Which boat could carry a cow?

Cyprus had no cattle, sheep, goats or deer of its own. Its native large mammals, the dwarf hippopotamus and the dwarf elephant, were gone by the time farmers arrived, and wild boar, good swimmers, can manage about 10 km of open water (Vigne et al. 2014). Every domestic animal in the early villages came by boat, and the bones show them coming in stages. The pig came first, with hunters, before about 10,500 BC. Dogs, cats and the house mouse followed in the 9th millennium BC. Goats and cattle arrived around 8400 BC, fallow deer and sheep two to five centuries after them.

The animals are found in the island’s earliest farming villages of the Neolithic: Parekklisha-Shillourokambos inland from Limassol, Akanthou on the north coast, Ais Yorkis in the west, and Kissonerga-Mylouthkia near Paphos, where the settlers sank wells. Obsidian came too, from Anatolia, and more of it reached Akanthou than Shillourokambos, which points to the shortest crossing, from the Cilician shore to the north coast (Vigne et al. 2014). Cereals came as seed. A single basket goes a long way: at the sowing rates of traditional Anatolian farmers, 100 kg of grain sows between one and two hectares (Hillman 1973), so seed was never the bulky part of the load.

The mice are the clock. Mainland house mice kept arriving often enough to stop the island population drifting away from the mainland stock, and Jean-Denis Vigne and his colleagues read that as “at least two successful crossings each year, and in all likelihood many more voyages than this (perhaps even ten or more trips in a given year)” (Vigne et al. 2014). Later boats brought new blood. Within a few centuries Shillourokambos had smaller domestic pigs, larger sheep “probably coming from Syria”, and larger cattle, found also at Ais Yorkis.

None of this can be the work of drift. The currents round Cyprus run anticlockwise at a quarter to half a knot, flow less than half the time, and run across the routes from the mainland (Bar-Yosef Mayer et al. 2015). A current that weak and that fitful would take days to move a log 70 km, and would not aim it at Cyprus. A log washed off a beach does not land both sexes of six species, with seed corn and obsidian, again and again for centuries. The cargo proves planned and repeated voyaging by people who knew the way. About the boats themselves it proves only two things: that they existed, and that they carried calves.

The oldest boats found anywhere in the Mediterranean lay in the mud of Lake Bracciano, north of Rome, where a Neolithic village now called La Marmotta was drowned. Five logboats came out of it, dated between about 5600 and 5100 BC (Gibaja et al. 2024, find). That is some three thousand years after the first cattle reached Cyprus, at the far end of the sea, on a lake.

La Marmotta 1 is the one that matters. It is a single oak trunk hollowed into a boat 10.43 m long, 1.15 m wide at the stern and 0.85 m at the bow, its sides 65 cm high aft and 44 cm forward. Four low bands of wood were left standing across the floor as it was hollowed, stiffening the bottom, and the inside is burnt: its builders used fire as well as their polished stone axes and adzes (Gibaja et al. 2024). The other four are narrower and of softer wood. Paddles came from the same village, one of them 1.05 m long.

Older accounts credit Marmotta 1 with raised gunwales joined by mortise and tenon, and with a mast step. Neither appears in the 2024 publication, and both are left out here. Three wooden objects shaped like a T and pierced with two, three and four holes are harder to set aside. The excavators suggest they fastened ropes to a sail, to a stabiliser, or to a second hull lashed alongside as a catamaran, and they do not choose.

Nothing of the kind is known from the eastern Mediterranean, Anatolia or the Levant: we know of no logboat, paddle, boat model or picture of a boat there before about 5000 BC. The closest thing to boatbuilding evidence comes from Atlit-Yam, a village of about 7000 BC now under the sea off the Carmel coast of Israel, whose hafted and heavily resharpened axes felled trees and split green logs. The team that studied the tools write that the village’s chisels and adze “may have been used for water craft construction, but there is no unequivocal evidence” (Yerkes, Galili and Barkai 2014).

The single dugout on this page is drawn on La Marmotta 1 and a little shorter, as a boat of the same kind; the lines drawing lays the Italian outline over it. In 1998 a replica of Marmotta 1, Monoxylon II, was rowed some 800 km along the coast from Italy to Portugal by eight to ten rowers and a steersman (experiment); its best day was 57 km in fourteen hours, and Vigne credits it with more than 1,000 kg of carrying capacity (Vigne et al. 2014, after Tichý). It appears in the model as a dashed ghost. A replica tests an idea about a boat; it is no evidence for the boat itself.

People hollowed logs into boats with stone tools in Europe, Africa, Asia and the Americas. The oldest that survive are a little younger than the first Cyprus crossings. A small pine boat from Pesse in the Netherlands dates between about 8000 and 7000 BC. Dufuna, from a river bed in north-east Nigeria, is African mahogany, 8.4 m long and about half a metre wide, with sides “barely more than 5 cm” thick, and dates to the 7th millennium BC (Breunig et al. 1996). At Kuahuqiao in the lower Yangzi valley a pine logboat of about 6000 BC lay beside stone adzes, wooden adze handles and the flakes struck from them, where it had been worked (Liu and Chen 2012). On the Danish coast the Ertebølle fishers paddled boats of lime and aspen with small clay hearths aboard; Tybrind Vig 1 is 9.5 m by 0.65 m (Andersen 1994). The longest Stone Age logboat in northern Europe, Stralsund 1 on the German Baltic, is about 12 m long and only 0.6 m wide (Klooß and Lübke 2009).

Drawn at one scale, they show two things. Length runs from about six to twelve metres, eight to ten is normal for a big boat, and twelve is the known ceiling. Width is the hard part. Boats of lime, aspen, alder and beech are half a metre to 70 cm across. Only oak boats, La Marmotta 1 and the Neolithic boats from Paris-Bercy, reach 0.85 to 1.2 m (Arnold 1998), and a calf needs that width. Søren Andersen, who published the Danish boats, judged the slender Tybrind type “very difficult to handle” and fit only for sheltered, calm water.

Seagoing dugouts recorded by travellers were bigger, and they were seldom plain logs. Columbus, on 13 October 1492, saw canoes in the Bahamas “made from the trunk of a tree like a long boat, and all of one piece”, some carrying forty or forty-five men; Oviedo on Hispaniola saw forty or fifty aboard, in hulls flat underneath with no keel (ethnohistoric). Those came from Caribbean hardwoods far larger than anything on the Levant coast. The Nuu-chah-nulth of Vancouver Island hunted whales from cedar canoes nine to eleven metres long with a crew of eight, and they reached a beam of 1.5 m by filling the hollowed log with water, heating it with hot stones and forcing the sides apart (ethnographic). On the Gold Coast around 1680 the French trader Jean Barbot recorded cargo canoes of 35 to 40 feet, five feet wide; West Africa’s largest canoes, up to some 24 m with a hundred men, worked in calm water (Smith 1970).

The pattern holds across these records. The plain log is a river, lake and sheltered-coast boat. For open water, its users spread it, raised its sides with planks, paired it or gave it outriggers. Oak cannot be spread: we know of no case of an oak logboat being expanded (McGrail 1987). That leaves raising and pairing, and both are built here. The single dugout at 10 m sits near the top of the Stone Age range, long enough for eight paddlers and a steersman. A longer log adds paddlers and speed; it adds no width, and width is what a calf needs.

The first boats were built on the mainland shore with the mainland’s tools. By the 9th millennium BC the woodworker of the Levant had a flint axe with a ground and polished edge, about 120 g on average, in a straight haft; adzes hafted crosswise, which were becoming more common; wedges of wood and antler, flint blades and scrapers, and fire (Yerkes and Barkai 2008). Polished stone axes reached Cyprus early: they are found at Ayia Varvara-Asprokremnos, a site of about 8900 to 8600 BC, and were laid in a grave at Shillourokambos. Specialised adzes for woodworking appear on the island only in the Late Neolithic, so the Build mode equips its workers from the mainland kit.

We know of no Mediterranean or Near Eastern text that describes making a dugout. The method has to be borrowed, and the best witnesses are Europeans who watched it done in the Americas. On Roanoke Island, in what the English called Virginia, Thomas Harriot recorded the whole sequence in 1585–86 (Harriot 1590). A small fire was kindled round the root of the tree, fed “by little, and little” so that “the flame should not mount opp to highe”, until the tree fell “of yt owne accord”. The crown was burnt off, and the trunk raised on poles “laid ouer cross wise vppon forked posts” at a height where the builders could work it. They took off the bark with shells, then made a fire along the length of the log, “sauinge at both the endes”, quenched it, scraped away the burnt wood, lit a new fire, “and soe they continue sometymes burninge and sometymes scrapinge, vntill the boate haue sufficient bothowmes.”

Others saw variations. On Hispaniola Gonzalo Fernández de Oviedo watched the Taíno chop first with hafted stone axes, then burn what the blows had bruised, little by little, kill the fire and chop again. In Louisiana Antoine-Simon Le Page du Pratz (1758) describes rolls of earth mortar laid along both sides and across each end to keep the fire off the edges, and more mortar on any side the fire began to eat. André Pénicaut, on the Gulf coast in 1699, saw the fire put out “with thick mud”, the char scraped with cockle shells “as thick as a man’s finger”, and the hollow washed.

All of them are working with the same physics. Char insulates the wood beneath it, so a burn stalls after a centimetre or so and must be scraped back before the next fire will bite. The oak boats from Paris-Bercy were burnt and adzed in layers about a centimetre deep, with adze scars 3 to 6 cm wide running in rows along the bottom (Arnold 1998), and a team in Pennsylvania burning a pine log for two to four hours a day took off less than three-quarters of an inch a day. The Build mode follows that rhythm: many shallow burns, a low fire moved from bay to bay, wet earth along the gunwales, a blunt char adze and shells between fires, and the four bands of La Marmotta left standing as ribs. Unburnt bands like these may also have shown the builders how deep the floor had gone.

Fire helps, and the evidence allows a boat made without it: in 2003–04 two novices hollowed an oak roughout in thirty days with grooves and wedges alone (Rouzo and Poissonnier 2007, experiment). The finds under-record fire in any case, since adzing can remove every trace of it (Watts 2014) and decayed wood has been mistaken for char (McGrail 1987).

The work is long. Between 2017 and 2018 a Japanese team felled a cedar a metre thick with replica stone axes in six days, 12 hours 38 minutes of actual chopping and 36,459 strokes, and none of their replica Palaeolithic axes lasted much beyond 6,000 strokes (Kaifu et al. 2025, experiment). Our oak is bigger and harder. American replicas of prehistoric dugouts have taken 350 to 850 hours of burning, scraping and adzing each, copying canoes made mainly of cypress and pine (Yerkes and Koldehoff 2018). For the single oak dugout we estimate about 150 to 300 person-days, most likely 200: a crew of six working for one to two months, spread over a winter and a spring. The paired craft, with its second hull, beams, deck and pen, needs about 320 to 620. Nearly nine-tenths of the log ends as chips, char and smoke. The Build mode counts the days as the steps go by.

A weaned calf of seven or eight months weighs 100 to 150 kg; a grown cow, at least 500 (Vigne et al. 2014). Vigne argued that the calves must have crossed standing. “Keeping the ruminants without movement for more than 3 or 4 hours would have entailed serious physiological disorders,” he wrote, so “the PPN voyagers certainly did not transport the ruminants laying bound on the bottom of the boat; they had to make the voyage standing up”, and the crossing had to be made in about ten hours. Standing calves, in his view, needed a stiff, decked craft, and he set the single dugout aside.

The veterinary evidence supports half of that. Cattle tend to ruminate lying down, resting on the brisket (Schirmann et al. 2012). What harms them is lying on the side, where gas cannot escape the rumen and the animal bloats (Merck Veterinary Manual). Ruminants under anaesthetic are propped on the brisket for that reason, and feed is withheld beforehand to empty the rumen. Herders with small boats have long carried young stock lying and tied: the Irish curragh “will carry pigs or goats or sheep or calves with their legs tied, but not adult cattle or horses” (Dent and Machin Goodall 1988). A weaned calf fasted for a day, hobbled on its brisket with its head free and bedded on brushwood, is the way a herder would choose, and it may be the answer to Vigne’s problem. We know of no source that says how long such a calf can lie safely.

The single dugout is loaded both ways here. In the Afloat mode its draught, freeboard and stability are worked out from the hull’s own shape as each person, animal and basket goes aboard. The same hull takes eight paddlers, or a family with seed grain and piglets, or sheep and goats, or two calves lying hobbled on the floor, and the panel shows how much side stays above the water for each: about a quarter of a metre with any of them. Vigne’s own estimate for Monoxylon II was room for two or three weaned calves beside its crew.

Then stand the calves up. Their weight rises in a hull barely a metre wide, and when they shift their feet the boat heels, by about 10° in the model, where water would only come over the side at some 30°; load two standing calves and watch the freeboard and the roll. The wet oak of the hull is heavy and low and acts as ballast, so the log stays upright. Its plainest failure needs no calculation, and it is the one the panel warns of: nothing stops a standing calf from stepping over the side of an open log. Standing calves need a pen, and an open log has none. A grown cow does not fit at all. It is wider than the inside of the log.

Peoples who wanted more side on a dugout added planks. In the Lesser Antilles the Carib hollowed their seagoing pirogues “with stone axes and with a small fire” and raised the sides with planks (Du Tertre 1667); West African and north-west American canoe makers did the same. Sewing wood to a logboat is very old: a boat from Lystrup in Denmark had its cracks stitched through bored holes about 5100 BC (Andersen 1994). We know of no Stone Age boat with an added plank, so the extended dugout here, a plank a side sewn along the gunwale with bast cord, packed with moss and braced by three spars, is speculative. It raises the side the water has to climb. It does nothing for a standing calf, which can still step out.

For standing animals, the people who actually ferried cattle in logboats used two. On the Danish island of Hasselø in the nineteenth century, two parallel logboats were held apart by beams through both sides of each hull; four to six cattle or horses stood across them, forelegs in one boat and hind legs in the other, with men at their heads and four rowing (Rasmussen 1953, via McGrail 1987, ethnographic). Pack horses crossed the same way in Albania about 1900, five or six at a time. In Virginia in 1703 the Quaker traveller Thomas Chalkley wrote: “We put our horses into two canoes tied together, and our horses stood with their fore feet in one and their hind feet in the other.” Every one of these was a short crossing in sheltered water.

Other pairs carried a deck. On the river Miño in Galicia the barco de dornas had three timbers across two hulls, the outer two through square holes in both sides of each hull, pinned outside, under a platform of planks (McGrail 1987). In 1566, as the Spanish chronicler Gonzalo Solís de Merás recorded, the Calusa of south-west Florida came out in “twelve canoes, and two of them fastened one to the other, with decks covered with awnings of hoops and matting” (via Fitzpatrick 2013). Vigne suggested something of the kind for Cyprus: two or more dugouts lashed together with planks on top, so that calves could stand.

The paired craft here is built from those recorded parts. Two oak hulls of 9.5 m lie side by side a little more than half a metre apart. Two beams pass through both hulls near the ends and are pinned outside; three more are notched into the gunwales amidships, and a long pole lashed over all five stops the hulls working against each other. Split planks laid fore-and-aft make a deck, and on the deck stands a pen of woven hurdles, bedded with brushwood, where two calves stand crossways with handlers at their heads. The paddlers sit in the open hollows fore and aft. In the Afloat mode the pair is far stiffer than one log and takes the whole colonising load, people, animals and seed, with about 0.28 m of side still above the water. Its weakness is speed: the same paddlers now drive two hulls, a deck and a pen, and on the shortest route the model puts them about thirty hours at sea unless relief paddlers come too.

It is also the least attested craft on the page. We know of no prehistoric paired logboat; every recorded cattle ferry on logs is post-medieval. When Hannibal’s army crossed the Rhône in 218 BC it bought and built logboats and swam its horses behind them, a man on each side of the stern guiding three or four by their reins (Polybius 3.43). The only open-sea livestock carriers we know of in the ethnographic record are Polynesian double canoes, which were sailed and carried pigs, dogs and chickens, never cattle.

Bundles of reed float without any hollowing at all. At As-Sabiyah in Kuwait, a site founded between 5500 and 5000 BC, excavators found a clay model of a bundle boat and more than fifty lumps of bitumen with the print of reeds on one face and barnacles on the other: the coating of seagoing reed hulls (Carter 2006, find). Off north-west Corfu, fishermen paddled the papyrella, a raft of giant fennel stalks on a frame of cypress saplings, 2.1 to 2.4 m long and made for two men, until the Second World War; one survived to 1965 (Johnstone 1973). In 1988 Harry Tzalas built a six-paddler version a little over 5 m long and took it from Lavrion to the obsidian island of Melos: 72.5 nautical miles in eight paddling days at about 3 km/h, with eight more days stuck ashore by weather (Tzalas 1995, experiment).

The reed craft here is that six-paddler size. Water runs through the bundles, so only the stems give lift, and the crew sit wet. It is a credible boat for people. For calves it fails: a bundle raft bends with the sea, stays soaked, and has no stiff floor for a hoof. The reconstructed Magan reed boats in Oman showed “excessive flexibility”, and one of them sank at sea on its way to India. The Afloat panel rejects any calf aboard. Even eight paddlers, or a family with its seed and piglets, ride low in the water by the panel’s reckoning, and the sheep and goats, the cow and the colonising load all run out of room.

Hide boats are the other rival, and herders of cattle had hides to spare. Herodotus (1.194) describes round boats of hide on willow frames coming down the Euphrates to Babylon with “a live ass in each boat”, and the Irish curragh carried calves with their legs tied. In Norway in the 1970s Sverre Marstrander built a hide boat 6.68 m long from eight cow hides; it carried more than a tonne, and six paddlers drove it at 2.8 knots over a measured kilometre (Marstrander 1976, experiment). The hide boat here is larger, about 8.5 m, and would need the hides of some twenty cattle. It floats high and steady, takes every load on the panel except the whole colonising party, which runs out of room, and its weaknesses lie elsewhere. Skins waterlog and stretch unless dried between trips, a hoof can go through a hide floor, and we know of no hide boat, model or picture of one from the Neolithic Near East. The Near Eastern hide craft that are recorded, the round quffa and the kelek raft on inflated skins, are river boats.

The shortest crossing runs from Cape Anamur on the Anatolian coast to Cape Kormakitis in north-west Cyprus, about 70 km (69 km in Held’s measurement, 74 km in Bar-Yosef Mayer’s). Vigne prefers a start from the Adana plain, at least 80 to 90 km, using the current. From the Syrian coast to Cape Andreas is about 100 km. Sea level was perhaps 35 to 50 m lower than today, yet the shelves on both sides are so steep that the distances hardly change: in the words of Daniella Bar-Yosef Mayer and her colleagues, the change “has not played a major role” in the crossing (Bar-Yosef Mayer et al. 2015). A small island off Cape Andreas, if it existed, had gone under before the cattle came.

In spring, summer and autumn the wind blows mostly from the north-west, which from Anamur puts it on the paddlers’ quarter and from Syria in their faces. Land never drops out of sight. Simple geometry puts Troodos, at 1,951 m, above the horizon from a paddler’s seat some 160 km away, and the Kyrenia range at about 120 km; in clear air the mountains of one shore are in view until those of the other rise.

The experiments set the speed. Tzalas’s reed craft averaged about 3 km/h. Monoxylon II, rowed, made about 4 km/h on its best day. In 2019 Yousuke Kaifu’s crew of five paddled Sugime, a 7.5 m cedar dugout felled and hollowed with replica stone tools, 225 km from Taiwan to Yonaguni across the Kuroshio current in about 45 hours, steering by the stars, the swell and other natural signs; the boat carried only its crew, some food and water and two spare paddles, the crew were trained sea kayakers, and the escort ships passed them more water and food on the way (Kaifu et al. 2025, experiment). The only paddled craft known to hold the 7 km/h or more that a ten-hour crossing of the shortest strait demands are modern racing outriggers, whose crews are swapped every twenty or thirty minutes from an escort boat. A loaded dugout with one crew does nothing of the kind.

The crossing time in the Afloat mode comes from a speed model calibrated on those experiments, and it is labelled as our estimate. From Anamur it gives the hide boat about 17 to 21 hours, the single dugout about 20 to 24, and the paired craft about 30 with one crew of paddlers or about 24 with relief aboard. Every paddled craft on the page spends a night at sea on the shortest route. Vigne’s ten-hour limit appears on the timeline as his inference. If calves rode lying on the brisket, the limit may never have applied.

A sail would shorten the crossing, and Vigne considered one. Even then he reckoned on “hardly more than 7–8 km per hour”, ten hours or more. The sail is in any case far too late. The earliest pictures that might show one, a painted disc from As-Sabiyah and a clay model from Eridu with what may be a mast socket, belong to the 6th and 5th millennia BC and are disputed; the earliest certain sail is painted on an Egyptian jar of the Naqada II period, 3500 to 3100 BC, now in the British Museum; and sails reach the Aegean around 2000 BC. The model shows a sail only as a rejected what-if.

The bones rule out a single great voyage. Pigs, then dogs, cats and mice, then goats and cattle, then deer and sheep: the species arrived over more than two thousand years, at a rate Vigne puts at about one new species per thousand years rising to about eight, with fresh lineages topping up the herds and flocks long after the first landings (Vigne et al. 2014).

The arithmetic points the same way. If a single dugout carries its paddlers and a few passengers, or its paddlers and a pair of calves, then a farming community of a hundred or two hundred people needs a few dozen boatloads of people, and a breeding herd of cattle needs a run of cattle trips besides, all on fair-weather days in spring and autumn. That is a programme of several seasons or years. A demographic model of the earlier forager settlement of Cyprus concludes that a viable founding population ran to a thousand people or more, arriving in two or three large waves within a century (Bradshaw et al. 2024). It describes a different, older migration, and its dates are debated, but the scale of the problem is the same.

Vigne suggests that the crossings were in the hands of specialists, a group of experienced voyagers who knew the routes, the currents and the landing places, perhaps the stars, and who lived apart from the farmers they carried. The Afloat panel lets the reader set the load and count the trips. The villages that grew from those landings kept cattle for a while, then let them go. By the middle of the 8th millennium BC cattle had vanished from the island’s herds; the later villages of Kalavasos-Tenta and Khirokitia lived on sheep, goats, pigs and deer, and cattle did not return until the Bronze Age.

Every hull on this page is an argument. We know what crossed and roughly when. We know of no boat, paddle or picture of a boat from the eastern Mediterranean before about 5000 BC, and a single find of the 9th millennium BC would confirm or overturn everything here. We know of no prehistoric paired logboat. No source fixes how long a hobbled calf can lie safely, or whether some calves crossed unweaned, lighter but needing milk. The speed model rests on a rowed replica and a modern crew of experienced paddlers, and we know of no paddling experiment with a loaded craft lasting twenty hours. Big straight oaks near the departure coasts are assumed; the size of oak, Calabrian pine and cypress trunks on those coasts around 9000 BC is not established. After the first generation, some boats may have been built on Cyprus itself, of island oak or pine. Cedar is often suggested, but we know of no cedar timber from Cypriot contexts before the 5th century BC.

Several familiar pictures are rejected, and the last drawing shows them in the style kept for rejected ideas. Cattle swimming from Anatolia: swimming works for river crossings and narrow sounds, and wild boar manage about 10 km. A drifting raft: the currents are too weak and run the wrong way. One great ark. A grown cow in a single log, which does not fit. Standing cattle on a reed raft, which has no floor to stand on. A giant dugout of twenty metres with dozens of paddlers: no Stone Age logboat passes about twelve. An oak hull steamed and spread like a north-west American canoe. Rowers with oars in oarlocks, where every Stone Age find is a paddle. Pottery jars aboard, in a period without pottery: water probably went in skins, and seed in baskets of coiled cord coated with asphalt, like those preserved in the Nahal Hemar cave in the Judean desert (Bar-Yosef and Alon 1993). And the sail, three thousand years or more too early.

What remains is a cargo and a strait. The calves that reached Shillourokambos crossed 70 km or more of sea, in a craft that has left no trace, perhaps lying hobbled in the bottom of a log, and almost certainly through a night at sea.