Through a mirror, darkly: reflections on encoding archaeo-historical farmstead data

Written by Gísli Pálsson, postgraduate student at Department of Historical, Philosophical and Religious Studies, Umea University

It is a truism nowadays to say that an archaeological site is embedded in extensive networks of relations. But just what are the implications of this networked thinking, and how far do these networks extend? Historical sources enable the mapping of extensive reciprocal resource access arrangements, ownership structures and the use of coastal and highland commons; the landscape shows several generations of land divisions, and the archaeological record indicates extensive trade and mobility of materials. All of these strands of evidence lead back to the farm, and beg the investigation of an important question: what, exactly is a farm? Furthermore, how should a database of land use tailor its structure to the multiplicity of relevant sources? My Phd project – Storied Lines: tracing the tendrils of agency across Iceland’s medieval landscape addresses these questions from a data driven perspective by reviewing the implications of current theories of Iceland’s settlement.

Archaeology is a discipline driven by data. Archaeological datasets are, furthermore, characterised by issues that engender problems quite specific to archaeological data management. Scale, resolution, fragmentation, uncertainty, taphonomy, relativistic dating, to name a few of these issues, lead to complexities that data structures often struggle to cope with. This paper reflects on these complexities by considering structures of data necessary to encode the concept of a farm in Iceland‘s agricultural landscape in the author‘s database of land use (Landmarks). Specifically, the paper considers one key source for the design and structure of the database, Jarðabók Árna Magnússonar & Páls Vídalín, reviewing the way in which the concept of the farm unit can be derived from its description, and the implications that has for he structure of the database.

Hin rigtuga og fullkomna Jarðabók: farm-as-archive

My PhD project is based on a database I have made that accompanies the largest archaeological database for Iceland – Ísleif – but aims to contextualize the roughly 100.000 archaeological sites found therein by focusing on how land was used and how settlement units interacted. The database is structured around the farm as a basic unit, and uses the Ísleif database, several historic records, aerial imagery and archaeological survey data as its main sources. The time frame is, in some sense the entire history of Iceland, but begins by intensively mining the best source for agricultural land use in Iceland – Jarðabók Árna Magnússonar og Páls Vídalín. This spatially extensive but temporally shallow horizon is then used to model land use change by latching on to spatially narrow but temporally deep source material in select locations.
Jarðabók is a remarkable source. It is a census of every farm, both occupied and abandoned, and part of a very ambitious project outlined by the Danish Crown including population and livestock censi, completed in 1702-1703.

The property census was initially forecast to take less than two years, but in the end took 13, from 1702 to 1714, and after working together in the initial two years, its authors Árni Magnússon and Páll Vídalín divided the responsibility between them and relied on several assistants to complete the census (Guðmundsson, 1985). Páll and regional assistants undertook the majority of the fieldwork, while Árni, based in Skálholt, administered the project and handled the correspondence with the Crown. Broadly speaking the methodology remained consistent throughout, and although while the degree of detail varies across counties, certain core information is recorded for every farm in the country, both occupied and abandoned.

The methodology is outlined in a document signed by the two surveyors at Öxará, 18th of July, 1702 (Magnússon, 1916, 21). The composition of this rigtuga og fullkomna jarðabók was to include information of the ownership, value, tithe, property tax, rental fees, expected and actual livestock numbers, tenant farms and seafaring operations, pasture and grassland quality, environmental resources, access to external resources, as well as a host of other information (see Figure 1). This is an astonishing degree of detail compared to other property assessments done in 17th-19th century Iceland, where ownership, value, taxation and rent was usually considered sufficient (cf. Lárusson, 1982).

This source is ideal as a baseline for constructing a database of land use. It covers almost the entire country – the records for the four easternmost counties were lost in the Copenhagen fire of 1728, leaving records for 3560 of the 4020 farm assemblages in the country (see Figure 2; more on this notion of farm assemblage later). The consistent methodology allows for consistency in the data structure of encoding that would not be possible if the farms were surveyed with variable methodologies. The short time period taken to survey the country also aids the consistency of encoding, although 13 years is certainly a duration where significant change can take place. That said, a short timespan and consistent methodology by no means guarantee a straightforward translation from source to data. The next section teases out some of the implication of the concept of farm, what it means for our understanding of land use, and finally whether it is a suitable basic building block for understanding land use in the early 18th century as well as changes in land use through time.

Assemblages of networks

What are the spatial characteristics of a farm? Points can seem very inadequate in describing the typically complex architectural syntaxes of a homestead. At the same time, however, going beyond the point introduces representative biases based on the uneven quality of data we have for farm mounds in Iceland, where some have been excavated, others surveyed from surface observation, yet others surveyed using remote sensing, and finally some that have not been surveyed to any significant degree. Furthermore, the need to ‘go beyond the point’ is heavily dependent on the types of research questions one has for the material, and at least for now the research questions I have are based on the connections between farms and important places in the environment.

Polygons offer an alternative to the point as a representation of a farm property. They can show the dimensions of a farm property in a way that a point cannot, which can help understanding the relative sizes of farms, their access to resources as well as the genealogical relationships farms have to earlier, larger settlement units. The issue with polygons is that constructing them for every farm is simply impossible. A baseline for Icelandic property boundaries in the late 20th century does exist (Gísladóttir et al., 2014), but there is no reason why these would reflect early 18th century boundaries to a reasonable degree. Furthermore, the polygon introduces two biases. The first is that boundaries were dimensionless, clear lines fixed in space. This does not appear to have always been the case.

An example of this is the 16th century conflict between Grænavatn and Reykjahlíð, where the discovery of valuable minerals in a poor hinterland resulted in a legal case as Kolbeinn of Grænavatn disputed Þorsteinn of Reykjahlíð’s boundary markings which overlapped into the land he believed was his by right (Árnason, 2006, p. 15). Both farms had laid claim on the territory for generations without conflict, most likely because boundary disagreements never escalated to a legal dispute while the land was considered valueless. This is not surprising given the affordances of the Icelandic landscape: small patches of productive land are often seperated by large areas of sand banks and gravel – for all intents valueless land for the purposes of agricultura. In situations where boundaries fall within these valueless areas, the exact location of a boundary simply does not matter to landowners. That is, until something of value is found within these ares. One might rather think of zones of control, where ownership is indisputable close to the core of the farm – the farmstead itself – that then diffuses out to the margins.

So which is better, the point or the polygon? In fact, I use both. The single point does not represent complex geometries but can be seen as a proxy where the farm mound is representative of a whole farm. The polygon gives an idea of the historic boundaries of the farm while introducing some biases such as regarding the entire internal surface as homogenous, and the boundaries as dimensionless lines. In fact the reason why I decided not to put too much theoretical weight on either is that as I continued reading farm records I realized that I should be thinking of an alternative ‘main’ representation: the network.

Farm properties were not necessarily contiguous. The earliest full description of a farm’s assets, the Stafholt deed of c. 1140, shows an extensive spatial network (see figure 5). Some of these nodes are individual farm units acquired – or perhaps parcelled out of – Stafholt, others are upland common areas owned by the farm and leased out to subordinate tenant land holders, some represent resource access claims in the lands of nearby farms, and yet others are referred to simply part of Stafholt itself – spatially disarticulated from the main property, but nevertheless part of it.

Figure 5: The spatial mapping of a 1140 property deed showing the ownership and access that Stafholt (in black) has to other properties and commons.
Figure 5: The spatial mapping of a 1140 property deed showing the ownership and access that Stafholt (in black) has to other properties and commons.

Figure 5: The spatial mapping of a 1140 property deed showing the ownership and access that Stafholt (in black) has to other properties and commons.

Furthermore, the domestic economy of the farmstead appears to have relied on extensive networks of resource access and exchange, and these networks affected the movement of people and materials across the country. Some of these were stable, such as networks of primary and subsidiary farms, parishes, legal districts, as well as ecclesiastical and royal property networks. Others were more fluid, such as private property networks and resource exchange networks. These are not inconsequential: early land deeds regularly list the rights both to and of medieval farms, and the first comprehensive land survey of Iceland lists a number of farms that cannot survive without access to certain essential resources in neighbouring lands (See figure 6). This is an example of a coupled subsistence – neighbouring farms relying on access to essential resources in nearby lands. A clear indication is the resource utilization of driftwood. Mapping out the driftwood (and beached whale) access rights in the west of the country shows how extensive the spatial networks of resource utilization are for this resource. Driftwood was essential for the construction of Icelandic turf houses (Oddsson, 1638), and one can therefore argue that even single-farm subsistence agriculture relied on large, distributed networks.


What I argue for – and consequently how I have designed my database – is that a farm should not be thought of as a single productive unit but as a node in an assemblage of networks. This does not simply have theoretical consequences, but cyberinfrastructural ones. Data structures need a complicated syntax offered, for instance, by relational data models; they need spatial awareness to be able to generate new spatial understandings such as topological and topographical networks, and they need the flexibility to allow users to go beyond the intention of the data encoder (me). I have tried, to the best of my ability, to structure my data in a way to address all three implications.


Árnason, A. 2006. Kröfulýsing Fjármálaráðherra f.h. Íslenska Ríkisins um Þjóðlendumörk á Svæði 6. Reykjavík: Óbyggðanefnd.

Guðmundsson, B. 1985. Efnamenn og eignir þeirra um 1700. Reykjavík: Háskólaútgáfan.

Magnússon, Á. 1916. Embedsskrivelser og andre offenlige aktstykker. Copenhagen: Gyldendalske forlag.

Moretti, F. Graphs, maps, trees: abstract models for literary history. London: Verso.

Lárusson, B. 1982. Islands jordebok under förindustriell tid. Lund: Lunds universitet.

Karlsson, G. 2000. Iceland’s 1100 years: the history of a marginal society. London: Hurst & co.

Gunnarsson, G. 1987. Upp er boðið Ísaland: einokunarverslun og íslenskt samfélag 1602-1787. Reykjavík: Örn & Örlygur.

Gísladóttir, F. Brink, S. & Arnalds, Ó. 2014 Nytjaland. Hvanneyri: LBHÍ.

Oddsson, G. 1638. De Mirabilis Islandiae.

Teitsson, B. 1973. Eignarhald og ábúð á jörðum í Suður-Þingeyjarsýslu. Reykjavík: Menningarsjóður.

Testing the drone course

Yesteday we had a prequel or “test” of a drone course we would like to hold on a regular basis. It was a great success so it will probably return the next semester. Until then here are some images from yesterdays event.WP_20170517_006 WP_20170517_008 WP_20170517_019 WP_20170517_021