room with strongly rounded corners, fitted with a large round hearth embedded in the floor, and a load-bearing pole.

At Khizanaant Gora C1–2 house forms evolved and were now roughly square with rounded corners.81 Most had a rectangular vestibule attached to one side and a central hearth. The average main room had a living area of 8.75 m2, which is about half of that available to the occupants of the following phase. Other architectural developments occurred in Level B. These include a storage platform along the back wall, a clay threshold and the use of stone foundations for outside boundary walls, each of which are mirrored at Kvatskhelebi. Circular hearths were still requisite fitments.

Wattle-and-daub structures have often been associated with mobile communities and economic strategies that included either transhumance or a form of nomadism.82 Roger Cribb has argued that such homes reveal similarities in settlement patterns and house forms with tent communities, even though nomad tents have more open space around them than do these prehistoric Caucasian structures. Moreover, he pointed out that, based on ethnographic research, highlands tents can be quite substantial and secure structures, varying considerably in method of construction depending on the purpose of the settlement.83 The most common tent types are, in fact, composite dwellings that combine durable components with perishable material.A nomad tent, with its constrained internal area, also enables us to speculate on the social organisation within the household. Qashqai tents, for instance, are rather formally divided along the midline, with domestic and family activities taking place on one side of the tent and the entertainment of visitors in the opposite half. Other nomads, however, prefer a system whereby the space within and around the tent is utilised for different purposes and divided accordingly into domains by temporary partitions. Despite this intriguing comparison, we have yet to establish through zooarchaeological evidence the degree to which Kura-Araxes communities might have been mobile. Even so, it is possible that Kura-Araxes groups adhered to long-held familial customs concerning domestic structures and spaces, regardless of whether the houses were permanent or temporary.

Villages of Circular Structures

Round houses present us with a different character.They have been found over a wide area, but there is a heavy concentration in the Ararat Plain in Armenia, and in Nakhichevan. Further south, circular buildings appear in north-west Iran, most notably atYanik Tepe.84 Generally, these round structures were built

81Kikvidze 1972.

82Cribb 1991: 137.

83Cribb 1991: 84–112, 220–3; see in particular his tent types 4 and 5, fig. 6: 14.

84Summers 2013a.

with mud bricks set on substantial stone foundations, though in the higher altitudes, at sites such as Elar in Armenia and Galgalatli in Dagestan, only stone was used.85

Size varies greatly from houses with a floor area of approximately 12.5 m2 (2.3 to 4.0 m in diameter), as seen in the earliest levels at Mokhra Blur (Levels 11–9) and Kültepe Level II, to large structures with diameters measuring 13 m (Kültepe Level II) and 7.5 m at Ozni, where Kuftin found circular hearths embedded into the floor, an andiron, and a red-painted, rectangular incense burner.86 Most commonly, round houses ranged in size from 5.2 to 7.5 m in diameter. The general chronological trend appears to be a progression from small to large round houses, with rectangular annexes gaining popularity with the passing of time. Occasionally there are variations of the standard features, such as the pebble floor at Mchadijvari Gora.87

Early excavations at Shengavit exposed a large area, but fell short of a clear stratigraphic sequence, a problem that has been rectified by the recent Armenian-American project.88 Large exposures made during the early excavations have provided us with a good idea of the settlement layout. Twelve circular houses and one apsidal structure, inter-connected by walls, form a densely packed settlement with no obvious routes of communication, a situation recalling theYanik Tepe and Köhne Shahar (Ravaz) complexes.AtYanik walls were one mud brick thick, sometimes two, with thresholds higher than the house floors. A central pole supported the roof of the larger circles that measured up to 5 m in diameter and covered an area of 23.5 m2. Such a large expanse would have required a light superstructure probably comprising a domed roof of withies daubed with mud.89 Internal features included low mud-brick partition walls, platforms, and kitchen facilities always positioned to the right of the entrance.

The deep sequence at Mokhra Blur (Armenia) provides another very useful snapshot of the evolution of Kura-Araxes villages in the Ararat Plain.90 When Mokhra Blur was founded (Levels 11–9), the hamlet comprised small, freestanding round houses.Then, in the second cultural phase (Levels 8–4), the size of houses grew markedly as did the thickness of walls, now often two courses across; annexes, too, were often attached. The focal point of this settlement, however, was an impressive stone tower that survives to a height of greater than

85Khanzdian 1979: 151; Gadzhiev 1989: figs 2–6.

86Areshian 1978; Zhorzhikashvili and Gogadze 1974: 30–41, pls 39–63;Abibullaev 1961.

87Tsitlanadze 2008: fig. 4: 2.

88Baiburtian (1939, 2011) initiated investigations between 1936 and 1938. Sardarian (1967) continued excavations from 1958 to 1983, and since 2008, Simonyan and Rothman (2015) have renewed investigations. For a well-illustrated and informative catalogue of the Shengavit material held in theYerevan Historical Museum, see Badalyan et al. 2015.

89Summers 2013a: 15.

90Areshian 1996.

4 m. At the very top of the tower lies a basalt monolith 4 m in length, which in its day presumably stood upright.91

Round houses have also been found at Norabats (Armenia), a small hamlet measuring no more than 0.5 ha in area. The plan of the excavated area shows a handful of freestanding circular structures – two large (about 7.5 m in diameter) surrounded by six smaller ones.92 Apparently, some had curvilinear annexes. Importantly, according to Areshian, the distribution of artefacts in the buildings and annexes shows that the various areas had different functions.The internal area around the hearth was used for cooking and eating, whereas the annexes were primarily used as storage spaces, for the initial processing of food, and also for tool-making.The larger structures had a central hearth, while one of the smaller buildings (no. 8) had a hearth and oven partially built over by the wall, which according to Areshian reflects a foundation ritual.93 Circular walls have been also reported at Ada Blur (Level I),Agarak, Garni, and Dzyanberd.94

Twenty-eight circular houses representing the full range of sizes were exposed in the deep (9 m) Kura-Araxes deposit at Kültepe (Nakhichevan). They invariably had a round hearth fixed into the centre of the floor and occasionally a rectangular clay oven was set against the wall.95 Some houses were partitioned in two, though the function of the respective halves is unknown; others had a mud-brick bench running around part of the wall. Curious variants of this group are found in Georgia: the ‘keyhole-shaped’ structure at Tetri Tskaro, comprising a pebble-paved circular building with rectangular annex; and the semi-circular rooms with attached angular outhouses built of pisé from Samshvilde.96 At Garakepektepe, floors were painted red, recalling the earlier practice at Khizanaant Gora D.97

Stone and Mud-brick Rectangular Houses

Houses built entirely of field stones, or with walls of mud brick are similar in plan to the wattle-and-daub structures – square room with annex. They are dispersed over a wide area and do not appear to cluster in any one specific region, though there are quite a few in the Anatolian Upper Euphrates. Structurally, they can be freestanding, or abut each other, sharing a party wall. Amongst the most interesting sites is Pulur (Sakyol) now submerged beneath

91Areshian and Kafadarian 1975.

92Areshian 1978, 2005;Areshian and Kafadarian 1975.

93If this is correct, it provides the only instance of such a ritual. Another possible interpretation is that the hearth and oven belonged to an earlier building horizon that was superimposed by the building no. 8.

94Tumanian 2012;Areshian 2005: 79.

95Abibullaev 1965a.

96Gobedzhishvili 1978; Mirtskhulava 1975.

97Ismailov 1972: 480.

the waters of the Keban Dam reservoir.98 Hamit Koşay distinguished a thick Kura-Araxes cultural deposit with eleven building levels, the most impressive of which was Level X (Houses 70–90). Houses were adjoined and arranged in a radial fashion around the perimeter of the mound, conforming to the Anatolisches Siedlungsschema of the western regions.They were single-roomed and opened out onto a large central courtyard. Their fitments, however, are standard Kura-Araxes household elements – mud-brick benches and platforms, central hearths embedded into the earthen floor, and ground stone tools. Houses 79, 80, and 83 had rich deposits, leading to the attribution of ‘shrines’. This duality, which combines a south Caucasian sense of domestic space with a western Anatolian concept of village layout, is unique.99

Terraced Settlements

Sites located on the slopes of hills adopted a distinctive form of settlement planning that included terracing.Amiranis Gora, in the Samtskhe-Javakheti region, for example, is made up of no less than ten occupational terraces stretched along a mountain slope. At Harich (Armenia), three terraces were reinforced by massive stone walls, and at Chirkeisk (Dagestan), twenty stone houses were constructed on a terraced slope.100 Two sites provide the clearest evidence for these highlands sites: Chobareti (1600 m asl), located in the Samtskhe-Javakheti region, and Gegharot (2,100 m asl), in the Aragatsotn province of Armenia.101 A presentation of the Chobareti settlement will suffice.

Situated in a saddle on the slope of a mountain, Chobareti commands a strategic position, overlooking the valley below and, just over the mountaintop, the Kura River valley, not far from where it meets the Uraveli. Excavations have so far exposed four buildings (Structures 3–6) with stone built lower walls and possibly wooden superstructures. The floor levels of three fully exposed buildings were very well preserved.

One building (Structure 4) is unusually large (Figure 5.4(1)). Although its southern half was dug away when a pipeline corridor was constructed, its dimensions remain sizeable – it measures 15.3 m in length and its maximum preserved width is a few centimetres short of 4 m. It has a well-preserved, curvilinear north (back) wall built of small field stones set up against an earthen face. Two short internal walls compartmentalise the structure into three adjacent rooms.The interior of Structure 4 displays some of the standard fitments of a Kura-Araxes building: fragments of horseshoe-shaped andirons, a stylised horned animal moulded from clay, pale-coloured (monochrome) pottery

98Koşay 1976.

99Sagona and Zimansky 2009: 184.

100Chubinishvili 1963; Khachatrian 1975: 34; Gadzhiev 1987: 29; 1989: 26, 29. 101 Kakhiani et al. 2013; Badalyan et al. 2008; Badalyan et al. 2014: 152–69.

vessels, as well as black and red ceramics, which are in the minority and are otherwise largely found in the upper deposit. Judging by the number of ground stone saddle querns and concentrations of carbonised cereals in Structure 4, the inhabitants clearly processed food within its walls, especially in the eastern room. Grinding stones are heavy, tough tools, able to resist a fair amount of manhandling.Yet while some examples were found intact, indeed in situ, others are fragmented. This raises the interesting possibility that certain grinding stones were intentionally broken, a practice well documented in south-eastern Europe.102 Conspicuous breakage is not a practice that is generally associated with the Kura-Araxes communities, who, if anything, tended to leave items intact. Even so, it does lend itself to the idea that, in the Kura-Araxes world, utilitarian objects had a ‘life’ and were permeated with a ritual element.The eastern room also yielded a stone stamp seal with geometric incised pattern reminiscent of the Jemdet Nasr period, suggesting connections with distant lands.

Eleven radiocarbon readings and depositional levels confirm two occupational phases for Structure 4 at Chobareti (Figure 5.5). The earlier and more intense period, distinguished by a dark ashy soil, suggests the building was constructed around 3300–3200 cal BC.The two readings (Wk-37351;Wk-37352) were obtained from grain samples collected from the lowest floor. Structure 4 was contemporary to a series of pits and, to judge from a charcoal sample (Wk-34451) collected from a sounding in 2009, to Structure 3 as well. The three dates we have from the upper deposit in Structure 4 (Wk-34457, Wk- 34458,Wk-34459), again reckoned from grain samples, are later.They indicate that the building continued to be occupied between 3190 and 2900 cal BC.

Structures 3 and 6, both single-roomed houses positioned on the edge of another terrace, were also replete with material. Both had the usual internal fitments. At some point, a pit was dug into the bench of Structure 3 and several pots and the horn core of a sheep were placed in the depression.103 Across the entire floor were a considerable number of in-situ vessels, some positioned on the central hearth. Unlike Structure 4, preliminary analysis indicates that this room did not contain much grain; nor did it have the quantity of grinding stones that were found in Structure 4.

Structure 6, excavated in 2016, has a superbly built back wall of neatly fitted field stones preserved to a height of some 2.2 m, and measures 6 m across.The floor is covered with more than sixty ceramic vessels, many complete and in situ, providing us with a vivid picture of the spatial distribution of activities. Large jars were placed in the western half of the house, especially along the wall, while smaller items were positioned on the low bench that ran along the back wall. Importantly, Chobareti has provided secure, albeit provisional, information on the transition date between monochrome and red-black ceramics.

102Chapman 2000a.

103I would like to thank Rémi Berthon for the identification of the horn core.