A crew heaves against the hull, logs turn under the keel, and someone at the back picks up the last one and runs it to the front. That is how many people picture Vikings getting from one river to the next. Emperor Constantine VII, who described the business in the tenth century, speaks of dragging and carrying instead, and the scholar who surveyed the modern replica voyages found rolling over long distances unrealistic. The picture is not entirely wrong. It simply covers far less than it is usually asked to.
Read about the route from the Varangians to the Greeks and you soon meet the sentence that boats were hauled from river to river on rollers or logs.
By now there is more to go on than a plausible image. Since the 1980s, replicas from Sweden, Norway, Finland and Poland have travelled the rivers of eastern Europe. Alongside them stand an imperial description from the tenth century, saga passages, excavated timber sleepers and a canal on a Danish island. Together they point to no single method but to a toolbox.
Researchers call the place where a boat leaves the water a portage, from the French for carrying. The people involved had sharper words. In Scandinavia an eid (Swedish ed) is the natural isthmus or watershed, and a drag is the place where hauling actually went on. Gaelic tairbeart, behind Scotland's Tarberts, means a bringing-over. In Russian the place is a volok, a word tied to dragging. The Swedish maritime archaeologist Christer Westerdahl warns against reading every such place name as proof, because names migrate and get applied by resemblance.
Popular accounts blend things that differ technically.
True rollers are loose round timbers on which the hull moves forward, picked up behind and laid down again in front. Sleepers stay put. Scandinavians called them lunner, and the ship slides across them without their turning. Westerdahl describes them as usually cloven logs laid rounded side up, sometimes secured at the ends with stones or stakes. Wagons and wheels lift the hull off the ground. Carrying works only for light craft, in the extreme case after taking the boat apart.
A fifth option avoids the problem altogether. Only the cargo crosses, the ship stays where it is, and another waits on the far side. Much recent research ends up here.
The most important written source is not Scandinavian. Around the middle of the tenth century, Emperor Constantine VII Porphyrogennetos had a handbook compiled for his son, known today as De administrando imperio. Chapter 9 describes the annual journey of the Rus from Kiev to Constantinople. Our article on the Varangian route deals with the Old Norse-sounding names Constantine gives for the rapids beside their Slavic ones. What matters here is what the crews did with the obstacles.
At the first rapid the Rus put in to the bank, get out, strip and feel their way with their feet through the water, while others work the boat with poles at bow, midships and stern. That is no land route, but guiding a boat through the water. Only at the fourth rapid, which he renders as Aeifor, do the boats come out of the river. The cargo is unloaded, slaves in chains are led six miles along the bank, and the boats are partly dragged and partly carried on shoulders. Jenkins's modern English translation is still in copyright, so I give the passage in my own words.
Rollers are not mentioned. That does not prove they existed nowhere, since the text is no manual and covers one stretch of the route. It is still telling, because carrying on shoulders presupposes a certain class of boat. Nobody shoulders a ship of several tonnes.
Constantine calls the vessels monoxyla, literally "of one timber". According to him, tributary Slavic groups make them in winter, bring them to Kiev and sell them to the Rus, who fit them there with oars and other gear from their old boats. In an essay on inland traffic in Russia, Christer Westerdahl stressed that the word really does mean a dugout, whose sides could be raised with added planks. His conclusion goes further. No single vessel travelled the whole way from Scandinavia to Constantinople, he argues, and craft were changed along the route.
That shifts the original question. It no longer asks how a longship got across a watershed, but whether one ever stood there at all. The famous hulls from graves and harbour basins are seagoing ships with keel and mast step. They were made for crossing the Baltic, not for the shallow upper reaches of a Russian river. The archaeologist Rune Edberg, who took part in such voyages himself, notes that light, flat-bottomed, locally made boats were standard on the eastern rivers in historical times, and that Scandinavian travellers most likely used the same. Søren Sindbæk, cited by Edberg, reads sites such as Staraya Ladoga, Timerevo and Gnezdovo as places of storage and reloading at the turn between summer and winter transport. That is a well-argued reading. No find secures it for each individual voyage.
Edberg gathered the attempts from 1983 to 2006 in the journal Situne Dei. He says openly where a tractor helped.
The Swedish researcher Erik Nylén began with Krampmacken, about eight metres long and built after the boat find from Bulverket on Gotland. According to Westerdahl the structure there is dated to about 1120 and the boat to about 1200, so strictly speaking post-Viking. The first attempt in 1983 ended at Poland's eastern border. In 1985 the boat went up the Vistula, where the crew managed only about ten kilometres a day despite hard labour. Across the Dukla Pass in the Carpathians, 500 metres above sea level, they pulled it on wheels, partly with a horse, and finally reached Constantinople by way of the Ondava, Bodrog, Tisza and Danube.
The 16-metre Norwegian Havørn, a two-thirds scale copy of the Gokstad ship, tried the Daugava and Dnieper in 1992. At Jēkabpils in Latvia a 300-metre rapid running at eight knots cost it three days. It then rode a truck to the Dnieper at Mogilev and went on mostly under engine.
The Swedish Aifur, nine to 9.5 metres long depending on the source, with a hull weighing around 800 kilograms and about nine people aboard, takes her name from Constantine's fourth rapid. In 1996 the Lovat became impassable above Kholm in low water, the crew had no horses and accepted a tractor. She crossed the second watershed on simple wheels built on the spot, helped by members of a Smolensk club. More voyages followed up to 2001. Edberg concludes that the middle and upper Lovat, part of the classic route, was not a viable option for boats of this kind.
Later came the Finnish Heimlösa Rus on the Don and Volga (2000), the Swedish Himingläva (2004), which had oxen pull her across a Caucasus crossing of about 1,150 metres, and the Polish Welet (2006), which used locks, motor and truck as needed. Edberg explicitly rates the ox haul as more a sporting achievement than evidence for the tradition.
These voyages produced the sharpest verdict on the subject. In connection with Havørn, Edberg records that the popular image of crews rolling ships and boats long distances on logs came across as unrealistic. He adds more generally that keeled vessels proved relatively heavy to handle and that it is never easy to move ships far across land.
The reasons are not hard to see, though what follows is mechanical reasoning rather than a measured result. A clinker hull is light and flexible. On separate rollers, its weight rests on a few points of a narrow keel. The path has to be level, rollers sink in a bog, and uphill anything not held rolls back. A weight comparison gives the scale. Aifur weighs about 800 kilograms, while Hjemkomst, a large American replica, is listed at 16 tons in a register of Nordic replicas. The gap shows how little a trial with a small boat says about a large hull.
The replicas have their limits too, and those were named early. As early as 1985 Westerdahl published an essay asking whether the Krampmacken project rested on an illusion. In a later essay of 1992 he measures the boat against the basic rules for reconstructions and finds it too small at eight metres, since he assumes the ships of the Rus were larger. He rejects the idea of one and the same ship making the whole journey, all the more since Constantine himself names locally made boats.
A second problem is that today's rivers are not those of the Viking Age. Nylén thought they had hardly changed in a thousand years. Edberg disagrees and points to dams, dredging and straightening. A Dnieper rapid photographed in 1913 now lies under a reservoir. If the Lovat was too shallow in 1996, that says little for certain about the year 900.
The experiments also mostly lack the helpers. Modern teams travel with nine people and little time. A busy hauling place in the tenth century could look different. Westerdahl argues that people at such sites both assisted with transport and controlled passage, pointing to the huge cemetery of Gnezdovo near Smolensk, on the crossing between Dvina and Dnieper.
Then there is winter. Westerdahl thinks some portages may have been used even more in winter than in summer, and calls the winter half-year the best time for collecting furs. Edberg points to Sindbæk's thesis and to the many sleigh parts from Novgorod. For the middle and upper Lovat he reckons with travel on foot, on horseback or by sleigh. None of the summer voyages could test that.
Striking out every roller would go too far. Sverris saga, a nearly contemporary king's saga, tells how King Sverre of Norway, during his rise, had 40 men fetch ships from the Randsfjord and 40 men fell trees in the forest. The ships were then hauled in one go over five miles to Mjøsa with the freshly cut trunks as rollers. That at least is how David McCullough reports it in his 2000 Glasgow dissertation, drawing on Sephton's translation (pp. 20–22); I have not checked the saga text itself. It is the one passage that favours the rolling image over a longer stretch. The saga stresses, however, that such a haul had never been accomplished there before, which already hints that it was no routine. By then the Viking Age also lay about a century in the past.
Local tradition knows round timbers too. For Draget near Nynäs, Westerdahl quotes a report of a boat dragged across on round wooden pieces.
McCullough also tested it with his own hands. For his dissertation he took Borgundknarren, a replica Viking Age cargo vessel, to the isthmus of Mavis Grind in Shetland and ran three hauling trials there. His captions show a hull shored with props and "rollers" (the quotation marks are his), a harness, and a split in the keel from the first trial. I could not read the chapter with the results, so I give no figures. The split alone shows how vulnerable the one part is on which everything rests.
The finds lead towards fixed tracks more than towards rollers. Westerdahl cites excavated remains of sleepers on the Dragseidet in Hamarøy, about 3.5 metres apart, though dated only to the twelfth and thirteenth centuries. In northern Norway, 67 hauling places have been recorded. At West Tarbert in Kintyre, wood came to light in the 1990s, ploughed up according to a 2024 BBC report. Scotland's Canmore heritage database apparently records the same pieces: six oak timbers from a drainage ditch, the longest 4.96 metres. They were conjectured to be rollers or remains of a hauling surface, but tree-ring matching produced no date. The BBC says the wood was being reanalysed in 2024; I found no result. The same report mentions, in general terms, split timbers covered in slippery seaweed. For West Tarbert that is not attested.
The written sources mostly show kings at war, as Westerdahl himself points out. In 1064, according to tradition, Harald Hardrada took the smaller vessels of his fleet past Trollhättan to reach Håkon Jarl on Lake Vänern. In 1263 men of Håkon IV dragged boats overland from Loch Long to Loch Lomond, an episode our article on the Battle of Largs places in context. The saga says boats; the place and distance (Arrochar, a mile and a half) and the word light appear only in a Scottish local history. With Harald it is the smaller vessels, in 1263 not the ships themselves. Near Birka, McCullough lists two short crossings, Södertälje at about 100 metres and Draget at 500 to 1,000 metres, the order of magnitude at which hauling paid off.
Best known is Magnus Barefoot in 1098. The most useful sentence of the scene comes at the end, almost as an aside.
"…he had a skiff drawn over the strand at Cantire, and shipped the rudder of it. […] The king himself sat in the stern-sheets, and held the tiller; […] and there is a small neck of land between it and the mainland of Scotland, over which longships are often drawn."Snorri Sturluson, Heimskringla, Saga of Magnus Barefoot, ch. 11 (Laing's numbering); public-domain English translation by Samuel Laing, excerpts
Snorri wrote a good hundred years later, and the story itself is considered doubtful. The aside about longships, by contrast, describes what was apparently thought normal in his day. It is one of the few hints that larger warships crossed short necks of land too. It does not say how.
Closer to the North Sea lies the Schleswig isthmus. Hedeby lay at the head of the Schlei, which leads to the Baltic. In the west, ships came from the North Sea up the Eider and Treene to Hollingstedt. Between them lies only a strip of land, for which figures range from under 15 to about 18 kilometres. Crossing here saved the dangerous voyage around Jutland.
Excavations from 1995 to 1998 revealed two landing places at Hollingstedt and many Viking Age finds from the Rhineland and Norway. The site counts as the North Sea harbour of Hedeby and later of Schleswig, and in the twelfth century the bank was reinforced with timber platforms. The World Heritage portal and the entry by the state archaeology office describe the crossing as goods being shifted to ox carts and moving on along the Ox Road. In this account, ships stay on their own side of the isthmus.
Popular texts and the English Wikipedia, by contrast, say that ships too were pulled overland on a corduroy road. I found no such track for ships along the full distance in the archaeological accounts. That cannot be clearly proven. There is, however, a written source for a single case. According to Westerdahl, Saxo Grammaticus reports that King Svend Grathe secretly brought ships into the Schlei in 1151 and hauled them overland to the Eider to attack his rival Knud Magnusson and his Frisian allies. Saxo remarks that this brought him more trouble than gain, though Westerdahl leaves open what is meant by that. I have not seen the Latin text for this. Westerdahl explicitly cites the case as exceptional, which argues against a regular ship route. Elsewhere, too, often only the cargo went by land, as Westerdahl notes for German traders between Dvina and Dnieper in the thirteenth century.
The most radical answer to an isthmus is on Samsø. The Kanhave Canal cuts the island at its narrowest point, between Sælvig Bay in the west and Stavns Fjord in the east. It is just over 500 metres long, about eleven to twelve metres wide and about 1.5 metres deep. The sides were lined with horizontal planks held by driven posts. The sampled timbers are beech and oak, including oak diagonal braces. More than 11,000 cubic metres of sand were moved.
The dendrochronology, or tree-ring dating, is among the best anywhere. The Danish National Museum's 1996 report dates oak samples to summer 726, another to 729, and the beech samples probably also to 726. Some overviews give 727, which is a discrepancy. The canal is therefore about 70 years older than the Lindisfarne raid and no Viking work in the narrower sense, but the project of an older power able to command a large workforce.
For our question it is doubly interesting. According to the Danish heritage agency, it was not open to the west, so ships were hauled or moved on wooden rollers over the last stretch. A towpath runs alongside, and the Danish encyclopedia names flat-bottomed boats as its users. These are interpretations, not finds of rollers. The agency speaks of warships of the snekkja type, which cannot be proven for the year 726. Evidence of trade is absent, the agency points to ship maintenance and speaks of a temporary naval harbour. Either way, the canal shows that someone before the Viking Age preferred digging to hauling ships across the entire island.
The most spectacular ship on land is also the least believable. Under the year 907, the Old Russian Primary Chronicle says the Greeks had closed the harbour. Oleg then had his warriors build wheels and set the ships on them. With a favourable wind they raised sail and bore down on the city by land. The Greeks asked for peace in fear, and Oleg hung his shield on the gate.
Scholars read this almost universally as legend. The text was written some two hundred years later, Byzantine sources are silent on the campaign, and in 2021 Aleksandr Romensky classed the episode in Studia Ceranea as folkloric-epic material. He sees behind the story rather an entry into Byzantine military service after 904. Some researchers, among them Grégoire and Dolley, regard the whole campaign as invention, while others take the historical treaty of 911 as a firm core; see our piece on the Rus treaties. McCullough drily notes that sailing on wheels lies beyond the capacity of most vessels.
Romensky even finds the wheeled ruse conceivable because of parallels, but cannot rule out that the chronicler lifted it from his sources. A comparison with 1453 is useful, when Mehmed II had ships brought over the hill into the Golden Horn, as his contemporary Kritovoulos describes. There the ships stand on large cradles, the track is laid with beams, and soldiers and capstans pull on cables. Crews raised sails here too, but as theatre, not as propulsion. Eight stadia as an affair of state show how much effort such a haul really took. I know of no link between the two stories.
Did Vikings really haul ships over land? Yes, several sources say so. Short necks of land, boats of manageable size and kings on campaign are the most common setting. Long distances with large hulls are not attested as normal practice; Sverris saga tells of a five-mile haul as an exception.
How many people did it take? No figure can be given for the Viking Age. In Westerdahl's estimate, the full crew of a large rowing ship could probably manage short stretches itself, while longer ones called for animals or local helpers. From late tradition at Listeid he cites two horses for a boat of 40 to 50 barrels (about 125 litres each).
Where can you still see any of this? Traces of the Kanhave Canal on Samsø are visible in the landscape. Nothing of Hollingstedt's Viking Age harbour shows above ground.
Why not simply build canals? A canal like Kanhave paid off only on short stretches and only for a power able to organise a large workforce. On Russian watersheds the scale would have been quite different. No such works are known there from the Viking Age.
The Varangian route · Vikings in the East · Longships · Oleg of Kiev · The Rus treaties · Hedeby · The Danevirke · The Ox Road · Magnus Barefoot · The Skuldelev ships
The Varangian route · The longships · The Vikings in the East – Varangians, the Rus & Byzantium · Oleg of Kyiv · Viking Shipbuilding.
The sun wheel is one of humanity's oldest signs – around the world, since the Stone Age. Precisely for that reason …
Read more →Seiðr is the ritual magic of the Old Norse world, used by a seeress – the völva – to learn hidden things in trance and …
Read more →The svefnþorn, the "sleep-thorn", is the instrument with which, in the Volsung tradition, Odin puts a valkyrie to sleep …
Read more →Search for "Viking symbol birth month". You get tables: January the hammer of Thor, April the Vegvísir, October the …
Read more →