A Khirokitia Household

Dikaios called the buildings of Khirokitia tholoi, the word archaeologists use for beehive tombs, and he gave them domes. “During the first period”, he wrote in his final report of 1953, “the tholoi were mostly of small size with a dome of mud-brick or pisé on a low masonry substructure.” His proof was one house, Tholos XLVII, whose wall leans inwards. He read the lean as the start of a vault and restored a dome 3.70 m high over a room some 3 m across. Later guidebooks added cones of thatch carried on central wooden poles.

Visitors who climb the hill today pass five round houses built beside the Visitors’ Centre, and they are flat-roofed. The Department of Antiquities built them in the 1990s, with the plans, dimensions and materials supplied by the French mission that had been digging the site since 1977 under Alain Le Brun. Sophocles Hadjisavvas, who described the work, wrote that the roofs were taken from “the imprints on fragments of earth from a collapsed roof” (Hadjisavvas, in Le Brun 1997). They are flat-topped cylinders of pale earth plaster with a rim of river pebbles round the roof edge, a detail for which we know of no excavated parallel.

The replicas are also a composite. Two of them copy S.117 and S.125, houses from the east side of the hill and from an earlier phase, and they stand beside a copy of a stepped entrance that was built later, on the west side. The household in this report sits where that entrance really was, at the same building level, and every part borrowed from elsewhere is marked as borrowed.

Khirokitia stands on a small, steep hill in a deep bend of the Maroni river, which wraps round it on the north, east and south-east. The only easy way on is from the west, across a saddle to the higher hills beyond. The village spread in a semicircle round the steepest and highest part of the hill, from the summit, at about 210 m, down to the foot, and it covered about one and a half hectares over all its phases. It gave its name to the Khirokitian, the last stage of the Aceramic Neolithic in Cyprus, from about 7000 to 5200 BC: farmers and herders who made no pottery, carved vessels in stone and buried their dead under the floors of their houses.

Dikaios exposed a long, massive structure running up the slope and across the saddle for about 185 m. He called it the Main Road, while admitting that “at first sight some stretches of it give the impression of a town wall”. Le Brun showed that it was a wall, Mur 100, and that it was the second of three. The first, wall 435, rebuilt as 423, closed the earliest village on the east. Mur 100 doubled it on the west and replaced it. Then, at the level the French team calls III in the west sector, the village grew downhill to its greatest extent, and a new wall, 284, was built about 12 m further west, enclosing a strip of land about 10 m wide. Wall 284 is “particularly well built”, preserved about 3 m high and up to 2.5 m thick, and it has been traced for more than 60 m (Le Brun, in BCH 1982 and 1984).

The line of each wall follows abrupt breaks of slope, so the levelled village stood well above the ground outside. That fact shapes everything that follows. Wall 284 “was built at the same time as the dwelling units adjoining it, whose walls are included in its very structure” (Le Brun, in BCH 1982). The houses of this report are bonded into the wall, and the wall is a terrace edge as well as a boundary.

In a short paper of 1992 called “Monter, descendre” (going up, going down), Le Brun published a plan and a drawing in perspective of one small area of the west sector at level III. It is the only published drawing of Khirokitia that shows houses, the village wall and the elaborate entrance together, at one moment. On it, wall 284 runs from north-west to south-east. Against its outer face stands the stair block of the entrance he called access 2. Against its inner face, packed close, are round units labelled S.97, S.99 and S.102, two more that carry no number, and between them a small patch of flat stones about a metre across.

Le Brun’s season report for 1980 describes a group at this level: “five of these units delimit a space of small dimensions whose floor is plastered”, left bare in one circle for a grain-grinding installation, locus 289 (BCH 1981). The locus number is not written on the plan, so our identification of the patch of stones as locus 289 is an inference. Its size, level and zone all fit, and so does the ring of houses round it.

The houses and the entrance come from one excavated area and one building level, drawn after Le Brun 1992, figures 2 and 3. The positions of S.97, S.99, S.102 and the wall are scaled from that small plan to within about half a metre; the two unnumbered units to within about a metre. The fifth unit round the courtyard is known only from the excavator’s description, and we have placed it ourselves. The steps down from the wall top into the village were never excavated. Some details come from other houses at Khirokitia and are marked as borrowed where they appear: the roof build-up (S.85, the same level, position unknown), the door height (Tholos V), and the optional windows (S.125).

Dikaios’s compound of XXVII, XXVIII and XXIX has a clearer story of use, and the east-side group of S.117, S.122 and S.125 is the one the replicas copy. Neither has a published plan that also shows the wall and its entrance, so both appear here only as parallels.

Every layer of the model is labelled excavated, inferred, reconstruction, speculative or rejected, and each says where its evidence comes from: this group, another named unit, or building practice in general. Rejected layers appear only so that they can be compared. In the drawings, line styles do the same work, and anything beyond the edge of the published plan is dashed.

The 3D model opens on the best-supported household. Its options switch in the alternatives: the dome, a corbelled door head, windows, a sleeping platform, painted walls and floor, and the fifth house round the court, which is on by default because the excavator’s text requires five houses, and can be switched off because its position is ours. The cutaway slices the whole group horizontally, like the published plan, so that every house opens at once; the entrance lies below the lowest cut and stays whole.

At Khirokitia a single round building was a room. Le Brun’s household is “a compound of several of these circular units around an unroofed space, a sort of small inner ‘courtyard’”, with a fixed quern in the court for grinding grain and cooking done indoors (Le Brun 1997). Units range from 2.3 to 9.2 m across the outside and from under 2 to about 15 m² of floor. At level III they are small: the mean floor area is 4.80 m², and the units grew larger in the levels above. Le Brun allows at most one occupant to a unit, and he warns that not every unit on a plan was in use at the same time.

Our group has one large room and several small ones. S.97 is the largest, 3.3 m across inside with 8.55 m² of floor (Papaconstantinou 2006). A curved partition, drawn on the plan like a letter G, encloses a sector on its west side and opens to the east. We give it a small hearth platform beside the opening, a reconstruction following Le Brun’s general rule for the site: hearths are small rectangular platforms under a stone slab, “designed only to hold charcoal from fires lit outside”. Such hearths are found only in roofed rooms, and they make little smoke.

S.99 is the smallest, 1.6 m across and 2 m² of floor, with no partition. Le Brun noted the “absence of a door” on the side facing the wall; we have put its door on the far side. A room this size is usually read as a store, and that reading is ours. S.102 is D-shaped, its flat side lining the wall. On its later floors, at level II, it had a platform and a hearth, and we have moved them down onto its level III floor, which is unrecorded. The unnumbered unit east of S.97 had partitions shaped like a letter E. Of the unit to the north the plan shows only the southern arc, and the rest is our reconstruction.

Burials under house floors are common at Khirokitia, 254 in all. We know of no burial records for these units, and the model shows none; the practice is described under Neolithic and Chalcolithic burial.

Le Brun put the whole village at 300 to 600 people, “no doubt closer to 300”. Floor areas here fall well below the 10 m² a person that the anthropologist Raoul Naroll derived from living societies, even allowing for lofts, which is one reason the figure is low (see population and demography).

Khirokitia’s builders used what lay on and below the hill: blocks of pale limestone gathered from the surface, rectangular slabs 30 to 50 cm long and 10 to 15 cm thick, and dark rounded diabase cobbles carried up from the bed of the Maroni. Mortar was mud. Earth, mixed with straw and dried in the sun, went into walls either as mudbrick or as pisé, the loose field term Le Brun and Dikaios use for massed earth. We know of no sign of shuttering, so the earth was probably laid by hand in lifts, like cob.

Le Brun lists four ways of building a wall, and the commonest is a double ring: an outer ring of stone and an inner ring of pisé or mudbrick, set straight onto levelled ground with no foundation trench. The plan draws the larger units of our group with doubled lines, so they have that build, with a low stone part and earth above it. The small store, S.99, has a single earth ring on a stone footing. The only measured bricks at Khirokitia come from Tholos XXV, 25 to 30 cm long and 6.5 to 7.5 cm thick, laid with each joint over the middle of the brick below (Dikaios 1953, p. 169). Their width was never recorded. Two colourings can still be seen on the site; we use the whitish bricks with ochre joints, which show only when the model is taken apart.

Walls were thickened over time by adding inner rings, which Dikaios read as consolidation. XV(II) had a wall 1.20 to 1.30 m thick round a room about 2 m across, and XLV had three concentric stone circles, 2.10 m in all. The walls of our group are thinner, yet a small unit here still gives up close to half its footprint to wall. Sometimes flat stones were set on edge round the foot of a wall, to protect it “from erosion by rain water” (Le Brun 1997), and the model puts them on the court and passage sides.

How high the walls stood is not recorded for any house in this group. Elsewhere on the site walls survive to 2.30 m (Tholos IA, Tholos XXV) and, in the stone wall of Tholos V, almost 3 m. The one complete doorway, also in Tholos V, is 1.63 m high. A threshold, a door of that height, a lintel and enough earth above it to hold the lintel down give a wall-head 2.25 m above the court, and we use that height for every unit, having found nothing to set one apart from another.

The six units’ walls hold about 85 tonnes of stone and earth, 20 of them stone, including some 3,800 sun-dried bricks. Build mode counts the stone, earth and bricks for each step from the model’s own geometry, so the weight of the walls can be read off as they rise.

S.85 was a level III house in the same west sector as our group, burnt early in its life. The fire brought the roof down “en un seul bloc”, in a single piece, and floor 417 beneath it was entirely covered with fragments of earth hardened by the heat. Le Brun described them in his report on the 1983 season (BCH 1984). They are 5 to 8 cm thick. One face is flat and smooth; the other carries the negative prints of branches and reeds laid parallel to one another, and in the great majority of pieces that face lay downwards. He reconstructed “a flat roof, a terrace, made of a bed of branches covered with coats of pisé, of which Cypriot peasant architecture still offers examples”.

With no post-holes in the floor, Le Brun supposed that the roof rested on wooden crosspieces running across the line of the branches and seated on top of the wall, which he thought “tout à fait vraisemblable”, entirely likely, in a room only 1.85 m across. By 1997 his account had two crossed layers of plants, probably reeds, over a wooden frame, under several layers of pisé and earth. Dikaios had already published a second burnt roof, of timber and pisé over the corridor of Tholos IA, with sockets for its beam ends left in the wall. The sockets are 0.17 m high, the only measure of a roof timber from the site.

Our small units, spanning up to about 2.2 m, sit squarely on S.85’s evidence. S.97 spans 3.3 m and the unit to its north about 3 m, and giving them the same roof is an extrapolation. Le Brun makes it himself: “all the data indicate flat, terraced roofs.” The model gives S.97 and its neighbour thicker bearers, 15 to 18 cm across, and the smaller rooms 10 to 12 cm, laid about 55 cm apart. Over them go a layer of thin branches and reeds, a second layer of reeds laid across it, a first coat of straw-tempered earth 5 to 8 cm thick pressed down onto the reeds, and about 10 cm more of earth, smoothed flat and rolled over the edge of the wall-head. We know of no eaves, parapet, gutter, smoke-hole, hatch or ladder at Khirokitia, and the model has none; the base stones along the wall foot suggest that rain ran off the roof edge and down the wall. No roof has been measured in section, so the total thickness is ours.

The timber was cut without saws. The builders had axes of pecked and polished diabase, and bone chisels whose wear shows them struck with a mallet into dry wood (Legrand 2007). The bearers are therefore roundwood, bark on or roughly peeled, their ends chopped and crushed, their diameters uneven. The tidy peeled logs of the replicas are too regular. Charcoal from the site shows deciduous oak, ash, poplar or willow and plane throughout, with Calabrian pine (Pinus brutia) increasing in levels III and II; no roof beam has been identified to species. Pollen of reedmace and sedges from the site fits the prints on the roof fragments.

Roofing the six units takes 28 bearers, about 75 m of roundwood in all, some 95 m² of reeds and branches in two layers over 48 m² of roof, and about 14 tonnes of earth. Build mode counts them for each roof as it goes on, and the roof in section shows every layer at once.

Dikaios thought the inward curve of XLVII’s wall and the plate he captioned “springing of the vault” were “a convincing proof of the manner in which these tholoi were roofed”. Le Brun looked at the same wall and found no vault in it. A dome of stone or brick is normally corbelled, built of horizontal courses each set a little further in than the one below, which push nothing outwards. “Nothing of the sort is apparent here,” he wrote; “the wall in its entirety appears to have toppled over.” A mudbrick wall just to the north has courses that lean south at the same angle as Dikaios’s vault. The hill is steep, and the weight of earth pressing downslope distorts walls, most of all in their upper parts, while their lower parts tend to bulge out. The curve does not justify a domed roof, and the burnt flat roof of S.85 supplies the positive case.

The conical thatch of the guidebooks fares worse. It needs a central post, and no Khirokitia house has a central post-hole; Dikaios himself listed only two doubtful post-holes on the whole site. He had tried a thatch over a mud vault, carried on a post, for Tholos IA, and he called the post “not altogether satisfactory”. The pillars that stand in some of the larger houses were no roof supports either. Unit VII received its pair of piers only on the latest of its three floors, so it had stood and been roofed without them, and XVII, one of the largest rooms on the site at 5 m across, has none. Le Brun reads the piers as supports for platforms or lofts. At Kalavasos-Tenta, the Aceramic village in the next valley, Ian Todd left both a dome and a flat roof open.

The model keeps the dome as a switch, shaded as rejected; the plate below sets it beside the flat roof, the conical thatch and a speculative low sleeping deck.

Doors at Khirokitia are “usually around 0.5 m wide, and frequently with a paved threshold”, and they need not open at floor level: some are “reached by one or more steps” (Le Brun 1997). XLVII’s threshold stands 1.70 m above its earliest floor, faced with thin slabs of gypsum, with three steps down inside. Only Tholos V keeps a doorway to its full height, 1.63 m, with a head closed by corbelled stone. That door runs through a tall stone wall, and ours pass through earth above a low stone ring, where a timber lintel is the practical choice and impressions of wooden lintels survive over window openings on the site. The model gives each doorway three poles as a lintel, 0.50 m of width and the Tholos V height, and offers the corbelled head as an alternative. We know of no pivot stones at Khirokitia, unlike the Chalcolithic houses of the west of the island (see the Chalcolithic round house), so the doorways are left open.

Only one door in the group has a published position, and it is a negative: S.99 has none on the wall side. S.97’s door follows a short double line on the plan that runs from its wall towards the court; the others face the court because a household round a court would use it. All are reconstruction.

Windows were rare. S.125, on the east side, is “one of the rare examples of a unit which has windows”, three of them, about 0.45 m wide, at regular intervals (Le Brun 1997). Unit III has an opening 0.45 m high and 0.41 m wide, and Dikaios found two light-shafts, niches lit through a gap between the stone and the earth of the wall. Sill heights are not published. Our group’s walls do not survive high enough to say whether any had windows; the model can add three to S.97, borrowed from S.125 and shaded as speculative.

Inside and out, “the internal and external faces of the wall are covered by a whitish earth plaster”, and the same plaster runs on over the floor. We know of no lime plaster on any wall; lime appears once, in the bottom of a basin. Floors are earth “of yellowish or darker colour”, 2 to 8 cm a coat, “as a rule smoothed” (Dikaios 1953), and they were renewed again and again: XV(II) has twelve. The red earth render of Cypriot village houses and of the Chalcolithic replicas at Lemba has no place here.

Colour came as paint. The French excavations found their first red-ochre wall painting in a level III unit, and the report does not say which (BCH 1982); Dikaios had already found plaster painted red in X(III). In S.122 on the east side, both the wall and its notched piers were painted. The motifs cannot be read, so the model’s optional ochre is faded patches with no pattern. One floor, in XXXVI, kept traces of green paint, “clear evidence of the custom of painting the floors” in Dikaios’s view, though it is a single case; that too is a switch.

The court is small, about 3 m² as we have drawn it, and it was plastered. The plaster “was laid so as to leave free a circular zone for a grain-grinding installation”, about a metre across: a paving of flat stones wedging a fixed quern, “the whole ringed by small stones set on edge” (BCH 1981). A photograph of the installation, tagged 289, appears in Le Brun’s guide to the site, and it shows a large oval slab bedded in flat stones inside its ring. We have sized the quern from that photograph and set a hand stone on it; ground-stone tools are plentiful on the site, and this one’s place is ours.

In Dikaios’s compound of XXVII to XXIX, a living room and a kitchen with a central hearth open onto a yard with a grinding bay: a quern raised on a platform half a metre high, a stone tray at its foot to catch the flour, and a paved circle. Grain and flour were household business, done in the open at the centre of the group, while cooking happened indoors over charcoal. What the grain was, and what else went into the pot, is covered under food and diet.

Beyond the court the houses stand “separated only by narrow passages”, used for getting about and for dumping rubbish. Between S.99 and S.102 the gap is 30 cm or less, too narrow to pass. Carefully plastered open spaces with no fittings in them turn up elsewhere on the site, and Le Brun reads them as space kept free on purpose. The model carries a thin, patchy skim of plaster out from the court into the passages, which is our reconstruction.

Outside wall 284 the ground lay more than 2.8 m below the village surface (Le Brun 1992). Access 2 climbs that height “without creating an opening in the wall”. Against the wall’s outer face stands a massive block of stone, 10 m long and 1.60 m deep, “with carefully plastered faces”. A stair only 0.80 m wide is built into it, with treads 0.45 m deep of flat stones and pisé and risers of 0.30 to 0.40 m. By 1992 seven steps had been dug, three in a lower flight running straight at the wall and four in an upper flight turning along it. Further work found a third flight at right angles to the second. The middle one, in Le Brun’s words of 1997, lies in the core of the structure, “hidden from view”.

A walker approaches from the west, the only side the hill allows, and meets a stone structure that rises almost 2 m in front of the stair’s opening and prevents a straight approach; preserved almost 2.5 m high, it seems to lean against a projection of rock. The walker passes round it, climbs the first flight towards the wall, turns left, climbs the hidden flight inside the block, turns right, and climbs the third flight onto the top of the wall. There, S.99 stands directly ahead with no door, S.97 blocks the way north, and the gap between S.99 and S.102 is too narrow to pass. So the path turns right again and runs south along the wall top for several metres. Le Brun suggested that S.102 was built with a flat side to leave room for it, and he marked the idea “not verified”. The way down into the village was “no doubt” by “steps which have yet to be excavated”, so in the model everything after the arrival on the wall top is reconstruction.

Nine risers of 35 cm lift the path 3.15 m to the wall top and one more takes it down into the village, so the wall shows over 3 m of stone outside and only 35 cm inside. Three steps across a block 1.6 m deep leave room for treads of only 40 cm, and the third flight is our geometry. Le Brun was clear that the stair belongs to the block, “carried by a monumental construction” built for that purpose alone, and the model keeps all three flights inside it.

Le Brun called the arrangement unique in Cyprus and the Near East. Its one earlier parallel at a village’s edge, four steps set into a retaining wall at Beidha in Jordan, works on a different principle. An earlier way into Khirokitia, access 1 in the first wall, was a gap about 1.20 m wide whose plastered floors ran on outwards over a stone construction built against the wall; Le Brun thought it served only to bridge the drop and made no attempt to control who came in.

Whether access 2 was built for defence is harder to say. Le Brun wrote of the “enormous investment in time and labour” the wall represents. The stair is too narrow and too carefully plastered for flocks, and since no animal pens have been found inside, the animals were kept outside the village. We know of no evidence at Khirokitia that the village was ever attacked, and no sign of how the entrance was closed, if it ever was. The structure in front of the stair was either a free-standing bastion or part of an outer wall with an entrance of its own; that was still undecided in 1997, and the model shows the bastion. What the design certainly did was control movement: nobody could walk straight in, anyone arriving had to climb and turn three times in a passage 80 cm wide, and flocks stayed on the far side of a 3 m wall. The wider history of walls on the island is in fortifications in ancient Cyprus.

Build mode puts the household together in twelve steps, from levelling the terrace to setting the quern. Wall 284 goes up second, with the houses beside it marked out as lines of stone, because their outer rings are part of the wall’s body. The seasons attached to the steps are practical suggestions only: walls raised in the dry months, two or three weeks for bricks to dry, roofs closed before the first autumn rain.

We know of no published estimate of the labour behind a Khirokitia house or behind the wall, and the replicas have produced no published record of their building. The estimate in Build mode is ours, made from the model’s quantities and from general rates for earthen building, and it gives ranges.

On that basis the six houses and their court come to between 130 and 290 person-days of work. The 19 m of wall 284 in the model, with the stair block and the stone structure in front of it, come to between 550 and 1,150: each metre of the wall holds about 7.5 m³ of stone and earth, some 15 tonnes.

Whatever the figures, the wall and the stair block, over 3 m high and more than 2 m thick along a line traced for more than 60 m, dwarf the houses they enclose.

Nothing stayed built for long. Floors were replastered unit by unit, and S.97 was still in use at the level above. Wall collapses are recorded in unit V and twice in XV(II), and the low-walled houses of Dikaios’s commonest type were short-lived and built over. Fallen upper walls were often left in place and floored over, raising the floor by 24 to 45 cm at a time in XLVII, and sometimes cleared away “to avoid losing internal height”. At the end of level II part of the western edge of the village collapsed, and at level I a line of large upright blocks stood inside the line of wall 284, as much a retaining kerb as a fence.

We do not know which of these units were in use together. Floors were renewed separately, and the plan draws one level, which is a span of time. The fifth house round the court is known only from a sentence. The door positions are ours except for S.99’s missing one, and no height or build is published for any wall in the group. The third flight of the stair and the steps down into the village were still unexcavated in 1997, and the block in front of the stair may be a bastion or part of a second wall. The core of wall 284 has not been described: we fill it with rubble in mud, and the older Mur 100, a mass of pisé faced with stone, suggests earth as the other reading. Even north is a judgement: the arrow on the plan can be read two ways, and we chose the reading that puts the outside to the west-south-west, as the text requires.

We have no calibrated date for level III itself, only for the Khirokitian as a whole, about 7000 to 5200 BC. We do not know whether any of these rooms had windows. And the roof of S.97 depends on whether a burnt roof 1.85 m across tells us how people roofed a room 3.3 m across.

The pieces of S.85’s roof were found lying where they fell, nearly all of them face down, with the prints of reeds laid side by side on their undersides. The reeds burnt in the fire; the earth that was pressed onto them still carries their shape.