Some notes on earthworks and dykes in Iceland and in the North Atlantic

Introduction
A complex system of interconnected earthworks in the county of Suður-Þingeyjarsýsla, NE-Iceland, is discussed in Arch. Isl. vol 2 (Einarsson, Hansson & Vésteinsson 2002). These extensive, turf-built earthworks have been known from this area as well as from other parts of Iceland for a long time; the structures in Suður-Þingeyjarsýsla were discussed by Kristian Kålund as early as in the 1870s (Kålund 1879:164). As there appears to be some interesting similarities between the earthworks in Iceland and structures found in Orkney and Shetland and in W Norway, this paper gives a description of the structures in the latter areas, and suggests possible interpretations that could have some relevance for Iceland, as well.

Regarding the earthworks in Suður-Þingeyjarsýsla the preliminary, but highly interesting conclusion of the above-mentioned study is that they date from the Middle Ages and form a pattern which suggests they once functioned as boundaries between adjacent farms (the transverse earthworks) and between the farms’ homelands and the commons (the horizontal earthworks, see Einarsson, Hansson & Vésteinsson 2002:61).

Dykes and earthworks
Today, both the Orkney and the W Norwegian farming landscape are characterised by the many stone fences. But in the pre-improvement period the so-called hill dyke (ON utgardr) was the only fence used in both these regions. Still, the age of this system is not all clear, and the history and development of dykes in the areas in question is a rather complex one.

Some of the similarities between the traditional farming landscape in Orkney and Shetland and in W Norway stem from the fact that both these areas were part of the heath landscape so characteristic of Europe’s Atlantic coasts. Over much of this region variants of the infield-outfield system were present in the medieval period. The infield was a relatively restricted area of arable land kept under permanent cultivation by the liberal application of manure to it each year. This system was present wherever pastoralism predominated and there was abundant land for pasture.

The farming systems in Orkney and in W Norway were both variants of the open field system. According to one definition, the open field system consisted basically of four elements: First, arable land was divided into strips owned or tenanted by various people each of whom farmed a number of these scattered about the fields. Then, both arable and meadowland were pastured by the stock of the same farmers after the harvest and in fallow years. Thirdly, the pasture and waste were used by the farmers for grazing their stock, often with strict control on the numbers of animals allowed. Finally, all these activities had to be organised by a formal meeting of the farmers, either at a manorial court or at a village assembly (Taylor 1975:71; cf. Dodgshon 1980).

Farming in the heath landscape of Atlantic Europe required that an equilibrium was found and maintained between, one the one hand, arable land, and on the other, meadow land. It is this delicate balance of forces that makes the hill dyke ‘a line of fundamental importance’ (Thomson 2001:322). The hill dyke was ‘not a fortress wall that encapsulated the community, not a dividing line, but a link between the elements of the township’ (Fenton 1978:89).

W Norway
In traditional W Norwegian farming, there were no dykes in the ‘innmark’ (homelands), then. Smaller, raised stones in each end of the strip marked strips belonging to different farmers. In some areas (i. e. areas where the stock was moved to and from the byre each day) stone-lined cattle-roads (geil) leading from the farm buildings and to the hill dyke could be seen. Outside the hill dyke, a number of enclosures for tending animals might be found. But, as a rule, dykes were not used to separate different belongings. The exception was along the boundary between different tax farms. When the ‘innmark’ of two or more different tax farms lay together, a dyke was normally marking the farm boundaries. The same boundaries were not, however, marked by fences outside the hill dyke. The dyke dividing the ‘innmark’ of two farms must in many instances be a consequence of the so-called ‘odalsskipte’ regulated by the Medieval Gulathing Law. This kind of fence was constructed of either stone or turf, as were the hill dykes. Stone fences seem to have been almost exclusively used in more recent times, due to the nearly unlimited supply of boulders resulting from the use of heavier machinery and the cultivation of new lands associated with the improved agriculture of the 19th century. In the Middle Ages, however, the picture is not so clear. Some interesting instances of earthworks which probably date back to the Middle Ages, are known. The Fjotland farm in Kvinesdal was surrounded by an extensive earthworks (fencing off the infields from the outlands, as they can be reconstructed from an 1865 severance map), while really substantial earthworks are known from Ytre Egeland, Kvinesdal, and Sangvik, Søgne, all in Vest-Agder county, SW Norway (on a smaller scale, it might be mentioned that Norwegian churches and churchyards known to have been abandonded in the Middle Ages already, often are surrounded by earthen walls, for examples, see Brendalsmo & Stylegar 2003). Farm boundaries outside the hill dykes were usually not fenced until the mid-18th century.

This system was predated by a similar, but much more small-scale system in the Migration period (c. 350-550 AD). At some point, after c. 550 AD and maybe as late as the Viking Age, the hill dykes were moved and the infields made much larger. In several cases, the location of the farms also changed, and the total number of farms decreased.
The Norwegian agrarian historian Ottar Rønneseth has argued that this restructuring is in fact a medieval phenomenon, that probably took place as late as by the period of the Landslov (AD 1274). He further claims that this restructuring, whereby the hill dykes were moved so as to incorporate large areas of meadow lands, also involved the appearance of farm boundaries in the form of straight lines, so that the lands of each farm would make geometrical figures. At the same time, the farms were matriculated for the first time, so that each farm became an independent tax object (Rønneseth 2001). This locking of the farming landscape is of course related to the fiscal valuation of all settlement units that finds its expression in the so-called ‘boltall’; the normative land rent numbers divided into ‘bol’ units of different name, that the Landslov reckons with.

Orkney
Most of Orkney’s households were to be found in multiple-occupancy townships. However, there were a large number of single farms, the biggest of which were of township-size (Thomson 2001:315ff). The settlement structure within each township could be very complex. Grimeston in Harray, Mainland consisted of ’several smaller towns or districts, such as Isbister, Horraquoy, Beneath the Dyke etc’ (Sandnes 1996:66). This last ’district’ was actually surrounded by a separate dyke. The farms Vola, Nistaben, Behind the Town, and Windywa’s belonged to Beneath the Dyke (ibid.). The hill dyke divided the township from the common moors (Clouston 1932:350).

In Orkney too there are indications that townships at some point have been expanded. As Thomson writes,

‘in a number of townships there are references to an inner dyke known as ‘the auld bow’. We find the term being used in two senses; primarily it was applied to the dyke itself, and less frequently it was used to describe the land enclosed by the dyke (the old bu, the old farm). The second usage seems to imply that there was a stage of township expansion when the hill-dyke had been ”flitted out” to take in the grassland of the backs. By the time the ”auld bow” was recorded, the dyke seems invariably to have been ruinous, and sometimes its very location was uncertain. Inner dykes of this kind are not unique to Orkney; a similar old dyke is a regular feature of Shetland townships, and indeed an inner dyke was sometimes found in Scotland. The inner dyke, as long as it remained intact, created an intermediate zone, inside the hill dyke but distinct from the arable land; it allowed the backs to be used as a series of peripheral grazing enclosures’ (Thomson 2001:328).

What about the dating for the re-location of the hill dykes and the expansion of the Orkney townships, then? Much of the more substantial evidence for the age of the different types of dykes in the Scottish Northern Isles is from Shetland.

The small island of Papa Stour, where large-scale archaeological investigations have been carried out in recent years, can serve as a case in point. Here, it is clear that the present day two-fold division of the island, marked by a stone hill dyke, was established by c. AD 1300 (Crawford & Smith 1999). At that time, documentary, place-name and archaeological evidence indicate that settlement was limited to the eastern half of the island. In earlier periods settlement in Papa Stour was more widespread. Remains of prehistoric house sites and burnt mounds are found both in the hills and within the hill dyke (Carter & Davidson 1998:130).

Another Shetland site, that of the abandoned farm of Kebister in Mainland, have yielded more information relevant to this discussion. Here, several radiocarbon dates from dykes are available. A number of dykes were found at Kebister, including some that are being interpreted as a series of subsequent hill dykes (Owen & Lowe 1999). The ‘March Dyke’ is thought to have been the final hill dyke at Kebister. No dating evidence for this dyke is available, but Brian Smith believes the final hill dyke to have been constructed as late as the end of the 18th or the beginning of the 19th century (Smith 1999:20).

A second dyke, termed Dyke 2, may have served as an early hill dyke, enclosing an area of c. 12 ha (Owen & Lowe 1999:289). This dyke seems to have been built during the 12th or 13th centuries AD (Butler 1999:64). There is evidence for arable expansion at Kebister in this period, and another dyke, Dyke 3, might have been constructed in the same period to mark a subdivision of the arable land. However, there seems to have been an active farming community at the site already in the late Pictish and/or early Norse period.

The lesson regarding the age of the different kinds of dykes, summed up by Owen and Lowe, is worth quoting:

‘The assumption for too long has been that sub-peat dykes, of which many survive in Shetland, are usually prehistoric features. In the light of the evidence from Kebister, even allowing that some elements of the Kebister landscape may represent Iron Age survivals, and despite the evidence of the Shurton Hill prehistoric sub-peat dyke, this assumption now requires review’ (1999:290).

The date of Dyke 2 in Kebister, together with the evidence from Papa Stour, seems to suggest certain similarities between the development pattern in the Scottish Northern Isles and W Norway.

The ‘great chains’
However, there is yet another common aspect of the dykes in Orkney and in W Norway that should be considered relevant for the Icelandic case, as well. A late 13th century charter from Iceland mentions an earthwork fencing off the whole scattered hamlet of Selvogur in the SW part of the country from its outfield (Einarsson, Hansson & Vésteinsson 2003:70). And, indeed, dykes fencing off areas considerably larger than single farms are a feature common to W Norway and the Scottish Northern Isles.

In Orkney, two or three townships could also be surrounded by a common dyke. There are a number of examples for this; the dyke encircling the townships of Netherbrough and Russland in Harray, Mainland, being one (Sandnes 1996:83). Such ‘great chains’ seem to have been a rather widespread feature of the Orkney landscape, as in many instances the hill dykes of several townships were connected, such as can be reconstructed from Murdoch Mackenzie’s charts from 1750 (Stylegar in press).

‘The huge number of stone fences that is so striking for anyone who wisits Jæren, are almost all of them from the period after 1800. (…) The fences of olden days could run for several kilometres, and had no obvious connection with the farm boundaries of recent times. (…) They are usually only visible as long ridges, and only seldom are any stones visible above the surface’ (Aanestad 1911:555). Thus wrote the agronomist Sigurd Aanestad, looking back from his vantage point in 1911, at the very end of a period which had witnessed huge improvements in agricultural methods and practices and the resulting enormous changes to the settlement structure and the cultural landscape of his native Jæren.

These ‘fences of olden days’ were the hill dykes. In the more densely settled regions in W Norway, for instance in the lower-lying settlement districts in Jæren, the hill dyke of one farm might be linked to the hill dykes of a number of other farms, thus constituting a great chain that joined several neighbouring farms. In many cases, some or all of the farms within such a system of dykes had some sort of common arrangement for the exploitation of the different resources in the outlands – be it pasture, peat cutting, hunting, or fisheries.

There are a number of good examples from Jæren, as well as from the other SW and W counties. When it comes to the question of common use rights, an interesting case in point is Vik in Sogn og Fjordane county, where c. 20 farms had their homelands within one ‘bøgard’, i.e. hill dyke. In early spring and late autumn the gates in the hill dyke were thown open, and the animals from all the farms were allowed to roam free and graze on the arable lands in this whole area. This system lasted until the 1860s (Frimannslund 1961).

Some characteristic features seem to distinguish many of the areas within such ‘great chains’ – or the ‘infield complexes’, as one could call them. The farms located within one such dyke system have obvious similarities with the ‘large complex settlements’ known from Iceland and Greenland, see Vésteinsson, McGovern & Keller 2002. More often than not, the later parish churches are located inside these complexes. As for place names, it seems that the majority of names indicating rather late farm divisions (i.e. names ending in bø-, -hus-, gard-, tun- etc.) are found within the complexes. In Åseral, a mountain valley in the interior of Vest-Agder county, the farms Underberg, Åbø, Gard, Austrhus and Åsland are located in the vicinity of the parish church. A common hill dyke surrounded all these farms. In Lista, also in Vest-Agder, this was the case with the farms Huseby, Lunde, Torp and Hauge, plus a number of lesser settlements. Some of these names suggest a rather late farm division. But how late? In one case do we get some more substantial information regarding when one such complex was partitioned. A raised stone still standing in the dyke dividing Lunde from Huseby has a runic inscription stating that ‘her skipter morkunne’, i.e. ‘here is the boundary between the farms’. The inscription is dated to the early 12th century. Provided that the runic stone is as old as the dyke itself, this might indicate that the complex settlements were broken up in the Early Middle Ages (cf. Stylegar 2001).

Social and economic background
The Norwegian legal historian Absalon Taranger thought it likely that the original hill dykes fenced off whole hamlets (’grender’), even if a ’grend’ was comprised of several farms (Taranger 1907:272). In a similar vein, J. Store Clouston argued that many Orkney townships, judging from their great size and the written recordings concerning them, had never been single farms at all, but always considerably greater areas (Clouston 1932:347f.). But what kind of social and economic organisation can account for these complexes? After all, the hill dykes circumscribing one such complex must (for obvious reasons, i.e. the practical needs of animal husbandry) have been constructed in one, great operation (cf. Rønneseth 2001). It is clear that not only were a great deal of planning involved in this operation, but also – probably – a great deal of force, or the threat of force.

Farms lying withing the ‘infield complexes’ in Rogaland and Vest-Agder have yielded some of the most important Late Iron Age and Viking Age archaeological finds in SW Norway. This, together with the the characteristic features discussed earlier (dominance of ‘young’ name types, location of parish churches etc.) and the degree of cooperation necessary for the homelands-commons system to function within such ‘great chains’ of hill dykes, suggest that the W Norwegian ‘infield complexes’ are in fact large complex settlements that might have been organised on an estate basis, probably with a number of dependant settlements. The breaking up of these estate complexes could in fact, then, be related not only to the establishment of individual tax farms, but also to the transition from thralldom to tenant farming (Stylegar 2001:30-32). The same trajectory – i.e. an ’original’ settlement structure dominated by large, complex estate-like settlements – could be the answer to some of Clouston’s questions regarding the Orkney townships, too, and it is one possible explanation for ‘the considerably greater areas’ than single farms being fenced off in many areas in the Scottish Northern isles.

What is the relevance of all this for the Icelandic earthworks, then? Well, the suggested similarities between these earthworks and the great chains of hill dykes in W Norway and the Scottish Northerns isles, logically lead to a number of questions, the answering of which might help us to throw new light on the Viking Age and Medieval settlement structure in Iceland. I will conclude by asking some of these questions.

Are the preceived similarities between, on the one hand, the Icelandic earthworks and, on the other, the great chains of hill dykes in W Norway and in Orkney, superficial, or can they be shown to be more substantial? And, if the latter is the case, what kind of settlement structure and/or farming system can account for the similarities? Do the transverse earthworks represent a younger phase than the horizontal ones, perhaps indicating that huge areas originally demarcated by the horizontal earthworks, were breaking up? Is it possible that the dyke systems incorporated whole landnams? Was the Icelandic landnam society aristocratic and estate-based already from the outset – and was the ‘virgin’ landscape of Iceland organized after Scandinavian, specifically W Norwegian, models (as I suspect the Orkney landscape, by no means a ‘virgin’ landscape at the time of the Norse landnam, was, see Stylegar in press)?

References
Aanestad, Sigurd (1911), ”Norges ældste jordbruksformer”, Tidsskrift for Det norske
Landbruk, Vol 18, 554-557.
Brendalsmo, A. Jan & Stylegar, Frans-Arne (2003), “Runde kirkegårder”,
Fortidsminneforeningens årbok 2003, 164-176.
Butler, Simon (1999), “Fossil pollen profiles and landscape history”, in Owen, Olwyn &
Lowe, Christopher, Kebister. The four-thousand-year-old story of one Shetland
township, Edinburgh.
Einarsson, Árni, Hansson, Oddgeir & Vésteinsson, Orri (2002) “An extensive system of
medieval earthworks in Northeast Iceland”, Archaeologia islandica 2, 61-73.
Carter, Stephen & Davidson, Donald (1998), “Interpreting the soil landscape of Papa Stour”,
in Turner, Val (ed.), The shaping of Shetland. Developments in Shetland landscape
archaeology.
Clouston, J. Storer (1932), A History of Orkney, Kirkwall.
Crawford, Barbara E. & Smith, Beverley B. (1999), The Biggings, Papa Stour, Shetland. The
history and archaeology of a royal Norwegian farm, Edinburgh.
Dodgshon, Robert A. (1980), The origin of British field systems: an interpretation, London.
Fenton, Alexander (1978), The Northern Isles: Orkney and Shetland, Edinburgh.
Frimannslund, Rigmor (1961), “Hegn”, Kulturhistorisk leksikon for nordisk middelalder, VI,
287-290.
Kålund, Kristian (1879), Bidrag til en historisk-topografisk Beskrivelse af Island, II,
Copenhagen.
Lamb, Raymond G. (1983), “The Orkney Trebs”, in Chapman, John & Mytum, Harold (eds.),
Settlement in North Britain, 1000 BC-AD 1000, Oxford.
Marwick, Hugh (1995a) [1923], “Place-names of North Ronaldsay”, Selected Papers, 1,
Livingston
Marwick, Hugh (1995b) [1923], “Antiquarian notes on Sanday”, Selected Papers, 1,
Livingston.
Owen, Olwyn & Lowe, Christopher (1999), Kebister. The four-thousand-year-old story of
one Shetland township, Edinburgh
Rønneseth, Ottar (2001), Gard og gjerde. Faser i utviklingen av Jærens kulturlandskap,
Stavanger.
Sandnes, Berit (1996), Bimbister og Brae og Bilinga. En studie av norrøne stedsnavn i et
språkkontaktområde, Harray på Orknøyene, unpublished thesis, Trondheim
university.
Smith, Brian (1999), ”A short history of Kebister”, in Owen, Olwyn & Lowe, Christopher,
Kebister. The four-thousand-year-old story of one Shetland township, Edinburgh.
Stylegar, Frans-Arne (2001), Kulturlandskap, gård og gjerde på Jæren, in Rønneseth, Ottar,
Gard og gjerde. Faser i utviklingen av Jærens kulturlandskap, Stavanger.
Stylegar, Frans-Arne (2003), ”Plog og spade – gård og landsby. Synspunkter på
middelalderens agrare driftsformer og bebyggelsesstruktur i Vest-Agder”, Heimen 3-
2003.
Stylegar, Frans-Arne (in press), ”Township and ’gard’. A comparative study of some
traditional settlement patterns in Southwest Norway and the Northern Isles”, New
Orkney Antiquarian Journal, vol 4.
Taranger, Absalon (1907), Udsigt over den norske retshistorie. IV. Privatrettens historie,
Kristiania/Oslo.
Taylor, Christopher (1975), Fields in the English landscape, London.
Thomson, William P. (2001), The New History of Orkney, Edinburgh.
Vésteinsson, Orri, McGovern, Thomas & Keller, Christian, “Enduring impacts: Social and
environmental aspects of Viking Age settlement in Iceland and Greenland”,
Archaeologia islandica 2, 98-136.

Arkivert i:

• orkneys
• island
• shetland
• bosetning
• vikingtid
• middelalder
• arkeologi