial excavation is found about 70 m below the Giresun Kale, could be middle Byzantine—1in other words, eastern summit on the northeast side. The entrance 1s an oval the identification with Constantine Porphyrogenitus’ station opening about 50 cm wide, which is the mouth of a tunnel of Prine is not, archaelogically, impossible. The chapel probleading down into the rock for about 8 mat an angle of about ably represents that of the later monastery of St. George,°’

40°. The mouth faces roughly north and the cutting runs but whether it was also a fortified medieval monastery is southward. There is a small hole above the mouth, which problematical. The fortifications, rather than the chapel, are may have been used for fastening a lid, and a larger hole the predominant part of the buildings, suggesting that this below. D.C. W. was unable to explore the tunnel, but was site was originally a government fortification with a chapel told that it opened out to a width of about 2 m at a depth of within its walls. Its purpose was probably to act as a strongabout 8 m. The tunnel then changed direction, continuing for point from which to harry and divert any force threatening a further 8 m, where a blocked passage led off from it. The to besiege Kerasous and to cut off its supply lines to the secondary hole below the main entrance reaches the first south. It could also act as a signal fort. The salient feature of chamber where the tunnel changes direction. The tunnel the site, from whatever direction it is viewed, is the large leveled out at this point for a further few meters, where a round tower which gives the impression of fortifications

second blocked passage led off. much larger than in fact they are (pl. 64b).

,; 6.the Giresun, or Puga, made Adasi that the , , a. broken -— away since excavation The maximum extent ofwas the island and of Ares, which is The surface of the rock at the entrance suggests that it has ;

vegetation. .

tunnel was originally about 2 mislonger. The sides of Itthe . , 30 m ; roughly circular, about 250 paces (fig. 24). is about tunnel have been made uneven at intervals, as if to allow the oa: . ; . aboveThe sea level at its highestnow point.obscured There is no beach user; to getappears a grip ontothem. entrance byexcept, ;and the rock fall sheer into deepiswater on all sides —_ , , perhaps, to thedate north. The presentoflanding place, but to the Iteee is 6g: impossible the remains the church, . ; south;is.;de; southeast, where there is a small inlet in the rocks, Cuinet’s opinion that it is Byzantine®® is plausible. The rock. ; scribed in a sixteenth-century Greek portulan place cut tombs and tunnel may .;well have no connection with theas; the .; 10 where sailors could anchor their boats ashore. '° Two rings have been cut into the solid rock further west. This area, The perhaps from represent a Bronze Agewinds, shaft arwas , prob; . ; tunnel ; whichcould is protected the northwestern

church. . ; , burial, or even a trial search for water in preparation for one ;

, , ablyborings always the landing place of theof island, which can only be of the; ;great subterranean which are a feature / . the .; . besieged by small boats or skiffs, such as many summit fortresses in Anatolia. The alterations of ; ; 41Ottoman rapae ;hill , aokaAptia which attacked it in 1368.

direction in it could be explained by The a system of following the oo , rocky shore ts fairly low and clean swept for a varying natural rock fissures. These tunnels are generally regarded as point ; ; ; . the oo. ‘*Pontic”’ oe _ , distance to where thebrickwork contour.;rises. At this being in origin,inland but Middle Byzantine ; ; in island is completely by a boundary a: wall., The maaround the entrances to tunnels the castlesencircled of Koloneia ; . ; sonry is of roughly squared blocks laid in fairly (Sebinkarahisar), Eudokia (Tokat), and Amaseia (Amasya) , ,smaller - ; regular ; ; courses, with the irregularities evened up with stones shows still used as average water sources Byzantine . ,; ; to . . ; ; - that and they brickwere fragments. The size ofinthe stones 1s 20

times. of Gedik evidently held religious . a 30The x 10summit to 15 cm; mostKaya are of the local greyish black significance and we have suggested above that it might rep; , . conglomerate, but there are some good ashlar blocks of resent Gerasimos’ hermitage. 5. Gedik Kaya Kalesi 69. Triantaphyllides, Pontika, 45. The hill ends in a lower spur east of the Gedik Kaya 70. Delatte, Portulans, II, 31.

summits and about 150 m above sea level. The spur has a 71. Panaretos, ed. Lampsides, 76. On these skiffs, see A. Bryer,

‘Shipping in the Empire of Trebizond,” The Mariner’s Mirror, 52

68. Cuinet, Turquie d'Asie, I, 74. (1966), 3-12.

134 SECTION XIII imported sandstone. The mortar is of lime and pebbles and the second fragment could hint at the beginnings of the south the core of the walls is made of well-bedded mortared rubble. side of an apse and is a reminder, at least, that the island once The width of the wall varies along its length but averages one housed a monastery. Further details within the walls could meter. As is shown in plate 65a, there are two openings with only be gained by excavation.

semicircular arches, which are formed with neater stone There are two water sources. The larger is within the walls blocks (pl. 65b), and there are two semicircular towers on the to the west of the rectangular tower and provides fresh water

western walls (pls. 65c, 66). at ground level, currently used. The second is a small trickle The only visible sign of masonry which may be of an earlier outside the walls on the north side.

date than the rest of the walls occurs to the south of the The soil is rich and is at present cultivated by a single southernmost of the two western towers, where the wall is family who also keep sheep and cows. It 1s clear that a small. plastered with lime and powdered earthenware and there is if determined, colony of besieged people could subsist on it

the beginning of a barrel vault of ashlar blocks. for some years. The surrounding rocks are a major breeding A rectangular tower or keep (pl. 67a, b) has four stories ground for both seagulls and migrants, whose great numbers and now stands to a height of about 12 m. It has window provide an echo of the fearsome birds of the island of Ares openings which are slits on the exterior, about | m high and who beset the Argonauts. 0.07 m wide but opening out on the interior to a width of The walling of small but regularly coursed stones is similar 1.36 m. The walls are typically 1.22 m thick. There are no to that of Gedik Kaya Kalesi and of the presumed work of signs that wooden beams were used to reinforce them, but the Alexios II in Giresun Kale. The more elaborate masonry beam holes for floor joists are visible, both in this rectangular and windows of the rectangular tower are reminiscent of that tower and in the semicircular ones of the west wall. in the square tower at Kordyle, probably built, or rebuilt, by The only other remains above ground of walls within the Alexios III after 1362. It seems very probable that the forboundary enclosure are two fragments on the highest part of tifications on the island of Ares are fourteenth-century the island, to the north of the tower. One is so small that no Trapezuntine. information can be gleaned from it. An angle halfway along

Section XIV

CAPE ZEPHYRION AND KENCHRINA (?) DESCRIPTION Evliya mentions Purpolum on the west side, “which is a

. . . . 1 1 274

Geographical sources mention the cape and classical settle- small square castle, situated on a hill by the seashore, with a

ment of Zegbpioc, Zefalo, Zeffanol, Zeffallo, Zeffara commander and a garrison; the inhabitants are Greeks. We Zefano, Cefalo, and Zephyros until the sixteenth century. ! weathered the cape of Zemreh, where villages are to be seen

There is no doubt that the cape is either or both the twin)" the mountains. .

promontories now called Ulu Burunu and Cam Burunu and The Turkish | : 200,000 map marks two villages on the there is no reason why the tiny natural harbor still called west coast, Kalecik and Hisartistu. They are in fact names for

Zefre Liman should not represent Zephyrion itself. the same place, which consists of a narrow natural harbor The cape is a heavily wooded mountainous area; its high- and small coastal castle, and a village hidden in the steep est peak in the center rises to 546 m. There are a number of woodlands above. Local enquiry and two examinations of small villages along the rocky coast, linked only by well- the coast from Esbiye to Kesap (which entailed walking right preserved paved pack-animal ways in the Pontic style. In round the cape and a close circumnavigation by boat) sugmodern, and probably medieval, times the coastal road from 8°! that, apart from Hosios Antonios, this 1s probably the Kerasous bypassed the cape by turning inland southeast at only castle on the cape. Kalecik-Hisarustu castle is almost the “road’s mouth,” Apouootopiov (Yolagzi—the Greek certainly Evliya’s Purpolum, a word which appears to mask a name is perhaps a nineteenth-century translation of the "OW lost Greek name—perhaps ending in *-polis.” We tenTurkish, rather than vice versa), encircling the Kel Dagiand _‘@tively propose it also for the site of Kenchrina. Armelit Kale (described below), and rejoining the coast in The name of Kenchrina (Kexptva, Keyxptva, Keyypewv) the gorge of the Yagli (or Esbiye) Dere. At this point the road — SUBBESTS sorghum millet (Keyxpos) and the places called

is overlooked from the west by a medieval castle, now called Kenchreal in the Argolid, Corinth, and elsewhere. This Andozkalesi, which is slightly misplaced on the Turkish Kenchrina was an imperial castle with a small town (TOALX1 : 200,000 map. Andozkalesi is mentioned in no medieval viov)”; It Is mentioned only by Panaretos and Libadenos in source, but Triantaphyllides identifies it with the castle of | COmmection with the Trapezuntine civil wars of 1349-55. ‘Ootoc “Avtidvioc,? which suggests a medieval name. The Panaretos makes it clear that the place could be invested by position of this castle at one end of the road, Dromostomion land and sea and that Kenchrina was close to both Kerasous at the other, and Armelit Kale in the middle show that the 24 Tripolis.® In 1886 a Greek of Kerasous proposed that road may always have turned inland—no traveler reports Bulancak represented Kenchrina.’ Probably following this following the coast round the cape, a difficult operation suggestion, K iepert placed it on his map (with a query) on the which cannot be negotiated by wheeled traffic even today and Dikmen Tepe, 3 km south of Aivasil and 5 km southeast of which more than doubles the distance between Andozkalesi Bulancak—the site is described above." But nothing on

and Dromostomion. In fact most travelers, until this cen- | Dikmen Tepe suggests a small town and imperial castle: tury, went from Tripolis to Kerasous by sea, but the course furthermore a passage in Panaretos indicates (although not entails standing out so far that they could see little on either conclusively ) that Kenchrina lay to the west of Tripolis, side of the cape. There are only two records of any detail. A rather than of Kerasous. In 1355 {wo generals went to sixteenth-century Greek portulan, working from the west, Tripolis and then on to Kenchrina (cig tag Tpinodeis Kat

“ . wg .

mentions first a long beach (evidently that below Diizkéy), — &®S TOD Keyxprva), where in the context of the campaign

and then two unnamed places on the east side of the cape.° described in a preceding paragraph, Kerasous would prob1. Arrian, 24; Anonymous periplus, 36, Skylax, 86; Ptolemy, 4. Evliya (1644), I, 41. Geography, ed. Miller, 875; Miller, JR, col. 648; Kretschmer, 5. Libadenos, Periegesis, ed. Paranikas, 44; ed. Lampsides, 75, Portolane, 648; Thomas, Periplus, 249-50, 270; Delatte, Portulans, 81—84. II, 34; Black Sea Pilot, s.v. “Zefre Burnu,” 405; Ritter, Erdkunde, 6. Panaretos, ed. Lampsides, 71. XVIII, 833: Bzhshkean (1819), trans. Andreasyan, 38, ““Keciburnu”; 7. Paulides, ATP, 2 (1886), 196.

Chrysanthos, AP, 4—5 (1933), 86. 8. Kiepert, ZGEB, 25 (1890), 321, map; for Dikmen Tepe, see 2. Triantaphyllides, Phygades, 34; Bryer, AP, 24 (1961), 108. p. 125 above.

3. Delatte, Portulans, Vl, 34. 9. Panaretos, ed. Lampsides, 71.

136 SECTION XIV ably have been mentioned if Kenchrina had lain further west. area may have been about 40 paces wide at its greatest extent The balance is against an identification west of Kerasous. from northeast to southwest. A rock-cut cistern was reported Chrysanthos, apparently on the strength of a mention of on the north side, but D.C. W. was unable to find it in the Kenchrina by Libadenos (who was also there in 1355), placed dense undergrowth.

it without argument on Cape Zephyrion, where he stated The fort was evidently intended to oversee the road across that a castle stood in his day.'° Libadenos’ reference in fact —_ the cape and may have been used as a signal station. It was

gives no indication of the whereabouts of Kenchrina, but misty when D.C. W. visited it, but Dikmen Tepe, west of Chrysanthos’ assumption is a good one because the only area Kerasous, and Hag Tepe, east of Tripolis, should normally between Tripolis and Kerasous which is unaccounted for is be visible from it. that of the mountains and coast of Cape Zephyrion, and

because it explains the absence of Kenchrina on the por- 2. The Castle of Andoz, Holy Antony ; -

tulans, where the cape is naturally marked instead. The This castle Is best approached by fording the Yagh Dere remains at Kalecik-Hisarustu appear to be the only traces of below it (pl. 68a) —which Is usually less hazardous than a medieval castle on the cape and, we propose, may not only attempting the single-plank bridge whi ch runs across the represent Evliya’s Purpolum, but also fourteenth-century river mouth 500 m below—and climbing steeply through

Kenchrina. overgrown hazelnut groves to a ridge about 250 m high. Here two semicircular bastions of dry stone mark what is evidently

History the western, and probably only, entrance. The castle lies

almost due east-west, and is about ninety paces long and no Cape Zephyrion has no ancient or medieval recorded his- more than twenty paces wide at its broadest extent, reaching

tory. Kenchrina is probably the fortress in which the east over the river (fig. 25). It is in three sections. The first, Trapezuntine warlord Niketas Scholaris was imprisoned in western, enceinte is the largest, about 58 paces from the 1345;'' it was from there that he came to be installed as _ bastions at the southwest corner to the northeast corner. It is Grand Duke in 1349."* Alexios III exiled members of the very heavily overgrown and the north walls are difficult to Doranites and Xenites families to Kenchrina in 1351 and two distinguish. Those on the south side are almost straight and Doranitai were strangled in Kenchrina castle the next year.'* stand above steep rock. At the northeast corner is a rectIn the general rising of 1355 Alexios sent warships against the — angular vaulted structure of irregularly-coursed stones and

Scholaris in Kerasous—but he was in fact in Kenchrina mortar. Its external measurements are 4.84 x 3.72 m. It with his faction. Scholaris’ son was in Kerasous but managed stands about 3.10 m high from floor to vault and the interto escape to Kenchrina. Alexios, leaving his army inTripolis, _ nally plastered walls are about 0.65 m thick. There is a door sent for cavalry from Trebizond and invested Kenchrina by on the southern side, a window on the western, and a niche land and sea. The Scholaris faction paid homage to him but 1.75 m wide on the eastern. Most of the roughstone vaulting remained in the stronghold, but later that year two generals is intact, but the windows and door are caving in. It could be

captured Scholaris and his followers.'* The history of Trapezuntine work but the structure is neither a chapel no or Kenchrina during these years suggests that it wasarelatively _—_q cistern. The niche suggests that it is probably a Turkish obscure and inaccessible place, which received unexpected tomb rather than the tomb of a “Holy Antony.”

prominence as the Scholaris faction’s hideaway. The second section of the castle, beyond and below a rough curtain wall, is a finger about 25 paces long, which

MONUMENTS projects east along a rocky spur. This culminates in a third

1. Armelit Kale section, an outpost 8 paces long and 3 paces wide with, at its The road across Cape Zephyrion climbs to a height of eastern extremity, a comparatively well-built semicircular about 500 m at the pass above Kazanh. To the south a bastion. Below it the spur falls giddily down into the Yaghi conical hill rises to about 750 m. On its summit is the site of Dere. The second and third sections command views of the

what appears to have been a small fort. river and road for about 5 km to the south and on to where About 10 m below the summit, on the northeast, are they reach the sea, about | km to the north. random-coursed masonry foundations. A villager reported The function of the castle of Holy Antony asa watchtower that there had been walling, but that ithad beendestroyedby °° the road is Clear enough. The site is unsuitable for a treasure seekers, who had also dug here and there on the settlement in the crags below it, but one might speculate that site—as was demonstrated by signs of recent random exca- it may also have served as the acropolis of a local settlement, vation. There was a fair quantity of unglazed earthenware perhaps on the site of the nearby modern Esbiye. The r oute

sherds and fragments of ridged tiles (for measurements of southward leads eventually lo Alucra and Koloneia which, see Appendix) on the southeast of the site. The hill has (Sebinkarahisar ); there is said to be a mine nearby and the an abrupt flat summit of about six paces across; the walled Yaghi Dere delta, although not wide, is fertile. 3. A sixteenth-century Greek portulan indicates a church of 10. Libadenos, Periegesis, ed. Paranikas, 44; ed. Lampsides, 75; the Theotokos on Cape Zephyrion.'* We saw no traces of it

Chrysanthos, AP, 4—5 (1933), 86-87. on Ulu Burunu or Cam Burunu. 11. Panaretos, ed. Lampsides, 68. 12. Panaretos, ed. Lampsides, 69. 13. Panaretos, ed. Lampsides, 70.

14. Panaretos, ed. Lampsides, 71. 15. Delatte, Portulans, Il, 34.

CAPE ZEPHYRION AND KENCHRINA (7?) 137 4. Kalecik-Hisarustu Castle is modest in size, but the configuration of the ground imThis fort stands on the largest of a series of small rocky mediately to the east hints at further structures and the promontories projecting into the sea. To the north there is a roughly shaped stone, lime, and small pebble mortar with sheltered modern quay and beach where boats are winched pounded earthenware (the latter feature a Byzantine, or even up. To the south there is a rock, about 30 paces long, on Roman, commonplace, rather than Trapezuntine) brings it which stand traces of stone and mortar walling. The prin- into the category of older Pontic coastal stations. It is clearly cipal feature is a promontory, cut off at its neck by a wall intended to guard and service the little harbor. If this place about 20 paces long, which stands up to a height of 2 m in can be identified with Kenchrina, it is hardly worthy of the places (fig. 26; pl. 68b). This forms the east side of a rect- description of “‘polichnion.”’ The settlement attached to this angular fort, the north wall of which runs about 58 paces at fortified skala stands today, as it must always have done, 305°. Part of the west wall stands, but almost all evidence of about 100 m above and behind the fort. the south wall, along the edge of the rock, has gone. The fort

Section XV

SITUATION, IDENTIFICATION, AND HISTORY Kuruca Kale is one of the most impressive on the Pontic

Tpinodic, Tpinddets, Driboli, Tripolli, or Tripoli,! is the °02t (pls. 70a, b and 71a, b). - a

modern Tirebolu. Papamichalopoulos’ notion that the name In 1890 Cuinet stated that at Kur uca Kale “on voyait, ilya is derived from the three cities of Ischopolis, Argyria, and quelques annees, deux statues representant Jean Comnene ll Philokaleia? is hardly likely; the small unwalled town is et sa femme Eudoxie, ainsi que des inscriptions qui ont ete

clustered on and behind three rocky promontories which __‘fansportees a lentree du couvent de Notre-Dame-demore probably gave rise to the name, which first appears in Blacheraine (BAayepatvev).” °” He cites no source for this the classical geographers for what seems to have been a intriguing information and it is disturbing to find that no minor Greek colony. The name and natural features must tr aveler (including Fallmerayer, "' who Was an eager epi also suggest that ancient Tripolis, with its castellum men- graphist) mentions the statues or inscriptions (although tioned by Pliny,? must have stood on or very near the present Hamilton did note “some rudely carved stone over the gate-

site of Tirebolu. Kilise (Monastir)* Burunu, the headland way”),'* and that there appears to be no other reverence to one kilometer west of Tirebolu, is another possibility. Here what seems to have been a socal monastery onthe B aenerna Cuinet noted ‘“‘une caserne nommeée Pikila,> située en mer, et Cuinet, who seems to have ad first- and Knowledge © au fond de laquelle on remarque un mur bati dans l’eau. Ona Tirebolu, cannot simply have invented the information, and vainement cherchée a comprendre quelle pouvait étre, dans it is worth assuming, for the moment, that it IS Correct. The Pantiquite, la destination de ce mur.”’ © Selina Ballance thinks Grand Komnenos John I (1280-97) married Eudokia, that the rock-cut oil or wine press on Kilise Burunu and daughter of Michael VIII Palaiologos, in 1282; she died mn several large blocks of stone used in the walls of a local farm 1301. The alliance was of considerable importance, for it was and in the church which stood there ‘“‘seem to indicate the the first between Trebizond and Constantinople and John II

classical site of Tripolis.”” was obliged to abandon certain imperial pretensions to The eastern and western promontories of Tripolis were obtain it.'* He and Eudokia were probably depicted in wall fortified. The westerly castle is, with reason, called Curiik paintings 1n the now destroyed church of St. oreeory ° Kale (‘Ruined Castle”) (pl. 69a, b). The regular walls of this Nyssa, Trebizond,"* presumably in a style already estab enclosure stand an average of 2 meters or eight courses high lished for Trapezuntine imperial portraiture by Manuel i" and are faced with dark granite blocks. There is a southern Imperial statues are, however, another matter. Ihe only gateway. There seems no reason to dispute Cuinet’s opinion thirteenth-century (and last) Byzantine emperor known to that the work is Ottoman.® The easterly ““Kastro”’ was called have been portrayed in a statue was Michael VIII Kuruca Kale (“Dry Castle”), and in Cuinet’s day Greeks Palaiologos himself, who was shown at the feet of the called the ruins facing it on the mainland ‘“‘Anghelia.”° Archangel Michael, presenting the City of Constantinople. But, according to Pachymeres, this statue was of bronze;'® 1. Arrian, 24; Anonymous periplus, 36; Libadenos, Periegesis, ed.

Paranikas, 34, 36, 79; ed. Lampsides, 71, 84; Kretschmer, Portolane, re

648; Delatte, Portulans, 1, 237; IH], 34; Miller, 7R, col. 647; Mikrasiatic Studies in Athens contain information on all place-

XVIII, 823. available to us. 2. Papamichalopoulos (1901), 257. Cuinet, Turquie d’Asie, 1, 52, 10. Cuinet, Turquie d’Asie, 1, 52-53. |

Triantaphyllides, Phygades, 33-34. In general, see Ritter, Erdkunde, names in Tripolis, obtained from refugees; they have not been ascribes the same notion to Strabo, who does not, however, appear 11. Fallmerayer (Fragmente, 1840), 154—56.

to3. Pliny, refer to Tripolis. 12. Hamilton (1836), I, 257. Natural History, VI, tv, 11: Tripolis castellum et fluvius. 13. Cf. Miller, Trebizond, 28-29, 32-33, 45. 4. So in the first feuille of the engraved route of Hell (1846), 14. Finlay (MS, 1850), fols. 41-42; Ioannides, Historia, 236; and

attached in the album to his work. But there is evidence for a p. 226. monastery there: it was perhaps a mistake for Kilise Burunu. 15. Finlay (MS, 1850), fols. 38-39; cf. Talbot Rice, Haghia

5. Mexia, in Sakkas, Tripolis, 20. Sophia, 1.

6. Cuinet, Turquie d’Asie, I, 55. 16. Pachymeres, Bonn ed., II, 234 (referring to its damage in the 7. Ballance, Bryer, and Winfield, AP, 28 (1966), 256. earthquake of 1296). It was again noted in 1420: G. Gerola, “Le

8. Cuinet, Turquie d’Asie I, 53. vedute di Costantinopoli di Cristoforo Buondelmonti,” SBN, 3

9. Cuinet, Turquie d’Asie, 1, 53. The files of the Center for (1931), 275 f.

TRIPOLIS AND THE PHILABONITES RIVER 139 while that in Tripolis was presumably stone. In the important Ottoman stronghold, complete with a “kale Monastery of the Blachernai at Arta are two tombs of mem- cammi imami,”’?* but, unless the western Curuk Kale is bers of the family of Michael II Angelos, one of whom, John, intended, it is curious that it was the church rather than the was held as a hostage by Michael VIII Palaiologos in 1263.'’ mosque which survived longest. A possible solution would These tombs are decorated with figures in imperial costume be that the church was turned into a mosque, but how would which are in low relief, not statues. If the statues of John and Greeks still be allowed access to the Ottoman castle? The

Eudokia existed in Tirebolu, they were therefore very re- problem does not affect, however, our tentative identifimarkable, although this sort of sculptural representation was cation, but that depends solely upon the slender thread of not without contemporary precedent in Constantinople and Cuinet’s unsubstantiated statement and cannot be regarded Arta and all three examples are curiously linked by the as any more than hypothesis. person of Michael VIII. One wonders, however, if the Tripolis was certainly a personal possession of the Grand Tripolitan ones were true statues and not simply Hamilton’s Komnenoi, who regarded its castle as a safe refuge. One may ‘‘rudely carved stones.” That carving in low relief was known speculate that John II retired there during the brief usurpain thirteenth-century Trebizond is shown by the citadel re- tion of Theodora in 1285. During the civil wars the Grand liefs and the decoration of Hagia Sophia, behind which lies Komnenos Alexios III retired there from his more insecure both a Seljuk and Caucasian tradition in northeastern capital in 1351 and four years later left his empress there for Anatolia. But equally the example of her father’s statue safety.?* In 1404 Clavijo described it as ‘‘a large town’ and might have been in Eudokia’s mind. Here may also lie the the first (westerly) one in the jurisdiction of the Grand clue to the Monastery of the Blachernai. Another Byzantine Komnenoi.’* Clavijo did not land and is probably mistaken bride of aGrand Komnenos, Theodora Kantakouzene, wife in both respects: Tripolis was never very large and the most of Alexios IV, had in mind, it has been suggested, the nameof — westerly possession of the Grand Komnenoi would then the famous Constantinopolitan monastery of the Panto- have been Kerasous. krator when she rededicated its namesake in Trebizond. '® Of the medieval churches of Tripolis the Panagia on Kilise The cult of the Panagia Blachernitissa was even more wide- Burunu was destroyed before 1967; that in the eastern castle spread and, with the very special imperial connections with was only a memory before 1962;° and a third, St. John, has the Blachernai in Constantinople where Eudokia had lived, left no trace. It stood above the town and is said to have been what more appropriate dedication for a Pontic monastery turned into a mosque in 1875.7° In the neighborhood was the

founded by her?! dervish rekke of Sar Halifa.?’ apparently the only temple of

Where were the Pontic Blachernai? The supposition that its kind in the Pontos outside Trebizond. the “‘statues’” and associated inscriptions in the castle of Tirebolu stands 7 km west of the mouth of the Harsit Dere, Tirebolu indicate that John IT and Eudokia built, or rebuilt, which runs 76 km southeast to Torul. Pliny names it the the castle is a natural one. Within it stood a now destroyed ‘*Tripolis,” 78 and in the nineteenth century it was known as church which later Greek sources name as the Theotokos or the Tirebolu Su;?? it is in fact the lower reaches of the Kanis, Panagia.’° Its portico, elaborate rope-work reliefs, window which bisects Chaldia. As will be seen, however, it was almost moldings, distinguished masonry and high-drummed dome, certainly known as the Philabonites in the later Middle Ages. as Shown in Laurens’ lithograph (pl. 73), together with the Twenty stadia east of Tripolis, according to Arrian, stood fact that it was embellished with wall paintings, are all fea- ta ‘Apyvpta,°° apparently corresponding to modern tures of late thirteenth-century Trapezuntine imperial foun- Halkavala*' just east of the Harsit mouth, a silver mine

dations. The substantial ruins around the church, in which continued to be worked until about 1800.°? It may Laurens’ drawing, further suggest monastic buildings. Was even correspond with the evidently Pontic Argyria of the this the church of the Panagia Blacher[n](ai)on? There may Iliad.°° After Tzanicha,>+ the deposits of Argyria would have been a precedent for another monastery within an have been the only major local source of silver available to imperial castle in the case of St. Phokas at Kordyle; there was the Grand Komnenoi. That they may have used the mines certainly a precedent in the Chrysokephalos at Trebizond. The church itself was apparently still in use as an object of special pilgrimage by local Greeks in the seventeenth cen- 22. Gokbilgin, BTTK, 26 (1962), 334. tury; they abandoned it only in the nineteenth century.*' It is 23. Panaretos, ed. Lampsides, 63, 71. frankly difficult to reconcile this information with the fact 24. Clavijo (1408), 109. ; oe 25. Winfield and Wainwright, AnatSt, 12 (1962), 132-33. that by the early sixteenth century “Driboli” castle was an 26. Sakkas, Tripolis, 17; Schulze, Kleinasien, II, 184-85. 27. Cuinet, Turquie d’Asie, 1,55; F. Babinger, s.v. ““Tirebolu,” EZ.

17. D.M. Nicol, The Despotate of Epiros (Oxford, 1957), 198. 28. Pliny, Natural History, V1, tv, 11.

18. Laurent, AP, 18 (1953), 246-47. 29. Hamilton (1836), I, 255; Hell (1846), calls it the “‘Harkavala 19. See J. B. Papadopoulos, Les palais et les églises des Blachernes Dere” in the route map attached to the album to his work.

(Salonike, 1928), 107-21: and F. Diremtekin, “*Mintika (Blachernae) 30. Arran, 24. Surlar, Saraylar ve Kiliseler,” Fatih ve Istanbul (Istanbul, 1953), 31. Cf. Ioannides, Historia, 220.

193-222. 32. Hamilton (1836), I, 258—60, inspected the site.

20. Chrysanthos, AP, 4—5 (1933), 506; Ioannides, Historia, 219. 33. Homer, Iliad, Book II, line 857: TnAoSev €& “AAbBysc, S9ev 21. Chrysanthos, AP, 4—5 (1933), 506, 511; Sakkas, Tripolis, 19. Apyvpov éott yevéedin. The identification, for what it is worth,

Local Greeks resorted there on the feast of the Zoodochos Pege depends upon the relation of ‘AAUBns to Chalybes, the Pontic (which is likely to have been introduced after 1461; after 1870 the fair Chalybians.

was moved to Kilise Burunu). 34. See p. 309.

140 SECTION XV there is suggested by their concern for the defense of the But the facts are that the Philabonites was not used as a

Philabonites-Harsit valley. route by medieval travelers and traders and that, far from Argyria is dominated and protected by the great fortress of being a bandon, Tripolis was virtually isolated from its sup-

Petra Kale or Bedrama, 5 km to the south and on the east posedly natural hinterland. Indeed, imperial control and side of the Philabonites.**° This is evidently the Hétpmpa of | Greek settlement probably did not penetrate much further Panaretos*° and of the Acts of Vazelon. It is first mentioned inland than Petroma. Modern place names and Kiepert’s in 1268 in a deed witnessed by John the Good, Imperial map of Greek-speaking villages in 1890 are a very fair pointer Master of the Tent, who came from the villages of to the extent of medieval Greek settlement. It was virtually Petroma.*’ Of the three churches mentioned there by absent in 1890 and, apart from one slight indication,** there Bzhshkean there is today no trace,*® but spasmodic silver are no reports of medieval or modern Greek churches in the mining at nearby Israil Maden was still reported in 1967. valley today. South of Petroma no modern place names exist After 1461 the Tripolitanians are said to have escaped to with an obviously Greek background, with the interesting Petroma, where they were eventually starved out.*? The exception of Kavraz, five kilometers south-southeast of story cannot be substantiated, but Bzhshkean and Kinneir Tirebolu, and the Kavraz, a western tributary of the Harsit. have independent reports that a local amazon Derebey held Both are perhaps related to the famous Pontic family of out against the pasha of Trebizond in Petroma for six Gabras, which emerges in the late tenth century; its heyday, months before she surrendered in the first years of the nine- however, came before 1204 and Ttirkmen settlement.*>

teenth century.*° After 1461 Tripolis was administered as a Kale dependent

The whole of the Philabonites valley to the southeast of | upon Kurtin ndahiyesi, which administered the whole coast Petroma might, on geographical grounds, be expected to between Gorele (Koralla) and Giresun (Kerasous) and even, have been a dependency of Tripolis, perhaps as a bandon of —— according to Gékbilgin, as far as Ordu—although he gives which we have no record. But the valley was in fact im- no evidence for the last statement.*® This arrangement may passable to wheeled traffic until very recently, when it has well have arisen from Cepni settlement earlier, which made taken four years to dynamite a road into existence; until 1961 Kurttn capital of a Cepni nahivesi and devoid of Greek place there was no connection between Kiurtiin and the coast apart names. It is clear that the Philabonites and flanking districts from a precipitous pack-animal track which fords and re- of the valley had fallen into the hands of the Cepni by the fords the river. This makes a nonsense of Janssens’ observa- fourteenth century, leaving the Greeks only the coast and the tion that the Harsit 1s ‘“‘sans aucun doute le moyen le plus castles of Petroma and Torul at either end of the valley, with facile pour rejoindre la route de Perse a partir d’un port de la a pocket of settlement and pasture at Simylika (Sumuklu), Mer Noire orientale,’ *' and of his enquiry why Tripolis did which could perhaps best be protected from Torul, although not therefore become more important, “‘et lon n’en est que Alexios tried to do it from Petroma. But because the valley plus perplexe quand on consideére les avantages de la route de was blocked at either end, the Tiirkmens were in turn inHarsit.” *? His conclusion is that Tripolis is not a satisfactory hibited from further expansion until they could spill into port. The anchorage is indeed chancy; there are rocks, and a Cheriana. But it was they, rather than the inadequacies of the late Greek portulan warns of two shoals, although there is harbor at Tripolis, who probably stifled all commerce along shelter for small boats between the promontories of the the valley and ruled out the establishment of any bandon

town.*? there. Such circumstances might also explain the otherwise 35. It was called Bedrama in 1969, the name by which it was exceptional fact that a small inland settlement was given known to Kinneir (1813), 332. To Sakkas (Tripolis, 2), and to COMtrol of a large and important stretch of coast by the Hamilton ([{1836], I, 258), it was known as Petra Kale. Bzhshkean Ottomans: the Fatih would naturally have made good use of (1819) (cited in Lebeau, Bas Empire, XX, 492 note 2, but not ap- existing Muslims in Kurtun. But the victory of the Cepni

parently in ras. pacha 38 29), Proposed the somewhat nahiyesi was the culmination of a long process, which is made

se eymeton, oviowed by eens geen Seif ai clearer in a passage in Panareton Misled by Bzhshkean in this passage, A. A. M.B. identified the

eastern castle of Tripolis with Petroma in AP, 24 (1961), 108. It is In February the Emperor [Alexios III] set out against the no comfort that Janssens, Trébizonde, 120, fell into the same trap. [Cepni] by land and by sea. And about [4 March 1380] he

36. Panaretos, ed. Lampsides, 79: Tétpwpav. divided his army into two sections. He sent some 600 foot37. Vazelon Act 50 of 1268: tod av3evtiKot Kovptovvapiov soldiers away from Petroma and the Emperor himself took

"Ioavvov tod KaAOd Tov Gx0 ywpac Nétpwpac. On the kourti-

narios, see the Pseudo-Kodinos, Traité des Offices, ed. J.

Verpeaux (Paris, 1966), 180-81. 44. H. Kiepert, Map; cf. the settlement maps in Bryer, Neo38. Uspenski, Ocherki, 89. Hellenika, | (1970), 53-54; and in Bryer, Isaac, and Winfield, A P, 39. F. Babinger, s.v. ‘‘Tirebolu,’”’ E/. There were, of course, 32 (1972), 254-55. P. I. Melanophrydes, To Kuiovptotvv, PPh, 2

ballads about its heroic defense against the Infidel: see (12) (1937), 6-7. In 1967 A.A. M.B. walked from Tirebolu to Triantaphyllides, Phygades, 86. Gokbilgin, BTTK, 26 (1962), 335, Torul in three days, finding no sign of Greek settlement along the

states that Bedirme Kalesi became a castle-holding after 1461, entire Harsit. There is apparently no church at the significantly with a population of thirteen. It was known as Kale-i Bedirme. named K Oseliakgakilise. 40. Bzhshkean (1819), cited in Lebeau, Bas Empire, XX, 492 45. There is a Gavra just south of Vice, between Ardesen and

note 2; Kinneir (1813), 332. Arhavi in Lazia, and a Gavraz (now Kizilkavraz), 15 km east of

41. Janssens, Trébizonde, 21. Sivas: see Bryer, Gabrades, 187. 42. Janssens, Trébizonde, 21. 46. Gokbilgin, BTTK, 26 (1962), 329-30, 334.

43. Delatte, Portulans, 1, 237; cf. Cuinet, Turquie d’Asie, I, 54. 47. Panaretos, ed. Lampsides, 79.

TRIPOLIS AND THE PHILABONITES RIVER 141 command of the cavalry and another very large party of foot, We therefore propose that Sthlabopiastes lay on or close to crossed the country of those who live along the upper course of the site of Vakfikebir. Further, we propose below that a the DiraPwvitns River up to their winter camps,** and he — sgmewhat similar campaign conducted by the Grand Komdestroyed their tents and he slaughtered and he burnt and he set nenos John IV against the sheikh of Erdebil some eighty free many captives of ours, liberating ZIWDALKG and he years later, also came to grief at the same spot at the mouth of turned back and halted for a short while at L0AaBoniaotys. the Fol.°!

.; ; 49

The 600 who had set out from Petroma made a raid into ¢ rol. KotCavta and massacred and ravaged and burned; whenever

they came to grips with the Turks who were pursuing them, as MONUMENTS they fought their way down to the coast, many of the Turks fell. 1. The Church of the Panagia on Kilise Burunu

The Romaioi expected to meet the Emperor, but reached the The medieval church of the Panagia on Kilise Burunu was

shore first, fighting hard and slaying as they went. When they ;

got to the shore of Sthlabopiastes and did not find the Emperor arenes the nineteenth century and has been published there, as they had arranged, they were a little inclined to let elsewhere. themselves be pushed back and some 42 Romaioi fell. The 2. The Eastern Castle of Tripolis (pls 70a, b, 71a, 72a, b).

Turks and Turkish women who died numbered over 100. The eastern promontory on which the castle is built forms a rocky outcrop joined to the mainland by a low isthmus— It is the reference to local Gepni and to Petroma which now overrun by the new coastal road. The promontory has a

locates the campaign: hence, as has been pointed out more maximum height of 25 m. On the landward side of the than once,°° the Philabonites must be the Harsit. Alexios III ‘sthmus. outer walls seem to have extended over a conseems to have marched first along the upper reaches of the siderable area. But by the late 1950s the remnants of walling Philabonites (though not south of Torul), where the Cepni here and there between the houses were too sparse to allow of had their winter encampments, but they have obliterated a coordinated plan. The promontory provided a naturally other names mentioned in the passage. Kotzauta, perhapsa — defensible site, and with the building of the inner walls Tiirkmen name, appears to have been in the P hilabonites around it the castle must have become one of the most secure area, south of Petroma. Coban Kale, marked on the British of imperial fortresses Army map of 1901 on the west side of the Harsit, about The entrance into the inner castle is on the south side (fig. 12km from the sea, looks superficially like a castle, but closer 27, pl. 72b) and is now reached by a modern stairway which inspection reveals that it is no more than a rocky eminence, may be partly built over the site of the original approach—to devoid of artificial defenses. The nearby Koz Koy may con- the east of the stairway are signs of rock-cut steps. The ceivably mask the name Kotzauta, but apart from Petroma —youssoirs of the outer face of the entry arch are perhaps itself, Suma Kale, which lies on the Erikbelt Dere, a tributary nineteenth century and may replace the original ones which of the Harsit, 4km northeast of Ktrtun, is the sole castle that bore the imperial effigies discussed above. The voussoirs on we have found in the Philabonites area. For want of any the inner side of the arch appear to be medieval, and, as is other candidate, we tentatively propose the relatively wide common in Late Byzantine and Trapezuntine arching, have stretch of cultivable valleys around Kurtin and Suma Kale no central keystone. The doorway is about 1.75 m high. It

as Kotzauta. . . oo would be well protected by flanking fire from the projections

The pastures and now ruined village of Simuklu, attested on either side; there was also an outwork the form of which as Christian in 1717, 1733, and 1779, 9 km southwest of is not now clear. Kirtun and 44 km northeast of Sebinkarahisar, mark the At intervals along the walls are rounded and pointed

extreme limits and goal of the expedition and of projections which are now solid and would appear to have Trapezuntine settlement surviving in the Turkmen interior. always been so. There are no signs of loopholes. It appears Equally important is the whereabouts of Sthlabopiastes, the that defenders fired from the catwalk, using the projections start and finish of the Emperor s party and where the a. platforms for flanking fire. The projection west of the door P etroma troop was to meet him. The hame suggests that is exceptional because the upper masonry courses appear to escaping slaves were caught there; clearly it was on the sea, have been rebuilt and there is a trace of a round-arched certainly east of Petroma, and (from the abundant place window (pl. 71a) names further east) somewhere west of, say, the Holy Cape. At the west end is an additional fortification at a lower Given this long stretch of coast, one must look for a route level (pl. 72a). The walls here are 1.25 to 1.50 m thick and inland, other than from Petroma, which would have been there are the remains of four embrasures with narrow interpracticable for foot and cavalry and gives access to the upper nal and wide external openings. The shape of these embrareaches of the Philabonites. There is only one obvious one. It sures indicates that they are cannon ports. A Russian leads from the mouth of the Fol Dere at the Buy uk Liman of wheeled cannon of 1894 is still in place. There are similar modern Vakfikebir (whose medieval name is otherwise un- additions at Kordyle (Akcakale) known) up to Tonya, thence to Erikbeli and down to Kurtin At the eastern end of the promontory the rock has the in the heart of the Gepni lands of the Harsit—our Kotzauta. appearance of having been cut to a level seating for walls, but - 48. Xseipadtiac: the sheltered lower settlements of pastoralists; the shaping may be fortuitous and natural. Certainly there is cf. Bryer, DOP, 29 (1975), 139.

49. The identification of this place name, and its history, 1s

discussed in Bryer, DOP, 29 (1975), 197 and note 138. 51. See p. 153. 50. E.g., Chrysanthos, AP, 4—5 (1933), 32; Sakkas, Tripolis, 2. 52. Ballance, Bryer, and Winfield, AP, 28 (1966), 254-56.

142 SECTION XV no trace of Hellenistic or Roman masonry and of Pliny’s modernes. G. et H., petites niches garnies d’un rebord sculpté;

castle. I, coupole portee sur quatre pendentifs qui s’appuient sur

ght up . .

The wall surfaces are constructed of roughly-shaped quatre voutes, J, K, L, M, garnie intérieurement d’une espéce

blocks set in regular courses and of simple random-coursed de triple archivolte N... .O, porte laterale, sans doute celle par masonry. Small stones were used to fill the gaps between the laquelle arrivaient les seigneurs du lieu; P, porte ou fenétre blocks and the whole exterior surface has been brought up to ruinee, Q, grande porte d'entree, au dessus de laqu elle se

; Lo. ; trouvent trois fenétres pareille a celles de l’abside et disposées

a smooth surface with heavy lime pointing. The thickness of [like the three windows shown on the south wall in pl. 73]. R, the walls varies from ca.0.75 m to 1.25 m; they now stand 8 m petit portique supporté par quatre colonnes a chapiteau byat the highest. The stonework is a mixture of local basalt and zantin orné. Voute a aréte. Tout l’édifice est construit en petites

waterworn stones, presumably from the bed of the pierres de taille. Exterieurement une ornamentation 4a l'aide de Philabonites, and of yellow and red blocks of Oinaion plusieurs petites arceaux.5* limestone.

The interior of the castle exhibits no remains now above The plan and description tally well enough with Laurens’ ground level. The site of the chapel was pointed out to lithograph in plate 73, with the serious exception that Hell D.C. W. by the keeper of the lighthouse which now stands specifically indicates only three windows (the central one

within the castle. above the other two), of the same shape and arranged in the , same way as Hell indicates in a sketch (not shown here)°° of 3. The Church of the Panagia in the Eastern Castle of three windows above the portico. It is probable that Hell’s

Tripolis account is to be preferred to Laurens’ artistic licence.

Our knowledge of the appearance of this church is derived In plan, dimensions, and proportions the church of the almost entirely from the notes, drawings, and measurements Panagia at Tripolis, is strikingly similar to those of St. Philip of Hommaire de Hell, made in 1846 (from which fig. 28 Isa at Trebizond, and of St. Michael at Platana.°® All three are symmetrical and geometrical reconstruction by Mr. Richard basically rectangular in plan, with an abrupt narthex, highAnderson), and from a lithograph by Hell's traveling com- drummed dome over an almost square naos with vestigial panion, Jules Laurens, reproduced in plate. 73. They may be crossing, and a single apse lit by three windows. But in St. regarded as independent SOUrCES.” Hell S notes should be Philip and St. Michael the western part of the structure has read in conjunction with figure. 28. (His original system of been remodeled. The portico of the Panagia may be a clue to

lettering has been altered for greater clarity.) what originally stood at the west end of the other two churches. For its period it is unique in the Pontos, but in plan is very A, B, C, trois fenetres en plein ceintre, ayant 1.52 m. de like the belfries which were built outside so many nineteenthhauteur et 0.45 m de la recurs @ entree de labside 4 gauche - century churches,°’ although it stood no higher than the

se trouve un mur vertical D, qui s’eleve a 1.65 m. de hauteur; 11 .

est couvert intérieurement de restes de peintures. Cependant il door. It was, perhaps, a simpler example of the great porticos est postérieur a la construction de l’éedifice, car la ou il est en which stand outside St. Eugenios, the Chrysokephalos, and contact avec la muraille de l’abside, il recouvre des peintures: the Hagia Sophia at Trebizond, and which are such a feature peut-étre le prétre s*habillait-il derriére ce mur: on remarque of the architecture of the city. The semicircular niches, G and encore deux tables E et F; je ne sais si elles sont anciennes ou H, are unique in their positions in the Pontos, but liturgical

53. Hell (1846), IV, 394 and pl. xx, figs. 9-11; and plate in the 57. For examples, see Ballance, Bryer, and Winfield, AP, 28

album attached. (1966), 238-41, 282, pl. 4, figs. 2,9; Bryer, AP, 29 (1968), 122, pls. 53,

54. Hell (1846), IV, 394. 56, 58, figs. 12, 13: Bryer, Isaac, and Winfield, AP, 32 (1973), 146, 55. Hell (1846), IV, pl. xx, fig. 11. 261-62, fig. 34. So close in plan is the portico of the Panagia, 56. See Ballance, AnatSt, 10 (1960), 159-61, 164-67, figs. 13, 14, Tripolis, to nineteenth-century belfries, that it must be asked whether

17, 18. it is not in fact a nineteenth-century addition. Greek churches were

2 ga _

THE COMPARISON OF MEASUREMENTS AND DIMENSIONS IN METERS IS:

ges e./§BVitessliesoMessiscealsc& S\zeci\2 25 Be.2, ae: oP 25mIS

a2L2§ias= 5 § § 224 s2 5/8 SFO 6 FES SB LEQ ES A SHMICESIZEAIZEBSILAVIABE

St. Philip, 2.95 2.80 3.80 4.57 2.05 4.15 12.20 12

Trebizond

St. Michael, 2.55 3.05 4.12 4.85 4.85 Platana

Tripolis over 8

Panagia, 2.39 2.63 2.63 4.10 4.10 much 12

TRIPOLIS AND THE PHILABONITES RIVER 143 cupboards are common enough in other Pontic churches. might be narrowed down if our suggestion that the castle and Wall D, which so puzzled Hell, may have been no more than chapel of Tripolis are associated with John II (1280—97) and a later solid iconostasis, or part of it. The ropework decora- his wife Eudokia (1280-1301) is correct.

ya to , Brow Pwork 4. Petroma (pls. b, 75), ; comparable similar in St. on Philip, the74a, Chrysoke; ' ; ; This castle stands a summit about 400 m above sea level phalos, St. Eugenios (allofinwhich Trebizond), St. Michael, Platana, the east side falls almost sheer into the Philabonites. and on Cape Jason. But it is nowhere so elaborate; perhaps . ; tion, shown by Laurens in the foreground in plate 73 is

. oa depicted ; It can beitseen the mouth of the river and glimpsed Laurens has withfrom toofrom considerable enthusiasm. ; the , . ; . between the mountains Halkavala. On a misty .day Exceptionally adventurous low and high relief sculpture is a

oy:Philip, ; ; valley can be overlooked from the and castle from the sea to feature, of St. Trebizond, St. Michael, Platana, , both Loot about 4 km south of Petroma; it commands coastal the ;Panagia, Tripolis. St. Philip has a dog-tooth archoversee and ; the , ; and the valley routes and would effectively any threat what is; apparently a single-headed eagle. St. Michael’s ; ,;suggested ; . , , to the mines at Argyria, which we have may have exterior is decorated with recedinginblind arcades. Hell men4 , 59 .tions erassomething rere been exploited the later Middle Ages.similar inon thewhich Panagia, “exterieurement ; , ; 1: above are ; ; 9a. l’aide The rock the castle stands risesune strikingly ornamentation de plusieurs petits arceaux”’ (which . , va: steep woods. The only approach is from the southeast, leadare difficult to make out in Laurens’ lithograph), and further

; ae io . ingwithin. to stepsWith at thethis onlybackground, entrance (fig. pls. 74a, bof and relief work the29; “‘statues”’ ; i,75). ; So, The steps are largely built of squared stones, showing signs of John I] and Eudokia but reported by Cuinet in thisalso castle cut become , wear, in places are out of the rock. At the top the more plausible. entrance isSt. covered by a projecting bastion on north. This St., Philip, Trebizond, Michael, Platana, and the enthe ¥& Pro) 8 ar , and the adjoining wall have traces of a complete mortar Panagia, Tripolis, must be regarded as a group. Millet dates ; facing over the masonry. To the (west) of the St. ae Philip to the thirteenth century andleft Talbot Rice St.entrance Michael to the thirteenth-fourteenth century.°® These dates runs a vaulted tunnel, now about 1.75 m high, 2.25 m long,

y and 0.75 m wide. This has a substantial mortar fill in the

——_—_____— comparatively well-faced and well-laid masonry. From the allowed bells (and hence belfries) with the reforms of 1839 and 1856. wall running beside the tunnel, which has an internal drain, a

But when Laurens drew the Panagia in 1846, the portico was very shaft overlooks the entrance (pl. 74a). ruinous and for some years after 1839 local Muslim authorities were To the right of the entrance a recess in the bastion ap-

bitterly opposed to the building of belfries and porticos, especially in ;

conspicuous positions, such as that enjoyed by the Panagia. In a parently marks the opening to another tunnel. There IS a dispatch dated 16 April 1841, H. Suter, British Vice-Consul at blocked shaft or window on the north side of the bastion. Trebizond, describes the fate of a portico or belfry of a newly rebuilt The interior of the castle is no more than 25 m long, with

church in the city of Trebizond on Easter Day of that year: rocky outcrops and a slope down to the northern end. The

Attached to the main building was an external portico, supported des are |; lv sheerandd all Nf h:

by two small stone columns. As the Firman did not specify this, a sides are arge y sheer and Go not generally Ca or more than resolution was passed that it should not be allowed to exist. Its 4 low retaining wall, but there is a square bastion to the west demolition was therefore summarily decreed, and forthwith ac- and a semicircular one to the southeast. The average width of complished by the assembled parties amid loud and rejoicing cries.” the walls is 0.90 to 1.00 m. They are faced with comparatively

Panagia, Tripolis, is not therefore nineteenth-century work.

(PRO FO 195/173). There seems little doubt that the portico of the well-coursed granite blocks, whose average size is 58. Millet, BCH, 19 (1895), 454; Talbot Rice, Byzantion, 5 9-30 x 0.20 x 0.20 m. The core consists of a lime-and(1929-30), 66-68. Chrysanthos, AP, 4—5 (1933), 440, followed by pebble mortar with a large number of closely packed flat Ballance, AnatSt, 10 (1960), 161, names, however, Anna, daughter stones and some tile or brick fragments.

of Alexios UT (1349-90) and wife of the Treasurer John There are drainage holes in the northern wall and two

Mourouzes, as foundress of St. Philip (see p. 230). Chrysanthos cites i ; both filled in. Th lat the hich

loannides, Historia, 238, and Miller, Trebizond, 114-15, neither of Wells, OF cisterns, ot now hiled mn. tne we at the highest whom give a further reference. It is difficult to pin down the origins of point of the castle, in a rocky outcrop facing the entrance,

this story, which, unlike a suspiciously parallel explanation for appears to be unlined and is simply rock cut. The well just to the origins of another famous Pontic family and Phanariot dynasty, —_ the northeast of the entrance is lined with well-faced square

the Hypsilantai (supposedly descended from an alliance between stones. About two meters north of it is a large clay vessel,

Eudokia, daughter of the Grand Komnenos Manuel, and a Grand . ;

Domestic Constantine Xiphilinos-Hypsilantes), does not appear to sunk into mortar and embedded into the rock.

derive from where one would expect it: the fertile imagination The castle was thus equipped to withstand sieges. The of Athanasios Komnenos Hypsilantes, Ta peta tyv “AAwoiv, basic masonry is medieval, although some features may date (Constantinople, 1870), 10. It might even have a more respectable ancestry than the Hypsilantes story, for a lament on the loss of St.

Philip in about 1665 mentions a daughter of a Grand Komnenos 59. Hamilton (1836), I, 258: from near Argyria, “the view up the Alexios as its foundress: FHIT, ed. Papadopoulos-Kerameus, 150. river [was] very striking. About ten miles off [sic] I saw on the summit However, the fact that the Lament does not mention a Mourouzes Is of a lofty hill a remarkable-looking rock, which my guide called, ina significant, for the family only became prominent after ca. 1671 strange mixture of languages, Petra Kaleh, or rock-castle, the rooms (Hypsilantes, op. cit., 165; cf. Bryer, Neo-Hellenika, | [1970], 46). No and apartments of which were all said to be cut out of the solid

Mourouzes occurs in medieval Trapezuntine sources, let alone a rock.” A.A.M.B. failed to reach the summit of the castle on 10 ‘Treasurer,’ and Anna, daughter of Alexios III, in fact married September 1967, and again on 8 September 1969, but on the latter Bagrat V (VI) of Georgia. The story must be dismissed as ficti- occasion Dr. John Haldon and Miss Jane Isaac of the University of tious, and has been by S. Ch. Skopoteas, Oi ‘YwtAavtat. “H Birmingham were able to complete the ascent, and we are grateful to TpaneCovvtiakn Kataywyn tous, AP, 20 (1955), 194—99, on dif- them for the photographs and notes for the plan and description

ferent grounds. used here.

144 SECTION XV from the era of the Derebeys or, even, its last Amazon Suma Kale is about one and a half hours’ walk and perdefender. Considering the great technical difficulties of build- haps 400 m above Kiurtun to the north-northeast. It stands ing on the sheer rock, it is of remarkably fine quality. ona pinnacle high on the west slope of the Erikbeli valley and

5. Suma Kale (Kotzauta 2) (pls. 76, 77a, b) is about two hours’ walk south of the watershed pass at Between Torul and the sea, the Philabonites (Harsit) River ere h pinnacle (fig. 30: pl. 76) form tural |

runs for the most part through steep gorges which mark its By P BP ormis @ haturat daner descent of some 950 m. But here and there the valley opens fortress, which must have been impregnable so long as it had out and is flanked by gentler fertile slopes where some tribu- woe and watlr Supp ties, The wees ar nature of the edes of

tary stream flows into the Philabonites. One of the widest of straight edees for the base of th ° alle - Ta), The orm

these reaches is at modern Kirtiin, where the Erikbeli flows Bit €08 ae S (ph fra). ANE CONin from the north and (a little to the east) the Cizre comes in struction is of rough stones laid in random courses and there from the south. Both the Erikbelj and Cizre Rivers are is little to differentiate the rubble core of the walls from their shadowed by tracks running north-south which cross the outer surface. This surface was brought nP toa smooth finish Philabonites near the confluence of the Philabonites and with small stones and a heavy Pointing of lime mortar. Cizre by one of the few old bridges. The track south leads Wooden tie-beam s were used at intervals to strengthen the

through relatively fertile and (in the eighteenth and nine- wal ven here's no trace 0 Stringers. we vee

teenth centuries) well-populated country, past a reported per up ; na:

castle at Hidirilyas, midway between Emrek and the the top. There are no embrasures for firing from; for this a Miiskene Dagi.°° The track northward divides below Suma catwalk near the top of the wall must have been used. The Kale: one routes crosses the watershed over the ridge of the catwalk must have been a we oden structure set up agains! Alacadagi to come down the valley of the Akhisar River to the walls, which are not sufficiently thick to have allowed of a

the sea at a point between Koralla (Gorele Burunu) and masonry one. There was presumably a curtain wall right Besikdiizii (a modern town) which will be proposed as the across the spur, but it now remains only on the northwest

medieval Libiopolis.°' The other route traverses a low pass *! ©, Standing ' ae eee een from the lower

at Erikbeli and follows the Fol Dere past the church at Fol bail by d d ; h ,

Maden®? to the sea at Vakfikebir, which we have proposed as ailey by wooden ladders. It is separated from the main spur

, , . .; by a gap of about seven meters which would have been too

Sthlabopiastes. Suma castle thus guards a point at which two wide for a drawbridee. If there was anv connection it must tracks southward from the coast converge before crossing have been in the na hn of a wooden hawbei dee. supported the Philabonites to continue southward. The valleys around by diagonal struts from below. Such a bridge moat d he N e had

the castle are fertile; it could have been the residence of some ; me rt of it arranved as a drawbrid an be removed in feudal lord of area.However But its position in-any ; P bridge, S.S ; , ; time of the attack. there is and nowrelatively no trace of hospitable site suggest that it was more likely a government nor of anv barbican structure on the main spur. so it is fort controlling and protecting a north-south route, and 5 he h ; Pats , eventually failing, in the fourteenth century, to prevent in- possible that { ere never Was any connection t o the PINNacrs.

b i to the coastal vall ° It must be said, however, that climbing the pinnacle is today

cursions by Gepni into the coastal valley. exceptionally tricky, and the construction and use of the

castle would have required something more convenient than

60. Bryer, Isaac, and Winfield, AP, 32 (1972-73), 234. the present mode of access. On the flat-topped adjoining rock

61. See p. 157. oo! spur are remains of the robbed wails of what was probably an

62. See p. 157. outer bailey. The remains do not add up to a coherent plan.

|KEPAZOYZ . \ Y aey

0 100 200 300 METERS \ Wy —————— —————_—_————_————————— \ \ i:

GIRESUN es 0 rr KEEP AND PART OF A 9 Peeing . | ..XQ~we7| ~mo Po Sa . Ne . ‘CUS An . NF 5 ST, iy \ NO. . — — NL

UPPER CITADEL — NTS g © N\yf>s .

a) Sains ae , |5- Qo . SRY OA oe . a Are eae : on _ a ) ae NDE eo os \ ; é - ™“ - Oar ~ ; SN

; { Jf“ ee . MO | . f S / \ yn O ¥ ; y aN : , |.| < \ , CN oY YX O\ } 5S ‘ ‘g \ , p/Ca — i- i~re o, \ \,ON /3/ fy 7 ee a ,_ ‘ ‘.\* \..\ .\ | fof |aNar

21. Plan

Gedik Kaya Kalesi

near KEPAZOYZ, GIRESUN .. AN VE Gedik Kaya Kalesi S . ~~ > — WA >— Probable Plan of Church fo 7 "7 ~ - pan C7 / astots —\ .

~ S ~ coe |

: N ~ f ' a 7 a oa aa ey “ “ / a “ Le

-, ele ee a\—; _—— eS py cfr ————

A oO ae rR AOU SWF Nh awe , _

“a(jf/ ?\ \4 A. oN . / > 0_ SS 10aS 20SSMETERS KF METERS APPROXIMATE | \

22. Plan of Site 23.

° me Lf = Co SALES 2 Ly Ns, ‘Un CLE e

a

eco 6

tJ

go ur “Remains of Barrel Vault SAS

OG & ne yAWs NicheMooring ML\ Ne Ur gy Niche Fragment of Wall Sor \ pring

— Se Well” » SON ; ‘bleV2 { Ly co Ne rower, CURRIER A Landing Place WNAN EESR f

gin 2 Seep) APHTIAZ NHEOE

Nis Giresun, or Puga, Adasi APPROXIMATE THE ISLAND OF MARS

0 50 100 150 200 METERS

24. Giresun, Map

Se | : ~ : a . ¢ A OXLIOX . “A — | “ 2 \ ; ™“ / : _ \ue ANTQNIOZ ~ee u —— a Ky | ~ BO Y 7, ts, “aN. ANDOZ Oy ye ¢ oS eR & Castle ts Cs ie B: SAR My Ge | }

— / ee r ; SY ! ee OY Roh ad A

—7| i:poo ae o—YT ( (5lasSeSyo “We oo // oe

, ‘|ee!||— 1 KK cf.XX Sh L oo,WS . L\ ' Ss xX. rk Zs,be *ae\

. ae / , _ . A SS, « vv | ~— . / va ye SS. SOR = — fi aN Wa a re : es ] A = S/)} a a“?

6 ge FE 4 \ _ee. \aN . \"+WELL |: _. . Cy ey overgrown- Heavily 4Pm AN( \ re) Doo, —.- . SF WFPssiB E 4 \\

\ oe ? KEXPINA ZE@YPPIOZL AKPA \

2 ee4> | ZEFIR BURUNU LZ - NSLON ENTRANC

Fey a os Pv O ».. 10SN 20| tt30J (APPROXIMATE) 40 50 Meters NI pp 2? -Meters 25. Plan 26. Kalecik-Hisariistii. Plan of Site with Fortifications

ea , °°.‘ TPIMOAISZ, TIREBOLU

Le Eastern Castle

f 5 a7 GES WAN ~

SSS ee nel IN po ay

SSS SSAA) ON YW a ‘Lighthouse FES BRAVES 9,

o St SN ite o urc » an Bihar PS NES, (Mod — Wawa. , Sis J

NI . Filling) / Sy YM yo ff,

DY wan we y APPROXIMATE a 0 50 100 METERS 27. Plan of Site

TPITIOAIZ, TIREBOLU Church in Eastern Castle (after Hommaire de Hell)

SEE Bl ee >

op ee TENS @ C a ar Seo min Oo NO eal Q | IM | WK | pe

—anne NgosUe [D\ “a NO ! | ny fi |

N

0ht 2 3YY 4 5 6 FF 7 8 9——~ 10 METERS 28. Plan

YY ii ; }/

Nay | eee / \ tree | ae / > Nie “Sy~/Possipé OUFER WALL

s aeit¢@ en x ry tn------~- 7 \ 0 SU = WS Ve Yr, an \y oe || ON oN Me \\ \/ ON STEEP SLOPE

cee f Reconennnn nanan A

v=) ~ 4 wc _ SSS © SS LON ed)

{— Jom —a2 WELL ~~FS \ eBKf iy —— \ (gl ie ML —— oS G CE OY; ; as = = prAIn: Mee ue

a 2 Gy immeasene EY , Se : A —

an \ mae BND? AMPHORA SET = Ty Vr te mie a us ‘ _

\ WELL _— Z oor aS —Ne \ Wo BLOQKED Oe Ss I ot \ == — we) x S wiINpows SSA oS ‘eae S SS 2,

oS = WN Se SS | ON \ ~\ \‘al eC a.==S ~~ 9 yy ae?vas yo. ‘\

ES NST Sr a METPQMA, Petra

0 Approximate 10 20 Meters or Bed rema Kale 29. Plan of Site

“a \

Nar mas IN > } me fs n ye a 4

_ Rock Spur ay iN ‘,

(Probable site “yy 7 \ of timber bridge) y, 4 ay SA Pinnacle a),

ey We Vg

SC Ce ae a> ae F AF DR Fo TO

Ef YF

APPROXIMATE

O 10 20 30 40 50 Meters 30. Plan of Site

Section XVI

It is almost certain that Koloneia never lay in Trapezuntine retains the old name of Koloneia,> which naturally still hands. But it lay within the economic, rather than political, designated the area. More curiously, Ibn Bibi and inscripsphere of the Grand Komnenoi, for the area produced alum. tions on the later coins of Eretna retain the Armenian version As in the nineteenth century, the alum was probably mined it: Kogoniya.® By the eighteenth century local Greeks were

by Greeks. A proportion of it was exported to Europe still calling it Koloneia but had also picked up a Turkish through Trapezuntine territory, initially via Trebizond itself, name, Karahisar.’ A further complication came in the nineand then (after the Chaldian routes became dangerous), via teenth century, when the place became, so far as churchKerasous. Hence the inclusion of Koloneia in this Study. men and some antiquarians were concerned, Nikopolis.® Nikopolis itself lies at Piirk, 25 km to the southwest.°

IDENTIFICATION It seems clear, therefore, that Turkish Karahisar, found An inscription originally found in a church in the castle of nl the pourteenth century, 4 a sation On popuiar

Sebinkarahisar must identify the site with KoAdveia beyond alte ied nine .. Beck enh ab le het “kL reasonable doubt. It refers to a John, imperial strator and pence y applied on y to the black castle a ove t e Gree ’ droungarios of Koloneia. The letter forms as we know them rmenian, and Turkish Neen of on oneia Ogontya. a

from published facsimiles suggest a ninth- or tenth-century oe hecah Were ap arena ( wn ae cast °) date.’ An unpublished seal of John, imperial strator and ef Sar he “c sar (orienta io” Are ye emens paraphylax of Koloneia can be assigned to the same period i pan c 10 B: h ne e” of Afyonkarahisar which

and could well refer to the same person.” ues ie e ver or me re century or so, town and castle Mavpoxaotpov emerges as an alternative name for ave been called Sebinkarahisar. Koloneia from the eleventh century. Attaliates calls the DESCRIPTION

theme and t Koloneia, but the castle perched above the .

s own xo 3 an © P great black basalt rock of Koloneia town Mavrokastron.° By; The the twelfth century the Continuator ,; ; La. rises abruptly out of a rich plain, the floor of which is at about 1450 m above sea

of Skylitzes refers, apparently, to both castle and town as ; ;

Mavrokastron.* although Anna Komnene characteristicall level, and is surrounded on three sides by arid, alum-streaked

° 8 y mountains and by an exit to the Lykos valley to the south.!!

1. The inscription, which was removed to a now lost Greek The rock is about 1,000 m long, 500 m wide, and 160m high.

church of the town after 1868, has been independently reported four It stands nearly sheer, “‘like an island alone in the midst of a times: O. Blau, “‘Aphorismen alter und neuer Ortskunde Klein- waving sea of flowers,”’'” except for a very steep slope to the

Asiens,” Mittheilungen aus Justus Perthes’ geographischer Anstalt

... von Dr. A. Petermann, 11 (1865), 252; Triantaphyllides, Pontika, 5. Anna Comnena, Alexiad, ed. Leib, III, 76. 114; Taylor (1866), 294 (with facsimile); X. A. Sideropoulos, Iept 6. Ibn Bibi, trans. Duda, 151-52, 306, 344; J. H. Mordtmann, s.y. thc €v Mixpa@ “Appevia NixondAswc, CPSyil (Parartema), 17 “Karahisar,” EJ.

(1886), 135 (with facsimile); S. Bénay, ‘‘Quelques inscriptions 7, E. C. Colwell, The Four Gospels of Karahisar, \ (Chicago, chrétiennes,” EO, 4 (1900-1), 93-94: P. Le Bas and W. H. 1936), 33-94, gives the most vivid picture of life in 18th-century Waddington, Voyage archéologique en Gréce et en Asie Mineure, Sebinkarahisar, through colophons of venerators of the Karahisar Inscriptions, III (reprinted, Hildesheim, New York, 1972), 431, no. Gospels; see also Bryer, Isaac, and Winfield, AP, 32 (1972-73), 2381814g; Cumonts, SP, IT (1906), 296, 302; the same, “‘Les inscriptions 43. chrétiennes de l’Asie Mineure,”” Mé/Rome, 15 (1895), 286, 294. P. 8. Oikonomides, Pontos, 87-90; Ioannides, Historia, 199; SiderLemerle, “L’histoire des Pauliciens d’Asie Mineure d’aprés les sour- opoulos, loc. cif. (note 1); R. Janin, s.v. “Colonia,” DHGE.

ces grecques,” 7M, 5 (1973), 60 note 24, gives a corrected ver- 9. Cumonts, SP, II (1906), 302-17. sion (save for the last word). An uncorrected version, based on 10. B. Darkot, s.v. ‘““Karahisar,” JA; A. D. Mordtmann, AnatoBenay, Cumont, and the facsimile of Sideropoulos, would read: lien: Skizzen und Reisebriefe aus Kleinasien (1850-1859), ed. + Tij¢|Matpn|xfic obcilac Gvapyle Adys,| bate &) to oo F. Babinger (Hanover, 1925), 437—42. This Koloneia must be disdovo | Imavy Pao[AKo] otp[atop]t [kai] 5pw|vyaptjo KwAovijas. tinguished from Cappadocian Koloneia, which by the time of

2. Dumbarton Oaks, Shaw Coll., no. 73. Nicetas Acominatus (Choniates) had become Ak Saray: #) vov

3. Attaliates, Bonn ed., 125: ... tob Mavpoxdotpov ppovpion, Tagapa Aéyetat (Bonn ed., 689; cf. 72, 541). gig TOV “AppeviaKdv tonwv éxi AdMov KEIWEVOD ByNAOd Kai 11. See the Tiirkive Jeoloji Haritas:, Samsun sheet (Ankara,

SVOKATEPYAOTOD. 1962). 4. Skylitzes, Bonn ed., 679. 12. Taylor (1866), 293.

146 SECTION XVI southwest. This gives access from the town, or vicus, up to the dismissed two Latin inscriptions, reported by Taylor, should

to the castle.'° read:

principal, and what appears to have been the only, entrance be noted.'® The first was a fragment in the citadel, which Koloneia is the most formidable of Pontic natural strong-

holds, but the areas which it controls are hardly worthy of ISTIL

its scale. The vulnerable and low-lying Nikopolis was in fact JPEROI a more important communications center and commanded

more extensive lands. Koloneia is more of a kleisoura. The second, ‘‘a Latin inscription of Pompey” on a granite Hidden in its cul-de-sac, away from the Taulara-Satala high- slab, had disappeared from the side of the outer gateway a way, Koloneia figures in none of the /tineraria. It would have year or two before Taylor’s visit of 1866, and does not seem

been near impregnable in summer and simply unapproach- to be otherwise noted. The name certainly indicates the able in the winter months, when it is snowbound. It is frankly settlement of a Roman colony—Procopus states that it was

a place of refuge rather than of strategic importance. It Pompey who entitled it Koloneia. One might speculate that was no doubt this that attracted the separatist Paulicians the local alum mines were important enough for the Roman to the area in the seventh to the ninth centuries. Koloneia government to declare it a colony and hence rule it directly. could not command the relatively distant Lykos valley, but it The first literary references to Koloneia do not come until could protect the local alum mines and the precarious alum two letters of St. Basil, which also contain the first mentions route north to Kerasous. But this function can hardly have of a bishop there.'? Koloneia was at first a suffragan of been important in Byzantine times, when there 1s no evidence Nikopolis, became an independent archbishopric at about that the alum mines were in fact exploited, and was hardly the same time as its erection into a theme in the ninth century, successful in Trapezuntine times when the route, but not the and finally a metropolis without (except in one list) suf-

mines, was controlled by the Grand Komnenol. fragans from the late tenth century. Its position 1n the lists The present road runs north past the former Armenian declined from 3lst to 68th, but it is mentioned until the village of Tamzara to a gorge about 10 km from fifteenth century. Nine bishops are known from the late Sebinkarahisar. Here, in the nineteenth century alum was fourth century until ca. 1030. In 1390, it was brought, with worked by Greek villagers from Katoymptov (Gedahor) and other neighboring sees which had been more or less abanKoitvovk (Goyntk), principally in the pinkish-colored cliffs doned, under the jurisdiction of Trebizond. By the sevenon the west side of the road where there are traces of the teenth century it was simply a curacy of the Trapezuntine largely open-cast mining. The alum was washed at various monastery of Soumela, but became the metropolitan see of points further along the gorge.'* This is the only area known Nikopolis itself (of which Koloneia had begun life as a to have been worked recently and we see no reason why it suffragan bishopric) in the nineteenth century. ”° should not also represent the classical and Trapezuntine The region was, however, identified with heresy from the

mines. seventh century, when the Georgian Katholikos Kyrion History learned of Nestorianism from a priest of Koloneia.*'

Constantine-Sylvanos, the earliest Armenian Paulician here-

So prominent is the stronghold of Koloneia that it could siarch, founded a Church of ‘‘Macedonia”’ at the castle of

hardly have escaped the attention of the Pontic kings or Kibossa, near Koloneia (cic Kifoooav tO Kdotpov, failed to have been one of Mithridates’ seventy-five treasure- nmAnotov KoAwvetac)*? in the 670s; soon after a Church of

fortresses whose water holes were blocked and walls razed by ‘‘Laodicaea”’ was established at Kuvoc yopav (KoinoPompey. It is a candidate for Sinoria.'° Procopius notes that choritai).2°> Grégoire’s identification of the latter with the it was built ‘“‘by men of ancient time,” '® and its five massive famous Pontic castle of Kainochorion’* is a happy one, rock-cut tunnels and cisterns (together with rock-cut water runnels, collecting pans, and steps) are typical of other 18. Tavlor (1866), 295 Pontic examples examined by Jerphanion.'’ The present 19. St. Basil Letters 195and 227, in PG, 32, cols. 708 and 852-57. walls stand in what is evidently ancient rock-cut prepared 20. Gelzer, Texte, pp. 551 no. 86, 571 no. 101, 585 no. 27, 599 no. seating, but there 1s no certain trace of pre-Roman masonry. 69, 608 no. 78, and (where there is mention of a suffragan) 641 no. 68;

This might confirm that Pompey indeed destroyed the an- George of eae ed. Selzer p. 3 3L, Quien herokls

cient fortress, were it not for the fact that Procopius main- ynekdemos, ec. Monigmann, p. 27 No. Ws.05 Le Suen, Us 1, COM: tains an opposite story: that Pompey actually strengthened 429-32; Laurent, CS, V (1), p. 630 no. 814, and an eighth, unnamed, bishop in Petrus Siculus, Historia Manichaeorum, PG, 140, col.

the place. This is difficult to substantiate, for no certain trace 1281; C. Astruc, W. Conus-Wolska, J. Gouillard, P. Lemerle, D. of Roman masonry survives either, but before the notion is Papachryssanthou, and J. Paramelle, “Les sources grecques pour l’histoire des Pauliciens d’Asie Mineure,” 7M, 4 (1970), 47; M&M,

13. Evliya (1644), II, 206, mentions, however, ‘three strong A&D, 1, 199-201; II, 154; Bryer, Isaac, Winfield, AP, 32 (1972-73),

gates.” 240-42; Lemerle, TM, 5 (1973), 59-60. 14. Cumonts, SP, II (1906), 299; Bryer, Isaac, Winfield, AP, 32 21. N. Garsoian, The Paulician Heresy (The Hague, Paris, 1967),

(1972-73), 244—52; information from members of the Faculty of 145, citing Orbelian. Geology, Ankara University, visiting the site in 1969; the mines are 22. Petrus Siculus, PG, 140, col. 1280; Astruc and others, 7M, 4

marked on the Turkish 1 : 200,000 map. (1970), 43. 15. Magie, Roman Rule, Il, 1224. 23. Petrus Siculus, PG, 140, col. 1297; Astruc and others, TM, 4 16. Procopius, Buildings, II, 1v, 6—9. (1970), 61. 17. Jerphanion, Mé/USJ, 13 (1928), passim. 24. See p. 147.

KOLONEIA 147 but he suggested K oyulhisar or ““Geusuk”’ for Kibossa.*° We kleisoura.°* There is some doubt as to when precisely the

know no place corresponding to ““Geusuk”’ (Gozuk ?), and themes of Koloneia and Chaldia were created out of Koyulhisar is already overendowed with possible classical Armeniakon, following the pattern of smaller tenth-century and Byzantine names.*° Furthermore, Peter of Sicily, our units, but the reign of what may well have been Koloneia’s early tenth-century source, specifically located Kibossa near first doux (or, possibly, tourmarch), the spatharios Kallistos, Koloneia, while Koyulhisar is nearer Nikopolis. The fact can be dated exactly. According to a version of the Life of the that Constantine-Sylvanos was taken for his execution south Forty Martyrs of Amorion, composed in ca. 845-46,

of Koloneia may perhaps be interpreted as meaning that Theophilos appointed the iconophile Kallistos doux of Kibossa did not lie in that direction. At all events Kibossa Koloneia in his will, in an attempt to rid the capital of him.*°

must be sought in the mountains round Koloneia. No Theophilos died in 842. Kallistos was betrayed by his modern place-name seems to correspond with it, so one must Paulician soldiers to the Muslims and died with (but is not

look for a castle. The most appropriate (and only major) numbered among) the Forty Martyrs on 6 March 845. A candidate is Sisorta Kale, the ‘Castle in the Mists,” 33 km strategos of Koloneia is mentioned in 863,°° which is often west-northwest of Sebinkarahisar.2’ The Paulicians evi- taken to be the first indication of the existence of the theme. dently favored remote and inaccessible strongholds such as That Koloneia had a doux in ca. 842—45 is, however, beyond Sisorta, but there are no grounds, other than those of geo- doubt; that it was therefore a theme then seems a very

graphical likelihood, for identifying Sisorta Kale with reasonable assumption.

Kibossa. Constantine Porphyrogenitus describes the comparatively

Constantine-Sylvanos of Macedonia-Kibossa was con- small theme as running from east of Neokaisareia and indemned to death by stoning by the imperial official Symeon cluding 1) noAic ABpaBpaxnvav,?’ Mount PadraKxpoc in ca. 683 and was in fact killed by one of his disciples, (either Fraktin, 70 km south-southwest of Aziziye or, more Justus, south of Koloneia. Symeon apostatized to the Paul- probably, the Karacam Dagi), NikomodAtc (Purk), and icians and, as Symeon-Titos, reestablished the Church of Tegpikn (Divrigi).°? The theme had sixteen unnamed strongMacedonia-Kibossa; he was killed in turn before 687. holds (of which Sisorta Kale was surely one), began as the Koloneia remained a Paulician center until as late as the tenth in the listings and figures in them until its demise in the ninth century and was clearly very diverse ethnically. In ca. eleventh century.°° 853 another Paulician leader, Sergios-Tychikos, was killed The castle was perhaps repaired against the Arabs in the by ‘o.... TCaviwv 6 ano kaotéAAov Tig NixonoAewc.?® The early tenth century; Ioannides noted the date 902/3 above a name Tzanion is a reminder of the Caucasian Tzannoi and of gate,*° and perhaps the founding of the castle church by the later Chaldian Trapezuntine family of Tzanichites of | John, imperial strator and droungarios of Koloneia, can be Tzanicha,’’ and the passage indicates that Nikopolis was ascribed to the same period.

still a fortified castle. But its surviving artificial defenses, In 1057 Koloneia was in the hands of the famous which appear to be almost entirely Justinian’s work, made no Nikephoros Katakalon Kekaumenos, who raised its theme more than a simple castle. The exposed position of Nikopolis levies in support of Isaac Komnenos’ successful rebellion.*'

naturally led to its superseding, administratively and Romanos Diogenes passed it in 1068, and in the following ecclesiastically, by Koloneia during the Arab invasions, from year it was in the hands of the Norman-Italian mercenary

which it never recovered. captain Crispin.** The area was then probably denuded of According to Procopius, Justinian made considerable efforts to strengthen Koloneia and to give the region pros- 34. Oikonomides, Listes, 349 note 345. perity in the process.°° The earlier, blocked, gateway to the 35. V. Vasilievski and P. Nikitin, ““Skazaniia o 42 Amoriiski’kh castle can perhaps be attributed to this period. Justinian and muchenikakh i tserkovnaia sluzhba im,” Mémoires de l'Académie Hierokles refer to Koloneia as an urban center.*' Its relative —/mperiale des Sciences de St.-Petersbourg, 8th Ser., 7 (2) (1906), remoteness probably preserved it from the worst of the Arab 22-39: cf. Garsoian, Paulician Heresy, 126~27, where Theophilos’

, ; ; ; appointment is mistakenly said to have been before 842.

raids, but it was briefly captured by Yazid ben Usaid al- 36. Theophanes Cont., Bonn ed., 181, 201, 283. Sulami in 778°? and its surrounding villages were destroyed 37. On which see p. 169. by Sayf ad-Dawla in 939-40.°* Koloneia became a 38. Constantine Porphyrogenitus, De Thematibus, ed. Pertusi, 64, 73-74, 141-42; Honigmann, Ostgrenze, 52, 56-59, 62, 184: Oikonomides, Listes, 349. Among seals not cited by Pertusi are a

25. H. Grégoire, ““Précisions geographiques et chronologiques number in the Dumbarton Oaks Collections: nos. 58.106.684 and sur les Pauliciens,” BAcBelg, 33 (1947), 297 and note 1, 298. 58.106.688 (kommerkiarioi of Koloneia and Armenia); 55.1.3788 26. Miller, 7R, col. 675; Cumonts, SP, II (1906), 290-94; Anniaca, (the hypatos Leo, krites of Koloneia); 55.1.1558 (Michael, imperial

Danae, or Talaura. spatharios, strategos of Koloneia); 55.1.1577 (Niketas, sekretes, 27. See Tarhan, Map; and the anonymous article in Tiirkive krites of Koloneia); Shaw Coll., no. 71 (Christopher, imperial protoTurizm, 5 (27) (August 1965), 54—55 (our pagination). spatharios, strategos of Koloneia); Shaw, 72 (Andreas ?, krites of the 28. Petrus Siculus, PG, 140, col. 1301; Astruc and others, 7M, 4 hippodrome and of Koloneia); and Shaw, 73 (John, imperial strator, (1970), 65. paraphylax of Koloneia, noted on p. 145). 29. Cf. Bryer, BK, 21—22 (1966), 174-95; 23-24 (1967), 129-36. 39. Gelzer, Themenverfassung, 102.

30. Procopius, Buildings, III, 1v, 6—9. 40. loannides, Historia, 199: a.m. 6410.

31. Justinian, Novel 31 (1); Hierokles, Synekdemos, ed. Honig- 41. Cedrenus, Bonn ed., II, 625; cf. P. Lemerle, “‘Prolegoménes a

mann, p. 37 no. 703.3. une édition critique et commentée des ‘Conseils et Recits’ de 32. Mordtmann, s.v. ““Karahisar,”” E/. Kékauménos,” MAcBelg, 54 (1) (1960), 37-38.

33. Vasiliev, Byzance et les Arabes, II (1), 46, 90, 117, 289—90. 42. Attaliates, Bonn ed., 105; Laurent, Byzance et les Turcs, 65.

148 SECTION XVI Byzantine troops and may well have been disaffected; the polishing, and for pigments in painting. It is locally used existing Armenian element was strengthened by refugees from today for tanning leather. These purposes do not require

the east during the century. Koloneia presumably passed to great quantities of alum. There do not appear to be any the Turks in, or very soon after, 1071. In 1106 it was briefly in specific references to a Byzantine alum mining industry, the hands of Gregory Taronites (alias Gabras ?), doux of perhaps because the demand was so small. This does not Chaldia.+* Gregory was allied to the emir Danismend and it mean that Pontic alum was not still produced, but it was is possible that the area was already Danismendid territory. hardly important. After, or even before, their fall in the 1170s, Koloneia seems The situation was changed from the twelfth century, when to have passed not to the Seljuk conquerors of the the European textile industry began to require alum ona very Danismendids, or to the Mengucekids of nearby Erzincan, large scale, mostly for fulling.°° It came at first almost enbut to the Saltukids of Erzurum; at any rate Mehmet ibn tirely from, or through, Egypt.>' A charter of 1236 is the first Saltuk was able to grant Andronikos Komnenos, the future definite indication of the mining and export of Anatolian emperor, a castle near Koloneia in the years 1173 to 76/77.*¢ alum—1in this case to Cyprus.°? Rubriquis found the trade in

Was this castle either (and both) Kibossa or Sisorta Kale? Genoese hands in Konya, and Vincent of Beauvais noted The freebooting Andronikos would have favored a strong- that it was found near Sivas (Sebasteia)—perhaps meaning hold on the very borders of Saltukid and Byzantine lands, Koloneia. The Zaccaria company obtained its famous alum which ran along the Paryadres, such as Sisorta. In 1201/2 the concession at Phokaia in 1275. Most surprisingly, there is no

Mengiicekids succeeded the Saltukids in Koloneia.*” What evidence that alum had been mined there before.°* The arguments that have been put forward to show that the question to bear in mind 1s how much “‘Phokaian”’ alum was Grand Komnenoi held it for a while thereafter are highly in fact exported through there from Koloneia. The first hint

dubious and may be discounted.*° of a Euxine rival to Phokaia comes in ca. 1275, when a The Mengiicekids were vassals of Konya, but after the prostagma was issued in favour of the Zaccaria company collapse of the Seljuk state, Koloneia passed to the descen- restraining Genoese from importing alum from the Black dants of Eretna and then to the Turkmen confederations. Sea. A Greek alum ship was seized soon after. The embargo

In 1408, after Timur’s invasion, it was in the hands of was lifted in 1304.°* During the period of the embargo, G6zeroglu; ten years later it went to the Karakoyunlu however, there is evidence that the Genoese were shipping Tirkmens, and in 1459 to Uzun Hasan of the Akkoyunlu alum within the Euxine, that it was handled through confederation. Perhaps Uzun Hasan regarded it as part of | Trebizond, and that it originated in Koloneia. In 1289 comes

the “Cappadocian” dowry of his Trapezuntine wife the first mention of the Genoese shipment of “‘allume di Theodora; it has also been suggested that the Karahisar rocca di Colonna’”—a load of 500 cantara. In 1290 the Gospels reached Koloneia as a result of the alliance.*’ export of alum through Trebizond to Caffa is twice Mehmed II took Koyulhisar in 1461, but prudently skirted recorded—one a consignment of 50 cantara.°°

. - . . , Commerce, Il, —71.

Koloneia on his way to Trebizond; he only took the fortress 50. Hevd. C 1 565_71

m 1478 after the battle of Jercan (Derxene), assigning its 51. C. Cahen, “L’alun avant Phocée. Un chapitre d’histoire écon-

alum mines to his treasury. omique islamochreétienne au temps des Croisades,”’ Revue d’Histoire ‘Allume di rocca di Colonna.’ Pontic alum mines, Economique et Sociale, 41 (1963), 433-47.

almost certainly of Koloneia, are first mentioned by ed. R 52.R. ue Marsevics Cyprus meaty o Vinneb ck (0 ieecti ny 49 Bi those rat egni Hierosolymitani, Rohricht (Innsbruc eniponti], as. s ‘yptic at for eurecon's ola en) medicinal , ad gold 1893), 280 no. 1071; and H. E. Mayer, Marseilles Levantehandel und

° ein akkonensisches Falscheratelier des 13. Jahrhunderts (Tubingen, 1971), appendix 10. .

43. Cf. Bryer, ““Gabrades,” 176. 53. Rubriquis (ca. 1240), ed. Dawson, 218: “At Iconium I came 44. O. Jurewicz, Andronikos I. Komnenos (Amsterdam, 1970), 79. across several Frenchmen and a Genoese merchant from Acre, by

45. B. Darkot, s.v. ““Karahisar,”’ JA. name of Nicholas of Santo Siro, who, together with his partner, a 46. The “evidence” consists of: 1. the misreading of the inscrip- Venetian called Boniface of Molendino, has the monopoly of alum tion cited on p. 145 note 1, by Blau, /oc. cit. (note 1 above), as from Turkey, so that the Sultan cannot sell to anyone except these concluding ... 6 do0bAo[c] “Imavv[ov] Bao[iAéac] Tp[aneCodvtoc] two, and they have rendered it so dear that what used to be sold for Ka[i] “Po[paimv]—i.e., the Grand Komnenos John I (1235-38), or fifteen besants is now sold for fifty.” Vincent of Beauvais ts cited John II (1282-97); 2. the statement of the unreliable Evliya (1644), by Mordtmann, s.v. ““Karahisar,” El: ... aluminis minera iuxta II, 205, that Sebinkarahisar “‘fell into the power of the Greek Princes Sabastiam quae valet unam argentariam. We are unable to trace the of Trebisonde”’; and 3. the statement of Dupré (1807), 11, that the reference. Schiltberger (1402), 43, speaks only of the fertile vineyards gate of the castle had a double-headed eagle above it, and appeared of Koloneia. Cahen, art. cit., 440—1, writes: “A vrai dire, ignore s’1l

to have been built by the Genoese. The Genoese are locally and peut se trouver aucune attestation de l’exploitation byzantine [of erroneously credited with building many Pontic castles. The double- alum] a l’€poque macédonienne, ou antérieurement. ... il est peu

headed eagle, far from being a symbol of Trebizond in the 13th vraisemblable que ce soient les Génois des Zaccharia qui l’aient century (when a monocephalous eagle is twice displayed in the Hagia découvert [alum at Phokaia], bien qu’on ne semble pas avoir de Sophia, Trebizond), was a common Seljuk and Tiirkmen symbol. It preuve d’une exploitation anteérieure.”’ This makes one wonder is found, for example, on the Cifte Minare of 1253 in Erzurum. The whether, like those supposed products of modern Hong Kong, eagle may have stood above the present (13th-century ?) gate of which in fact come from mainland China, Phokaian alum was in fact

Sebinkarahisar castle; there is no trace of it today. produced entirely in Phokaia, and whether some Anatolian alum,

47. Colwell, Four Gospels, I, 11. like “Colonna” was not renamed after its Phokaian port of export.

48. Hasan Tahsin Okutan, Sebinkarahisar ve civari ((Sebin- 54. Dolger, Regesten, UI, 65 no. 2016, 66 no. 2020; Heyd, karahisar], 1949), 42-109; Mordtmann, s.v. ““Karahisar,” E/. Commerce, I, 438-39.

49. Pliny, Natural History, XXXV, vu, 184. 55. Balard, Sambuceto, nos. 574, 813.

KOLONEIA 149 The alum came from three Anatolian regions, of which than the Phokaian, just as Evliya was later to eulogize it.° “Colonna” produced “‘‘allume di rocca,”’ the finest quality. Pegolotti also states that its yield was the same as that of “Colonna” ts, as has long been recognized, Koloneia, a name Phokaia: Koloneia produced 14,000 Genoese cantara a still in use in the fourteenth century. But Heyd’s identifi- year—about 700,000 kg or 684 English tons. cation of Koloneia simply with ‘‘Karahisar” has led modern This quantity is substantially more than the total Italian commentators to confuse it with Afyonkarahisar (Akroenos, tonnage out of Trebizond in a year—which can rarely have Nikopolis), southwest of Ankara and many miles from the exceeded 200 tons. The contracts of 1289/90, admittedly

sea.°° when there was an embargo on exporting alum out of the

In the 1290s Koloneian alum was evidently taken east Black Sea, amount to 26.9 tons at the most. That the Italians through Cheriana and Chaldia and then north down to were interested in Kerasous is evidenced by their reprisal raid Trebizond. By the 1340s, after the Euxine embargo had been on the alum skala in 1348.°* But neither Venice nor Genoa lifted and Turkmens had made this route dangerous, it was had an officer there and alum is not mentioned in any of the sent direct to the sea at ‘“‘Chisende,” as Pegolotti’s handbook Trapezuntine-Italian trading treaties. Yet, if Koloneian makes clear: ‘““Allume di rocca di Colonna ene il migliore alum was so fine and was exported through Kerasous in such allume che si lavori, e lavorasi in Turchia dentro al mare, e quantities, it may be expected to be one of the economic fae scala a Chisende di Turchia dentro al mare alla marina mainstays of the Grand Komnenoi, rivaled only by the presso di Trabisonda, e viene 7 giornate infra terra; e fanne il Zaccaria brothers. Clearly it was not. detto luogo per anno in somma secondo dicesi da 14 mila The explanation must be that, while Pegolotti is probably

cantara di genovesi.”’ >’ correct in detailing the quantity and quality of Koloneian

“Chisende,” the ‘scala’? near Trebizond, must be alum mined in the 1340s and in stating that there was an alum Kerasous (‘““Chirisonda”’ and variants on the portulans),°® route to Kerasous, that road must have been only a minor

through which alum was also exported in the nineteenth outlet for it, depending upon Genoese embargoes and century.°”? There are remnants of at least five hans on the old upon Trapezuntine-Turkmen relations in the intervening Koloneia-Kerasous road,°° corresponding to Pegolotti’s mountains—although one may imagine Turkmens being seven days of porterage—a laborious business but evidently somewhat puzzled by their loot if they raided an alum

easier than taking it on the Trebizond highroad. caravan. But other indications suggest that the bulk of The Phokaian alum concession ended in 1458. Four years Anatolian (and Koloneian) alum was handled through the later Pius II, by discovering even better alum at Tolfa, in the emporia of Sivas and Konya before it reached Cyprus and Papal States, entirely altered the commercial situation and the European routes. Here one wonders how much alum was able to finance his ill-fated crusade from it.°’ But from exported from Phokaia in fact came from Koloneia. Finally, the mid-thirteenth until the mid-fifteenth century the bur- it seems likely that, although the mining of alum round geoning European textile industries were heavily dependent Koloneia had existed on a small scale since classical times, upon massive imports of Anatolian alum. Theoretically, large-scale exploiting of it was developed not by Byzantines Koloneia should have made the Grand Komneno’i a fortune, or Trapezuntines, but by the Turks, and was probably stimu-

even though it was not actually in their hands. lated by European as well as local demand. The reason for this is that Pegolotti states more than once

that Koloneian alum was the best in the world, better even MONUMENTS 1. Mavrokastron-Karahisar of Koloneia (pls. 78a, b-83a) We distinguish here: the castle, or the main enceinte of 56. Heyd, Commerce, 1, 438: followed by, e.g., P. Laven, walls; the citadel, or the inner structure near the northwest

Renaissance Italy, 1464—1534 (London, 1966), 63 and map. . 57. Pegolotti, Praticu, ed. Evans, 349 (in one incredible version, corner; and the keep, or the tower which stands above the the sum of ‘14 mila mila cantara” is indicated); cf. also 43, 293, 306; citadel and castle walls.

and Agathangelos Xerouchakes, To gurdpiov tij¢ Bevetiac peta The single enceinte of castle walls surrounds most of the

tg AvatoAns, EN-Et.Kpnt.2n., | (1938), 60. summit of the rock of Koloneia and is close to Evliya’s a Suet (1838), 435-36. Brant (1835), 220. estimate of 3,600 paces in circumference.°* The walls follow 60. The old road runs parallel and to the west of the newroadand ‘Natural features, particularly along two spurs which run is still used weekly by the postal jeep; it started perhaps originally at south and south-southeast. There are occasional bastions at Armutlu, 5 km west of Sebinkarahisar (Barth, /oc. cit.). On this weaker points and the rock 1s often scarped. The walls have road there are two hans at Asarcik. one at Kanlihan. one between heen repaired and rebuilt many times and there are numerGucese and Kirttn, and another at Tashan, near Mesudiye. In 1866 , ; ; it took 18 hours; only horses and mules could negotiate the 2,100m ous changes of construction—walling with header and Egribel Pass, and the carriage of alum in bulk can never have been stretcher beams, herringbone masonry and random-coursed easy. The new road was built in 1873, and, although it is often blocked, is more commonly used today: Cuinet, Turquie d’Asie,1, 24;

G. Stratil-Sauer, ““Verkehrsgeographische Bemerkungen zur Stadt 62. Evliya (1644), IT, 206. Trapezunt im Ostpontische Gebiet,” AP, 29 (1968), 318-19. Both 63. Panaretos, ed. Lampsides, 68. On the Genoese cantara, see E. roads traverse exceptionally fine yavla landscape, but remain an Schilbach, Byzantinische Metrologie (Munich, 1970), 188. ordeal for the jeeps which attempt the old road and the buses which 64. Evliya (1644), II, 206. Other, briefer, accounts are in the

maintain a precarious service along the new one. Anonymous (1805), III, 296: Morier (1809), 334-36; Anonymous

16—17. 388; and Taylor (1866), 293.

61. J. Delumeau, L’alun de Rome, XV°-XE¢ siécles (Paris, 1962), (1826), 223-30; Fraser (1835), IH, 352-53; Aucher-Eloy (1836), II,

150 SECTION XVI masonry. The ashlar masonry round and near the former gular blocks laid in regular courses, but after the first line of

gate is in striking contrast to the rest of the walls. intermittently laid header beams the masonry degenerates to The former gate is now blocked (pl. 80a, b). It stands about random coursing. midway on the west side of the castle, about 40 meters north On the south side of the castle is a stretch of wall with of the present gate. It is reached by a track which follows the regular ashlar stone coursing. Some parts of the walls, ususcree in such a way that it is largely invisible from the town ally in the middle courses, have herringbone masonry. Two below to the west and even from the walls above. The gate is fragments of ornamental carving were located among some inserted in the northwest side of a wide bastion, and 1s uncoursed stonework. overlooked from either side. There could well have been The towers and bastions are both round and rectangular. further defensive arrangements in the interior of the bastion, A tower on the southeast side has a complete circular wall of but this is now filled with rubble almost to the arch of the masonry, but most of these bastions are open on the inside: gate. The exterior walls round the gate have ashlar masonry: they may have had wooden closure walls, floors, and roofs. It

well-faced nearly square (60 x 60 cm in the lower courses, is likely that walls, bastions, and towers had crenellations decreasing with height) blocks of granite with a little lime- and catwalks, but none are evident now. and-pebble mortar. The interior facing has been largely strip- Passing through the present gate into the castle, one finds ped where it is exposed, revealing a mortared rubble core that the walls follow two spurs to the south between which a containing pulverized brick or earthenware, with substantial depression leads up to the citadel and keep at the northwest use of long tiles. An interior lintel about 2.5 m long has now corner. In the depression are some traces of what could be the fallen away. The arch of the gate is made of two semicircular chapel from which came the inscription of John the strator courses of eight voussoirs each; there are no keystones. The and droungarios. In Cumont’s day the church was already stonework of the area surrounding the gate is reminiscent of ‘‘terriblement dégradée”’ °° and nothing useful can be made that of the northwest wall of the castle of Neokaisareia and in of the presumed remains today. However, cubes of glass striking constrast to the more roughly built round bastions mosaic found in the area, suggest that it was a church of some and rubble walls of the remainder of the enceinte at Koloneia. importance; the fragmentary blocks of ornamental carving The former gate and ashlar masonry may perhaps be as- noted in the uncoursed masonry of the curtain wall may signed to Justinian’s rebuilding. It may also have been the come from it too. gate repaired in 902/3. Above the interior arch there 1s a niche Below this site are five great tunnel cisterns, cut into the which may have contained the inscribed date reported by rock at an angle of 45°. Four are more or less blocked, but the

Ioannides. fifth, on the southeast side, below the citadel, has a wall The former gate has not been systematically blocked, cu- enclosing the entrance. A rock-cut stairway leads to the

riously suggesting that it was still in use when the present gate cistern proper, which may represent the oldest part of the (pl. 82b), on the southwest side of the bastion, was built or castle. The entrance is framed in masonry, of which the lower remodeled. This is a substantially larger construction pre- courses of the sides and the archway itself are of bricks laid in

ceded by traces of rock-cut steps and paving stones. It is thick layers of mortar (pl. 83a). Cumont regarded this work flanked by round towers. Some of the purplish ashlar facing as being “assez récent.”’ °° It is certainly later than the cutting

stones of the gate are reused—D.C. W. suggests that the of the cistern, but the brick is Byzantine, perhaps Justinianic whole facing could be Ottoman. The gate itself is framed with (see Appendix).

astragal molding, some of which survives on the south side The citadel, at the northwest summit, is roughly rectanand is pointed (somewhat like the gate of Alexios II in the gular (pl. 81a). It is approached through an eastern gate with west wall of Trebizond). The relief of a double-headed eagle a flattened segmental arch amid ashlar masonry, whereas the which travelers noted near the keystone in the nineteenth other walls are of rougher work. In the opinion of D.C. W., century has now gone. Above the arch 1s later masonry with the gateway to the citadel and the present gateway to the random coursing of three builds: two successive sets of bat- castle are of the same build, Turkmen or Ottoman. In the tlements and a final blocking of the upper series when the opinion of Taylor, the present gateway to the castle is Seljuk wall was last heightened. These repairs argue for a relatively and the gateway to the citadel Roman. In the opinion of early date for the original version of the present gate; the A.A.M.B., the molding, arch, double-headed eagle, and possibly Seljuk double-headed eagle might point to a reused stones of the present gateway to the castle point to a

thirteenth-century date at the latest. Turkish or Turkmen presence there between the late eleventh The round tower on the south side of the present gate has and the fifteenth century, and the citadel (in its present form) fairly well-cut rectangular stones laid in regular courses. The to an Ottoman date before the mid-seventeenth century. round tower on the north side is perhaps of a different build, The keep, in the northwest wall of the citadel, 1s a massive

having smaller blocks (the largest is about 15 x 25 cm). octagonal tower, about 12 m in diameter, about 27 m high These blocks are unevenly cut and laid in random courses. At and with walls about 1.5 m thick (pl. 79b). Beam-holes vertical intervals of about 2 m, courses were laid with indicate four stories. The walls are faced with rough stone wooden headers at intervals of up to 0.5 m.; it was impossible and mortar, with cut ashlar blocks only round the quoins and to ascertain if wooden stretchers were also employed. pointed windows. D.C. W. believes that its construction may The walling to the north of the north tower is initially of the same build, with headers and stretchers. The lower 65. Cumonts, SP, II (1906), 301. courses of the wall between the two gates are of large rectan- 66. Cumonts, SP, II (1906), 300 and photograph.

KOLONEIA 151 also be related to the citadel and the present castle gateway. have been described elsewhere. The village church at Cumont (inexplicably) regarded it as being the oldest tower Koinouk was of some decorative distinction and was built in in Mavrokastron-Karahisar of Koloneia.°’ A. A. M.B. be- 1841 during the episcopate of Christopher Phytianos of lieves that the stonework, windows, interior staircase, and Chaldia, archbishop of Nikopolis (i.e., Sebinkarahisar). The general appearance of both keep and citadel indicate that, Phytianos family provided, significantly, not only a series of whatever their foundation, they are of largely Ottoman con- metropolitans to Chaldia, but also mining masters to the struction. The keep is described by Evliya (as a heptagon),°® area.’! and there is a simple plan of the citadel by Sideropoulos.°? 5. Kayadibi Clearly, the whole site requires further investigation The monastery of the Theotokos (Meryemana) lies in the

l le si ires fi tigation. i,

2. Birogul (pl. 83b) rock face of Kayadibi, 7 km due east of Sebinkarahisar

A fragment of walling stands in the cemetery of Birogul, castle. It was traditionally founded by St. John 6 2 km east of Sebinkarahisar castle. Too little 1s left to de- ‘Hovyaotns, bishop of Koloneia (481-90), whose feast on termine the nature of the building, but the thickness of the 3 December was celebrated there until recent times. It was walling and a few fragments of ridged tile (see Appendix) also the home of.a wonderworking icon of the Virgin which

3. Duman Kaya . , ; 75 . The conspicuous rock of Duman Kaya stands at the a , ; ;

indicate that it was Byzantine. brought pilgrims every 15 August, many of whom went on to venerate the Four Gospels of Karahisar in Sebinkarahisar

. itself during the period 1575—1848.’* Further evidence for . ; ' ; monasticism inbetween Koloneia is provided by the fact that was a crossing of the Kelkit Purk and Sebinkarahisar. On it; — .itscesummit Koloneian monk that John Tzimiskes chose as Patriarch Taylor found “the remains of of recaptured a cell, the abode . 73 sea Christian Theodore II (970-76) Antioch. many years ago of ascetic, and some small re; 1370 The remains the monastery of the ,Theotokos, servoirs forpresent water, hewnof out of the rock. ; 14 ; which are published elsewhere,’* belong almost entirely to

4. Katochorion (Gedahor) and Koinouk (Goyntk) the nineteenth century. Like so many Pontic monasteries, it The alum mining area, about 10 km _ north of consists of a great cave, halfway up a rock face, in which Sebinkarahisar, was centered on Katochorion, Koinouk, monastic buildings and a small church were accommodated. and their surrounds, but their churches are of the nineteenth century, as are probably those at Turpcu and Licese, and 71. Bryer, Isaac, and Winfield, AP, 32 (1972-73), 226-27, 242-52. 72. ActaSS Maii, Ill, pp. *14—*18; Cumonts, SP, II, 296—97; Colwell, Four Gospels, 1, 12 -13.

67. Cumonts, SP, II (1906), 301 and photograph. 73. Habib Zayat, ““Vie du patriarche melkite d’Antioche Chris-

68. Evliya (1644), II, 206. tophore (976) par le protospathaire Ibrahim b. Yuhanna. Docu-

69. Sideropoulos, CPSyl/l (Parartema), 17 (1886), 135. There is a ment inédit du Xe siécle,” Proche-Orient Chrétien, 2 (1952), 359; Leo

simple plan of the town and surrounds attached to Barth (1868). the Deacon, Bonn ed., 100.

70. Taylor (1866), 292. 74. Bryer and Winfield, 4 P, 30 (1970), 371-75.

Section XVII

DESCRIPTION AND IDENTIFICATIONS the portulans (which he did not employ at all) places a The 59 km coast, from the mouth of the Philabonites near [otapHo¢ tod Aytov Evyeviov here, corresponding with the

Argyria and the borders of the most westerly known modern Cavuslu Dere and coastal settlement.” It is just Trapezuntine bandon near Kordyle, exhibits interesting possible that the name of the St. Eugenios River and settleproblems of shifting settlements and names. Basically, there ment survives a the old Turkish name of Eynesil. are three capes divided by two bays. The westernmost cape, The ancient Libiopolis, in the eastern bay, survived during the Kara Burunu (its classical name is not known; it seems to the Middle Ages; we suggest that it IS represented by modern

have been called the Kara Burunu from the late fifteenth Yuvabolu (rather than Fol), with Its nearby castle. In the century) is the least conspicuous. The eastern cape, the ‘Iepov portions it Is described as Utopoll, Vupolh, Viopolli, "Akpotnp.ov, is substantial and still keeps a memory of the rope’, Vio poh, Livopohl, and AtpLonoAts, probably bename “Holy Cape” in its modern name of Yoros (Yeros) coming Evliya’s Popoli, from where “numerous ruined casBurunu; it shelters the great bay of Trebizond itself. The tles” could be seen.” Our Gesme Onti Kale, near Yuvabolu,

99 4 oS ee .

central cape is called today Gorele Burunu, or (after the appears to be the medieval Libiopolis. oo castle on it) Kale Burunu, or (after local towns) Gorele Kale, Further along the bay, the major settlement of Besikdiizu,

or Eynesil Kale. which has mushroomed from the nineteenth-century Sarli

The settlements which lie between the capes have shifted Bazar, is very recent as is its twin town of the bay, further east considerably, although often keeping ancient names. The 0” the Buyuk Liman, Vakfikebir, which has grown out of the table following shows the situation. In classical times there nineteenth-century Fol. But Vakfikebir may have had a seem to have been settlements at Philokaleia and Koralla in ™edieval precedent, and the mouth of the Fol Dere has the western bay, and at Libiopolis and Kerasous in the already been proposed as the fourteenth-century Sthlaboeastern. Today the main towns are Gorele (Eleve) in the first piastes.” Three kilometers east of Vakfikebir is the Kirazlik, bay and Besikdiizii, Vakfikebir, and Iskefiye in the second. or Kireson (Kerason) Dere. This name may reflect one

There are indications of how the classical pattern evolved of the most the, survivals on the coast: a memory of into the modern one. During Byzantine or Trapezuntine the Kerasous, t ree days west of Trebizond, wer © times Philokaleia became, it seems, Laytos, Laitos, Lairos, Xenophon reviewed his Ten Thousand and round that they Lartos, Laito, Laites, or Lavtos; a late Greek portulan strips veo actually 8,600. Were it not lor the fact t in the ould

the name of its Italian definite article and calls it by what may also appears IN the An ONYINOUS PerIp lus, geograp ers woul

well have been its real medieval Greek name of Aét6c,!— perhaps ignore the evidence of distance and ascribe this

perhaps hinting at the eagles which circle the mountains Old oF eet Net hatit to the Kerasous of oiesun. behind the place. This medieval name may survive in the old But It Is ot aknkebir that its ee oases many Turkish one for modern Gorele: Eleve. Between modern medieval source—appears to survive to ay. 7 Gorele and the ruined settlement on Gorele Burunu (which The name, but not the settlement, of Old h Kerasous surely represents ancient and medieval Koralla), medieval survived; at the next headland east of it Isa wate lower and portulans place Sconigeni, Sco. Vigeni, and a Chavo de San probable settlement, the ruins of which are still visible, but its

Aughen.? The name (apparently of a river, settlement, or name, variously Gallite, Kalita, and Gelida Kale, probably headland) does not appear in other medieval sources, but masks a medieval one which corresponds with no classical Chrysanthos, who relied on local Greek names rather than 3. Chrysanthos, AP, 4—5 (1933), 85 and map. 4. Pliny, Natural History, V1, 1v, 11; Kretschmer, Portolane, 648;

1. Arrian, 24; Anonymous periplus, 36; Miller, 7R, col. 647; Delatte, Portulans, 11, 34; Clavijo (1404), 109 (the identification with Kretschmer, Portolane, 648; Delatte, Portulans, 1, 34; Chrysanthos, Fol here is the translator’s); Evliya (1644), IT, 41; Kinneir (1813), 333 AP, 4—5 (1933), 85; Sakkas, Tripolis, 250; Kinneir (1813), 332; (‘‘celebrated for its delicious wine’’); Hamilton (1836), 251; Ritter,

Hamilton (1836), II, 253-55; Bryer, AP, 24 (1961), 108-9. Erdkunde, XVIII, 817: Bryer, AP, 24 (1961), 109. 2. Arrian, 24; Anonymous periplus, 36; Clavyo (1404), 109; 5. See p. 141. Gokbilgin, BTTK, 26 (1962), 334-35; Evliya (1644), II, 41; Kinneir 6. Xenophon, Anabasis, V, 11, 2; Anonymous periplus, 36; Kinneir (1813), 333; Hamilton (1836), 252; Cuinet, Turquie, 1, 47; Bzhshkean (1813), 333-34; Hamilton (1836), 250-51; Chrysanthos, AP, 4-5

(1819), trans. Andreasyan, 39; Bryer, AP, 24 (1961), 109. (1933), 85.

FROM THE PHILABONITES TO TRIKOMIA 153 place name. Bzhshkean, followed by Ritter, suggested that side of Bostan Burunu. Biiyukliman (“Great Harbor’’), the ‘“Klida Kalesi’ held the metaphorical keys (KAeidEc) to old name for Vakfikebir, is a puzzling one, for there is no sign something.’ We do not like the notion but can think of no that there has ever been a harbor there. The two modern

better explanation. towns of Besikdiizu and Vakfikebir are prosperous market-

Moving further east, Iskefiye (Ckagia) appears to be a ing centers for the rich hinterland of flat quaternary terraces. recent place and name, although it had, in the nineteenth The inland reaches of the Fol Dere run through fertile agricentury, a ruined monastery of the Holy Taxiarchai (the cultural country which supports a third township at Qoavia Archangels), of which we can find no trace but which may (Tonya). Higher up the valley to the south there were iron indicate a medieval occupation of the site.® Finally there is mines and a Byzantine church near Fol. By the modern track the great Holy Cape, the castle of which is not on the cape up to Tonya it is 23 km, and there are two gorges to traverse, proper but near Kiiciik Mersin, perhaps Bordier’s Marcin.” but it seems that the medieval route would have run along the With Kordyle, the final and most easterly site on this heights above the gorges. Tonya itself lies at about 750 m stretch of coast, we enter substantial recorded history. Its above sea level in an irregular bowl where the valley widens name Kopovan is also that of a kind of tunny fish, yet out, forming a broad stretch of gently sloping agricultural another reminder of the ancient importance of the annual land. A castle is reported near Tonya, but neither author has gyration of the tunny round the Euxine.'° Kordyle kept its tried to verify its existence. At the south end of the bowl, classical name into the Middle Ages, when it was sometimes about 4 km above Tonya, the track again runs into a gorge also called after the monastery of St. Phokas which it housed; for 3 to4 km and then opens out through fine canyon scenery

it is the modern Akcakale (Akcekale)."' for a further 8 km up to Maden. Here the valley slopes are The numerous castles and forts along this stretch of coast gentle and there is even a small, flat, sedimentary valley raise the possibility that there may have been a signaling bottom with rich soil. Near the church at Fol Maden there is system—from Kara Burunu to Koralla to Kalita Kale a han anda small hamam, built of random-coursed masonry (Gelida Kale) to the Holy Cape. But the fort on the Holy set in lime-and-pebble mortar. The presence of the han Cape cannot be seen from Kordyle or Trebizond. Signals are strongly suggests the use of this valley as a route across the therefore more likely to have been sent inland at this point, mountains. A kilometer or two above Fol Maden the track via the fort on Hidir Nebi, which is visible from both climbs up through forests to a low pass at about 1,200 m at

Trebizond and Koralla.'? Erikbeli, and from thence it is an easy route down past Suma

Koralla has precipitous slopes in its eastern flanks, making Kale'* to the Philabonites at modern Kurtun and on south (until the building of the modern highway) the passage of the to the Lykos valley in Cheriana. Or there 1s a good track coastal road difficult and tortuous. On its western side the mounting upward from the Erikbeli pass and over the ridges hills rise gently from the sea to form the bounds of the bay of in a southeastern direction to the Zigana pass. From Tonya Koralla, giving considerable stretches of cultivable land in there are also easy routes across the Karadag ridges east of

the hinterland of modern Eynesil and Gorele. There is no Visera and to the Trikomia valley. easy route inland from the bay as a steep range of mountains This configuration must be borne in mind when considerto the south isolates it from the gorges of the Philabonites. ing Chalkokondyles’ account of the invasion of Cuneyd in

This is perhaps the reason why Koralla—although larger the years 1456-58. The Cvync ... AptaBiAnc—sheikh of than most coastal settlements—never attained any great Erdebil—collected forces against Trebizond from the east importance in the history of the Pontos. A castle was reported and south and from Samsun (?) (a0 te Lapiov). John IV to D.C. W. some 10 to 15 km south of modern Gorele in an moved out to meet him, however, not by taking the highway area which has place names such as Kaleboynu and Kaledibi, south from Trebizond but by going west to Kordyle. Here he

and there is a Kilisebeli inland on the Gorele Dere. Such learned that the sheikh was established in a natural strongnames suggest that there may be monuments in the hinter- hold (evidently still further west), the defile of Meliares, land of Koralla bay which neither author has explored. called Kapanion (thv KAgtoovpav ... tod Mediapn, to The next, eastern, bay of Vakfikebir contains only one Karnaviov AEyouevov). John’s troops moved forward (1. e., sheltered harbor, near the village of Yuvabolu on the eastern west or south), presumably in the direction of Kapanion. This would mean that they traveled across or round the Holy

7. Evliya (1644), 41; Hamilton (1836), 249; Ritter, Erdkunde, Cape. John’s pansebastos attempted to rendezvous with XVIII, 815: Bzhshkean (1819), trans. Andreasyan, 39; Bryer, AP, 24 John’s t bv sea. but a high wind prevented him landin

(1961), 109 and fig. 7. OnN'S TOOPs DY sea, DUE a THE P eC 8 8. Triantaphyllides, Phygades, 45; cf. p. 163. marines. The sheikh took the chance of killing the pan-

9. Arrian, 24; Anonymous periplus, 36; Kretschmer, Portolane, sebastos, his son, and about thirty of the now isloated troops.

406: Bordier (1609), 117. , , .

648; Delatte, Portulans, 1, 35; I, 237; Evliya (1644), II, 41; Ritter, The remainder fled back to John (our evidence that the Erdkunde, XVIII, 814-15; Hamilton (1836), 249; Black Sea Pilot, Emperor had not been on the expedition), who abandoned

10. Cf. Strabo, Geography, XII, m1, 19. Kordyle and returned to Trebizond by ship. The sheikh then

11. Anonymous periplus, 36; Miller, 1R, col. 647; Kretschmer, set up camp in the monastery of St. Phokas, Kordyle, while Portolane, 648, Panaretos, ed. Lampsides, 81; Clavijo (1404), 110; | John’s troops made their way back to the safety of the Gokbilgin, B7T7TK, 26 (1962), 315; Kinneir (1813), 334; Hamilton capital, which the sheikh finally invested.

(1836), 248; Delatte, Portulans, 1, 35. Bzhshkean (1819), trans. It is clear from this account that the sheikh had taken a

Andreasyan, 40; Ritter, Erdkunde, XVIII, 814; Chalkokondyles,

12. See p. 164. 13. See p. 144.

Bonn ed., 463—64; Succi, Trebisonda, 291; Bryer, AP, 24 (1961), 109.

5) | = Zz. Ay 3 & | 154 SECTION XVII

wa | ‘=e rr Z, = w q |M ts = | i) | ta bn — om .B) | n= 22 W2 os 11% 0|:6 ie |S 0/5 oO = Os Y if (1) = ae} by a 3 o 8) ae} Oo 8 =a OOOO 'S, wl > OAlM So) 'S > < ie © 4 c S eB) ~ iAals & =DZ & c 2 /R iE 2 .® 2| | oun) “< . or |= | — = = — © (1) mm ZF w 2 23 bey * 2 a] | 8 8 v, a\. |e | ls es |. |& » le 2~ aCF on0/0 asiSI “O|3 Ss 30/3 vo |G x)2-|Gola ©(2Mi|gsis 4sc) os/§ OD noS a| |sis td SI E Th oS 3

LL, &(£mA a =|PFF » 2~ SEL coO=gs) Pa 23 os~18 |~ x an YealO no) >Oole.

< OQ.2o = Pah5| Omaroon S5ea-Sole. S19

= Seg S 2 | ‘a == SS 2 a) rs aS 9 6 3) S| N = = Ss 3 | wt es ‘a 2 fa So of | Q~

ve) ~< AOU S)

O Like all the banda of the Empire, Trikomia became a almost certainly the modern Mayer, which in 1432 seems to nahiye dependent on Trebizond soon after the Ottoman have been settled with refugees from Limnia,'® the latter conquest; its defters give some clue to its earlier situation. In

having passed to the Turkmens half a century earlier. the years 1521 to 1528 its 106 settlements had a total In his only listing of the banda of the Empire, Panaretos Christian population of 4479 households and a Muslim of states that Trikomia was ravaged by the bubonic plague in 413 households. Of these, nine settlements were timar hold1382.2° One may speculate that he may also have the ings, twenty-one were assigned to the imaret of the Yavuz medieval name of the Kalenima Dere itself as the river tov Sultan Selim’s Pontic Greek mother, ‘‘Gulbahar,” in aytov Tpnyopiov.?! On 26 October 1363 Alexios III, who Trebizond, and seventy-four to the reservists of the military establishment in the citadel of Trebizond. ’° It is tempting to

Uspenskij-Beneshevich text reads: ... tov yoviKév (fav) av see in these proportions a reflection of the relative local (SwmBG5@ to THPK...) Ev TO yupio LaAapi tO yovikov pov ...; but military, ecclesiastical, and imperial holdings in the bandon

the microfilm of the Leningrad MS, fol 106¥, in fact reads: ... tiv before 1461; at least they give some idea of the population YOVIKEGY NH(Ov) év ow Bade TO TH PKOLNOLOV év to yopio oadAapl scale of a Trapezuntine bandon.

TO YWVIKOV HOD ... (1.e., “in the bandon of Trikomia.’’) ; ; ; ; ; 9. Laurent, AP, 18 (1953), 261: cf. Panaretos, ed. Lampsides, 69, In microcosm, the history of Trikomia may be epitomized 71.

10. Laurent, AP, 18 (1953), 262, 269. 21. Panaretos, ed. Lampsides, 75.

11. Laurent, AP, 18 (1953), 262, 267, 269. 22. Panaretos, ed. Lampsides, 44. 12. Laurent, AP, 18 (1953), 262. 23. Another candidate for the place of origin of the Hypsilantes 13. Laurent, AP, 18 (1953), 262. (Xiphilinos ?) family. There are similar places near Kerasous and 14. Laurent, AP, 18 (1953), 266, 269. Ophis—see Oikonomides, Pontos, 32.

15. Laurent, AP, 18 (1953), 262, 266, 269. 24. *“*Khordogop,” which appears only on the British | : 250,000

16. Laurent, AP, 18 (1953), 271-72. map of 1901, and on that of Lynch (1901). It may, however, have

17. See p. 197. strayed from the more famous Chortokopion (Hortokop) in

18. Laurent, AP, 18 (1953), 267. Matzouka, to the west bank of the Kalenima Dere in these maps. See 19. avddic év th Mayépy tovg Autviwtas tov Katneitny, tov also Lermioglu, Ak¢aabat (see note 6), 7-8.

TCavKapny, tov PofaAnnyv kai tov KataKxadAov. 25. Due south of Yoros Burunu and due west of Akcaabat.

20. Panaretos, ed. Lampsides, 80. 26. Gokbilgin, BTTK, 26 (1962), 310; Vryonis, Decline, 355.

162 SECTION XVIII by that of a single village, Chorobe. Laurent, who published from Dryona, further east) were still in favor in February the bull of 1432, did not raise the question of the identity of 1418, when a kyr Theodore Doranites was Alexios IV’s the trv Tpiceiav otdolv ths XopoPrc (elsewhere in the protokynegos and special envoy,*? but may have fallen with same bull, XmpofBn) in it with the GAov Xwpiov thv that Grand Komnenos. At all events, the chorion of Chorobe Xapopnyv of the bull of 1371, published by Lambros, while may have reverted to the crown, for by 1432 John IV (and in Ditten, who discussed the XmpofBiov of the interpolator of the name of his murdered father Alexios IV) confirmed the Chalkokondyles, was unaware of Laurent’s work, and none half-stasis of Chorobe in the possession of the newly re-

of the three scholars discussed the whereabouts of the founded monastery of the Pharos as a former estate. The place.*’ Admittedly one bull refers to a half-stasis and the paroikoi who now went with the half-stasis of the monastery other to a chorion, but Horovi, near Mucera, is the only included Sebastos the Laz, Galitanes, Matriskiotes, and the appropriate candidate for both in Trikomia. Possibly the sons of Koukouloukas.°* The probability is that the halfstasis, like that of Moutzoura, was a dependent of the stasis remained in the hands of the Pharos until the end of the chorion of the same name—even so it cannot have been far Empire, for the bull of 1432 remained valid as late as 1460. away. We propose, however, that the stasis and chorion may There is a good possibility that Chorobe passed thereafter, be treated as the same place and that, as in similar cases in the with other confiscated monastic lands of suppressed houses, Acts of Vazelon, the terms are not used very strictly. A hint to the vakif of the Yavuz Sultan Selim’s mother, which still that this is so 1s provided by the fact that the whole chorion exists.

had only three dependent peasants in 1371, while the half- In the history of Chorobe one might glimpse the use to stasis had four or more in 1432—a reversal of the expected which crown lands were put—to reward great servants of relative position if the terms are taken strictly. Both bulls state or endow monasteries. Although lands granted to a refer to an earlier situation and, by conflating their infor- great servant of state could be hereditary, the political vicissimation, a tentative history of the village and its dependent tudes of the fourteenth century made it highly likely that

crofts can be proposed for the century 1332-1432. estates granted out would soon return to the crown—for Before about 1332 the chorion of Chorobe was evidently example, Panaretos mentions that after one rising, in 1342, part of the crown lands. In about 1332 the Grand Komnenos ‘‘a strict investigation followed and there were many confisBasil granted it to a member of the Doranites family. It was cations’ >* of the property of the nobility. The estates of subsequently confiscated, presumably returning to imperial Grand Duke John the Eunuch, who endowed both Soumela

hands, perhaps when leading members of the Doranites and the Pharos, probably reverted to the crown in 1344.%° family were executed in 1352.78 At some subsequent date Property endowed to the church was less likely to revert, but the half-stasis passed into the hands of the Grand Logothete here the abandonment of the monastery of the Pharos later George Scholaris (fl. 1363-—64)?° and through him, perhaps, in the fourteenth century proved an exception. The result was

to the monastery of the Pharos. But the monastery fell on that Chorobe, or parts of the village, changed hands nine or hard days and closed. Chorobe must have returned to 1m- ten times within the century 1332—1432. perial hands; at any rate, in September 1371 Alexios III Modern Horovi sustains perhaps twenty families and the

restored the chorion to the amyrtzantarios kyr George lands of the village and its dependent crofts and grazing Doranites, elsewhere styled Judge of the Imperial Court and could hardly support more. The rewards of office of the of All Trebizond and Oikonomos of the Great Church,*° major ministers of the Empire were evidently modest. But the from whose father it had been confiscated forty years before. modest settlement of Chorobe has had further and singular

Doranites was invested with ‘“‘all the chorion of Chorobe of fame thrust upon it, for, as Ditten has pointed out, the forty-two nomismata, that is to say, the possession and entire evidently Trapezuntine interpolator of Chalkokondyles conuse of all different rights and privileges over the establish- fused its spelling with that of the rather more substantial city ments, farmyards, esochoraphia, exochoraphia, mountainous of Rostov.°© Perhaps he was a Trikomian, for the same places and watered lowlands, rivers and fields, together with interpolator gives us evidence for a final medieval placeall grazing and pastoral lands, to have and to hold...” as an name in the bandon. In 1429 John IV chartered a Genoese immunity—the nearest Trapezuntine equivalent of a lay ship from Caffa to attack his father, Alexios IV. He landed at

pronoia grant. Doranites’ children could inherit ‘‘all the St. Phokas, Kordyle. Alexios IV moved out to “Ayavtoc.?’ chorion of Chorobe with its people.” “The people” (not Because the interpolator of Chalkokondyles describes this, termed paroikoi) included Tzakas, Xanthos, and Mou-

goules.*! The Doranites family (which may have come —

Arch. Seld. B.S6, and was published independently in E. L{obel], ‘“‘“A chrysobull of Alexios III Grand Komnenos,” The Bodleian

27. Laurent, AP, 18 (1953), 262 (cf. 269); Lambros, NE, 2 (1905), Quarterly, 3 (1921), 140-43; cf. Lambros, NE, 3 (1906), 486; 197; H. Ditten, ““Die Korruptel X@poBiov und die Unechtheit der Papadopoulos-Kerameus, VizVrem, 12 (1906), 144—47.

Trapezunt und Georgien betreffenden Partien in Laonikos Chalko- 32. Iorga, N&E, I, 273: Domenico lhirtheodoro Doraniti, dicto kondyles’ Geschichtswerk,” Studia Byzantina, ed. J. Irmscher Prothoconigeo, ambassiatore dicti domini imperatoris.

(Halle, 1966), 57-70; and Chalkokondyles, ed. Darko, I, 122. 33. Laurent, AP, 18 (1953), 267, 269.

28. Panaretos, ed. Lampsides, 70. 34. Panaretos, ed. Lampsides, 67: kai yéyove d1wypos Bapvc Kat

29. Panaretos, ed. Lampsides, 74—75. Gpnayr TOAAN.

30. Vazelon Act 121 of the 14th century. 35. M&M, A&D, V, 278; Laurent, AP, 18 (1953), 246. 31. Lambros, NE, 2 (1905), 197-98. The bull, with two-page 36. Ditten, Joc. cit. in note 27.

autograph of Alexios III, is pasted into the cover of Bodleian MS 37. Chalkokondyles, ed. Darko, II, 219~20; Bonn ed., 462-64.

PLATANA AND THE BANDON OF TRIKOMIA 163 perhaps fortified, palisade as a proasteion, commentators 6. A nineteenth-century church at Demirci Koyu (Buyuk have assumed that it was a suburb of Trebizond. In the Acts Fiz) has been published elsewhere.*?

of Vazelon the term is used inWe its have common Byzantine sense of Lo . , , th . y 7. Ahanda.** proposed above its identification anazelon estate, however. Midway between Kordyle and Trebi, , . . with Chalkokondyles’ Achantos. Ahanda ts in fact _the name zond, and just west of Akgaabat, by the coast road which ; , of aIVnumber ofIV settlements along the northwest Alexios and John would have taken, isofstream Ahanda, which , i of ; Platana. Mr. James Crow, the University of Birmingham, we propose as Achantos, where John IV had his father - i, ws assassinated in his tent bv night visited the site in 1971 and reported the tradition of a church y men below the modern mosque and above the road at Lower

MONUMENTS ;

Ahanda: here was a piece of plaster with painting and a yellow glaze sherd in the embankment of the road, together

I. P latana For tress. with tile and bone fragments. The fragment of painted plaster The British 1:250,000 map of 1901 alone marks a fortin —_ was 0.11 x 0.07 x 0.04 min size. The plaster was lime with the northern quarter of Platana by the sea. There is no black flecks (perhaps burnt chaff). The colors included black,

evidence of it today. a haematite band, and a green blue band. In Upper Ahanda, 2. The medieval church of St. Michael, Platana, has been 65 minutes inland, were the remains of a chapel, oriented al

published elsewhere? (pl. 96b). 272°,9.7 x 5.6 min size, with walls of 0.75 m. Its single apse

was very shallow. The masonry was of the local granite,

3. A later church, to the west of St. Michael, referred to by _ partly shaped. On the south side pulverized earthenware was

Talbot Rice, has been destroyed.*° used in the mortar.

4. The medieval church in Orta Mahalle, to the north of St. 8. Five further churches were locally reported in the hinter-

Michael, has been published by Selina Ballance.*° land of Platana, none of which have we investigated. The 5. The medieval church at Visera, described by Talbot Rice reports concern: the ruins of a church at Mersin, a church al and Winfield and perhaps referred to by Janssens, has been Limlo, near Mersin, by a lake about one hour's walk inland:

all but destroyed?! (pl. 95a, b). a ruined church at Isturkiye, west of Akcaabat: and a church at Aynadogaz, about half an hour inland from Kordyle.

38. Hamilton (who states that it had wall paintings) (1836), I, 9. Mimeras (Mimera, OF Sivr Kale) lies about two hours 246-47; Ritter, Erdkunde, XVIII, 813; Talbot Rice, Byzantion, 5 Walk northeast of Ozyisa-Akpinar and three hours from (1930), 66-68 and pl. 21; Ballance, AnatSt, 10 (1960), 164—66, pl. Akpinar, between Ortahisar and Samara. Sivri Kale (which xxib, figs. 17, 18; Winfield and Wainwright, AnatSt, 12 (1962), 131: means “sharply pointed castle’) dominates the end of a Ballance, Bryer, and Winfield, AP, 28 (1966), 258-60, pls. 14, 15 (of series of hills called the Mehmet Aga Daglari which rise up

the evidently medieval mosaic floor, on which see also Talbot Rice, thof th nt where the Kalenima Dere divides into t

Haghia Sophia, 87); Janssens, Trébizonde, p\. xxv, fig. 49. The church SOUL OF te poi w ere the Ka enima ere IVI eS in © two was restored in May 1846, and may have been that raided in 1877: streams, both of which have their source in the high ridge of see Bryer and Winfield, AP, 30 (1970), 279. Hamilton and Ritter give the Haska yavlas. Thus the castle could control two valleys its dedication as St. Michael: Papamichalopoulos (1901), 256-57as_ (fig. 36: pls. 95c, 96a).

the Taxiarchai, and Chrysanthos, AP. 4-9 (1933), 503 and pl. 83, as The shape of the walled enclosure is dictated by the the Archistrategoi. Its own inscription of 1846 gives a single

Archistrategos as patron. These are all variants on a dedicatory summit on which the castle stands, forming a rough rectangle

theme. In AP, 24 (1961), 111, Bryer, following Panaretos, ed. of which a corner is missing on the southwest side, where Lampsides, 64, 118, identified it with the Incorporeal One the walls skirt the edge of steeply falling rock. This indent ( Aowpartoc), where, on 30 August 1332, the army of theeight-year- measures about 21 x 10 paces. The largest extent of the old Grand Komnenos Manuel II defeated Bayram Bey, taking many losed | 6 v: ‘n lencth and (at th horses. On reflection, however, it seems unlikely that Bayram Bey enclosed area Is OMly some paces in length and (a c could have penetrated quite so far into this area then, and more eastern end) 26 paces in width. In 1958 the southwest and likely that the place should be identified with the “Ac@patoc at the southeast walls stood to a height of about 5 m. but by 1967

head of the Matzouka valley, whose crown lands were given to most of these had gone and only a small section of the nous | ae azeon Act 104 of the 13th century, Bryer, DOP, 29 southwest wall remained to a height of 1 m. The thickness of 39. Talbot Rice, Byzantion, 5 (1930), 66-68: Winfield and the walls varies from | to 1.5 m. They are constructed of Wainwright, AnatSt, 12 (1962), 131. Talbot Rice described it as a roughly-cut blocks laid in random courses with a mortar of domed structure, of the 18th century. Although the 18th-century lime, sand, and pebbles. Within the thickness of the walls cathedral of Chaldia at Argyropolis (GimUshane) was domed, post- wooden stretchers were used as a reinforcement, but there is

Byzantine churches are exceedingly rare in the Pontos before the , c den head Th teri d heavil 19th century, and domed churches even rarer. Talbot Rice tended to No sign 0 Woo en cacers. ine exterior sides were Neavily

err (as in the cases of Kaymakh, Sachnoe, and Vazelon) by placing pointed up with lime mortar to form a smooth weatherlate medieval monuments in the period after 1461. Is it possible that resistant surface. There may have been an outer ward on the

church “‘of no littletrace interest,” as late as 1901 ; ;was thedescribed ridge,asbut ofandwalling was. Papamichalopoulos noted only two churches in the town (see note

found.

38), it belonged to the very modern period. It is impossible to be sure. —_——40. Ballance, AnatSt, 10 (1960), 167, fig. 19, pl. xxuc; Janssens, 132. It is apparently the church referred to in Janssens, Trébizonde,

Trébizonde, pl. xxvi, fig. 50. 18 (cf. also p. 232).

41. Talbot Rice, Byzantion, 5 (1930), 68, pl. 22; Chrysanthos, AP, 42. Bryer and Winfield, AP, 30 (1970), 230-33, pls. 63-65. 4—5 (1933), 503, pl. 82: Winfield and Wainwright, AnarSt, 12 (1962), 43. Lermioglu, Ak¢aabat, 8.

164 SECTION XVIII The lack of worked stones or of any sign of smoothing or Villagers have reexcavated a rock-cut tomb. It lies on an cutting of the natural rock suggests that the site is not earlier eastwest axis and measures 1.75 x 0.35 x 0.40 m. Around than the Middle Ages. The modest size of the fortified area is the tomb the rock surface has been levelled; some of the work paralleled in numbers of Pontic forts. A date within the later could represent seatings for a wall. Byzantine and Trapezuntine periods seems likely; it was At the south end of the ridge is a cutting which may have once associated with the fourteenth-century amyrtzantarios been made to serve as a water pan, and to the north of the Theodore Sampson. The heavily pointed external wall sur- tomb is a deeper cutting which may also have been a cistern, faces recur at Suma Kale, Kurum Kale, Kovans, and Kog but contained human bones in 1962. (qq.vv.), among others, and the heavy pointing together with Rock-cut tombs are not uncommon in the Pontos. Those the wooden stretchers may be noted as possibly early of St. Sabbas in Trebizond have a claim to be Mithraic; they

features. are usually taken to be pre-Byzantine. Like the Gedik Kaya

10. Akpinar, or Ozyisa Kalesi lies on the eastern slopes of tombs, near Giresun, those of Hidir Neb are cut in the Pen the Kalenima Dere valley, about an hour’s comfortable walk and there are no adja cent structures to assist their dating. from Polita along a track running up the slopes to the east of The other cuttings might Suggest, ul they represent wall seat: the village. It is on a rock which is less steep than the sur- HES, the site ofa Mithridatic castle. But the mortar, tiles, and rounding slopes and commands a view southward up the sherds indicate a Byzantine occupation; as we have seen, Kalenima to the long ridge of the Karadagi. The path of a there was a kastrop hy lax in 1432. medieval route up the valley may be marked by a good wide The site is of some importance mn local folklore. At the

track which runs by the site. south end of the ridge the footprint of Ali is pointed out; he

There are only a few scattered foundations of rough-cut stepped to the next summit, a lew kilometers to the south. uncoursed stones set in a strong lime mortar. The enclosed The bushes of the site are hung about with the rags of the area is roughly oval, about 15 m across. A villager reported sick; we saw also offerings of food and ° slaughtered cock-

entrances on the east and west sides. erel. As elsewhere in the Pontos, the villagers congregate The lack of any level seatings for the walls or of well- annually on 6 May to picnic and play games. The festival dressed blocks militates against a very early date, but it could tes place on a piece of nearby high ground called Hizir

be later Byzantine or Trapezuntine. As with so many Pontic - co. ;

castles, its last defender is aid, in local romance, to have been , Hidir Nebi is the only site in the Pontos which the authors

a lady. The castle, though modest, was reputed to have visited together. withstood many sieges, until artillery was brought up to the 12. The Sera valley runs parallel to the Kalenima Dere and higher land, called Ayana, to the east of the site. Here there to its east; its form was changed considerably in recent times

were Said to be ruins, which were not inspected by us. by a vast landslide which has formed a fine lake. Until 1959 11. Hidir Nebi (Metinkaya Kale) stands on the northern there was a church to the west of the lake immediately to the summit ridge of the Karadagi at about 1,540 m, the site is right of the road where it tops a TSE. The church looked reached from the eastern side by climbing a steep wooded medieval but unfortunately investigation of it was postponed escarpment from Salari. The summit is a flat ridge about until too late. About 100 m below the site is the ruin of a

20 x 40 min size (pl. 94). rectangular church with rounded apse, about q x 4m in There is now no trace of an outer wall around the ridge, Size. There is no trace of painting and the building could be

but random digging by local treasure seekers shows con- early nineteenth ri ay ; > valk ab

siderable quantities of lime-and-pebble and_lime-and- To | ° west and a out twenty minutes Was above pulverized brick or earthenware mortar. Roman or Byzantine Derecik is a cave chapel into the wall of which water stoups ridge tile fragments were common, and local tradition had it ane carved out. It had a now destroyed masonry apse, b ut the

that there was once a church there—which may perhaps be place is still locally revered and the rags of the sick are

confirmed by the appearance of a fragment of lime plaster prouent to i. hurch h f the c: bearing red, yellow, and white painting. Among sherds on h hb, wnuren Was soporte’ to the nan . i iL ,

the surface were glazed fragments of yellow and light green eaurens ut was not visited. A castle 1s reported’ at Kalecik, a

Byzantine ware. long way inland up the Sera valley.

_ 30

NN ge

we . fol ee CO

TT ——— Sie . a Qu ? < 52

Ow

¢ ae 3, / = —, a > \\ 7 lp iy

A Wt eddi “Sg AG > hy SK 2 \ iG - < 3% wn .

\\a aes == eS a paee L yovdid DOR = eee 7 h \ ra =~ % 1 ee

< __) CO %, \ \ \ 3 J 72, ‘| ARO 5S > HY

< ON Se 2 is aa \ Sere LE oD) eS a) < = SSL

=e NS {— 4 ey - - . a LL \ / \ , ana OK

of A Yrs Yo *

_ \ \ |,-2 A “ A 5 an . id ros aes \ ’7, .|§_. ; » ~.| / E> pes vs > oy \ \ = - ; Q

(‘ ,aSW Y a A : . * — : ‘ ; \ 4 ‘ é ie , ( CL NN... —— ce ; | APPROXIMATE Sy O 10 20 30 METERS SO OO OOOO Eee

32. Plan

A (\ N1.08 Ken Qe M. ~ -Wall 4.5 M. high « \

yoo \ Lmy| x

‘ Average Emplacements“ | 0 thickness . ae } — 19th ‘Cen. Gun ~~. \| WN

| __} > . — o (an ,

A ny Ae

WELLFP — . ae ‘4 TOWER VF / - . ; ie) ENTRANCE- rp oN | an

~4 N ,| ; \ , - me a

f KOPAYAH ‘SS Wa -

fs Akca Kale FORTRESS AND SITE(?) OF

QO 10 = 20304050 METERS MONASTERY OF ST. PHOKAS 34. Plan

|

IEPON AKPOTHPION MIMEPA, TPIKQMIA am ROS BURUNU ORTAHISAR, SAMARA

Al | = FO RT «“"\ Holy Cape: Castle

SN ‘4 4 | \’ \ i ; a , \ -7oN © YY ¢ | YS KY y Ca ey . \ af le :~RES AN ~ : NY SW. an om Ly ; LON. J ‘ , SN } > GO ‘SL J ae ae : Sy / on S ] VA 4

SFP S & ™ | , OS“ ) Ns y Sy nn ‘if 4 Yet 7 ae |fm i~ \ot\raaL i ai 4, js ff & 7(ff Hi y Vs

\ | fj er SR yer Neen / yo OOS ( APPROXIMATE “o a.CL // Approximate

\ \\ tt tt J 33. Plan of Site 36. Plan

ri_

!aad h Ft ~ rnd a= 1 | i | _.. \) | _ 'CHURCH ms | " fw 5(an .|) |l ii i ——) at FL ( AT /} } Cee)

FOL MADEN TAPZOS 0 5 METERS CHURCH at ASAGI TERSUN

35. Plan 37. Plan

_~ | \\ \

’ Chrysokephalos. Trebizond fell briefly to the Seljuks after

St. Eugenios and his little army of martyrs were sup- 1071. Popular guidebooks to the city make the Sekiz Direk posedly put to death under Diocletian. In fact, when and how hamami, near the sea gate of the Lower City, a ‘Seljuk his tale and cult came to Trebizond is in doubt. But the later bath.” The structure is certainly venerable but cannot, by Miracula retain a memory of a cult of Mithras on Mount any stretch of historical imagination, be ascribed to the Minthrion, which there is no reason to doubt. The martyr transitory Seljuk occupation of the city; it is much more overthrew the great statue of Mithras; there are probably reasonable to seek the contribution of the Gabrades, semiMithraic tombs in Boz Tepe, and the cult was locally re- independent dukes of Chaldia in the late eleventh and twelfth membered as late as 1438. The earliest indisputable evidence centuries, to the defense of Trebizond against the Seljuks.*° of a church in the city does not come until 884/85—Belisarios’ The period of the Empire of Trebizond is marked by great supposed local foundations cannot be justified. The earliest building activity, often well documented. When its final and evidence for a bishop, or at least a prelate, of Trebizond present defenses were completed by 1324, the city consisted comes in 253/54—the story of St. Andrew’s visit is a of three connected but distinct walled enclosures, which we medieval notion. No doubt there were Christians before the have called the Lower City, the Middle City, and the Citadel third century in Trebizond, but the fact that the faith took (see figs. 41 and 42). Each section, climbing from the sea hold comparatively late (by contrast, Sinope’s St. Phokas below, was smaller and higher than the last. The Lower City was supposedly martyred under Trajan), is demonstrated by at near sea level is about 134,500 sq. m in size. Eating into its Trebizond’s initially being listed under the mother of Pontic southeast corner and 20 m above it is the smaller Middle sees, namely, St. Gregory Thaumatourgos’ metropolis of | City, of about 67,200 sq. m. Finally, 950 m inland, 50 m

Neokaisareia.*8 above sea level and 40 m above the ravine beds below, lies the

The city had to wait until the sixth century for an iden- tiny Citadel, of only about 19,200 sq. m. tifiable revival of prosperity when, under Justinian, it again In 1223 the Melik (probably Mugith al-Din Tugrilsah of became a supply base for Armenian campaigns. Procopius Erzurum) made three unsuccessful attacks on Trebizond. states that Justinian built an aqueduct, called the aqueduct of Lazaropoulos’ account of the attacks gives invaluable inthe martyr Eugenios (the earliest historical reference to the cidental evidence of its appearance then, which is summacult of the patron) and that he restored most of the churches, rized in fig. 41. First and most important, he makes it quite which had become decayed. A famous inscription over the clear that, although the Lower City was walled in his day (the eastern (Tabakhane) gate of the city, repeated in a probably 1360s), it had not been in 1223. Hence, classical and later copy in St. Basil (pl. 168b), stated that the Emperor Byzantine Trebizond consisted of the Middle City and Justinian completed the restoration of the buildings of Citadel only, lying inconveniently 400 m from the sea and Trebizond through the agency of Bishop Eirenaios in indic- harbor. The Melik’s first attack was from the “‘Upper Road’”’

tion 5, year 480 (the Trapezuntine series of years began in (i.e., that leading south from the citadel over Mount 63/64); the date is therefore indiction 5, A.D. 542.°° Minthrion) and St. Longinos; and the third began along the whole undefended shore between St. Barbara to the west and 37. Zosimus, I, 33; Bonn ed., 32: ... tig mOAEMG dbo Teixeot —_ St. Constantine and the ‘Old Arsenal.”’ Certain places menmepretanupevns. Cf. Gregory Thaumatourgos, PG, 10, col. 1037f. tioned are known, some are unknown, and the remainder can

The inscription of Diocletian, now lost, is in Finlay (MS, 1850), , ,

fol. 414; Paranikas, CPSyil, 29 (1907), 269: and CIL, III, Suppl., be located, more or less securely, from the account. Known No. 6746. Ammianus Marcellinus, XXII, vin, 16. Notitia dignitatum, sites include the ‘“‘Upper Road,” the southern gate of the

ot eek. naan ithe ails oaralae ; * bo sulos-Kerameus Korte or Citadel, the Chrysokephalos, the ‘““Western River,” FHIT. 1.32: W. Blawatsky and G. Kochelenko, "Lecultede Mithra St: Evgenios, St. Barbara, and, probably, the Arsenal of sur la cOte septentrionale de la Mer Noire,” in Etudes préliminaires Daphnous. Unknown monuments include St. Eustratios, St. aux religions orientales dans |’Empire Romain, ed. M. J. Vermaseren, Longinos, and St. Constantine. Place names that can be

VIII (Leiden, 1966), 20-21; Tafur (1438), 130 (referring to located from the account include the Epiphaneia or Citadel Trebizond as “Samothrace,” on which see A. A. Vasiliev, “A note on square, the Gate of St. George of the Limnians in the north-

1933), 138 31 On the te vend ofS OOO | vee 8 AP, 4-5 east corner of the Citadel (indicating that the Citadel was 39. Procopius, Buildings, II, vu, 1. The Tabakhane inscription

(CIG, No. 8636), was first recorded by Tournefort (1701), II,

175-76. Other readings are in Texier and Pullan (1839), 190; rs

Hamilton (1836), II, 409; Ioannides, Historia, 230. The St. Basil and Geschichte von Trapezunt unter Justinian dem Grossen,” BZ, 30 Tabakhane versions are recorded in Paranikas, CPSyi/l, 29 (1907), (1930), 385-86.

297; Bzhshkean (1819), 69, 77; trans. Andreasyan, 48, 52; 40. Constantine Porphyrogenitus, De Thematibus, ed. Pertusi, 73, Triantaphyllides, Pontika, 51; and Zacharia (1838), 312. The 137-39. On the ‘Seljuk Baths,’’ see Succi, Trebisonda, 235-37; and Tabakhane inscription no longer survives. Cf. A. A. Vasiliev, “Zur here, p. 196.

THE CITY OF TREBIZOND 183 already curtained off from the Middle City), and the Gate of terminus post quem for Alexios’ building of the eastern wall

St. Dynamis leading north from the Middle City.*! of the Lower City. The wall follows a logical course down to It is clear that the classical, Early, and Middle Byzantine the sea; Lynch noted an outer ditch at its northern end which city, perched on its rock, was very small—no more than was to be paralleled outside the western wall. 86,400 sq. m. Hence, as in 1223, the Chrysokephalos and the The stimulus to build a more ambitious enclosure on the Middle City were a place of refuge rather than a primarily western side was perhaps given Alexios by the sight of the residential area. The suburbs were unprotected. It is also Venetians and Genoese fortifying their own bases in the clear that St. Eugenios was fortified: the Me/ik made it his eastern suburb in the 1310s. The work became imperative by camp in 1223 and it required siege engines to reduce in 1319 when “the Sinopitans started a great fire which entirely 1340.4? The monasteries of the Hagia Sophia and the Pharos devastated the beauties of the city, both inside and out.” °° were similarly to be defended with towers and walls. Also Lazaropoulos ascribes the building to Alexios and indeed outside the walls was the market place, burnt down by the two inscriptions on the western wall name the Emperor; one invaders in 1223 and 1456.*° In 1609 Bordier placed it in the mentions his Master of Works, Constantine, and the other is western suburb, but, as Clavijo and others indicate, the shops dated 1324.°' By 1324, then, Alexios II had completed the lining the “imperial way” between the walled city and the enclosure of the Lower City and could face Italians, Turks, Meydan formed the real commercial quarter of Trebizond. and Sinopitans more securely.

An archontic quarter for the town palaces of the nobility— Excluded from the new enceinte was the Church of St. where they could be kept under house arrest—is mentioned John ths Métpac, endowed, according to a now lost inscripin 1350;44 it could perhaps have stood between Mount tion, in 1306.°? Its parish was to be known as the “E&q@tetya Minthrion and the ‘imperial way” in the eastern suburb, and the church as St. John ‘““Without the Walls.’’ °° One of where wealthier nineteenth-century Trapezuntines and the inscriptions of Alexios II on the west wall refers to the European consuls had their houses. There were natural dis- enclosure as a movptCtoc, or burc—an Arabic word found asters, such as a possible earthquake in 1347, and the wooden also in the bourtzai of Nauplia and Modon in Greece; the houses within and without the walls were revaged by fire in inscriptions of 1302, 1306, 1314, and 1324 are remarkable for 1243, 1303, 1319, 1341, and 1456.*° The suburbs were always their demoticisms and arabicisms. By 1430 the Lower City vulnerable, but it was the formidable walled city which saved was designated as the —EEmKaotpov; its normal Turkish name the pocket Empire on at least five occasions, when in 1223, became the Asagi Hisar. In 1644 Evliya named its gates (later

1297, 1336, 1341, and 1456 Turkish invaders were unable to confirmed by Lynch), working from the southwest: the

take it.*° Zagnos or Imaret Kapisi (referring to Zaganos Pasha, the The events of 1223 and 1297 showed that the old classical Faith’s renegade Christian Beylerbey, and to the Imaret of and Byzantine walled city was too small. The enclosure of Giulbahar, respectively), which leads over the Zagnos Kopru the Lower City, including part of the western ravine, was into the Middle City; one or two gates called the Sut Kapisi

the achievement of Alexios II (1297-1330), who also (Lynch’s ‘‘Sotke’”’); a Molos Kapisi (Evliya’s ““Mevluz,” a strengthened Kerasous.*’ This great work increased the word which rightly puzzled him) by the harbor; and in the walled area from 86,400 sq. m to a respectable 220,900 sq. m. northeast corner Mumhane Kapisi, to which Lynch adds the It is not clear when the work began. Two inscriptions, one a Pazar Kapisi. Today’s main western “‘gate”’ was cut through malediction of 1302 and the other a regulation for the corps __—‘ the wall by the Russians in 1916-18; Succi ascribes this

of night watchmen in Trebizond, dated 1314, are recorded endeavor to the Emperor Justinian in ca. 540.°4 from a section of the wall which appears to have served as a The Middle City is called the kaotpov by Panaretos in sort of public notice board (fig. 42, pl. 109b).*® The exact 1303, 1362, and 1378; in the bull of 26 October 1314 there is whereabouts of this section, by an Armenian dyer’s house, is reference to the ecclesia beate Marie Crisocofole sita in unclear, but it was certainly part of the eastern wall of the Castro Trapesonde. In 1430 Chalkokondyles describes it as Lower City.*? The date 1302 might therefore be taken as a the péya Kdotpov and in 1432 the bull for the Pharos monastery refers to the whereabouts of the metropolitan church 41. Lazaropoulos, ed. Papadopoulos-Kerameus, FHIT, 120-22: (i.e., the Chrysokephalos) as in the naAatov kaotpov—even Cahen, P-OT, 125. We are grateful to Fr. Jean Darrouzés,A.A.,for a century after Alexios II had built the lower walls a distinccorrespondence on the interpretation of the passage in FHT, taking tion was being made between the “old” and “new” kastra.

the Korte (literally “‘tent™) as the Citadel proper.

42. Panaretos, ed. Lampsides, 65. ——__———

43. Lazaropoulos, ed. Papadopoulos-Kerameus, FHIT, 120-22; the city wall of which it formed a part had been destroyed. This

Chalkokondyles, Bonn ed., 466-67. suggests a site near Mumhane Kapisi.

44, Panaretos, ed. Lampsides, 69: ta tav GpyOvtwv OoTITIa. 50. Panaretos, ed. Lampsides, 63. 45. Panaretos, ed. Lampsides, passim, Chalkokondyles, Bonn 51. Lazaropoulos, ed. Papadopoulos-Kerameus, FHIT, 120-22;

ed., 466-67. Fallmerayer, OF, I, 132-33; Finlay (MS, 1850), fol. 36%; Paranikas, 46 Lazaropoulos, ed. Papadopoulos-Kerameus, FHIT, 120-22; CPSyll, 29 (1907), 297; Mordtmann, CPSyi/l, 15 (1884), Parartema,

Chalkokondyles, Bonn ed., 466—67; Panaretos, passim. 75.

47 Lazaropoulos, ed. Papadopoulos-Kerameus, FHIT, 120. 52. Millet, BCH, 20 (1896), 496-97. 48. H. Grégoire’s version in BZ, 18 (1909), 490-99, is to be 53. Chrysanthos, AP, 4—5 (1933), 460—63, 622, 747, 792.

preferred to those in Fallmerayer, OF, II, 79; Papadopoulos- 54. Chalkokondyles, Bonn ed., 466-67; Evliya (1644), II, 45; Kerameus, CPSyil, 17 (1886), 116; and Paranikos, 29 (1907), 300-1. Lynch (1893-98), I, map opp. p. 13 (followed by Chrysanthos); 49. The inscription of 1314 was removed to the Archaeological Ritter, Erdkunde, XVUI, 883; Succi, Trebisonda, 221-22; local Museum, Istanbul (where it is still), in or just before 1908, because information.

184 SECTION XX But by the early sixteenth century the Middle City had Uspenskj’s interpretation of this and of the portraits is that become simply the ‘“‘Miso-kastori,” and its normal Turkish the inscription refers to Alexios I (1204-22), first Grand name is the Orta Hisar, Panaretos states that Alexios III Komnenos, son of the Sebastokrator Manuel, son of the strengthened the Kastro in 1378, but Miller’s contention that Emperor Andronikos I Komnenos (1183-—85)—a sort of this is borne out by an inscription seems to be based on a genealogical claim to Empire—and that Alexios was buried confusion with one of the inscriptions of Alexios II on the in the chapel. It is now impossible to comment on this western wall. The only known inscription from the Middle interpretation; certainly a southern gate was in existence in City appears to be that of Justinian over the Tabakhane 1223, a year after Alexios’ death and certainly the permuGate. The site is suspended between two bridges with their tation of names and titles can point only to Alexios I. One attendant gates: the Zindan (or ‘“‘Dungeon’’) Kapisi leads to could only wish for a corroborative transcript of the inscripthe Zagnos Kopru on the west and the Tabakhane Kapisi tion. The second, uncorroborated, inscription is recorded by leads to the Tabakhane Kopru and “imperial way” on the the normally unreliable Paranikas; it consists of the name east. To the north the Gate of St. Dynamis leads past the and titles of the Grand Komnenos Basil (1332—40) and came Cifte or Gavur Hamam to the Lower City and in the south- ‘from the palace.’ The third is equally tantalizing. Miller, east corner the Yeni Cuma Kapis: (originally perhaps the citing a Trapezuntine parish magazine of 1916, not now St. Eugenios Gate) and the Gate of St. George of the available to us, states that ‘‘an inscription on ‘the tower of Limnians lead into the eastern ravine and the Citadel Michael Comnenos’ to the west of the palace preserved the

respectively.°> late Emperor’s name”’—1.e., that of the Grand Komnenos

Panaretos distinguishes between the Kastro, or Middle Michael (1341-49). Finally Fallmerayer noted on the now City, and the Citadel by referring to the latter as the KovAac destroyed Kule boyu an inscription of the Grand Komnenos

(i.e., the Turkish Kule), in 1337 and 1351—Lazaropoulos John IV, dated 1460. This has troubled historians of had described it as } KOpty in 1223 and reveals that there was Trebizond because John IV is usually taken to have died in

a court called the “Emigdveia. This could be the “castle 1458. Lampsides’ solution is that he died in 1460. An even square” referred to by Fr. Odoric in 1318 and part of the simpler one is suggested by the fact that Fallmerayer alone palace—we suggest at its southern end (fig. 44). The entire was able to read the date A.M. 6898 = a.D. 1460 and it Citadel seems to have been a governmental enclosure of — appears to have been very high up the tower and damaged; which the palace formed part. There are a few literary and perhaps he was mistaken. From this evidence put together epigraphic indications of the late medieval development of — onecan suggest, at the most, that the Citadel and palace were the area, which from earliest times until the eighteenth cen- repaired or built in the periods 1204-22, 1223-35, 1332-40, tury was associated with successive rulers of Trebizond and 1341—49, and before 1460. There is nothing inherently un-

must be built upon the accumulated debris of countless likely about any of this evidence, but as none of it can now be administrations. Panaretos indicates that in 1376 the za- checked (and the provenance of much is not good), it is a dati was called the ‘‘palace of the emperor lord Andronikos question of taking or leaving it. At the least, there is some the Grand Komnenos” and that it was possible to fall to — justification in rejecting the evidence of all four Citadel one’s death from it. The western halls stand over forty meters _inscriptions.°° above the ravine and are indeed perilous. The Andronikos ts It is with relief that one can turn from the dubious value of more likely to be Andronikos I Gidon (1222-35), second the Citadel inscriptions to a final astonishing source for the Grand Komnenos, who defended the Citadel against the = appearance of the Citadel in the early fifteenth century, for Melik in 1223 and was buried in the Chrysokephalos, than Bessarion’s account of the fortress of the Grand Komnenoti is the later and more transitory Emperors Andronikos II the most detailed description of any Late Byzantine palace. (1263-66) or Andronikos III (1330-32). The next four Bessarion’s Encomium on Trebizond suffers from an almost scraps of evidence are epigraphic: all present problems and total avoidance of proper names, but, as the Trapezuntine no inscription now survives. In 1916 Uspenskij noted ina __ diplomat Basil, he must have know the palace and, stripped chapel, probably in the Kule boyu at the southern end (pl. of not very much hyperbole, his account is factual enough.*’ 132), a painted inscription beside traces of imperial portraits:

... OTOVAOIAIMO ... BACTOKP

_. MHOH...A 56. Panaretos, ed. Lampsides, 65, 69, 78, 124 note 4; AOV Lazaropoulos, ed. Papadopoulos-Kerameus, FHIT, 121; Odoric

(1318), Hf, 245; Uspenskii, Ocherki, 34-42; Chrysanthos, AP, 4—5

HMONKVP (1933), 459-60; Millet and Talbot Rice, Painting, 114-45;

5 ...ANAPONIKS Paranikas, CPSyll, 29 (1907), 297; Miller, Trebizond, 56, 96 (citing HO | % | CEBACT OK, 13 November 1916); Fallmerayer, OF, I, 68, 193; II, 103; KPATOPOC... loannides, Historia, 232. Finlay (MS, 1850), fol. 354, noted in his MANOVHA diary: “I give the inscription as given by Fallmerayer for I could not pretend to be sure that he was right & saw no reason for his copy in correct [sic]. I had no telescope & my sight is bad.”

55. Panaretos, ed. Lampsides, 63, 74, 79; ASL, 13 (1884), 517; 57. Bessarion, ed. Lambros, NE, 13 (1916), 188-89; trans. (here Chalkokondyles, Bonn ed. 466-67; Laurent, AP, 18 (1953), 268; used by kind permission) Cyril Mango, The Art of the Byzantine Gokbilgin, BTTK, 26 (1962), 297 (the ‘‘Miso-kastori” was an excep- Empire, 312—1453 (Englewood Cliffs, New Jersey, 1972), 252-53. tionally small mahalle with only 41 households in ca. 1520); Evliya See also O. Lampsides, “‘Zu Bessarions Lobrede auf Trapezunt,”’ (1644), II, 44; Lynch (1893-98), I, map opp. p. 13; Ritter, Erdkunde, BZ, 35 (1935), 15-17; the same, ‘‘Die Datierung des “Eyxa@ptov

XVIII, 883; Miller, Trebizond, 69. TpaneCovvtoc von Bessarion, BZ, 48 (1955), 291-92.

THE CITY OF TREBIZOND 185 The dwelling of the emperors is set up in the present acropolis who in attacking it have done so to their own detriment.°' High and is itself no less than an acropolis, surpassing as it does all up, at the end of the building, there appears a covered imperial other buildings by the strength of its walls and the variety, size dais®? having a pyramidal roof supported on four columns. and beauty of its construction. Its west wall is common to the This too, is screened all round with white marble, roof and all,

acropolis and the palace, and serves the same purpose to both and it separates the emperors from their subjects as with a up toa height of two storeys; from there upwards, it extends for barrier. It is there in particular that the emperor makes his the sake of the palace alone and towers above the wall of the appearance, that he conducts business with his ministers, conacropolis by almost the same measure that the latter rises above verses with ambassadors, speaks and is spoken to.®* Farther on

the ground.°® The walls facing in other directions, being ad- is another imperial hall®* of outstanding height and width, equate in point of height, thickness and other respects, extend covered with a roof and having columns all round. In this all the way down and, while they take away more than half the building, which is decorated with paintings, is a flight of steps acropolis, they add this area to the palace, and are by them- so as to raise the emperor aloft, and it is here that he is wont to selves sufficient to resist the oncoming enemy and to guard give splendid banquets for his ministers and other subjects.°° safely those that may be inside. They afford entrance by means On the left side one encounters a suite of many rooms, includof two gates and one postern,°? and for the rest are securely ing one that differs from the rest: this is fenced by four equal constructed so as to exclude and ward off attackers. On either sides, like a frame, and contains memorials of the Creation of side is left an open space for rooms and the quartering of the the world and of the origin and history of man.°® On the rightemperor’s servants, while the palace rises in the middle and has hand side are many halls, vestibules, terraces, chambers and one entrance provided with a staircase of steps, so that the rooms separated by colonnades, running athwart one another, entrance is also a way up.®° As one enters, one straightway all of a measure that cannot be surpassed, each bigger than the encounters on one side splendid vestibules and halls of suffi- next and all constructed with unutterable beauty and due har-

cient beauty and size, capable of containing a great number of

people, the halls being surrounded with balconies facing in all 61. Such secular decoration was common enough in places asdirections and exposed to all the breezes. On the other side is sociated with Orthodox courts —€.8.. the scenes of Justinian S eX-

; aelong ploits and in the beautiful Chalke and Manuel’s in the its Great Palace, the paintings stretched a very building, floor 7;;; ; a - in theout Hagia Sophia at Kiev, the legendary decoration ofenthe; palace tirely paved with white marble, while its ceiling shines with the of Digenis, and the actual decoration of Gagik’s palace on blooms of painting, with gold and various colours. The entire Aght’amar, and the despotal portraits in Mistra. See Procopius,

hollow [of the ceiling] gleams with shining stars in imitation of Buildings, 1, x, 5; V. Lazarev, Old Russian Murals and Mosaics the heavens and exhibits excellence and delicacy of painting. (London, 1966), 52-59; A. A. M. Bryer, ‘““Achthamar and Digenis All round, on the walls, is painted the choir of the emperors, Akrites,” Antiquity, 34 (1960), 295-97; S. Miranda, Les palais des both those who have ruled our land and their ancestors: also empereurs byzantins (Mexico City, 1965), 144-45, John Kinnamos, painted there are the dangers our city has undergone and those Bonn ed., 171-72; cf. P. Magdalino, “Manuel Komnenos and the Great Palace,” BMGS, 4 (1978), 101-14. In Trebizond the head of the dragon slain by Alexios II and exhibited as a trophy in the palace

58. The description begins from outside the palace. From the west might have been kept in this hall—see Lazaropoulos, ed. ravine the break in masonry, over half way up the west walls, Papadopoulos-Kerameus, in FHIT, 64. The laconic nature of the between the Citadel (“‘acropolis’’) wall proper and the palace wall early entries in Panaretos’ chronicle rather suggests that these wall and halls above is clear enough (see pl. 145a) and seems to be referred paintings and, presumably, inscriptions were a source for it—see

to59. here. Bryer, BZ, 66 (1973), 332 note 3. The two gates are the Gate of St. George of the Limnians in 62. Bra.

the northwest corner and the more formidable Kule Kapisi—- 63. The arrangements give the impression of quasi-ecclesiastical Lazaropoulos’ tOv tis KOptns mvAEwMvMV (ed. Papadopoulos- furniture: the throne beneath a ciborium fenced off by chancel Kerameus, FHIT, 120) in the south. The present northern entrance Is screens. It was presumably here that Clavijo (p. 111) met Manuel III a modern cutting. There are two candidates for the postern: blocked in 1404: **... the Emperor called for us ambassadors, sending horses doorways on the west side of the north curtain wall and below the to bring us to his palace. On arrival we found him in a chamber that

southern hall leading out into the west ravine—the latter blocking was off a gallery, where he received us very graciously: and after

seems to be ancient. talking with him for a space we returned to our lodging.... The

60. The Citadel is now entered and the rest of the description Emperor is a man well-built, tall and of a stately presence; he and his depends on whether Bessarion is viewing it from the south or north. son were dressed in imperial robes, wearing hats of a very high shape, That he is entering from the north is suggested by the following: the which had cordings of gold running up the sides with a great plume Gate of St. George would be the normal entrance from the city (the on the top made of crane feathers; further these hats were trimmed

southern gate is a military one and would require crossing a ravine with marten fur.”’

other than by a bridge); the palace stands higher than the first 64. Bra again. We suggest that this second hall is the southwest (service) court described and the area of the southern defenses of the one-—see p. 195 and pl. 135a.

Citadel overlooking the ravines (where one would expect the most 65. On banquets in the palace on the panegyris of St. Eugenios, important part of the palace to be situated, rather than servants’ see Lazaropoulos, ed. Papadopoulos-Kerameus, FHIT, 65-67. quarters) is at least 5 m higher than the northern gate; and finally the 66. “EvtedSev éni pév aplotepa GAAOIS TE SWHATLOIG MAUNMOAAOIC chapel stood at the end, coinciding with Uspenskjj’s equally modest EOTIV EVTIDYELV KAL TOV GAAOV Evi SLAMEPOVTL, TETPAOL HEV tdaI¢C chapel in the southern tower. The approach established, it is possible SI PNHLEVY NAEvpais Kal OYTHA TAGiOLOV C@COVTL DIOLVNHATE TE to identify tentatively some of the existing buildings. On the right THS TOV GAWV YEVEOWS PEPOVTL KAI WS THV APYNV Av3pwnos yEyove

(i.e., west) side, Bessarion encountered first a vaulted throne room YEVOMEVOSG TE MG TA KAD’ ADTOV Exititevoev. An awkward passage.

and hall and next a higher, arcaded, banqueting hall. These could Mango (loc. cit. in note 57), whose translation we use, notes “‘Le.,

represent the two surviving halls, first a vaulted building (pls. paintings illustrating the book of Genesis.” There is no indication 137b—139b) and to the south a higher structure with graceful win- that they are in fact paintings; the only surviving Genesis cycle in dows (pls. 135a—136a). Within its restricted space, therefore, the Trebizond is the relief on the south porch of the Hagia Sophia; there palace of the Grand Komnenoi follows the pattern of other Late are fragments of relief in the east wall of the Citadel (pl. 130a, b). But Byzantine palaces at Tekfur Sarayi, Mistra, and Nymphaion: first is it necessary for the ‘“‘memorials”’ to be depicted at all? The passage

stabling and servants’ quarters round a courtyard, then an upper surely makes better sense if this “fenced” room is regarded as a audience chamber and the imperial apartments furthest from the library, “containing” (in Miller’s free summary, Trebizond, 122)

entrance. ‘memoirs on anthropology and political history.”

186 SECTION XX

—,.5;

mony. There too, is set up a church decorated with beautiful later than most. The blocks are also found extensively paintings and adorned with sacred offerings which, if not very reused, relaid in mortar as wall facing or serving for doornumerous, are of outstanding beauty. What the church lacks in jambs and lintels, in many places in the fortifications, churc-

size it makes up in comeliness.° hes, and other monuments of the city. There seems to be a later version of this type of masonry where the blocks are still

After 1461 the Citadel became the seat of the new Pasha of much the same size but are less accurately cut and therefore Trebizond. It was garrisoned first by local levies and later by give a rougher general impression. a Janissary corps. In 1758-59 the Citadel, St. Eugenios, and In the following list the numbers refer to positions marked the Theoskepastos were strongholds in a local battle between on figure 41 of extant sections of classical masonry.

the Pasha with his Janissary garrison, and the local 1. A length of masonry several meters long, to the left of the Derebeys—virtually repeating the events of the 1340s.°° A Gate of St. Dynamis. recently published Portuguese report of 1766 states that the 2. The east jamb of the Gate of St. Dynamis surviving to a Georgian king Erekle (Heraclius) II invested the Citadel on height of about three meters; the T-shaped recess for a three sides with cannon and took it; the editor of the docu- locking device is sharply cut. On the adjoining “apse,” see ment seems unaware that Erekle did nothing of the sort, but p. 206 below. a possible threat from him and a probable one later from 3. and 4. Two short stretches of masonry, too inaccessible Russia may have led to the final strengthening of the defenses to study.

of Trebizond.°” . . 5. The curved bastion by the Yeni Cuma Kapisi which has

Before the archaeological evidence need be considered, been much patched but appears to be Roman in origin. therefore, the following firm statements can be made about = A short stretch which has been pointed with mortar to the defenses, gathered from literary and epigraphic sources: some extent but appears to be originally of Roman date. the city was walled before A.D. 257; it was restored probably 7. and8. A short anda long section now carrying an inner in the sixth century and possibly in the ninth and late elev- Byzantine wall, but originally representing the outer wall: enth; the Lower City was enclosed in about 1302-25; the see p. 189 below. Middle City was strengthened in 1378, the Citadel wasem- 9. One of the longest stretches, although only four courses bellished with a palace, and underwent additions and repairs high. It is almost continuous with the base of the projectthroughout the period 1204—1460; finally, further improve- ing semicircular bastion, also with Roman masonry at its ments should be expected in all the walls in Ottoman times. foot, which marks where the curtain wall between the The archaeological evidence for the walls will be considered Middle City and the Citadel joins the western rampart. next, which will be followed by the combined historical and The internal angle here, with a blocked arched doorway in archaeological evidence provided by the twenty-five other each wall, is of the somewhat rougher masonry mentioned secular monuments and ninety-six known churches and above and is almost certainly later than most, possibly

monasteries of Trebizond. third to fifth century®®* or even Middle Byzantine,°?>

while the long stretch on the north-south wall is of the

MONUMENTS ‘‘regular’’ type, as is the lowest part of the curved bastion

The Defenses wall. The relationship between the two periods is a difficult

point, as the “early” curved bastion base projects further

I. The Classical Walls (fig. 41) . than the “later” wall with doorways (see fig. 41). Possibly Mrs. Selina Ballance has contributed the account which there was a considerable change in the plan at a later date.

follows. . . 10. The “‘exedra”’ (pl. 111a). This unusual polygonal strucThe masonry of the Middle City and Citadel is of many ture forms the inner face of the curved bastion mentioned types, which are described in the next sections. One distinc- under no. 9. Its walls comprise seven sides of a regular tive type, visible in the Zaganos Kopriisi and a dozen places dodecahedron, five of them having round-headed niches, in the walls (see fig. 41) can almost certainly be considered as the backs of which appear to be later blocking: to the north

being of the Roman period—.e., first to third century A.D. there is a return wall with a further niche, and in the south This masonry consists, typically, of large blocks of a dull wall there is a short passage with a blocked doorway in its grey conglomerate up to 1.30 x 0.55 x 0.45 min size, laid west wall. It seems certain that from this point a passage or

very true and close-jointed, without mortar. There is no stair descended the short distance to one of the blocked , particular bond, but the blocks are used as both headers and doorways in no. 9. The inner structure had a semidome, stretchers. In general they are dressed fairly flat on the sur- now largely fallen, of large blocks in regular courses formface, but in some instances have slight bossing. Presumably ing a facing for mortared rubble. A curious feature is an the blocks are normally the facing for a mortared rubble opening, like a small doorway and apparently original, in core, but this is not visible save in one case, which may be the lower part of the semidome and directly over a niche; it

67. Ona chapel in the palace, see p. 215. is now blocked but might possibly have originally given 68. Bryer, BK, 26 (1969), 196. The pasha was more or less be- access to a rampart wall, with a timber stair for access. In

sieged in the Citadel in 1827, 1830, 1833. about 1944 an earth tremor brought down a floor which

69. L. Tardy, ‘‘Heraclius, roi de Géorgie, d’aprés les documents portugais de 1766,” BK, 27 (1970), 101-7. But no credence can be

given this propaganda report. See D. M. Lang, The Last Years of the 69a. Notes of Selina Ballance. | Georgian Monarchy, 1658-1832 (New York, 1957), 136-225. 69b. Notes of D.C. W.

THE CITY OF TREBIZOND 187 had spanned the building below the semidome, but this walls survives or is identifiable. Work in the mid- and latemust have been a late addition—possibly even Ottoman. third century, either against the Goths or at the time of The purpose of the ‘‘exedra”’ is enigmatic. Among the Diocletian may be expected, but the only part which looks ideas that come to mind are a nymphaeum (since, by the noticeably later—the “‘exedra””—may not belong to the forthird century, the water supply could well have been intro- tifications. Nor are we assisted by any other still extant duced by pressurized aqueduct, so that water would have Roman buildings in the city, and no Roman inscriptions been available even at this height), or, perhaps, an early survive in situ. A survey of classical materials reused in church with west apse, which, though unique in these churches and elsewhere (which we have not made) might parts, is not totally impossible. D. C. W. notes that a func- prove useful. The only certain point is that the masonry tion as a praetorium and a Justinianic date should also not described above is typically Roman. be excluded. Excavation would probably yield a great deal The three sections of Trebizond, Lower City, Middle City

more information. and Citadel, are examined next, in topographical rather than

nearly vertical rock face.

11. A short, inaccessible piece of walling above a in chronological order.

12. The Zagnos KOprisii. This bridge is more fully 2. The Lower City

described on p. 190 below. Its principal Roman feature 1s The Lower City walls, completed by Alexios II in 1324, are one pier, clearly visible though encased in later masonry described here beginning from the southwest tower up to the

(pl. 111b). It is of typical Roman masonry with, in this Zagnos Koprusu. See figure 42. case, some bossing. It is complete up to, and including, a The Southwest Tower 1s about 15 x 12 m in plan eastcornice at the level of the springing of the arch and has two west and north-south. Like the rest of the walls it is built of voussoirs still in position of its western arch and three of its roughly squared stones laid in regular courses. The external eastern. Most of the lower part of this high bridge (which is aspect has been altered by the substitution of rectangular now virtually a causeway with one relatively narrow Openings for round arches in the windows (pl. 110b). The arched opening for the small stream) 1s built of Roman interior has three storeys and a flat roof. blocks. The eastern part, running up to the Middle City The walls of the first, or ground floor are at least 3 m thick. rock, may well be contemporary with the pier mentioned There was a barrel-vaulted chamber with an east-west crown above, as there may originally have been five or six open and round-arched windows in the center of the walls at east arches with a solid causeway at one or both ends. The and west. The barrel vault has now collapsed or has been blocking next to the pier is probably very late classical but demolished except for the springing at the sides. It was of much of the rest is considerably later since the blocks mortared rubble and constructed with shuttering, of which (doubtless reused) are laid in mortar. The pier blocks the marks are visible. The door was in the north wall at the appear to have somewhat stronger bossing than the rest, east corner. There appears to have been no internal combut the structure varies throughout from block to block. munication between the first and second floors. 13. Immediately to the north of the bridge there is a long The second floor is roofed with a quadripartite vault, the stretch of masonry, running first north and then turning walls and supporting pilasters being faced with neatly-cut west with the natural cliff, which includes several blocked rectangular stones alternately flat and bonded in. The supor blind arches. It seems an extraordinary place to have porting arches are slightly pointed. All four walls had roundany sort of opening. At the external angle it is at least arched windows; that in the south wall has been blocked and

twelve courses high (pl. 110a). all have been changed to a rectangular form, on the lines of Finally, D.C. W. notes a few large and apparently Roman those in ‘““Theodora’s Bedchamber’” in the Citadel (see p. 195 blocks on the outer part of the east ravine, on the south side below). The door is above the first floor; since no trace of of the Tabakhane Kopru at foundation level. They might another door is to be found, it must have led out onto both

have formed the foundations of a barbican. walls; it is round-arched. Above the pilaster capitals (which From the evidence of this surviving masonry it is clear that are undecorated projections) there are square beam holes, the main rock on which the city stands—-that is, the Middle presumably used for scaffolding when the vault was conCity and the Citadel, but not the curtain wall between them structed. On the north side a stone staircase (possibly the —was fortified with walls round its entire perimeter in original one) runs upward from west to east and then turns Roman times. The similarity of the stretches which survive south in the thickness of the wall to emerge onto the roof. indicates one particular period of building. In only one place The second-floor chamber must have been an elegant room, (the western side of the Citadel at nos. 7 and 8) was the line perhaps assigned to the captain of the guard since the tower of Roman wall different from that of the main Byzantine adjoins the old gate. The flat roof is now much overgrown period, and even here a Byzantine wall was built ona Roman with daphne and wild snapdragon; there were battlements.

base. The tower is about 15 m high.

The only attempt which we can make to date this masonry The exterior is of regular courses of rough stone placed is to surmise that Corbulo may well have fortified his base with a flat surface outward, with smaller stones filling irregu-

and that the earliest stretches may therefore date to about larities. The interior core (probably of mortared rubble) A.D. 48. Half a century later Hadrian may have attended to could not be seen. The surface appears to have been well the walls as well as the harbor. This presupposes that, if the pointed over with mortar to produce a flat finish. The quoins city had been fortified before A.D. 48, no evidence of these are of well-squared large blocks in the usual long and short

188 SECTION XX alternation. There are thirty-one courses up to the base of the or brick fragments, and well-cut quoins of large blocks. The second-floor window on the north side, ten more up to the difference is that in this wall are beam holes for tie beams, in

base of the third-floor window, and twelve more up to the which unrotted wood remains. The lowest row is about 3 m present top—perhaps sixteen originally, totalling fifty-seven above ground and the headers are at about 3-m intervals. courses. At about twenty-eight courses up the northwest The second row (where only two holes are visible) is 1.50 m corner quoins and twenty-seven on the southeast quoins are higher up. Above that there are odd holes. projecting blocks which must have been sculpted. The form The rectangular projections, or towers, protrude about is not now clear but there may have been a lion, facing 3.50 m at intervals of about 16 m and are about 8 m broad at frontally, with very long front paws. On the north face of the the exterior (pl. 112a,b). Beam holes appear as in the wall tower, above where the wall meets it, there appears to be a described above, but the towers and intermediate walls are

join in the wall. devoid of slit windows. The wall mortar is of lime and sand

Wall from the Southwest Tower to Zagnos Koprisu. and the mortared rubble core includes a large quantity of The wall started from the northeast corner of the tower, beach pebbles. where its broken beginning survives. It must have stretched On the inner side of the wall the arcade of blind arches runs

about 100 m, but all trace of it has disappeared. from the Zagnos Kapisi to the first tower to the north. The Zagnos Kapisi (pl. 110b), about 8 m north of the tower, arches are pointed and about 8 m wide. must have been one of the principal gates. The opening 1s Houses have been built against most of the wall, and the rectangular and about 3 m high and 2 m wide. The ground interior of only two towers could be examined. One still level outside appears to have fallen about half a meter. contained a house, the other had contained one; both had Each of the jambs of the blocked door consist of two disturbed the masonry. However, there was no trace of stone massive blocks and two smaller blocks of the local con- or brick vaults and no internal wall to enclose the space and glomerate. The lintel is a single block of Proconnesian make it a regular tower. These “‘towers” were, like that at marble, about 2.50 m long and 0.50 m high; its thickness is Kordyle’!® and those in fourteenth-century fortifications in not visible. Above this is a high tympanum with a pointed the Balkans, almost certainly no more than hollow projecarch. Fragments of corroded stone of the old cornice may be tions from the wall and contained perhaps wooden floors and seen at the base on either side. It is possible, however, that the ladders. The thickness of the wall, where it could be as-

tympanum cornice above the lintel and its cornice are orig- certained, averaged 2 m. inal and that the stones now at the base once bore carvings Tower 4 (counting the first tower north of the Zagnos which have since been defaced. The keystone 1s represented Kapisi as tower 1) once boasted the shorter of the two by two separate blocks. Within the tympanum is a recess for inscriptions of Alexios II, in two lines of letters incised in a stone, probably that reported to have a device of a single- brick or pottery.’? This inscription having now apparently headed eagle on it, possibly that shown in the photograph of — gone, the best evidence of its appearance is provided by a 1893 in plate 113—in which case it had been taken further hitherto unpublished photograph taken by Millet in 1893 (pl. down the wall.’° About 1 m above the top of the arch are two 113). The plate is broken (one break runs through the inscriphorizontal courses of stone within a brick frame; within the tion); it also seems to be a double exposure, for the two frame was the longer of the two inscriptions of Alexios I] in carved blocks in its lower part reappear in another of Millet’s brick, or coarse red pottery where the lettering had curves; unpublished photographs (pl. 114), where they are mounted only fragments of some half dozen letters now remain. This against a black background, with a tape measure hanging

inscription was originally about 3 m long.’' between them—the two views are not identical for the leftAbout 5 m north of the gate and 4 m above ground there ts hand block has been moved 90°. In plate 113, the out-ofa slit window and another at the same height 5 m farther focus masonry to the right does not appear to belong to the north. On the interior side of the wall, at the gateway, was a photograph of the wall and inscription either. The two stone blind arcade of stone arches supporting the catwalk.’'* The blocks must not, therefore, be regarded as having the restonework projects about 0.20 m from the exterior base of lationship with the wall and inscription which plate 113 the south jamb. Support for a drawbridge is unlikely; there appears to show. That they are perhaps associated with the may have been a wall of some kind starting at this point. wall is suggested by a number of points: obviously Millet The West Wall. North of Zagnos Kapisi there are more took the exposures one after the other and would therefore slit windows (apparently seven) at about the same height and have been working in this area. Lynch records a carving of a intervals, save that the intervals between the northern two single-headed eagle above the Zagnos Kapisi and below the extend to about 6 m. The slit windows run to a point where longer of the two inscriptions of Alexios II, where there is the wall recedes about 3 m inward. The distance to the first now a recess,’** and also a now apparently lost doublerectangular projection ts about 15 m and from the southwest headed eagle on tower 4. A carved animal projects from the tower to the projection about 100. At the first projection the southwest tower. Thus, we have tentatively assumed that the wall stands 8 to 10 m. Its masonry is similar to that of the sculptures recorded by Millet come from the west wall. southwest tower, with lavish pointing, fillings of small stones The blocks (pls. 113, 114) are of considerable interest.

70. Lynch (1893-98), I, plan opp. p. 13. 71b. See p. 158.

71. See SeeSee p. 183 above. Tla. Seep.p.183. 183.72.72a. p. 183.

THE CITY OF TREBIZOND 189 They are a pair, about 0.33 m high and 0.26 m wide, ap- The right-hand block in plate 113, visible again at the parently square in section. The base of each of them shows a bottom right of plate 114, is more enigmatic. One face shows

double ropework decoration and the top is indented as if to a thickly maned lion, his long tail between his legs, his receive a further block. The left-hand block is carved on at forepaws raised. Behind his neck, and broken off, is the head least three (and probably four) sides, the right-hand one on at of another beast. Possibly the lion is attacking an antelope, least two sides. They could therefore have formed part of two but behind him, coming from the left side, moves a beast freestanding piers or, perhaps, square pilasters; but their which has something of the body of a horse and the head of a

function 1s obscure. sheep—the body being on the left-hand side of the block and

The blocks are carved in comparatively high relief (per- the head on the same side as the lion, above his rump. haps up to 4 cm). The left-hand block of plate 113, shown The eagle and hare and the lion and antelope (if that is again twice in plate 114 at the top and at the bottom left, what is depicted on the right-hand block) are common depicts familiar Byzantine motifs. One side (shown only in Byzantine motifs. Daniel in the Lions’ Den is one of the Old pl. 113) displays an eagle and a hare. The eagle turns its head Testament scenes which (like that of Jonah and the Whale) to the right; its wings are outstretched and its body is marked are common enough in Early. Christian iconography but

with a crisscross pattern, each resulting diamond having a become increasingly rare thereafter—their appearance at small hole in the center. Its tail is splayed out behind and Aght’amar in the tenth century has sometimes been taken as beneath the body of the hare, which it clutches in its talons. an archaism. Yet, the style of the Trapezuntine sculptures The hind legs of the hare hang to the left; its head to the right places them in the later, rather than earlier, period and may

is difficult to make out. perhaps be associated with not only the work at Aght’amar

The other two sides of the block, visible in plates 113 and and Ani but also certain tenth-century Georgian reliefs in 114, and probably the fourth side, too, represent Daniel in Tao.’° Byzantine sculpture is far from being uncommon, but the Lions’ Den. On the side opposite that with the eagle and much of it has an undeniably freakish air; this is true of the hare Daniel is shown in the conventional manner, standing in Genesis cycle on the Hagia Sophia and certainly true of the the orans position. He is dressed in trousers, with an over- two little reliefs described here for which plates 113 and 114

skirt reaching down to the knees, a fur mantle crossing over provide our only evidence, since no trace of them now his chest. He is nimbed, but it is not clear if he is also wearing remains. a Phrygian cap. Except for the fact that the overskirt reaches To return to the west wall, there 1s a stretch of about sixty to the knees rather than to the thigh, his costume and stance meters between towers 5 and 6, through the middle of which

are identical with those of the Daniel in the early tenth- passes a road (the Maras Caddesi) made by the Russians in century relief on the church of the Holy Cross, Aght’amar. ’? 1916. This long stretch without a tower is curious. Lynch’s

Fifth-century sarcophagi in Ravenna and the Lateran plan shows that there had been a gate (one of the two Museum show a lion in a sitting position on either side of | Stitkapisi or Sotke Kapisi Gates) where the Russians widDaniel.’* At Aght’amar and in other versions of the same ened the entrance, and one or two large blocks of stone scene—a probably fifth-century mosaic in the Beirut in the wall by the gap further confirm the fact. It is true that Archaeological Museum and a relief from Ani, probably of | the Zagnos Kapisi is guarded by only one tower, but this the tenth century, now in the church of the Holy Apostles, gate would have had no flanking towers at all. Kars—the artist has solved the problem of showing a basi- North of tower 6 the wall disappears; the base of the north cally two-dimensional lions’ den by representing the two face of this tower has apparently reused Roman blocks, but lions which flank the praying Daniel, hanging, as it were, by the possibility that there was some sort of classical wall here their hind legs, their backs along Daniel’s sides and their should not be excluded. heads and forepaws to the ground as they lick Daniel's feet Between towers 9 and 10 are about fifteen meters of un(in the Trapezuntine example the tongues are emphasized). interrupted wall. Taking this as a standard pattern, we have

This is an awkward angle and usually entails making the therefore conjectured the existence of towers 7 and 8. lions rather smaller than Daniel. Instead, by moving the Midway between towers 9 and 10 and about four meters bodies of his lions back onto the sides of the block, our above ground are the jambs of a blocked door or window. sculptor has produced a neater design. Only the full-face The jambs rise to about 1.25 m; their upper parts have been heads and manes of the lions appear beside Daniel; the destroyed by the robbing of facing stones. To the west lay an smoothly molded bodies are carried round onto the sides. open space with Moslem marble stellae gravestones within

Above Daniel is an inscription: only UIOU ... CO can be which, about fifty meters from the wall, are traces of made out. Beneath the head of the left-hand lion one can medieval foundations. If they are connected with the blocked

read: TEOPIT fo} apparently the proper name. archway in the wall an outer fortified court could be surmised.

. . . , an . 1 : , S ; ; ;

73. Lynch (1893-98), II. fig. 142; Sirarpie Der Nersessian, The second Siitkapis: or Sotke Kapisi, noted by both Aght'amar. Church of the Holy Cross (Cambridge, Mass., 1965), Evliya and Lynch, lay below tower 10. Today there is a gap in

BE Bock R Goebel. and A. Heidenreich. The C , the wall through which runs the Kale Kapisi Caddesi, lead(London, 1962), pls. 49,50. Seealso K. Weitemann, Catalogue of the 118 '0 the Sotke Caddesi or Sotke Kapist Caddesi. No trace Byzantine and Early Medieval Antiquities in the Dumbarton Oaks

Collection, Il (Washington, D.C., 1972), 33. A Daniel is shown in 75. See D. Winfield, ‘Some early medieval figure sculpture from mosaic on the south side of the sanctuary in Hosios Loukas, Greece. north-east Turkey,” JWarb, 31 (1968), 33-72.

190 SECTION XX of the gate remains. Only the south wall of tower 10 survives: m. The machicolations stand 1.20 x 0.60 m and are 0.60 m

the remainder is part of a modern house built against it. apart. There is presumably a stone or brick vault below the Below tower 10 runs a stretch of about forty meters of wall present floor level but excavation would be necessary to which curves more or less due north. There may have beena __ establish its form. The distance between the present second

tower 11 as a corner tower, but this section of wall is too floor and the top of the machicolations is about 3.70 m. The

badly damaged to make us certain of its existence. mortar of this upper part is of lime and small pebbles, The wall then turns east. About twelve meters east of the whereas the mortar of the west wall contains larger pebbles. corner is a blocked door. Ithasashallowround arch madeof This upper storey has smaller stones for facing and less two large stones and the jambs are of large well-cut blocksof — regular courses than the west wall; it is perhaps contemlocal stone. The door is about 0.80 x 1.70 m in size. Unless porary with the repair and the gate. there were steps leading up to it, the ground level has fallen From this point east, the line of the walls is now lost; they here by half a meter. This and a second small door high in the turned south near the Mumhane Kapisi (which lay within wall, are difficult to explain. They would only have weakened bastions) and resume again south of the Pazar Kapisi. A sea

the defenses. The thickness of the wall at this northwest tower stood on a projection of the north-south wall at the

corner is about 1.60 m. northeast corner, near the beginning of the eastern arm of the The wall then runs east 60 m and north 10, then east 10 Molos (pl. 1 16a, b). Rising from a battering of comparatively and north 8, and finally east 8. The stretch contains a tall, large mortared stones, this tower appears to have had three thin, round-arched window, about | to 1.50 x 0.20 m, with faces to the north; the upper stonework is akin to that of the the remains of an arch of a second window about 2 m to the west wall. It had a series of comparatively large windows. ’*4 west. Both windows are about 7 m above the original ground The Outer Walls. Lynch noted outer walls and moats level. The arches and facing are formed of neat stones. A running north-south from a point south of the junction of the modern window to the west of the second probably replaced east wall of the Lower City with that of the Middle City toa a third original one, making the series a piece of regular point just south of the Pazar Kapisi; and from the Zagnos

fenestration. Kapisi to a point just south of tower 6 outside the west wall.

The gate called Molos Kapist is in the last stretch of 8 m The eastern outer wall cannot now be traced, but the western (pl. 1 15a). It 1s round-arched and above it is a pediment in outer wall runs at about ten to thirteen meters from the inner which a plain oval cartouche was probably intended to take walland is about 0.80 m thick. It does not appear to have had

an Ottoman monogram (rather than the golden head of any towers, and in 1963 only its foundations were visible in Constantine of local oral tradition), for the masonry of the places. Between the intersecting Maras Caddesi and the seagateway is much later than the wall. The outer ground level shore the line of the outer wall is lost and was not noted by here rises about 1.50 m above the threshold of the gate. Lynch. But it may well have continued on the west side of the

Within the later arch is another brick arch made of one present Resadiye Caddesi, for the width of this paved road course of bricks; they vary in thickness from 2.25 to 4 cm. corresponds to the distance between the two walls in the Within this, and about 0.20 m higher, the thickness is again upper stretches.

arched in stone. ; ,

The Molos Kaptsi tower and the windows west of it are 3. The M idle City . ,

; The Koprisu. The bridge over the western probably Byzantine but may be Zagnos later Alexios II’s work, ;pte: ravine from thethan now blocked Zagnos Kapisi to the Zindan for the wall here is of smaller stones and looks as if it might be , ; —— ; Kapisi in the Middle City is of four or five periods (see pls. a later repair—although not necessarily Ottoman. 117a, b, and 118a, b) A complete tower jutsa bridge north and of the Molosblocks, Kapisi.having i. re , 1. There of west large embossed It, is. oo. reached by anwas external staircase beginning east of the ; 75a , perhaps eight arches altogether at the lower level.’°* The door to the second floor. The lower walling of the exterior is ; ; the ; ; arches were round and all5 the stonework, including heavily buttressed to a height of La about m on the north and , ; , voussoirs, were fittedlooks without as exterior present westSo, faces. Aremains recessedofstone on the west face if itmortar. had ;Atae the two arches can be clearly distinguished on had an inscription. The upper courses of the north faceof stillthe ; stream. Above the first La i. the south side, to the west stand to within a few centimeters of the original height and aor . ,;

. ; ,; row The of arches a simple jutting carry machicolation. west face was has two narrow openings . ; cornice or string course, and higher up in the masonry, about five meters below the

with more or less rounded arches; the north face has three, ; Lo. , present bridge level, a second jutting section of stonework though the arch shapes are not clear. The east face hassince an | ; ,;all the ; ; may have marked the top ofInternally, the bridge—but opening with a pointed uncurved arch. these ; , ; , ,; surrounding masonry medieval be ceropenings widen out to allow space is forlate a bow and arrowone or ; -cannot :, tain. The original bridge (or possibly aqueduct) probably

even a smallconsisted cannon. The large internal arches arein allheavy curvedembossed ;f.b (pl. 115b) of a double row of arches Pp" ; ; are masonry. There seem to have some headers and There beam holes tie beams thebeen walls of the tower .they ; . : stretchers, but, for perhaps as atnresult of later repairs, and in the repair wall (if such it 1s). The thickness of the wall . ok

. ; , appear to haveis been rather irregularly distributed. on the second floor tower m; at the internal width ;;, ; , 2.of Thethe structure was1.30 repaired a later period. The earlier of the tower from east to west is 6 m. The windows measure arches were blocked with good masonry. again without 1.80 x 1.20 m on the interior and 0.90 x 0.40 on the ex- 6 Y a8 terior; those on the west face are rather smaller, 0.65 x 0.20 75a. See p. 228.

THE CITY OF TREBIZOND 191 exterior mortar and with irregularly placed headers and Lynch's plan and description of the citadel of 1893 remain stretchers. These fillings can be seen on the south side, to the most useful; Talbot Rice’s plan and description of 1932 the west of the stream and have less pronounced bosses. To are sadly inaccurate; figure 44 shows a plan made in 1969 this period may also belong the remains of an arch over the which, though not perfect, is probably the most accurate yet

stream, at a high level. There is trace of only one brick published. ’’ arch; it is constructed of a double row of bricks employed By combining the archaeological and literary evidence, a as voussoirs framed by a flat row around the outside. This tentative periodization of the growth of the citadel can be looks like a Justinianic or Middle Byzantine repair. Apart proposed. It must be emphasized that the table following ts

from this, there is no sign of brick in the bridge. no more than a hypothesis and that we have examined only 3. On the north side, to the east of the stream, is a facing of what masonry 1s visible, that 1s, above ground.

medium-sized rectangular blocks of stone—neat work Type A, classical masonry, has already been discussed. ’*

which could be Middle Byzantine. Its presence in the Citadel shows that the outer wall of the 4. An amorphous mass of coursed, roughly squared ma- Original acropolis probably ran along the present northeast sonry constitutes the greater part of the structure of the walls and may have reached as far south as the tower of John bridge. This is certainly Trapezuntine and Ottoman. IV; certainly the classical outer wall was the present inner 5. Present work includes the widening of the road and steel wall in the southwest corner, below feature 22, commonly

cantilevers. called (for no evident reason) ‘“‘Theodora’s Bedchamber”’ or

Bessarion states explicitly that both bridges overtheravine = Kuizlar Saray (pl. 133a, b). were made of wood.’® If his statement is accepted, he must Type B is also classical but somewhat later (we suggest that refer only to a wooden bridge proper supported by masonry the dividing line might be the Gothic destruction of 257) and of stages 1 to 4. Otherwise, the stages correspond fairly well is represented only by the “‘exedra” in the southwest corner of to those periods in which it has been postulated, from non- the Middle City which has already been discussed.’? Lynch

archaeological evidence, that Trebizond experienced _ notesat this point a blocked door between the exedra and the reconstruction ’°# (p. 186 above). The first stage would be Citadel which on his plan becomes a sort of gate and is first- or second-century work. The repairs of the second considerably enlarged in Talbot Rice’s version. There is no stage could be Justinianic and may have been necessitated by gate, but there is indeed a blocked door in the south corner of

the sack of 257. The third stage could be ninth century. The the ‘texedra’’—or, rather, at least two entrances opening fourth is Trapezuntine and Ottoman. It must be emphasized onto a low external platform. They give perilous access to the that this can only be a tentative periodization in this most ravine below, but it is difficult to see how the Citadel proper

extensively repaired and reconstructed part of the defenses of (which at this point appears to belong to one of the

Trebizond. fourteenth-century types E) could have been reached. There Below Zagnos Koprisu the west wall of the Middle City may have been arrangements such as a wooden catwalk has classical or Early Byzantine foundations (pl. 119a,b). which are now lost. Above them is masonry of well-cut medium-sized blocks, Type C. We have taken as our model of this type certain probably of the Middle Byzantine period. There are stone features which were already existing at the time of the relieving arches with a fill of small stones. It is conceivable Melik’s attack of 1223, and were described by Lazaropoulos that there was a series of four or five arches here. (fig. 41), especially the northwest hall (feature 28), which is In the west corner of the north wall of the Middle City are apparently the oldest structure within the Citadel and which the remains of an arch composed of comparatively large we have already tentatively identified with the first hall, that bricks which could be Justinianic or later. It probably of the throne in Bessarion’s encomium (pls. 137b—139b). covered a staircase to the Lower City—at the modern road The northwest hall is a slightly waisted rectangular buildlevel below is a rock cave which may have been associated ing, substantially constructed; the walls average 0.75 m in with it. If the arch was part of a doorway at the original ground level of the Middle City, then the Middle City stood, 77. Lynch (1893-98), I, map opp. p. 13; followed by Chryat this point, at least 5 m below the present ground level. The santhos. Talbot Rice, JHS, 52 (1932), 47—54, suffers from a number floor of the Chrysokephalos in the center of the Middle City of defects because it relies on Uspenskij rather than on any original

is about 1.50 m meters below the present street level. source. Thus garbled elements of Bessarion’s economium (some North of the Chrysokephalos lies a small square incorpo- attributed to Panaretos and none to Bessarion) appear, and Lynch is rating a garden wall which is of Byzantine build (pl. 121a). Lynch, rather than Uspenskij on Lynch, would have shown. The This wall has a single window which ts faced on the east side tower of John IV is referred to as the tower of St. John and attributed

; “re quite unjustifiably berated for errors—which an actual reading of

with Oinaion stone in the flat manner and on the west side to John II (1280-85). The article is reprinted in D. Talbot Rice,

wit beck. It might be part ofthe monastic buildings of the PLS dr amd ncn, Caled Suds anon 7 Chrysokephalos, the enceinte of which may well have choked Mr. John Haldon, Mr. Robert Keeney III, and Miss Jane Isaac. It

the entire width of the Middle City at this point. was checked on the spot by Mr. James Crow, who reported minor distortions in the northeast corner (where the northwest corner of

4. The Citadel (fig. 44, pls. 120-148b). the gate, feature 5, should be shown asa right-angle). A. A. M. B. has

Finlay’s sketch of 1850 is published in figure 43, but conflated his notes on the masonry with those of D. C. W. and of Mr. Haldon, to whom we are most grateful.

76a. See p. 186. 79. See p. 187. 76. Bassarion, ed. Lambros, NE, 13 (1916), 186: see p. 185 above. 78. See p. 186.

192 SECTION XX

5G

CONCORDANCE OF MASONRY TYPES AND FEATURES IN TREBIZOND CITADEL

2|? E2 3] ? E2 H 4|? E2 6/°7?1A Cc? H 7 |? E3? 8 A D3 E3 H 9 D3 E3 10 D3 E3 11 D1 12 D1D1 G 13 14 D1 15 | ? D3 E3 , 16 A D3 E3 |F : 17 D3 E3 H 18 D3 E3 H 19 D3 E3 H 20 D3 E3 H 21 |? G? 22 A C D3 E3E3GH 23 | ?| A? D3 24 D2 H

Type Letter: ?1A |B/C {D1/D2 ;,D3 | El |E2 |E3 |F}G /H

Feature No: 1 | ? E2

25 D3?| El | El H 26 | ? E3? 27 | ? D2? H 28 C {DI 29 El? |El? El?| El? | E3? H 30 | ?| A? | E3? H 31 | ? E3? 32 B El? | El? | E3? H SUGGESTED PERIODS OF TYPES

? Unknown

A Before 257

C Before 1222 Citadel described D1-3 13thcent., esp. 1223-35: The Palace of Andronikos I by Bessarion B After 257 ! The Korte | The Palace and

El-3 14thcent., esp. 1302-24, 1332-49

F 1458/60: The Tower of John IV G Ottoman, before Evliya’s account H Modern, after Millet’s photographs of 1893 and refacing of the 1960s

thickness. In its final form, at any rate, it had two storeys. On Oinaion stone at sill level. Two windows are now visible; if the lower floor there were six round-arched windows (one there were any others, they have been obscured by features 27 was possibly a door, hence requiring a flight of external steps) and 29. From the northest corner of feature 27 a roundon the principal, east, fagade. The south facade has a round- arched doorway leads to the upper floor of the hall.

arched door and window and the north facade two round- In its present form the interior of the hall 1s divided into arched windows. The west facade, later clad, seems to have two bays by a heavy groined vault, but it is clear that this is a

originally constituted part of the west wall of the Citadel. later remodeling as it cuts partially across the lower winHere in the upper storey, there are round-arched windows of dows. Originally, there may have been a wooden floor. two different periods of construction. One build consists of The neat ashlar blocks, header and stretcher facing (as in ashlar blocks, laid partly as headers and stretchers, perhaps the Hagia Sophia, No. 112), and curious voussoirs (as on the the vestiges of a projecting tower. The other is of smaller door of St. Akindynos and in the porch of the Nakip Camii, blocks of local stone and has a single horizontal course of Nos. 57, 53 respectively), all point to an early date for the first

THE CITY OF TREBIZOND 193 period of the building. The second period, characterized by porch of the Hagia Sophia;®* the drapery is more convincthe Oinaion limestone course and by almost as good ma- ingly rendered and there is a certain distinctive rhythm in the sonry, cannot be dated much later. Tentatively, therefore, we two pieces, which are probably associated. The Old suggest an early thirteenth-century date for the first structure Testament subjects perhaps hint at an earlier rather than and a later thirteenth-century date for its partial remodeling later date—neither scene is represented elsewhere in the

(type D1).8° Pontos, or at Aght’amar, but both are common enough (for

The clean masonry of the remodeling of the northeast hall example, in the mosaics of Sicily®*). Their dating, however, and its Oinaion stone course leads one to similar work in the would not serve to date the angled tower (13-14) any more main walls of the east side of the Citadel, from features 12 to than would the Fatih inscription on the other side of it.

14 (pl. 128a, b). The wall here, which in bastion 13-14 Types D2 and D3 appear in two similar, but distinct, achieves a height of almost 30 m, is of one build, without sign stretches of masonry associated with D1, but there is no of classical foundations, characterized by very neatly cut evidence for determining in which order the types D should rectangular blocks and the use of little mortar and dis- be placed chronologically. D2 (pls. 136a, 140a, b) is a homotinguished by a single horizontal course of Oinaion stone geneous block, represented by feature 24 (and perhaps 27), running for about 30 m along the stretch of feature 12 and similar to that found in the equivalent position on the west about halfway up the wall. Toward its southern end is in- side, but of less well-laid, rectangular stones and mortaring. serted a block with relief decoration. This shows the upper It is surmounted by a series of very large windows (averaging half of a circular rope-work design with what would origin- 1.46 x 3 m in size and about 1.6 to 1.8 m apart), directly ally have been five rope-work roundels on each side. People overlooking the ravine. Beneath the fenestration is a blocked who live beneath the walls at the point maintain that the now door leading out on to the ravine, about 0.6 m above ground eroded stone once bore an inscription. No sign of it 1s visible level and 1.38 x 2.42 m in size; it has a blocked tympanum and no traveler remarked it, but the notion is plausible (pl. 147a). The lintel is at present about 4.5 m below the inner

enough. A stone waterpipe projects from buttress 11 and ground level and there is no sign of interior access to the there are other water outlets along feature 12 (pl. 128a). door, which appears to be part of the original construction. The angled tower, of the same masonry type D1, is in fact a By 1376 these windows shown in plate 140a, b, would have hollow bastion (features 13—14) about 2 m thick at the top been, as they are today, the only windows of the palace which

(pls. 128b, 131b). There are signs of internal wooden struc- directly overlooked the ravine (features 22 and 28, which tures. Three blocks are inserted high up on its faces: on the were also fenestrated, were encased by further exterior northeast face there is a block with an inscription com- walls). We suggest that it was from one of these palace memorating the Fatih’s conquest of 1461;°' on the southeast windows that the Despot Andronikos fell to his death in face there are two Christian reliefs. The photographs repro- 1376: if we are right, this part of the wall was almost certainly duced in plate 130a and b were taken through telephoto built by Andronikos [I (1222-35), who is, therefore, also lenses, which necessarily distort the scenes, but the subjects probably responsible for equivalent building (D1) on the east are clear enough. The right-hand plaque shows a raven perch- side.®° ed on top of a sinuous ten-branched tree, offering in his D3 is a third type of masonry which is associated with the beak a roundel of bread to Elijah who is squatting, enveloped other thirteenth-century ones (pls. 125a, 126, 127, 132, 134b, in a fur cloak, his right hand to his cheek, his left emerging 140b, 145b, 146a, b). It is almost invariably found connected beneath the garment (1 Kings 12:4—7). The left-hand plaque with another kind of masonry above it, of the type which we shows a nimbed angel clothed in a flurry of drapery driving have called E3. D3 and E3 are, in association, the most Adam and Eve out of the Garden of Eden (Gen. 3:24). The important masonry types in the Citadel. They have much in Protoplasts are covered with skins to the buttocks and Adam common: rather roughly-faced rectangular blocks (less looks back at the angel who pushes him. The entire field of cleanly cut than in D1), laid in regular courses and sporting the background is decorated with a vine scroll in very low comparatively high crenellations, which, in the case of D3,

relief. are entirely blocked by E3 work. But D3 work is distinThe work is more distinguished than that of the capitals guished by a darker stone and slightly less mortar; E3 by a discussed above®* and of the Genesis cycle on the south lighter stone and abundant pointing or mortar laid to achieve a smooth finish (pls. 125a, 126, 132, 133b, 134b, 140b, 146a, 80. Although the masonry is not comparable and although it is, b, 147b, 148b). E3 is akin to the masonry of Kordyle and of strictly speaking, outside the Citadel, feature 6 can probably be the watchtowers and small castles of the fourteenth century

assigned - a period pelore O28 aso. ms : tne Sale of which stretch west; D3 is closer to D1 and D2, for which we east corner of the Middle City into the east ravine. As will be seen N2¥€ Suggested a thirteenth-century date, from pl. 123a, the present entrance is no gate but simply a square We propose, then, that the main waist of the Citadel (1.e., hole in the wall, which at this point is a good 13 mhigh. But perched _‘ features 12 to 14 to the east and features 24 to 28 to the west)

above it is a curious and now much altered vaulted structure, part of was probably built by Andronikos I, after the Melik’s siege which employs long bricks with abundant mortar. It is clearly of 1223. If so, the evidence of D3 shows that the remainder of

Byzantine, probably antedating the Empire of Trebizond. It would be tempting to identify it as a chapel, but there is no eastern apse or 83. M. Alpatov, “Les reliefs de la Sainte-Sophie de Trebizonde,”’ surviving appropriate decoration. Its purpose is therefore enigmatic. Byzantion, 4 (1927-28), 407-18.

81. Lynch (1893-98), I, 21. 84. SeeO. Demus, The Mosaics of Norman Sicily (London, 1949).

82. See p. 189. 85. See p. 192.

194 SECTION XX the Citadel (perhaps not then belonging to the palace proper) E2 1s the masonry found in the north curtain wall, which

was surrounded by a less formidable wall. presents special problems (pl. 124a). It is basically of one A photograph of the east wall (features 8 to 10), taken build, of a rougher E, or fourteenth-century, type. If it was before 1893 (pl. 126), shows clearly the different stages of built in the fourteenth century as an entirely new feature it masonry. The northeast section (feature 8) has deteriorated would be pointless to identify gate 5 as the Gate of St. George severely since, while the section south of feature 8 and north of the Limnians mentioned in Lazaropoulos’ account of the of feature 9 has survived more or less intact, and the south- Melik’s attack—besides, the gate itself is largely Ottoman east section from feature 10 has now almost gone. Since (pl. 123a). However, there are a few, but enough, larger 1893 damage seems to consist in the disappearance of a blocks standing on the naked rock at its foot to suggest an polygonal tower, the removal of a long stretch of upper wall, earlier foundation, although one cannot hazard a guess as to and the unblocking of upper crenellations around feature 8 its date. The western half of the wall is backed by houses and (pl. 125a, b); an addition seems to have been a rounding off of it is impossible to estimate its main thickness—which is feature 8 at the northeast corner. Plate 126 shows the two about 0.8 to 0.9 m at the top. There are two blocked doortypes of masonry more distinctly than they appear today. ways, features 2 and 3. Feature 2 is unremarkable. Between it Halfway up the wall 8 to 9 and below the later crenellations is and feature 3 is a rock-cut drain which slopes down from the a lower wall of darker stone, surmounted by the “ghosts” of wall and apparently turns in at its foot. The blocked door 3 six or seven crenellations. This lower wall would not have has a semicircular tympanum edged with carefully cut blocks stood much higher than the inner ground level of the Citadel. of Oinaion stone; the entire door is 1.37 x 1.70 min size and Taking the upper and lower walls as D3 and E3, the same its quality is comparable to the D! work in the northwest wall

pattern may be identified elsewhere in the Citadel. (pl. 121b). It may represent Evliya’s blocked entrance. The Lynch’s plan shows that the southern tip of the Citadel modern entrance to the Citadel, at feature 4, was cut after was defended in a highly complicated manner, with a double 1896. To the east of it the curtain wall is accessible on both entrance through a narrow passage. There is not enough sides. It ranges in width from 1.2 to 1.7 m and 1s topped by a

evidence to hazard a date for most of this work, but north battlement (against attack from the Middle City) Uspenskij’s photograph of the tower of John IV and its about 0.5 m wide, with a south catwalk 0.7—1 m wide. surrounds, taken in 1916, brings us back to familiar territory The northeast and only gate, which we have tentatively (pl. 139a, b). Here there were at least four masonry types: a identified with that of St. George of the Limnians,°*° consists classical base followed by the dark D3 carrying crenellations. of a hollow tower, with an outer, westward facing arch and and surmounted in turn by the lighter E3 blocking the crenel- an inner, northward facing entrance (pl. 123a, b). It 1s clear lations. But instead of continuing to its own crenellations, E3 that the bulk of this structure is later than the wall proper of is abruptly cut off from the masonry of the tower proper by E2, for the joints with it are square. The slightly different size more than twenty-four well-laid courses of squarish blocks of blocks used for the main masonry, the well-cut quoins, with abundant mortar, rising to a windowed chamber (per- which are not found elsewhere in the Citadel, and the joggled haps the chapel)—noted by Uspenskjj and seen also on the arch of the outer entrance point to Ottoman work. (The left of plate 126—-before the topmost crenellations. This we roughly blocked arch above the outer entrance arch is ap-

assign to our next type, F, the work of John IV or his parently not the “ghost” of an earlier gate but part of the successor in 1458-60. The whole tower stood about 21 original build and intended to strengthen the lower arch). meters above ground level at this point and maybe 50 meters The rusting iron door, noted by Lynch,?®’ is still on its hinges

above the bed of the eastern ravine. and is of no great antiquity; the inner, wooden, lintel cannot E3 surmounts D3 more or less continuously from this be very old either. But in its present form, the gate appears to point on, as far as feature 20. As we turn the corner at feature be that noted by Evliya in 1644; indeed, it can be no other.*®

18, Uspenskijy’s photograph of 1916 (pl. 134b) shows the This does not mean that there was not a gate in this place distinction better than the modern view (pl. 134a). Between before. The entrance through the E2 wall, although now very features 18 and 19, masonry type D3 climbed at an angle of | badly damaged, has surviving facing on the east side and the about 10°; its eight successive crenellations are clear enough, beginnings of what appears to be a stone arch; both are but the coursing, as in E3, remains horizontal. What is more consistent with the period of the curtain wall itself. As we interesting is that the now familiar distinction between D3 have said, the Gate of St. George of the Limnians which the and E3 can be seen in the inner (originally classical) wall at Melik attacked in 1223 cannot be the present inner entrance features 22 and 23 (pl. 133a, b). Thus the outer wall was at through E2 masonry, still less the outer entrance through this point a thirteenth-century addition, defended in turn by Ottoman masonry. We are faced with three possibilities: the a heightened inner wall; both were crenellated. At the cir- first is that the Gate of St. George of the Limnians 1s, in fact, cular bastion 19, E3 simply added to D3 (pl. 146b), but at the the Gate of St. Eugenios to the northeast (which was cerlarger bastions 20 and 25 (pl. 145b) E3 encased, and enlarged tainly standing in 1223); the second, that the Citadel was on what was probably D3—-for obvious reasons it is difficult much smaller in 1223, with a curtain wall running approxi-

to be sure of :the masonry, and thegate whole area is further 86. Called , aL | « Papad | “leinner . Called a “small ” in Lazaropoulos, ed. Papadopoulos1960s (ol 1374) E3 continuce srobably north» A nh 38 ae Kerameus, FHIT, 120; the identification of it with feature 5 i made 7 by Chrysanthos, AP, 4—5 (1933), 69; see p. 225.

here it supplants not D3 but a more recent type of build 87. Lynch (1893-98), I, 20.

which we have taken to be E1 (pl. 148b). 88. Evliya (1644), II, 44.

a s+ 7Ne8 xss 2

U + ~ rm) vt * OD + N nN z. 02 . > s WO y, a Ss :o _ Ps OeSEE ‘38 @ = a Sy 4 COn ao g a\ 2 S 22 L. = < x ae § Ss \ E35 > F + 3SE sy +

az ws ve s ~F 2= ECG 2caf« 2 @z0o% — 1 Oo EgN gz 2 © ane 2 \i, 522200 SE Bx 8 eea eS -a.,\.: Sz

ON CQ er" Co een - 8- pe so ny Ry 1N Oo mi LOU os0~2» =| 53o D N = it + wey 3eo 8°8 Z>% 8% S825 Ggbgur dE 2 2 O > oo S = SC) ws ¥o23= q = uv ng --3 yy — —_— SY) E £ £ Beh gE ye 80 ) Fe, ? ww) — *4% NG: s3 eS) a! = 3 Ope ok ~~ : . —! ete ~~ 8 Pat O