1The Kirmir river, a tributary of the Sakarya/Sangarius, lies northwest of Ankara in Galatia , at more or less 1000 m above sea-level, on the Central Anatolian high plateau. The upper Kirmir river forms a narrow valley roughly 100 km in length, from mountainous Kızılcahamam in the northeast to where the Kirmir is joined by the Süvari and the Ilhan/Karboğaz in the southwest (Fig. 1). Beyond this point, the terrain changes, the river valley becomes wider and less clearly defined, and the historical topography is relatively well known, with various ancient sites and toponyms[1].

2In contrast, the upper Kirmir river or Siberis, as it was called in antiquity, has attracted little scholarly attention so far. Its canyon-like southern ingress was guarded by Tabanoğlu Kalesi (Fig. 1), a Hellenistic/Galatian fort that has been identified as ancient Peium[2]. One third up the valley, the southern cliff is riddled by a large rock-cut cave complex that is today known as Inönü Caves. Apparently defensive in character, the inaccessible complex also includes what seem to have been Byzantine chapels on the second and third of four floors. The chapels are each identified by a templon or sanctuary screen (Fig. 2. 3)[3]. Further up from the Inönü Caves, the valley becomes wider and, in addition to easy passage in an otherwise difficult, mountainous terrain, also contains many patches of well-watered, flat land. Numerous Turkish hamlets and their farming activities, in particular the cultivation of rice[4] and, more recently, vegetables, attest to the fertility of the soil, even if the small dimensions of the fields inside the narrow valley appear ill suited to modern, industrialised agriculture.

3However, small dimensions would not have been of great concern in pre-modern times, and it seems unconceivable that ancient farmers gave the valley a miss, particularly considering the unfavourable, arid conditions in the surrounding highlands and mountains. The Life of Theodore of Sykeon that took place in the region around the turn of the 7th century records various settlements with threshing floors and vinyards, some of which were likely located in the upper Kirmir/Siberis river valley[5]. Such settlements would have been similarly small as the modern hamlets, as no one point of the narrow valley could have supported any larger agglomeration. Such small ancient settlements in Central Anatolia normally consisted of little more than mud, possibly with some field stones for foundations, but without lime mortar or any other lasting architecture[6]. This can partly explain why the settlement archaeology of the upper Kirmir/Siberis river is as obscure as most everywhere else in rural Galatia and Central Anatolia, although the fertile valley offered particularly favourable living conditions.

4Another part of the explanation may be sought in a dispersed settlement pattern without a clearly defined nucleus that, over time, would form a settlement mound or hüyük and thus stand out as an identifiable archaeological monument[7]. We shall come back to this point at the end of the paper. First, we need to report a series of new discoveries from a recent archaeological survey of the upper Kirmir/Siberis river valley[8]. The discoveries became possible because the monuments in question are unusual in so far as they were not built with mud but employed lime mortar and ashlars or are rock-cut caves. Exceptional as they are, these monuments still allow some general deductions concerning the settlement pattern and history of the valley, as shall be argued in the conclusion of this paper.

5As previous knowledge was scant, the survey had to rely mostly on local information and aimed at establishing an overview of the settlement pattern, archaeology, and art history. This was an urgent desideratum, as the local information was about to be lost, following rapid rural change, the disbandment of traditional structures, and the departure of old inhabitants, all of which used to preserve the historical memory. The survey accompanied this process by documenting Byzantine monuments as they became visible and accessible, for example because a Turkish building was abandoned, but before all would collapse and disappear. This approach profited from repeated visits over the course of several years and flexible low-tech that could be variously applied, depending on whatever archaeological evidence happened to become available on any occasion. Basic manual surveying was best-suited to efficiently document and map the relatively simple rural architecture and agricultural installations.

6The monuments under consideration here mostly date from the Byzantine Period, the last time before the modern era that the valley appears to have been settled intensively. With the exception of a Roman statue base, their dates range from the Early Byzantine heydays (5th/6th centuries) through the Arab invasion period (7th to 9th centuries), prosperous Middle Byzantine times (10th/11th centuries), until the arrival of the Turks (11th/12th centuries). The presentation starts with what seems to have been an Early Byzantine basilica church at Güzelçiftlik in the centre of the valley. Then follow a large Middle Byzantine cave house complex at Mahkemeağcin , a small Middle Byzantine farmstead at Değirmenönü , and a Middle Byzantine cave chapel and possible hermitage at Dikmen (İndere), close to the upper and lower ends of the valley, respectively. Two fortifications against the Arabs and the Turks, Tabanoğlu Kalesi and Alicin Manastırı/Çeltikçi Kalesi, defended the two main entrances to the upper Kirmir/Siberis river valley. Finally, an appendix presents an inscribed statue base for an ancient Galatian ruler.

7The Turkish hamlet of Güzelçiftlik (Turkish: Beautiful Farmstead) lies in the centre of the upper Siberis/Kirmir river valley, about half way down from the source of the river to where it is joined by the Süvari and the Ilhan/Karboğaz streams and the valley takes a different character (Fig. 1). To the west, above Güzelçiftlik, lie two ancient fortifications in close proximity to each other, the Küçük Kale or Small Fort and the Büyük Kale or Large Fort on Dikmen Hill (Fig. 4)[9]. Seen from the south, Dikmen Hill appears as a promontary, where Büyük Kale towers over the river, but the northern side slopes more gently and forms a flat saddle, before the terrain rises steeply again towards Mount Kurumcu. The saddle is occupied by an abandoned Turkish farmstead that appears to overbuild the remains of an Early Byzantine basilica (Fig. 5. 6. 7. 8. 9. 10). A second ruined building, apparently also of Turkish date, lies about 150 m to the northwest and incorporates an inscribed grave stone from the Early Byzantine Period (see below Fig. 14)[10].

8The farmstead is contained in a large rectangular corral of roughly 55 × 70 m. The southeastern part of the corral is flat, the rock appears to have been cut back, and it contains a large, rectangular, rock-cut cistern (Fig. 5). The farm building proper consists of a house and stables, and the rubble masonry of the stables is partly based on, and follows the orientation of, ancient ashlar walls (Fig. 6). To the east, behind the stables, more ashlars form the regular outer face of a rounded wall (Fig. 7). Inside, the stables re-use ancient mullions to support the wooden trusses of their tin-roofs (Fig. 8). The walls of the stables and the farm house contain numerous ancient spolia, ashlars, cornice blocks (Fig. 9), and voussoirs (Fig. 10).

9All ancient parts are made of andesite that was commonly employed in local church buildings (Fig. 11)[11]. This includes a small rectangular box that was found standing outside the rounded wall to the east of the stables (Fig. 12). An ancient lion relief on a rectangular slab that is today exhibited at Güzelçiftlik and said to have been found at the farmstead may have formed the lid of the box (Fig. 13). In this case, the box likely served as ossuary. Lions were common guardians of ancient graves in the region, including ossuaries[12].

10What appears to be an Early Byzantine grave stone is re-used as corner stone of the second Turkish ruin, ca. 150 m to the northwest of the farmstead (Fig. 14). The front of the rectangular slab is engraved with a large, square panel that is subdivided into four smaller squares. An inscription starts above the panel and continues in all four squares. It reads – above the panel – [Καλο]ποδίνο(υ) – in the upper two squares – Πολυ-χρονίου – in the lower two squares – πρ(εσβυτέρου) κε σ-[υ]νβίο[υ]. This may be translated as »(Grave of) Kalopodinos son of Polychronios the priest and his wife«[13].

11The plan of the remaining ashlar walls below and behind the stables may be reconstructed as that of a Christian basilica from the Early Byzantine Period. The apse has a diameter of about 7 m, the southern aisle is ca. 4,5 m wide, and the overall width of a three-aisled basilica would amount to roughly 20 m, with a 10 m wide nave. These are considerable dimensions that compare to some of the largest and most important churches of the region, for exampel the pilgrimage church of St Michael at Germia in southwestern Galatia[14]. The ashlar masonry seems extraordinary, too, and, in this region, is again attested for some of the same exceptional churches[15], whilst most rural buildings appear to have been built from rubble and mud, same as most Turkish houses, which can explain why they have since disintegrated[16]. The large mullions (Fig. 8) may have devided the nave and the aisles, as was customary in the region and beyond, including metropolitan churches[17]. The cornice blocks with dentils (Fig. 9) and fluted voussoirs (Fig. 10) compare to some of the aforsaid major churches in the region and beyond that were likewise built and decorated with similar cornices and fluted voussoirs[18].

12Thus, while the findings appear compatible with a major Early Byzantine basilica, such seems puzzling at first, considering the remote rural location. A settlement with a sizeable Christian community that would justify the building of a large church is not in evidence. Instead, the unusually large and rectangular corral, the flattened bed rock, and the rock-cut cistern, none of which is related to the church, may go back to an earlier, ancient occupation of the site, possibly a rural sanctuary with a rectangular temenos or precinct[19]. This could have been connected to the neighbouring ancient fortifications and would also help to explain the church, as other pagan sanctuaries are known to have later been replaced by churches[20].

13Assuming that the pagan sanctuary was the religious centre of a population that lived dispersed throughout the valley, the same seems conceivable for the Early Byzantine church, too. The narrow valley with potentially rich, but definitively small patches of arable land would seem to lend itself to dispersed settlement only, and the location of the site about halfway down the river appears suitable as a central place for a riverirne community. Pooling the richess gained from the well-watered flat lands and sunny slopes and vinyards could explain the large size and ambitious masonry and stone carving of the church. In addition, the site may have been frequented from further afar by people travelling long distances up and down the valley[21]. A rural location outside any settlement, inside of which burials were normally not allowed in anicent and Early Byzantine times, can also explain the grave monuments in close proximity to the church[22]. The ancient lion ossuary may have continued in use as, or been re-used for, Christian burial[23].

14The Turkish hamlet of Mahkemeağcin borders a patch of flat land in the upper Siberis/Kirmir river valley (Fig. 1)[24]. The Turkish houses are skirting a cliff that borders the valley on the east side and, at one point in the middle of the hamlet, contains a series of artificial caves (Fig. 15)[25]. Some caves were partly hidden behind Turkish buildings or buried in accumulated soil, until the complex was excavated and turned into a local attraction in the 2010s. At some earlier stage, some caves seem to have been used as stables, as is attested by secondary niches that could have served as troughs (Fig. 19. 21. 26. 27. 39). The western and lower cave rooms no. 1 to 7 include several wine presses and appear to have constituted an agricultural complex. At a higher level, the eastern cave rooms no. 8 to 16 would seem to have been sophisticated living quarters, including a chapel and a kitchen. Taken together, the caves formed a major agricultural estate with elite-style housing. Comparable rock-cut estates and houses elsewhere in Central Anatolia indicate a Middle Byzantine date.

15Cave rooms no. 1, 3, and 6 are monumental wine presses of a kind more often attested for Middle Byzantine Anatolia (Fig. 16. 17. 18. 19. 20. 21)[26]. Other such rock-cut wine presses occur in clusters of up to a dozen caves[27], probably because a lot of grapes needed to be processed at the same time, as they would otherwise go bad. Typically, each room ends in an elevated, rock-cut basin that served as treading floor. The front wall of the basin has a central hole for the juice that was collected in a corresponding, rectangular cavity in the floor of the main room (Fig. 19. 21). In front of and flanking the treading floor on both sides, two rock-cut flights of steps enabled the pickers to climb up to the high basin and empty their baskets with grapes inside. At first, the grapes would have been crushed through treading them by foot[28].

16Later, once the pomace did not yield any more juice to treading feet, the precisely cut, rectangular treading floor would be closed with wooden boards and pressed[29]. Pressure would be applied via a long wooden lifting boom that was anchored in a small central niche in the back wall of the treading floor (Fig. 16. 17. 19. 20. 21). The loose boom could rest in a central notch on the front wall of the basin (Fig. 16). The other end of the boom would be screwed down with a wooden spindle and a stone weight that stood in front of the treading floor, in the centre of the main room[30]. Byzantine stone weights with cross decoration are variously attested in the wider region, for example at the museums in Ankara [31], Bolu , and Yozgat (Fig. 22. 23). Turning the spindle required space and can explain why the main room had to be wider than the treading floor. Cave no. 1 is 8,5 m long, 5 m wide, and 5 m high, with the addition of a 2,5 × 4 m² large treading floor. Cave no. 3 measures 10 × 4–5 m², is 4 m high, and has a treading floor of 3,5 × 3,5–4 m².

17The lesser width of the treading floor may also have inspired its lower ceiling, which has no evident functional explanation, but makes sense aesthetically (Fig. 16. 19. 21). More aesthetical consideration is apparent, where blind arcades decorate the walls of the main room and a plain cornice marks the springing of the vaulted ceiling (Fig. 16. 17. 18. 19. 20. 21). Both elements would have required little additional labour as the cave was carved out, and the arcades that subdivide and thus measure the length of the walls may have helped with the correct alignment of the wooden press mechanism, the lifting boom, the spindle, the weight, etc. Holes on the walls next to the treading floor suggest that parts of the press mechanism were affixed there (Fig. 16. 17. 18). Bossages in the same positions may have served a related purpose and indicate that this was planned from the outset (Fig. 19. 20).

18Additional decoration is provided by central crosses on the front wall of the basin and above the treading floor of cave room no. 1 and, in the same position, on the front wall of the basin of cave room no. 3. The crosses in high relief are each inscribed in an arched niche. Comparable cross decoration is attested in other rock-cut press rooms (Fig. 42)[32] as well as on numerous press weights (Fig. 22. 23)[33]. It may have evoked blessing for the difficult and notoriously unpredictable task of wine making, similar to the blessing of the waters and all kinds of related agricultural and sea faring devices that still takes place today in many Orthodox communities on January 6th or 19th, respectively.

19The press rooms no. 1 and 6 are each flanked by, and connected to, a second, similarly large cave room, no. 2 and 5 (Fig. 15). Both these rooms have a high bench on one side and arched niches on the other (Fig. 24. 25). Cave no. 7 forms a third such room with a high bench on one side but has no arched niches and is not connected to a wine press (Fig. 26). The floor of cave no. 7 is lowered by one step along the side without bench. The purpose of all this is not entirely clear, but as some of the rooms are connected to presses and all of them are situated on the same lower level, they likely served related, agricultural purposes, possibly as stables for large animals like cows and horses, for whom the high benches and niches could have served as mangers or troughs[34].

20On the higher level, caves no. 8 to 16 appear to constitute an elite house of the kind that typically included lavish reception and living rooms, a chapel, and a kitchen[35]. Where possible, these houses were often arranged around a central courtyard. Such is not in evidence at Mahkemeağcin, where rooms no. 8 to 16 follow one upon the other along a fairly straight stretch of rock. However, a masonry-built forecourt seems conceivable, for example in front of caves no. 12 to 14 that may have had a rectangular overall façade before the outer parts of no. 13 and 14 collapsed (Fig. 15. 32).

21Caves no. 8 and 9 may have served as reception and living rooms. They have spacious square plans of about 40 m2 each and 5 m high, domed ceilings (Fig. 27. 28. 29). The walls are arcaded, the springing of each dome is marked by cornices, and diagonal ribs that rise from the corners and meet in the apex of the dome pretend to ribbed vaulting[36]. Caves no. 10 and 11 further along the rock have partly collapsed but appear to have been longitudinal and of middling size (Fig. 15). The interior walls preserve the usual carved decoration with various arcades. The back wall of no. 11 is centred on a small, window-like twin arcade in a high position and flanked by two lower cross medallions (Fig. 30)[37]. The north wall of cave no. 12 pretends to a gabled tympanum, although the room was in fact barrel vaulted (Fig. 31).

22Cave no. 12 is of similarly large dimensions as rooms no. 8 and 9, likewise with arcaded walls and cornice, but with a barrel vault, a long entrance passage, and with a special, triple-arcade and a central cross on the opposite, northern wall (Fig. 32. 33). The triple-arcade and cross are flanked by a small niche on the right or east side, which could conceivably have served for the prothesis, i.e. the liturgy of preparation, and thus identify the room as chapel, only that the neighbouring cave no. 13 appears to be a more likely candidate for this function, whilst carved crosses are omnipresent, cf. caves no. 1, 10, and 16.

23Room no. 13 has lost its south-eastern outer corner but is otherwise well preserved. It is 4 m wide, 5 m long, over 5 m high, and barrel-vaulted, which results in impressively steep proportions (Fig. 34. 35. 36. 37). The walls stand out for especially deep and sophisticated, twice recessed arcades. The long side walls to the north and south each have a low arcade above a tall one, as if the arcade of a basilica was followed by a row of clerestory windows. The shorter east wall has one taller, overarching arcade that reaches up to the usual horizontal cornice at the springing of the vault. The two inner of the four arcades each contain a low arch that could equal an apse window or a templon door. Above these lower arches, about half-way up the main arcades, follows a row of imposts, suggesting some sort of horizontal division. Roughly two thirds up the main arcades runs a more pronounced horizontal partition in the form of an arcaded frieze. It may be compared to arcaded Middle Byzantine templon epistyles[38] or to arcaded friezes on the outer facades of Middle Byzantine churches[39]. Above the frieze, the remaining part of the arcade is filled with what looks like a row of arched windows with imposts at the springing of the arches or the second storey of a two-tier templon of the kind that actually exists in some rock-cut churches of Cappadocia[40].

24Above the arcade and the cornice that marks the springing of the barrel vault, the east wall of cave no. 13 culminates in a tympanum with a large central cross inscribed in a wide arch and flanked by two smaller arches (Fig. 35. 36). The tympanum cross would appear to equal the crosses that commonly decorated the apse domes of Anatolian churches. Such apse crosses had been standard in Early Byzantine Asia Minor and continued to be depicted as main focus of many provincial Anatolian churches throughout the Middle Byzantine Period[41]. A second cross without an arched frame occupies the tympanum of the west wall (Fig. 37). All in all, the sumptuous decoration, which is reminiscent of a basilica, richest on the east wall, and culminates in the crosses that hover high above the steeply proportioned room, combine to make cave no. 13 seem the most likely candidate for the chapel that was usually contained in such large house complexes.

25Room no. 14 is 8 m wide and may once have been square and thus very big, but its outer face has collapsed (Fig. 15). The cave has the usual arcaded walls with a cornice on top but stands out for a pyramidal ceiling (Fig. 38). Such ceilings are well-known from other rock-cut house complexes and typical features of kitchens[42]. The pyramids culminated in chimneys and thus served as vents for the smoke of the hearth fires. Where the context is known, it indicates a Middle Byzantine date. A central doorway in the back wall of the kitchen leads into a smaller, windowless cave that may have served as larder (Fig. 39).

26Caves no. 15 and 16 formed two more large rooms, but both employ less ornamental carving as for example the chapel no. 13. Room no. 15 has mostly collapsed and cannot be reconstructed any more. The better preserved cave no. 16 has two arched doorways, a relatively low barrel vault, and a large cross off the centre of the east wall (Fig. 40).

27Overall, the spacious living quarters with opulent reception room, chapel, and kitchen correspond to what have been identified as Middle Byzantine elite residences elsewhere in Central Anatolia[43]. The large agricultural complex with numerous wine presses appears to confirm this and suggests sizable landholdings. The necessary labour force would presumably have lived in lesser mud buildings that have not survived[44]. The combination of extensive residential features with a substantial church and the large wine production facility indicate that Mahkemeağcin was the largest Middle Byzantine establishment of the Kirmir river valley.

28It is unlikely to be a coincidence that a substantial quantity of ancient ashlar masonry as well as an inscription have also been found here. Several Turkish houses employ ashlars that appear to be re-used ancient building material (Fig. 41). Normally, such village houses were built with rubble or mud bricks, not with ashlars. As to the inscription, it is almost certainly of Early Imperial date and commemorates the Tolistobogian leader Ortiagon (Fig. 80 and appendix). The inscription is carved on a statue base, and the monument must have been displayed nearby, probably in a residence belonging to Ortiagon’s family. The upper Kirmir valley was likely at the heart of the family’s rural estates from the beginning of the Celtic occupation of Galatia in the 3rd century B.C. at least until the 2nd century A.D. It is plausible that distant descendants of Ortiagon were still the most powerful regional land-owners in the Middle Byzantine Period. However, it may well have been Ortiagon’s landholdings that formed the premise for the cave house complex under consideration here.

29The 7th-century Life of St Theodore of Sykeon provides evidence for memories of the Galatian past and that the ethnic identity of the Galatians remained a factor to be reckoned with. When Theodore was called in to perform an exorcism at a village called Bouzaia, north of the Kirmir valley in the territory of the Bithynian town of Creteia (modern Gerede ), the demons who had possessed the villagers railed with Theodore and screamed back at the saint: »Why have you come here, iron-eater [a reference to Theodore’s gruelling training and preparation to become a holy man]? Why have you left Galatia for our district of Gordiane? We know why you have come, but we will not listen to you as the spirits of Galatia do. We are made of sterner stuff than they, and cannot easily be controlled«[45].

30Continuity of settlement and land holding will have been an important factor in keeping the cultural memory and identity of the Galatians alive throughout antiquity[46]. When the fortifications of the metropolis at Ankara were constructed by emperor Michael III in A.D. 859, one of the two commemorative epigrams carved on the walls beside the main gate to the citadel ended with an invocation of the city: Ἄγκυρα τερπνή, παμφαεστάτη πόλις, | πάσης Γαλατῶν πατρίδος σὺ λαπρότις. »Fortunate Ancyra, all-gleaming city, you, the splendour of the whole country of the Galatians!«[47]. The discoveries at Mahkemeağcin, extending chronologically over almost a millennium from the domains of a Galatian chieftain to the wine producing estates of a Middle Byzantine province, provide evidence that this continuity can also be traced in the countryside. At Mahkemeağcin, the continuous settlement tradition finds an explanation in the well-watered flat land that is as convenient for agriculture as it is rare in the valley and beyond and would thus always have been farmed.

31About 2 km downriver from Mahkemeağcin lies the Turkish hamlet of Değirmenönü (Turkish: In Front of the Mill) and, in its neighbourhood, what appears to be a rock-cut farmstead of the Middle Byzantine Period (Fig. 1). On lower ground, close to the agricultural land, a single wine press of the same kind as at Mahkemeağcin has been identified. Two more caves of undetermined form and function were reportedly destroyed in the 2000s. The preserved cave is half filled with rubble (Fig. 42. 43). It received light through an arched doorway and two windows in the east wall. The main room is 7,5 m long and about 5 m wide. The treading floor to the west is 2 m deep, 3,5 m wide, and has the usual small niche for the lifting boom in the centre of the back wall. The lower barrel vault of the treading floor is crowned by a central cross, the equivalent of the cross inside the treading floor no. 1 at Mahkemeağcin.

32Ca. 200 m further uphill, in a location that offers a view across the valley, lie rock-cut living quarters with a kitchen and a dovecot (Fig. 44. 45). The façade has collapsed entirely, but the kitchen is easily recognisable by its pyramidal ceiling and smoke vent at the left end of the small complex. The kitchen is only 4 m wide, half as wide as at Mahkemeağcin, which reflects the lesser size of the farmstead at Değirmenönü. Next to the kitchen follows a bigger room of at least 25 m² that may have been for living. The ceiling is about 3 m high and thus considerably lower than at Mahkemeağcin. The next room after that is only preserved in the shape of three small niches. On the right side, the complex is rounded off with a dovecote that would have provided the farmstead with manure (Fig. 46)[48].

33The dovecote and the wine press suggest that the small unit above Değirmenönü was contemporary to the larger complex at Mahkemeağcin, but constituted an independent farmstead in its own right. To judge by the number of wine presses, the landholding of Mahkemeağcin was several times as large as that of Değirmenönü. It seems remarkable that the two could co-exist in close proximity to each other, apparently without the one outshadowing the other. A smaller rock-cut complex comparable to Değirmenönü is known from Kırbaşı, formerly Gelegra, to the south of the lower Kirmir/Siberis river valley (Fig. 1). It combines what appear to be living quarters with at least one splendidly carved and vaulted (reception-?) room (Fig. 47. 48) and – at a distance of 25 m – one wine press (Fig. 49. 50)[49].

34The Turkish village of Dikmen flanks the İndere (Turkish: Valley of the Cave) stream, a small southern tributary of the lower Siberis/Kirmir river, beyond the area under consideration here (Fig. 1). However, some caves to the north and above Dikmen, just south of where the Siberis/Kirmir is joined by the Süvari and the Ilhan/Karboğaz, shall be considered in this paper[50], if only because they lend themselves to comparison with the caves at Mahkemeağcin and Değirmenönü. The caves at Dikmen appear to have been reclusive and in comparison confirm that those at Mahkemeağcin and Değirmenönü more likely served as farmsteads and elite residence. Otherwise, Dikmen is known for a hüyük or settlement mound further to the south, to be identified with the ancient city of Anastasiupolis[51], and for a Hellenistic/Galatian hilltop fortification[52], but the caves under consideration here have no obvious connection to either. The hüyük of Anastasiupolis lies at a distance of almost 10 km, as the crow flies, with various watersheds to be passed on the way. Only the fortified ancient hilltop is visible from the caves (Fig. 51), but the two sites are separated by an impassable ridge and also by a chronological gap of over a thousand years.

35The caves have an isolated hillside location above a steep and stony slope, in a barren area with neither tree nor brush, let alone arable soil. The cave rooms follow one upon the other along a rocky ridge of soft sand stone. Each cave is relatively small, with few distinguishing marks apart from occasional niches and windows (Fig. 52). The extend of the complex cannot be ascertained, because some entrances are today filled with accumulated rubble. However, one cave stands out through size and sophistication and can be identified as a cross-domed chapel. Its western façade has collapsed, but the north and east walls of a narthex are still standing and preserve traces of what appear to have been arcosolium graves arranged in two tiers, one above the other, on the north wall as well as on the southern part of the east wall, to the south of the door to the naos (Fig. 53).

36The steeply proportioned naos is cruciform, about 5 m long, 4 m wide, and 5 m high (Fig. 54). Each cross arm terminates in a conch, but the eastern arm is noticeably longer, has a prothesis niche on the north side[53], and only the eastern apse is polygonal, which makes for a complex pattern on the vault above (Fig. 55). The springing of the arches and vaulting is marked with a cornice (Fig. 56). The central dome appears to rest on a tambour. The steep proportions as well as details such as the prothesis niche and the tambour all point to a Middle Byzantine date[54].

37All in all, the cave complex appears unsuitable for agricultural purposes and related housing, but the small rooms could have conceivably accommodated monks, who may have sought out the remote and barren location because it lent itself to an ascetic way of life. Already The Life of Theodore of Sykeon knows of caves as dwelling places of hermits[55]. The sophisticated chapel may have been added by a donor who would have been buried in the narthex and prayed for by the monks[56].

38The fort of Tabanoğlu (also Tabanlıoğlu or Adaören ) occupies a promontory close to the southern entrance of the upper Kirmir/Siberis river valley, which it controls (Fig. 1). The modern name derives from a nearby Turkish farmstead. The fort was first built in Hellenistic times and may be identical with Peium, the treasury of the Galatian king Deiotarus, who ruled in the first half of the 1st century B.C.[57]. Among Galatian fortifications, Tabanoğlu Kalesi stands out for size (ca. 300 m × 50 m = 1,5 hectare) and quality of workmanship. Over half a millennium later, Peium was still known to St Theodore of Sykeon in the 7th century A.D.[58], and Tabanoğlu Kalesi appears to have continued in use throughout the Byzantine Period[59].

39The fort occupies a highly defensible promontory that is surrounded on three sides by cliffs and the meandering river, with only a narrow bottleneck on the east side for access (Fig. 57). The Hellenistic fortifications cut across and defend the bottleneck with several polygonal towers (Fig. 58. 59). They were built with dry ashlar masonry and included an arched gateway that was first renovated and later blocked in the Byzantine Period (Fig. 60). The earlier Byzantine renovation included the addition of a new, inner wall face that doubled the strength of the Hellenistic gate and also required a second, inner arch, most of which has since collapsed again. Later, the gate was blocked from the outside, where the Hellenistic gate and wall were completely hidden by the addition of a massive new face. At this later stage, the only remaining access to the fort appears to have been a smaller postern or sally port-style opening on the north side of tower B (cf. Fig. 73).

40Disuse during the Roman Period will have led to damage and partial collapse, and when the fort was revived in the Byzantine Period, this required extensive repairs and rebuilding. The pentagonal tower B, which appears to have been part of the original, Hellenistic layout, was rebuilt almost from scratch. Only the lowest layers consist of what may partly be original Hellenistic masonry, to judge by drafting on the southern corner that was discontinued in the Byzantine layers above (Fig. 61). However, even the drafted Hellenistic corner stones have been re-set in Byzantine lime mortar. The Byzantine phase includes two outside buttresses that would have lent additional strength to the Hellenistic foundations, as well as mortared, pseudo-isodomic masonry above.

41The Byzantine masonry of tower B re-uses ancient ashlars and is of varying quality, depending on the size and quality of the re-used parts. The lower, statically more important, and more visible part of the Byzantine masonry re-arranges the Hellenistic ashlars in alternating courses, either horizontally or vertically. The tall horizontal courses alternate further between long stretchers and short headers that reach backwards into the mortared rubble core and thus provide stability. The upper part of tower B is less carefully built with smaller blocks and yet smaller stones as occasional fillers. It also includes a central embrasure on each of the three frontal sides as well as a re-used Roman door stone and a finely carved Christogram on the southern side (Fig. 62).

42The pentagonal tower C seems to copy tower B, but may not be of Hellenistic origin, as all visible masonry is entirely Byzantine (Fig. 63). The façade is less well preserved, and what remains looks like the upper parts of tower B, i.e. more or less regular courses of mortared ashlar masonry with smaller fillers, but without the showy sophistication of the pseudo-isodomic alternations on the lower part of tower B. The building material is again entirely re-used, including odd fragments of brick as well as a Roman door stone and a rough cross medallion. The latter is placed in a low position on the frontal side of the tower, where it would have been visible from the approach to the postern in the north side of tower B. The short curtain wall between towers B and C that flanks the postern is also a Byzantine lime mortar construction with an outward facing of largish ashlars.

43The corner tower D is once more of Hellenistic origin and preserves half a dozen courses of ancient ashlar masonry where it is based on the bedrock (Fig. 64). An additional dozen courses of the same masonry, but less finely executed, continue eastwards to form a terrace wall along the northern edge of the approach to the fort. The upper part of tower D is similar to those of towers B and C, with moderately regular courses of re-used ashlars and smaller fillers. Additional stability is provided by a band of four to five brick courses. In a later, third phase, the outer corner of tower D was strengthened with a roundish buttress of rubble masonry.

44What appears to be the same, later, third phase is in evidence at various other parts of the fortifications, always employing pure rubble masonry only. Thus, the curtain wall between towers B and C appears to have been strengthened with a rubble buttress on the inner, western side (Fig. 59). South of tower B, it is the outer, eastern side in front of the former, Hellenistic gateway that now consists of an additional rubble curtain (Fig. 60). This rubble masonry continues southwards, envelopes tower A, forms two outer, semi-circular towers at the southeast corner (Fig. 64), and continues south of tower A up to its southwestern corner (Fig. 65).

45Moreover, the same rubble masonry was also employed for the building of an outer wall or proteichisma. This outer wall takes its northern starting point from the eastern end of the Hellenistic terrace to the east of tower D, from where it runs southwards in a straight line, all along and parallel to the main wall, at about 5 m distance to the east (Fig. 59). To the east of tower B and in front of the postern, the outer wall had a semi-circular tower. At the southeast corner, southeast of tower A, the outer wall climbs down the steep slope below the main fortifications so as to surround them also on the south side and to continue westwards along and above the cliff of the defensive plateau (Fig. 64). The rubble wall continues westwards beyond the bottleneck, where it becomes the only wall and forms round towers of its own (Fig. 66), thus fortifying the defensive plateau also on the steep southern side, above the river.

46To conclude, the Byzantine re-fortification of Tabanoğlu Kalesi appears to have happened in two distinct phases. The first Byzantine phase equalled a renovation of the Hellenistic fortification and re-employed essentially the same building material as well as some additional spoils, for example the Roman door stones. The most outstanding characteristic of this phase is the pseudo-isodomic masonry on tower B. It is not attested for the Hellenistic fort, but appears to follow a fashion in Late Antique and Early Byzantine fortifications of Asia Minor, for example at Pergamum in the 3rd century, at Aphrodisias in the late 4th century, at Miletus in the 7th or 8th century, and at Ankyra/Ankara in the 9th century[60]. At Tabanoğlu Kalesi, a refortification in this style may have been connected to the Arab invasions of Anatolia in the 7th to 9th centuries[61], when controlling the passage through, and the resources of, the Kirmir/Siberis river valley (water, food, and fodder) would have been of strategic importance[62].

47The third building phase with pure rubble masonry strengthened the earlier defences through the addition of substantial volume, in particular to the northwest of tower B and to the southeast of the same as well as all around the outside of tower A. This also resulted in the closure of the arched gate and reduced access to the small postern on the other, eastern side of tower B. The same rubble masonry was also employed to add the outer wall or proteichisma and thus considerable sophistication to the defences, and to fortify the southern cliff, possibly for the first time. This general re-enforcement suggests a fundamental change of the strategic situation. Outer walls were not normally employed against the annual Arab invasions of the 7th to 9th centuries but would seem to have been aimed at a less ephemeral enemy that was expected to stay and lay siege. This new kind of enemy were probably the Turks that arrived and stayed in the region from the 11th century onwards[63]. Other Anatolian defences against the Turks likewise employed outer walls[64] and were built from small rubble rather than large spoils[65].

48Çeltikçi, a Turkish village in the middle of the upper Siberis/Kirmir river valley, lies at the confluence of the Pazar or Çeltikçi river (Fig. 1). The Pazar river valley formed the only convenient connection to the northern highlands in an otherwise impassable terrain and in antiquity was traversed by a major road from Ankyra to the northwest[66]. About 5 km northwest of Çeltikçi, the Pazar river valley becomes narrow and is flanked by steep cliffs and mountains on both sides (Fig. 67). Three fortifications took advantage of this geography and jointly controlled the road and entrance to the Siberis/Kirmir valley. The smallest and lowest fortification is nestled in cliffs on the northern side of the Pazer river, immediately above the ancient road, and is today known by the misleading name of Alicin Manastırı (Fig. 68).

49Alicin Manastırı consists of no more than one long wall that fortifies a narrow, ledge-like cavity with no access from the sides or above (Fig. 69). The interior could only be reached with ladders from below and was practically unassailable but contained little space and no monastic infrastructure. The wall was built in two phases. A lower section without openings employs layered ashlar masonry with two bands of bricks that may be assigned to the Byzantine Period[67]. The upper section with up to four rows of windows is built without bricks, but with wooden tie beams on the outside of the masonry, which in Anatolia is typical of Turkish constructions[68]. Alicin Manastırı would have served as immediate refuge or redoubt for a small detachment of soldiers that guarded the valley and controlled the road, but on its own, the fortification and its garrison would have been too small to prevent any numerous and organized force like an Arab or a Turkish unit from passing by.

50However, Alicin Manastırı was in sight of two larger fortifications on the other side of the Pazar river and valley (Fig. 70). The latter are jointly known as Çeltikçi Kalesi with reference to the Turkish village on the other, southern side of the mountain (Fig. 67), but there can be no doubt that the fortifications were above all meant to guard the northern passage through the Pazar river valley. The mountain is today called Asartepe, reaches 1300 m, and offers distant views in northern, eastern, and southern directions. The top is occupied by a large fort that is known as Asar Kale. In addition, two lower, northern twin peaks immediately above the Pazar river valley are also fortified and referred to as Alt Kale[69].

51The taller of the two Alt Kale peaks forms a rocky outcrop that is topped by a round tower (Fig. 71. 72). The tower’s masonry consists of ashlars, some of which are re-used ancient blocks with chisel-draft and that alternate with bands of bricks, each with five courses. Additional bricks are sometimes employed as vertical spacers between the ashlars, thus forming a rudimentary version of what is known as cloisonné-masonry. In the context of Byzantine fortifications in Anatolia, this masonry dates the round tower of Alt Kale to the later Byzantine Period and the defence works aimed against the arriving Turks (cf. Fig. 73)[70].

52In addition to the rocky outcrop and round tower, a wall that is built with the same banded ashlar masonry and layers of bricks runs ca. 200 m down the eastern slope (Fig. 71). A second, like wall continues for about 300 m down the opposite, north-western slope, where it reaches and fortifies a second, lower peak directly above the Pazar river valley. The various walls did not add up to a defendable circuit but must have looked impressive from the road below. They would have served as a bulwark for troops that, whilst stationed at the fort on Asartepe, had to control the road in the valley below and, if necessary, provide relief for the small detachment at Alicin Manastırı.

53The peak of Asartepe is large enough for more extensive fortifications that enclose a rectangle of roughly 8500 m² (Fig. 74). On three sides, the fortifications take advantage of steep drops, only the north-eastern side slopes down more gently in the direction of Alt Kale. Each corner was defended by a tower, and the eastern corner may have had a pair of them, possibly enclosing a gate. Three additional towers were distributed across the long north-western and south-eastern sides (Fig. 74). Layout and building material in the form of large ashlars are conceivably of Hellenistic/Galatian origin[71], which would have helped the Byzantine re-enforcement of the site. However, all standing remains employ lime mortar and are thus clearly of Byzantine workmanship.

54The only structure standing above ground level is tower A on the east side (Fig. 75). The tower is almost completely intact, including some of its vaulting. It is polygonal in the Hellenistic/Galatian tradition and, at least on the outside, built with regular courses of large ashlars (Fig. 76. 77), suggesting that the Byzantines may have rebuilt an ancient tower with the original building material like at Tabanoğlu. However, the vaulting of tower A at Asartepe appears to constitute a separate building phase and employs smaller rubble as well as bands of single bricks (Fig. 78. 79).

55It seems impossible to ascertain whether the vaulting of tower A looks different only because it was built a year later or so, after the walls had dried out and hardened, and with smaller rubble and bricks rather than larger stones that would be heavier and more difficult to handle. Alternatively, the massive vaulting could be a later addition that replaced a wooden original or some more delicate earlier vault. Such was sometimes done to upgrade an older tower for defence against more advanced siege technology. In the case of Asar Kale, this might also explain why only tower A survives above ground, if it was the only one that was upgraded with massive vaulting. Such a massively vaulted structure would withstand the elements better and longer. Tower A was maybe the most obvious candidate for such an upgrade, as it overlooked the least defensive approach from the direction of Alt Kale (Fig. 75).

56If so, the upgrade of tower A most likely occurred in the later Byzantine Period, when other earlier towers elsewhere in Anatolia were similarly upgraded with massive vaulting[72]. The round tower at Alt Kale confirms that Çeltikçi Kalesi was defended against the Turks. However, the walls of tower A and the way that they appear to have been rebuilt on a polygonal Hellenistic/Galatian plan and with ancient building material, compare to the earlier Byzantine phase at Tabanoğlu, probably from the period of the Arab invasions in the 7th to 9th centuries. In conclusion, it may be suggested that Asar Kale, too, had ancient origins, was first revived against the Arabs, and renovated a second time against the Turks.

57The ancient chronologies of Alt Kale and Alicin Manastırı, if any, remain to be established, but in the Byzantine Period all three fortifications combined to control and defend the road and passage through the Pazar river valley. The combination was necessary, because each fortification on its own would not have served the purpose; Asar Kale was too far removed to effectively guard the valley floor, whilst Alt Kale and Alicin Manastırı were too small to hold out against invaders. But with the backing of the larger fort and garrison on Asartepe, the smaller fortifications and detachments that flanked the valley and road on both sides could not be locked in and ignored. Any invader would have to deploy considerable manpower, time and effort and practically beseech and lock in, if not take, Asar Kale in order to gain save passage along the Pazer river.

58As a most general conclusion, the various monuments would seem to confirm that the upper Kirmir/Siberis river valley was prosperous and populous throughout the Byzantine Period. In particular, the relatively large and well-built Early Byzantine basilica at Güzelçiftlik, Middle Byzantine cave house complex at Mahkemeağcin, as well as Early through Late Byzantine fortifications at Tabanoğlu Kalesi and Çeltikçi Kalesi all imply considerable means and sizeable workforces.

59A second general conclusion concerns the settlement pattern, which appears to have been dispersed at all times. For the Early Byzantine Period, this is suggested by the basilica at Güzelçiftlik, outside any known settlement and the only church of its kind along the whole valley. It may have served as a central monument of a riverine society that did not form any single agglomeration large enough for such an ambitious church building, but that was willing to pool resources in order to attain grandiose monumentality nevertheless. The Middle Byzantine chapels of the Inönü Caves and the rock-cut house complex at Mahkemeağcin appear minute in comparison and attest to the small size of each settlement.

60Thirdly, the contrast between the large Early Byzantine basilica and the small Middle Byzantine chapels indicates that the social order of the valley community changed over time. The Early Byzantine society that seems to have pooled resources for the building of Güzelçiftlik basilica was likely made up mostly of self-employed farmers that were relatively equal in their possessions. The Life of Theodore of Sykeon implies such a comparatively egalitarian society[73], and evidence from elsewhere in Byzantium points in the same general direction[74]. Each Early Byzantine farmstead would thus have been smaller, with fewer and less elaborate buildings, than the Middle Byzantine cave house complex at Mahkemeağcin, which can explain why the earlier farms have not been traced.

61In contrast, the large rock-cut installations at Mahkemeağcin that will have required considerable labour to build and run attest to the Middle Byzantine ascendance of a rich landowner, who would have towered high above most other inhabitants of the valley[75]. Such elite estates with sophisticated reception rooms and chapels of their own are well-attested throughout Anatolia[76] and correspond to a general change in Byzantine society that was increasingly dominated by land-owning aristocrats[77]. In the same scenario, the Middle Byzantine funerary chapel at Dikmen (İndere) appears as a typical aristocratic status symbol[78]. A smaller farmstead like the one at Değirmenönü may have owed its survival at least in part to the limitations of the narrow river valley, where the amount of arable land is mostly too small for the establishment of large estates. In addition, the fertile soil and ample water may have favoured labour-intensive small-holdings, whilst large estates were more often invested in extensive livestock farming[79].

62Rural fortifications like Tabanoğlu Kalesi and Çeltikçi Kalesi were normally built and maintained by the Byzantine army and not necessarily related to any civilian settlements[80]. The one objective did not necessarily exclude the other, though, and whilst the fortifications probably had the primary purposes of controlling passage through the valley and protecting its water, food, and fodder – essential resources for any military on the otherwise arid high plateau, they may also occasionally have sheltered civilians, first from Arab and later from Turkish raids. In addition, aristocrats with leading positions in the army may conceivably have taken particular interest in fortifications that would protect their estates[81].

63Overall, the upper Kirmir/Siberis river valley appears to combine a number of unusual features with an otherwise typical Byzantine settlement history. Exceptional is the geographical setting that, whilst allowing for small-scale prosperity, a limited strategic importance, and rock-cut architecture, also seems to have implied a diminutive dispersed settlement pattern. Typical are the architectural features of the various monuments and how they reflect a changing society, from relatively egalitarian with large communal churches in the Early Byzantine Period to aristocratic dominance with small private chapels later on.

64An ancient marble found at Mahkemeağcin (Fig. 1)[82] can be identified as a statue base and bears an inscription naming the best-known Galatian leader of the 2nd century B.C., Ortiagon, tetrarch of the Tolistobogii. The letters are carved on the front face of a rectangular block that consists of grey-veined white marble (Fig. 80). The block is 16 cm high, 71 cm wide, and 72 cm deep, with a plain fillet moulding, about 1 cm high, around the top edge. It was designed to stand on a pedestal or in the niche of a built structure, perhaps an exedra, and has a rectangular inset on top to accommodate the carved stone base of a statue, which will have been marble, not bronze. The block has a triangular chip at the back, a broken corner at the front left, which has removed the first letter of line 2, and a larger damaged area at the front, which has destroyed the middle of line 1 and part of the last word in line 2 of the inscription. The letters are a little more than 2 cm high.

Ὀρτιάγων Ο[. .]. .[- - - c. 11 - - -] υἱός

[Γ]αλατῶν Τολιστοβωγίων τ̣ε̣τ̣ράρχης.

Ortiagon, son of O. . . , tetrarch of the Galatian Tolistobogians.

65In the first line after the second omicron, two letters were removed by the break and are followed by the bottom of a vertical hasta and another letter trace. This would be compatible with reading »Ὀ[ρτ]ιά[γοντος]«, but this restoration leaves a gap of about five letters to fill before »υἱός«. Perhaps we should conjecture a patronymic that was etymologically related to Ortiagon and incorporated other elements. Galatian Celtic nomenclature included many lengthy compound forms (Rossolittanos, Adiatomaros, Comboiomaros, Kommontorios, Eposognatos), but a name of approximately sixteen letters would be longer than any of these.

66The tetrarch Ortiagon was active in the early 2nd century B.C., but the inscription almost certainly dates to the Imperial Period. The style of filiation, »Ὀρτιάγων Ο . . . υἱός«, mirrors the Latin formula, e.g. C. Iulius C. f., and shows Roman influence, which would not have been present before the second half of the 1st century B.C.[83]. The letter forms also suggest a Roman Imperial date. They are finely carved and elegant, including four-barred sigma, alpha with a broken cross-bar, beta with the bottom larger than the top, and eta whose cross-bar, with pronounced apices, is detached from the uprights. The most distinctive feature of the letters are the overlapping ends of the upright and transverse strokes of alpha (both at the apex and the cross-bar), gamma, lambda, and sigma. The transverse cut of nu does not reach the tip of the uprights.

67The aristocracy of the Galatian Tolistobogian tribe cultivated a Greek epigraphic tradition. The earliest example is the only Hellenistic inscription from Galatia, the grave monument for the younger Deiotarus from his burial mound at Karalar, dating to 43 B.C., carefully but less finely carved than the Ortiagon text, with broken-bar alpha and a distinctive form for pi, but otherwise using simple unadorned letters without apices[84]. The mid-1st-century A.D. text for a Tolistobogian priest of the imperial cult found at Sinanlı has broken-bar alpha and eta with joining cross-bar[85]. Both seem to be earlier than the new Ortiagon inscription, whose letter styles and well-spaced lay-out resemble those on inscribed statue bases in Ancyra of the Hadrianic–Antonine Period: for C. Iulius Saturninus, C. Iulius Scapula, and P. Pomponius Secundianus, provincial governors of the A.D. 130s or 140s[86], or for members of the Ancyran aristocracy at this period, some of whom were descended themselves from Tetrarchic families[87]. The Ortiagon statue base can thus be tentatively dated to the Hadrianic or Early Antonine Period, but the 1st century A.D. should not be excluded.

68Ortiagon was a significant figure in Galatian history. He was the leader of the Tolistobogians and in 189 B.C. led his tribe and the fighting men of the east Galatian Trocmians in a losing battle on Mount Olympus against the Roman forces under Cn. Manlius Vulso[88]. After the defeat, Polybius reports that Ortiagon attempted to unite all the Galatians under his sole command and testifies to his qualities: »he was munificent and magnanimous, his conversation was both charming and intelligent, and, what is most important among Gauls, he was brave and skilled in the art of war«[89]. In 184 B.C. Ortiagon joined forces with Prusias, king of Bithynia, in resisting Attalid attempts to control Western Asia Minor. but the alliance was defeated by the forces of Eumenes II at an unlocated site in Bithynia called Mount Lypedron[90]. The Attalid victory was celebrated in thank-offerings to Eumenes recorded in an inscription from Telmessos (Fethiye) in Lycia[91].

69Livy’s account of the battle on Mount Olympus, drawn from Polybius, reports that Ortiagon escaped from the battlefield[92], but his wife Chiomara was taken captive. Chiomara is the subject of a famous story that she was violated by her captor, a Roman centurion, negotiated her own ransom, contrived that the Roman soldier was beheaded as the ransom gold was being paid, and presented his head to her husband when she returned to him[93]. Polybius also reports that he, Polybius, had met Chiomara at Sardis , and admired her temperament and intelligence (»τό τε φρόνημα καὶ τὴν σύνεσιν«). She was surely his main informant both about Ortiagon and about herself.

70According to Suda, Ortiagon had a son called Paidopolites, who held the office of dikastes [94] , a detail that appears to be authentic, as Strabo’s description of the constitution of the three Galatian tribes indicates that each of them was divided into four sections called tetrarchies, each with a tetrarch, a dikast, a military commander and two subordinate commanders[95].

71The inscription from Mahkemeağacin is the only ancient source to indicate that Ortiagon held the title of tetrarch. Strabo implies that the office of tetrarch dated back to the time of the Celtic invasion of Asia Minor in 278 B.C.[96]. Tetrarchic ancestry was an important calling card for the Galatian aristocracy of the Roman Period, when leading citizens of Ancyra, including C. Iulius Severus (I. Ankara no. 72–74. 76), Ti. Claudius Bocchus (I. Ankara no. 82), and Ti. Claudius Gentianus (I. Ankara no. 84) gave prominence to their descent from earlier Galatian chieftains[97]. The monuments at Ancyra referring to tetrarchs date to the first half of the 2nd century A.D., and provide an important contextual reference point for the Ortiagon inscription. It is likely that his commemorative statue was erected at a period when the leading figures of Ancyran society were reviving memories of their tribal forefathers.

72The obvious explanation for the find spot of the new text is that it marks the location of Ortiagon’s family’s former tribal residence. The upper valley of the Kirmir Çay was in the north-west part of Tolistobogian territory. Mahkemeağacin lay north of the residences associated with the Tolistobogian Deiotarus, the most important Galatian leader of the 1st century B.C., respectively at Blucium/Karalar and Peium/Tabanoğlu Kale (Fig. 1). No Galatian fortress has yet been located in the immediate neighbourhood of Mahkemeağacin, but the re-used ashlar masonry in the village buildings (Fig. 41), which might have come from a Hellenistic fortress or from a later prestige building owned by the family in the Imperial Period. The statue of the family’s most famous forebear must have been placed on prominent display in the property.

The upper Siberis/Kirmir river valley is well-watered and fertile, but also narrow and small. It was thus always farmed and settled, but never on a large scale. This paper reports half a dozen monuments from the valley, including an inscribed statue base for an ancient Galatian ruler from Mahkemeağcin, an Early Byzantine basilica church at Güzelçiftlik, a large Middle Byzantine cave house complex at Mahkemeağcin, a small Middle Byzantine farmstead at Değirmenönü, a Middle Byzantine cave chapel and possible hermitage at Dikmen (İndere), and two fortifications against the Arabs and the Turks, Tabanoğlu Kalesi and Alicin Manastırı/Çeltikçi Kalesi. Overall, the upper Kirmir/Siberis river valley appears to combine a number of unusual features with an otherwise typical Byzantine settlement history. Exceptional is the geographical setting that, whilst allowing for small-scale prosperity, regional strategic importance, and rock-cut architecture, also seems to have implied a diminutive dispersed settlement pattern. Typical are the architectural features of the various monuments and how they reflect a changing society, from relatively egalitarian with large communal churches in the Early Byzantine Period to aristocratic dominance with small private chapels later on.

Das obere Siberis/Kirmir-Flusstal ist gut bewässert und fruchtbar, aber auch eng und klein. Es wurde folglich immer bewirtschaftet und besiedelt, aber nie im großen Stil. Dieser Aufsatz stellt ein halbes Dutzend Denkmäler aus dem Tal vor, darunter eine beschriftete Statuenbasis für einen antiken galatischen Herrscher aus Mahkemeağcin, eine frühbyzantinische Basilika in Güzelçiftlik, eine prächtige mittelbyzantinische Höhlensiedlung in Mahkemeağcin, ein kleines mittelbyzantinisches Gehöft in Değirmenönü, eine mittelbyzantinische Felskapelle und mögliche Einsiedelei in Dikmen (İndere) und zwei Befestigungen gegen die Araber und die Türken, Tabanoğlu Kalesi und Alicin Manastırı/Çeltikçi Kalesi. Insgesamt scheint das obere Kirmir/Siberis-Tal einige Besonderheiten mit einer ansonsten typischen byzantinischen Siedlungsgeschichte zu vereinen. Außergewöhnlich ist die geographische Lage, aus der sich einerseits begrenzter Wohlstand, regionale strategische Bedeutung und Felsarchitektur ergaben, andererseits anscheinend aber auch eine kleinteilig verstreute Siedlungsweise. Typisch sind die architektonischen Merkmale der verschiedenen Denkmäler und wie sie eine sich verändernde Gesellschaft widerspiegeln, von relativ egalitär mit großen Gemeindekirchen in frühbyzantinischer Zeit hin zu aristokratischer Dominanz mit kleinen privaten Kapellen in späterer Zeit.

Yukarı Kirmir/Siberis Vadisi olağan dışı sulak ve verimli ama aynı zamanda dar ve küçüktür. Bu nedenle vadide her zaman tarım yapılmış ve yerleşilmiş olmakla beraber kullanımlar hiçbir zaman yoğun ve büyük ölçekte değildi. Bu makalede vadide yer alan yarım düzine sıra dışı anıt tanıtılmaktadir: Bunlar arasında Mahkemeağacin’dan bir Galat hükümdarına (tetrarkh) ait yazıtlı bir heykel kaidesi, Güzelçiftlik’te bir Erken Bizans Dönemi bazilikası, yine Mahkemeağacin’daki görkemli Orta Bizans Dönemi kayaya oyma yaşam alanları ve işlikleri, Değirmenönü’ndeki küçük Orta Bizans Dönemi çiftliği, Dikmen/İndere’deki Orta Bizans Dönemi mağara şapeli ve muhtemel inziva yeri, Araplara ve Türklere karşı kullanılmış Tabanoğlu Kalesi ve Çeltikçi Kalesi ile Alicin Manastırı olarak adlandırılan iki savunma istihkamı bulunmaktadır. Bütünüyle bakıldığında Yukarı Kirmir Vadisi’nin kendine has istisnai özellikleri bunların dışındaki tipik Bizans yerleşme tarihiyle bütünleşmiş gibi görünmektedir. Olağan dışı bir coğrafi konumda bir taraftan sınırlı refah düzeyi, bölgesel stratejik önem ve kayaya oyma özgün mimari örnekler ortaya çıkarken, diğer taraftan da görünüşe göre küçük ölçekli, dağınık bir yerleşme düzeni oluşmuştur. Erken Bizans’ın büyük cemaat kiliseleriyle, nispeten eşitlikçi toplumsal döneminden, küçük özel şapellere sahip aristokratların egemenliğindeki geç zamanlara kadar, farklı anıtların mimari özellikleri ve bunların değişen toplum yapısını nasıl yansıttığı tipiktir.