occupies a powerful defensive position overlooking the Altınova plain. Although it was the region's main settlement for much of the middle ages,
Harput (aka Ziata) was an obscure, purely local stronghold throughout the classical and early medieval periods. Its only claim to fame was that in 359 it was the most important among several minor forts captured by a Persian expedition into Roman territory. It played no major role in the Roman-Sasanian wars of late antiquity, nor in the Arab conquest. It was "not incorporated into the elaborate system of frontier defense laid out by the caliph
Harun al-Rashid at the end of the 8th century". Even as late as
Qudama ibn Ja'far's systematic account of the Arab marches facing Byzantium between 928 and 932 it was unmentioned and must have remained purely local in importance. It seems to have first gained significance when the Byzantines conquered Anzitene in the 930s. During the early middle ages, Harput shrank in size. From its larger Roman-era site, with ten-
stadia-long circuit walls, it "contracted and perhaps retreated to the powerful position occupied by the medieval castle". In the 10th century, the fortress was expanded to its present size. From this point on, Harput became the main center in Anzitene and the surrounding region in general. Its population grew as the threat of warfare, Turkic raids, and
brigandage drove many people to seek safety near its formidable castle. The main castle apart from Harput in 10th-century Anzitene was Dadima (or Dadimon), at present-day
Tadım. It was an important military objective of Sayf al-Dawla's campaigns of 938 and 956. This castle was the medieval descendant of an important ancient settlement.
George of Cyprus, writing in the late 6th century, appears to implicitly label Dadima a city since he does not list it as either a fortress (
kastron) or a town (
polichne). He also implies that it had recently been elevated to the provincial capital of Armenia IV, succeeding
Martyropolis, the capital under Justinian, after the southern flanks of the Armenian Taurus were detached and made into the new province of
Upper Mesopotamia. Dadima remained the ecclesiastical capital of Armenia IV at least through 692, when its bishop Elias signed the canons of the
Quinisext Council. During the period of Arab domination roughly lasting from 640-938, Tadım declined to a small town, leaving Arsamosata the lone major city in the plains region. The reason may have been proximity to danger, with the Arab-Byzantine frontier now lying west of Tadım. Some of its population likely relocated to the Arsamosata area. Sometime after 956, Tadım sank into total obscurity, although its name survives to the present day. As of 1900, remains at Tadım include two traces of circuit walls: one enclosing a tepe 60 feet high, possibly representing the medieval site, and the other 1 mile west on the plain, possibly representing the ancient site. Some 8 km south of Tadım is
Hoh, which as of the 19th century was one of the principal villages on the plain. As of 1900, foundations of an apparently "Roman" or "Byzantine" building were identified atop a tepe 80 feet high and 200 feet in diameter; the tepe was apparently unfortified. Hoh is probably the place called
Hūrī in the al-Mutanabbi account. (This was first proposed by Canard and accepted by Howard-Johnston "for want of convincing alternatives - but with considerable hesitation".) Although it is listed as a
balad or city in the text, it was probably obscure in both classical and medieval times and was probably just a particularly large village, probably unfortified according to Howard-Johnston. On the other hand, Sinclair identifies the site of Aşağı Huh with the fort called "Chachon" that Basil I attacked in 873. In any case, it was probably fortified earlier in its history, since the Armenian Geography attributed to
Moses of Chorene in the 6th or 7th century lists a castle called Horē as one of two places in Anzitene canton.
Arsamosata in the northeast was a major city with a mixed Armenian and Assyrian population. It was located at the present-day site of Haraba. The
Syriac Orthodox bishop of Anzitene resided at Arsamosata. Tadım's decline under Arab rule left Arsamosata as the lone major city in the plains region. Arsamosata likely received an influx of new residents from both Tadım and the Malatya plain. Around the 11th and 12th centuries, however, Arsamosata went into decline, contracting from "a sprawling ancient city" into a smaller city contained within the walls. It remained a city until at least the end of the 12th century - the Syrian Orthodox bishopric is known until 1199 - and was probably inside the old city walls, although not effectively protected by them. After the local garrison was finally withdrawn, perhaps in 1234 when Harput was conquered by the Seljuks, Arsamosata's population "dispersed to villages on the plain and nearby hills". Other than Arsamosata, the castle of Salām was probably the only major site in the northeastern part of Anzitene at Sayf al-Dawla's time. A possible identification is the large stronghold of Shitar Kale near
İçme, which overlooks the plain from a height of 1000 feet and commands the Arsanias valley. Along with the village of al-Muqaddamiyyah, Salām lay on Sayf al-Dawla's route from Harput to Arsamosata in 938. The poet
Abu Firas refers to a battle "in the district of Salām" at one point, implying that Sayf al-Dawla's victory over the pursuing Byzantines took place near here. Sayf al-Dawla never targeted the highlands of northern and western Anzitene, "doubtless deterred by the danger of Byzantine guerrilla action in favorable terrain", so less is known about this area during the 10th century. Only 3 castles are mentioned in 10th-century sources:
Aşvan, al-Tell, and al-Minşār. The castle at Aşvan apparently put up no resistance to Sayf al-Dawla's forces in 956 - they occupied it without trouble and turned it into a base camp. It stood on the south bank of the Arsanias, at a convenient place for launching boats and rafts, and it controlled a ford. One candidate for its location is the mound called Aşvan Kale above the modern village of Aşvan: it is on the river bank, 4 km upstream from a major ford, and the name is identical. However, excavation of the northern half of the mound revealed no evidence of occupation between 400-1000 - nor any indication that the site was ever fortified during its occupation. Instead, according to James Howard-Johnston, the castle, along with the ancient and medieval town, was probably somewhere on the plain just to the west, immediately by the ford. Here, it would have been better positioned to guard the southern approach to ford on the road from Harput than the more defensible but distant mound above the present-day village. Alluvial action and human quarrying may have removed all visible traces of an early medieval castle. (However, according to Stephen Mitchell's hypothesis, there was never a
castle of "Hisn Aşvan" - rather, there was just a mound called "castle" in the 10th century just the same as it is today.) Al-Tell castle was probably in the hills west of Tadım, probably no more than 10 km or so away. Sayf al-Dawla sent a detachment to siege al-Tell in 938 while he was besieging Dadima. Canard proposed that al-Tell is the same place as the Tell Arsanas mentioned by
Ibn Hawqal; this is accepted by Howard-Johnston "for want of convincing alternatives - but with considerable hesitation".) The exact location is uncertain, but the coordinates Ibn Hawqal gave for Tell Arsanas would place it about 18 km west of Harput and about 24 km from the Euphrates ford at the village of Hammam. Tell Arsanas would have overlooked the main route southwest from Harput towards the ford at Tomisa. If al-Tell was indeed here, then Sayf al-Dawla's actions here would have been intended to hold the route and perhaps provide a distraction from his main actions at Tadım and Harput. The castle of al-Minşār was almost certainly at the large rock by the village of Kale on the mountain called Muşer or Mişar Dağ near the east bank of the Euphrates. This site "had been fashioned into a castle by hewing out platforms and steps and constructing walls in the prehistoric period". The site is just upstream from a ford on the Euphrates - itself upstream from the main one at Tomisa - which was used by Sayf al-Dawla in 953 to make a rapid retreat when his forces were almost trapped on the opposite bank. There was presumably a minor route running through the Anzitene highlands that he used then. 4 km south-southeast of modern-day Aşvan is the late medieval castle of
Taşkun Kale, which is on the northern edge of an upland basin. Anthony McNicoll tentatively dated Taşkun Kale castle and the medieval phase of its nearby church to 14th century and concluded that the settlement around it must have been fairly large given the presence of scattered potsherds and traces of stone walls. The name "Taşkun" is vaguely reminiscent of the ancient Roman border fort of
Dascusa - as does the nearby city of
al-Ashkūniyyah, a provincial capital across the river from Aşvan. There is a small castle on a small hill overlooking the village of
Keluşağı 12 km west of Tell Arsanas; it should probably be distinguished from al-Tell. It commands a second, more minor route west from Harput which is north of the main one. Based on local accounts of Roman coins being washed out from the spring rains, Huntington assigned this site to the Roman period. It may have also remained in use during the early middle ages. The ancient settlement of Mazara, whose site is now occupied by the modern city of
Elazığ, was a road station on the road linking Melitene to the upper Euphrates basin. After the classical period, it no longer appears in historical records, indicating that it declined to insignificance or was abandoned altogether. A large settlement at the mouth of the Zerteriç valley appears to have been built at some point during classical antiquity. It does not seem to have featured in Sayf al-Dawla's campaigns and may have dwindled to a village or been abandoned altogether by the 10th century. The town of Kalkas mentioned by Ibn Hawqal, which corresponds to the ancient Cholcis, was at the NE end of Hazar Gölü. When Sayf al-Dawla invaded through here in 956, the account makes no mention of him dealing with any Byzantine stronghold, so Kalkas does not appear to have been used as a military base during this period. ==See also==