The Marki House Through Five Centuries

For a thousand years before Marki was founded, Cypriots lived in round houses: a single circular room on a low stone base, walls of hand-laid mud, a hearth in the middle of the floor and a flat roof of timber, reeds and earth. Nobody in Chalcolithic Cyprus is known to have made a mudbrick.

Within a few generations of 2400 BC that way of building had gone. The new houses were rectangular and had several rooms. Their walls were of bricks shaped in wooden moulds and laid on footings of dry stone, and the rooms were grouped round a yard. They arrived with the other novelties that open the island’s Bronze Age: the first systematic working of its copper, cattle and donkeys, the ox-drawn ard, cemeteries of chamber tombs outside the village, new pottery such as Red Polished Philia ware and new household tools (Webb 2014). Who brought them is argued over, and the argument returns near the end of this report.

Marki answers a different question. Its excavators, David Frankel and Jennifer Webb, found that the same idea of a household held from the village’s foundation to its abandonment: “a walled rectangular courtyard with small enclosed rooms built of mud-brick on stone foundations located toward the rear” (Frankel and Webb 2006b: 289). What changed over five centuries was how open that courtyard was to the neighbours. The rectangle came first, with the first settlers. The closing of the household came slowly, and that is what the model shows.

Marki-Alonia lies on gently sloping ground above a ten-metre cliff on the south bank of the Alykos, where the chalks and marls of the central plain meet the dark pillow lavas of the Troodos foothills. The houses stand on havara, the soft whitish marl of the region. The river is dry for all but the winter months, and the nearest copper ores lie 5 to 10 km away (Sneddon 2002: 21–22). Frankel and Webb think the founders were “perhaps … a few dozen people moving away from a more substantial agro-pastoral village to establish a base closer to the copper sources of the northern Troodos” (2006b: 290). Survey found no Chalcolithic settlement on the spot: Marki was a new foundation, with no round-house village under it.

The Australian Cyprus Expedition, directed by Frankel and Webb, dug here from 1990 to 2000. About 1,500 m² of a site of some five or six hectares were excavated completely and another 500 m² cleared to the tops of the walls, “by far the largest exposure of any site of the period in Cyprus” (Frankel and Webb 2000). The first volume of the final report, in 1996, called Marki “An Early and Middle Bronze Age Town”; the second, in 2006, called it a “Settlement”, and the change of word was deliberate. The dig also settled an old quarrel about the start of the Early Bronze Age: the Philia deposits lie under the Early Cypriot ones, as Porphyrios Dikaios had argued against James Stewart, who took Philia for a regional variant (Webb 2014).

Across the excavated area the excavators identified 33 households, which they call compounds, over the whole life of the village, and divided that life into nine building phases lettered A to I. Each phase is “a palimpsest of perhaps three generations” (Frankel and Webb 2006b: 290). Today the site is farmland near the modern village of Marki. We know of no visitor facilities there, and the finds are in the care of the Department of Antiquities.

Mudbrick houses do not last. Frankel and Webb quote a study of recent Cypriot villages in which such houses “rarely lasted as long as a century” (Christodoulou 1959, cited in Frankel and Webb 2006b: 289). At Marki they were pulled down and rebuilt again and again, but the new walls were “normally built directly on surviving wall-stumps or offset a half-wall width to one side” (ibid.). The stone footings grew upwards with each rebuilding, which is why a household’s boundary can be traced through phase after phase. Building on the old stumps may also, the excavators suggest, mark “continuity of ownership and control of space”.

This report follows the ground of Compound 6. Compounds 6 and 7 are the longest-lived on the site: they held their boundaries from Phase C to Phase H, for “three or four centuries” (Frankel and Webb 2006b: 300). The plans of each phase are traced from the excavators’ published phase drawings, to within about 30 cm.

The house shown is a composite. The earliest state, labelled Philia, is not an earlier version of Compound 6. It shows the neighbourhood that stood on this ground before Compound 6 was built: the shared yard of two older households, Compounds 3 and 4, with their freestanding storeroom on the plot itself. The excavators argue that the families carried on (household 3 becoming Compound 7, household 4 becoming Compounds 6 and 8), but the buildings did not. From Phase C onwards the plot is Compound 6 itself.

The model shows five states out of the nine phases: Philia (Phase B), Early Cypriot I (C), Early Cypriot II (D), Early Cypriot III (E and F, which differ only in minor repairs) and Middle Cypriot I (G and H). The dates on the phase slider are rounded midpoints, worked out on the assumption that the nine phases were of about equal length between 2400 and 1800 BC; the excavators give no date for any single phase. Specialists also place the boundaries of the periods themselves differently, by up to a century.

The first settlers left little on this ground. In Phase A, about 2400 BC, the main housing seems to have lain just south-west of the excavated area; here there are two fragments of wall and “a complex array of post-holes in open space”, which Frankel and Webb read as animal pens and work areas (2006b: 290).

By Phase B two households, Compounds 3 and 4, stood side by side. Each had its own roofed rooms of stone and mudbrick, and the houses were already rectangular. Between them lay a single yard with “no physical barrier” in it (Frankel and Webb 2006b: 291). The yard held “a substantial bread oven” that suggests baking on a large scale, and the debris of flint-knapping and of antler and shell working. “Four sets of post-holes suggest the presence of small pens, sheds and a light fence.” In one corner stood a freestanding room with fragments of pithoi, the big storage jars, and vat-like vessels with perforated bases. On the southern edge a pithos had been sunk under the floor with a baby buried in it, and the excavators take it to mark a boundary. In the model the yard covers about 141 m², more than three times the 47 m² of roofed rooms around it.

The households had private roofed rooms and shared the yard and what stood in it. Frankel and Webb read the arrangement as “a survival mechanism appropriate to a newly established pioneer community, perhaps numbering only 40 people” (2006b: 301). On this reading, a village of a handful of households shared its oven, its pens and its work space because none of them could yet manage alone. The houses themselves show that the newcomers already knew how to build: mould-made brick appears here at once, and some of it was fired, which at Marki happens only in Philia to Early Cypriot II buildings (Webb et al. 2013). At least one Philia hearth was rectangular and set against a wall.

The sharing did not last. Within about a hundred years of first settlement, in the excavators’ phrase, the village had become one of “self-contained households” (2006b: 301).

In Phase C the yard was built over. In place of Compounds 3 and 4 stood “two large discrete compounds, 6 and 7”, of roughly equal size, about 100 m² each including their walls. They shared “a substantial common wall”, and there was “no direct access between them” (Frankel and Webb 2006b: 293). The excavators see in this “a desire to establish and maintain private space and ownership of resources”. Party walls, then, did not arrive late in Marki’s history: the two neighbouring families stood back to back from Early Cypriot I, and the wall between them was in use for the next three or four centuries.

Compound 6 put its three roofed rooms on the west side of the plot and its courtyard on the east, walled on three sides. Webb describes the Early Cypriot I pattern across the village: “substantial stone walls were built to enclose courtyards on three sides, leaving fairly wide entrances” (Webb 2014). The published plan of Compound 6 shows no wall on the south side of its court, and the model leaves it open. A curved feature against the north wall of the court is probably an oven; ovens still stood outdoors in this phase, and the courts still held pens and sheds, which the excavators record in Phases B and C only (2006b: 299). The model’s pen here is a guess at where one might have stood. In this phase the house has about 31 m² of roofed rooms and a court of about 42 m².

One neighbour was treated differently. A small household to the north, Compound 8, was reached through “an opening which led directly from the large courtyard of Compound 6 to the smaller courtyard of Compound 8” (Frankel and Webb 2006b: 293). Its only way in lay through the court of Compound 6. The excavators call it an offshoot of 6, and derive both from the older household 4.

In Phase D the household turned its plan inside out. “The older eastern courtyard was now fully enclosed and roofed over to form two rooms, taking in previously open space on the south, and the north-western area, which had previously been roofed inner space, was opened up to become the courtyard” (Frankel and Webb 2006b: 294). The new east room was wide, and the excavators found the stone base for a post in the middle of it.

The village around it was filling up. A formal north–south laneway was laid out along the east side of the compound, and Frankel and Webb point out that this “must have involved considerable re-negotiation of spatial boundaries and, presumably, of property rights” (2006b: 294). Compounds 6 and 8 were now both entered by a short passage off the lane, and some new compounds of this phase had no courtyard at all. There was no plan behind any of it: the excavators find “little suggestion of large-scale planning”, only growth by “an on-going process of inter-household negotiation” (2006b: 300). Roofed space in Compound 6 rose to about 49 m², and the new court was about 27 m². The door between Compound 6 and its offshoot stayed open.

Frankel and Webb suggest that the families of Compounds 6 and 7, the oldest on the site, “are likely to have exercised some social dominance” (2006b: 300). What they point to is modest: a private passage of their own, which elsewhere in the excavated area only Compound 29 had, and the means to set up an offshoot household. Nothing in the fittings or the rubbish of any household suggests “significant differentials in wealth and status” (ibid.).

Phase E is the model’s default state, and it is the best-documented. “Most of the walls of Compound 6 remained in place, but the eastern room was remodelled, so that it was now entered from the west instead of the south. The floor level was lowered and a new suite of built-in furnishings constructed. All three interior rooms were now provided with hearths” (Frankel and Webb 2006b: 295). The old opening into Compound 8 was “closed off”. From now on each room had its own fire, and the offshoot household had its own front door.

The house was entered from the lane by a narrow passage surfaced with gravel. At its end stood a short wall. Screen walls of this kind appear in the later phases at Marki, “apparently to conceal hearths from exterior view”, and are best preserved in Compound 29 (Frankel and Webb 2012: 496). Someone standing in the passage could see a strip of the court and nothing more. Inside, “interior rooms in most cases led one from another, achieving increased privacy with access depth” (Frankel and Webb 2006b: 299). Hearths, wall benches and bins stood only in the inner rooms, and ovens, which had been built in the courtyards in the earliest phases, “regularly appear in interior rooms from Phase C onward” (ibid.).

A summary of the 1990s seasons describes a house of this date in terms that fit Compound 6 closely: a roughly rectangular building of three main rooms, entered from its north-east corner by a narrow gravel-surfaced passage, with hearths built into benches against the outer walls, a rectangular mudbrick bin divided into compartments, a mudbrick oven, pebble-lined pot stands and “a well-constructed stone base for a post to support the roof”. The screen wall’s height and the positions of the hearths are reconstruction.

Frankel and Webb read the whole trend as “an increasing need or desire for inter-household privacy, perhaps linked with an increased emphasis on private property, intergenerational inheritance and the control and manipulation of space” (2006b: 299). In Phase F, nearby, the village gained its “first standing ruins” and its first households of a single room (2006b: 296–297).

By Phase G, around 2000 BC, Marki was past its peak and its centre was drifting south. “In Compound 6 the courtyard of Phases D to F reverted once more to interior space and some existing rooms may have gone out of use” (Frankel and Webb 2006b: 297). The published plan colours the old east room as open ground; the model keeps its walls standing and takes its roof away, which is one reading of several. The house now has about 42 m² of roofed rooms around an open area of about 34 m². Compound 8 was abandoned “as standing ruins or open space”, and Compounds 6 and 7 were now “divided from one another by standing ruins as well as by open spaces” (2006b: 298, 300). A new five-room house nearby, Compound 29, was built into older walls and entered from the east by a narrow passage, like Compound 6.

After Phase H the village thinned out. The final occupation, Phase I, is only a couple of rooms in the excavated area, and trial trenches 200 m to the south show that people lived there for about another generation (2006b: 298). Why the village was given up, around 1800 BC, “remain[s] unclear” (Frankel and Webb 2001: 123). The next settlement in the valley grew up at Marki-Reximon, 1.3 km to the west.

Every wall at Marki began as a footing of dry stone: two faces of unshaped stones with rubble between them and earth packed into the gaps, but no bed of mortar (Frankel 2000). The stones were “collected, for the most part, from the nearby river bed or surrounding hills” (Frankel and Webb 2012: 493), so the footings mix dark rounded cobbles of diabase, gabbro and pillow lava from the Alykos with angular white chalk from the hills to the north and east. Footings survive “well over 1 m” high (Frankel and Webb 2000), but that is usually several phases stacked on one another. For a single phase the model gives the footing 60 cm above the floor, four or five courses of cobbles, enough to keep rain splash and runoff off the brick. Outer and party walls are 55 cm thick in the model, partitions 45 cm and the screen 32 cm. These are close to the published plans and to the measured walls of Alambra, 8 km to the south-east (Coleman et al. 1996).

Above the stone went brick. The Marki bricks were made in moulds, as their consistent size and shape show (Frankel and Webb 2006a), and those in Early Cypriot III and Middle Cypriot walls “were not fired” (Webb et al. 2013). Their size is harder to pin down. A recent survey gives a Marki brick of “60 × 10” cm (Amadio 2023), but its Alambra entry, “70 × 32”, turns out to be a width and a height, since the Alambra bricks could not be measured for length; the Marki figure is very probably a partial record too, a length or width and a thickness. A study of Cypriot building reports that the Marki and Alambra bricks share a standard size larger than those of Middle Cypriot Erimi-Laonin tou Porakou, which are 40 × 14 × 12 cm (Amadio 2017, citing Philokyprou 1998). The model therefore uses a brick of 50 × 25 × 10 cm with 2 cm earth joints: 10 cm thick as in the Marki record, and just the size for two rows of stretchers, or one brick laid across, to fill a 55 cm wall. A different brick changes the count and leaves the volume of the walls the same.

With that brick, the Early Cypriot III house needs about 3,500 bricks: 44 m³ of earth dug near the plot, some 1.3 t of chaff to bind it and about 14,000 litres of water, weighing about 75 t when dry. The footings take some 4,100 stones and weigh about 56 t.

How high the walls stood is not known from Marki. The one Early Bronze Age argument for wall height on the island comes from Sotira-Kaminoudhia on the south coast, where stone walls fell over with their courses still in order. Stuart Swiny concluded that some “stood to at least 2 m and probably closer to 3 m in height”, and a burnt vernacular house he recorded at Old Paramali was 2.5 m high inside (Swiny 2003). The model raises the walls of the roofed rooms to 2.4 m; the evidence allows anything from 2.2 m to 3 m. Walls round the court carry no roof and stop at 2 m; the screen wall is 1.8 m, high enough to block a standing adult’s view.

No roof survives at Marki. Burnt collapse shows that roofs were made of timber, rushes and mud, and stone bases for posts survive in the middle of rooms. “Interior spaces, especially the larger rooms, often have central posts to support the roof” (Frankel and Webb 2012: 495). The same authors state that “so far as we can tell, buildings were only one storey in height” (2012: 493).

The posts follow a rule. At Marki roofed rooms “very rarely exceed 4.5 m in width”, and spaces 6 m or more across are read as courtyards (Frankel and Webb 2006a, as summarised by Falconer and Fall 2013). The east room of Compound 6 is about 4.7 m across inside, and the excavators found the stone base of its post. The south room is about 4.8 m across, and no post is drawn in it, so the model adds one on a stone base. Both posts carry a beam along the room, and the roof poles span from the walls to the beam.

The layers above the poles follow the flat roofs of traditional houses at Sotira and Old Paramali that Swiny recorded (Swiny 2003), an analogy from thousands of years later, used because nothing earlier survives. Round poles about 12 cm thick lie across the short span, 16 to 30 cm apart. Over them go reeds or branches at right angles, brush such as thyme and spiny burnet, 10 to 13 cm of earth laid dry and trodden, 5 to 7 cm of chaff-tempered mud plaster and a thin skin of tamped clay, renewed every year. Such a roof is 20 to 40 cm thick, weighs about 500 kg a square metre and falls gently towards one long wall to shed the rain.

In the model the whole package is 40 cm deep over the wall-head. The three rooms of the Early Cypriot III house need 49 poles, 240 m of them in all, and 14 m of beam on the posts. The finished roofs weigh about 29 t, some 460 kg a square metre, of which 11 m³ is dry earth. Swiny reckons a seasoned cypress pole at about 8 kg a metre, so a 4.5 m roof beam weighs 36 to 40 kg: two to a donkey, and two people to set each one in place. We know of no saw from Early Bronze Age Cyprus. Poles were felled with axes, stripped and adzed at the ends; the Philia levels at Marki produced no stone axes or adzes at all, since metal had already replaced them (Webb and Frankel 2007). Pine and oak grew in and around the valley (Adams and Simmons 1996), but we know of no identified roof timber from Marki.

The model gives the roof no eaves and no parapet, since we know of no evidence for either, and no roller or ladder. Whether families worked or slept on their roofs, and how they climbed up, nothing found at Marki can tell.

Marki’s doorways were 60 cm to 1.10 m wide. Their thresholds were simple, made of reused limestone blocks and the stubs of demolished walls, and the long monolithic sills found at Sotira “have not been recorded” here (Swiny 2003, citing Frankel and Webb 1996: 58). What Marki does have is pivot stones, many of them scored with rings of wear round the socket where a door swung for years (Frankel and Webb 2006a: 11). So its doors were real doors, hung on a post that turned in a stone. The model gives each doorway a slab threshold, a pivot stone inside on the hinge side, a leaf of split and adzed boards and a lintel of three poles. The height of the doorways, set at 1.7 m, is a guess. We know of no windows. The walls do not survive high enough to show them, and the taller walls at Alambra show none either, though windows there “may have occurred at a higher level than is preserved” (Coleman et al. 1996).

Floors were hard-packed mud, “often sloping and uneven”, smoother in the inner rooms than in the courts (Frankel and Webb 2012). They rose as rubble was spread at each rebuilding, and sometimes builders cut them down by 20 to 30 cm. The lowered floor of the east room in Early Cypriot III is set 25 cm below the court in the model, which means digging out about 4.2 m³ of earth.

Built from the ground, the Early Cypriot III house in the model takes about 230 tonnes of stone, earth, timber and water to be dug, cut, carried and placed, about 21,000 litres of it water. Scaled from the rates measured when archaeologists built a round house at Lemba in the 1980s, with estimates where Lemba has no figure, the work comes to about 330 person-days: 95 for the footings, 45 for making the bricks, 65 for laying them, 45 for the roofs, 45 for render and plaster and 25 for floors and fittings. A working party of five would need about 65 days, best spent in the dry months, when bricks dry and earth roofs can be laid (a practical suggestion). The rate for making bricks, about 80 a person-day, is an estimate; we know of no measured figure from Cyprus. Real rebuildings cost less, because they reused the old stumps and much of the old stone.