Intercropping in the Merovingian period? Legume finds from Harestadkrysset, Norway, in their wider Scandinavian context

In 2021–2022, excavations in advance of roadbuilding at Harestadkrysset, southwest Norway uncovered archaeological remains from various periods, including houses from the Merovingian period (570–793 AD). In the largest of these houses, the wall ditch and one of the postholes produced charred remains of large legume seeds, tentatively identified as cultivated legumes (cf. Vicia / Lathyrus / Pisum) and dated to 579–645 cal AD. These seeds are unusual in both their large size and their concentration: although vetches are occasionally found as single examples in wild/weed seed assemblages in Rogaland, this is by far the largest number of legume finds from a single house dating to earlier than the Medieval period. This suggests that legumes were cultivated, and that their inclusion in the archaeological record results from their collection and processing. This paper sets these finds in their broader context, comparing finds of legumes from both Rogaland and elsewhere in Scandinavia. The find is also compared with historically-attested agricultural practices, including intercropping of oats and peas in order to improve soil fertility. It is possible that this find represents a local innovation or change in agricultural practice in the face of climatic changes in the Late Antique Little Ice Age, and this paper also compares the Harestadkrysset finds with other archaeobotanical material from this period in southwest Norway.

In AD 536, a large volcanic eruption, likely in North America or Iceland, caused a dust veil event, resulting in a rapid downturn in global atmospheric temperatures (Büntgen et al., 2020). Subsequent eruptions in the course of the next decade, along with a solar minimum worsened and prolonged this cooling. This period is known as the Late Antique Little Ice Age (LALIA, Büntgen et al., 2016), and has also been proposed as inspiration for the apocalyptic Fimbulvinter of Norse mythology (Price and Gräslund, 2015). At the same time, the Justinian Plague, which began in AD 542, caused significant population decline in southern and central Europe, and is likely to have had at least indirect social effects on Scandinavia (Loftsgarden and Iversen, 2024). This combination of climate change and pandemic, and its ensuing social and environmental impacts, is often known as the 6th century crisis, and has been suggested as “the worst time in history to have been alive” in Europe (Gibbons, 2018, p. 733). In Norway, this period marks the end of the Migration Period (AD 400–570) – a time of population increase and agricultural expansion – and the beginning of the Merovingian period (AD 570–793). The date ranges assigned in Norwegian archaeology to time periods mentioned in the text are given in Table 1.

This transition is marked by a clear change in the archaeological record. In Rogaland, the region on which this paper is focused, many of the abandoned farms still visible in the landscape were occupied in the Roman Iron Age (0–400 AD) and Migration period (400–570 AD) (Bjørdal, 2016). In the ensuing Merovingian period, there is a decline in both the number of settlements and the number of finds (Petersen, 1933, 1936; Vetrhus, 2017), specialised ceramic production ceases (Rødsrud, 2016), and there is a downturn in fine metalworking (Kristoffersen and Røstad, 2020). It can be inferred from the climatic data and the changes in the archaeological record that the cold weather also affected agriculture, leading to lower productivity of arable fields and worse conditions for livestock – especially cattle which are vulnerable to cold and wind, and produce less milk under these conditions (Pedersen and Widgren, 1998, p. 255; Westling, 2025, p. 27). Crop failures and reduced agricultural production would also have led to a decrease in population. With a smaller population and a less clement climate, it is likely that populations adapted by modifying their arable agriculture practices to focus on more cold-tolerant crops and less labour-intensive practices. Sara Westling has noted that overall in Rogaland there is both a decrease in the quantity of cereal finds generally and a decrease specifically in finds of oats (Avena sp.) in favour of hulled barley (Hordeum vulgare L.), with a slight increase in wild nuts and berries and occasional finds of rye (Secale cereale L.) as a possible experimental crop (Westling, 2025; Westling et al., 2022). It is also likely that the increased reliance on outfield, and especially marine, resources seen in the Viking Age (793–1030 AD) and onwards in Scandinavia (Indrelid et al., 2015; Mooney et al., 2022b) has its roots in the Merovingian period.

This paper presents archaeobotanical evidence from houses dated to the Merovingian period from Harestadkrysset, Rogaland, southwest Norway, with reference to the climatic, environmental and societal changes of the 6th century crisis in Scandinavia. I focus in particular on a rare find of charred legume seeds that I interpret as a possible agricultural adaptation to challenging conditions for arable agriculture. These finds are also situated in their wider context of other legume finds in Scandinavian archaeology.

The archaeobotanical material presented in this paper comprises charred plant macrofossils deriving from excavations at the site of Harestadkrysset, Randaberg municipality, Rogaland county, southwest Norway (Figure 1). The excavations were conducted by the Museum of Archaeology, University of Stavanger on behalf of Statens Vegvesen (the Norwegian roads authority) in advance of road construction related to the Rogfast tunnel project (Demuth et al., 2024). Harestadkrysset was a large, multiperiod site: the site as a whole is briefly presented below, but it is only the Merovingian material that is presented in detail here. The archaeobotanical sampling strategy and the methods employed in the processing and analysis of the plant macrofossil samples are also presented in this section.

Excavations at Harestadkrysset 2021–2022

Excavations at Harestadkrysset took place over two field seasons (from April to September) in 2021 and 2022, and covered an area of 18,593 m 2. In this area, 485 archaeological features, including postholes, pits and graves, were recorded and excavated. This number was slightly lower than estimated before excavation: the area lies in agricultural land and the archaeological remains were impacted by extensive ploughing. Radiocarbon dates for features and structures at the site produced results ranging from the Neolithic (4000–1800 BC) to the Early Medieval (1030–1200 AD), but the majority of features were dated to the Late Neolithic (2300–1800 BC), Early Bronze Age (1800–1200 BC), Pre-Roman Iron Age (500–0 BC/AD) and Merovingian period (Demuth et al., 2024, p. 82). For date ranges for these periods in Norwegian archaeology see Table 1. The features mostly comprised the remains of timber-built houses and outbuildings, in the form of negative features (especially postholes) visible in the natural underground once the topsoil had been stripped away.

As outlined above, archaeological material from the Merovingian Period is rare in southwest Norway, so it is unusual that of the 60 samples sent for radiocarbon dating from Harestadkrysset, 19% (11 samples) returned dates belonging to this period. These samples derived from a range of features including an isolated hearth, a cremation grave, and features related to four houses (Demuth et al., 2024, p. 87). These features and structures are described in more detail below.

Processing and analysis of archaeobotanical samples

Material for radiocarbon dating at Harestadkrysset was mostly recovered from bulk soil samples for macrobotanical analysis. A total of 202 samples were analysed from the Harestadkrysset excavations, from contexts dating to periods from the late Neolithic to the Merovingian period. These samples were processed using a Siraf-type flotation machine (developed after Williams, 1973). Flots were collected on a 500 μm mesh, according to standard procedures at the Museum of Archaeology, University of Stavanger (AM-UiS), then air-dried before sorting and analysis (Bakkevig et al., 2002; Demuth et al., 2024). The heavy fraction of the samples was collected on a 1 mm mesh, rinsed and air-dried, and sieved through meshes of 4, 2 and 1 mm. Finds including plant remains, bone, shell and artefacts were collected, recorded, and analysed by a relevant specialist. For more detail see Demuth et al. (2024).

Flots, comprising charred plant remains and other light material, were sorted and analysed using a Nikon SMZ-1500 stereomicroscope at magnifications of up to 112.5×. Identifications were made by comparing the archaeological material with published reference material (Anderberg, 1994; Beijerinck, 1947; Berggren, 1969, 1981; Bertsch, 1941; Cappers et al., 2006; Dombrovskaya et al., 1959; Griffin and Sandvik, 1989; Jacomet, 2006; Katz et al., 1965, 1977; Korsmo et al., 1981; Neef et al., 2012) and with a modern reference collection held at AM-UiS. Nomenclature for scientific names of wild taxa follows Mossberg and Stenberg (2018), while nomenclature for scientific names of crop plants follows Zohary et al. (2012). English common names for all taxa follow Stace (2019).

The full results of this analysis are available online in the project report (Demuth et al., 2024) and are not repeated here. Instead, I present a detailed account of the samples taken from features dated to the Merovingian period. Further information from the published report is used as comparative data to contextualize the Merovingian finds.

Results of analysis of samples from Merovingian features

Two structures at the site were securely dated to the Merovingian period: House 1 and House 10 (Figure 2). These were timber-built longhouses of which only negative features such as postholes remain, and it is these features from which the samples derived and from which material was taken for radiocarbon dating (Table 2, Figure 2). Postholes from timber-built houses have been shown to contain material from the active phases of the buildings to which they belong and the formation processes of these assemblages have been discussed in detail elsewhere (Beck, 2017; Reynolds, 1995; Verhoeven, 2010). The results of the plant macrofossil analysis for Houses 1 and 10 are presented in this section. In addition, the results from the analysis of archaeobotanical samples from a cremation grave (Grave 2190) dated to the Merovingian period are also presented. Although this feature represents plant use in a ritual rather than domestic sphere, it still gives an idea of the crop plants and other taxa present at the site in the Merovingian period. The full results of these analyses are presented in the Supplementary Material. Dates from the Merovingian period were also returned from postholes belonging to two other structures at the site: House 3, in the northernmost part of the site, and House 7, about 60 m northwest of House 10 and 20 m southwest of House 1. These have not been included in the material presented here as there is uncertainty around their dating: features in both structures also returned dates from the Bronze Age (Demuth et al., 2024). House 3 in particular was also poorly preserved, with only a few postholes present, and it was not possible to determine if all of these belonged to the same house (Demuth et al., 2024, pp. 52–53).

House 1

House 1 was located in the central part of the site. Radiocarbon dates on charcoal and Avena sp. grains from the structure returned results from the Migration period and Merovingian period (Figure 2, Table 2). The phasing of the house to the Merovingian period was based on the principle that the youngest date applies (Demuth et al., 2024, p. 44), but the house may have been in use for a longer period. Seventeen samples were taken from features belonging to House 1: sixteen from postholes and one from a pit. The latter sample (116) contained no charred material other than small charcoal fragments. Generally speaking, the posthole samples also contained few plant macrofossils: the average number of identified seeds per litre was only 0.6, and several of the posthole samples also produced no identifiable charred seeds (87, 89, 91 and 115). Grains of oats (Avena sp.) were found in several samples, along with two grains of hulled barley (Hordeum vulgare L.) in sample 96. Other macrofossils found in the samples included hazel (Corylus avellana L.) nutshell fragments, and seeds of blackberry (Rubus fruticosus agg.), raspberry (Rubus idaeus L.), heath-grass [ Danthonia decumbens (L.) DC.], clover (Trifolium sp.), fat hen (Chenopodium album L.), knotweed (Persicaria sp.) and corn spurrey (Spergula arvensis L.).

House 10

House 10 was located in a plateau in the southern part of the site. It was dated to the Merovingian period based on dating of charcoal and seeds of Vicia/Lathyrus/Pisum from postholes and the shallow wall-ditch surrounding the house (Figure 3). Thirty-eight samples were taken from features belonging to House 10: eleven from postholes, two from remains of floor layers, one from a pit, twenty from sections through the wall ditch, and four from external features connected to the house.

In the samples from postholes, the average macrofossil find density was 0.73 seeds per litre. Finds included grains of oats and wheat (Triticum sp.), and seeds of raspberry, grasses (Poaceae), sedges (Carex sp.), buttercups (Ranunculus sp.) and clover. Additionally, in sample 149 a large legume seed (cf. Vicia/Lathyrus/Pisum) was found (see below for more detail). The find density of the samples from floor layers was somewhat higher, at 4.64 macrofossils per litre. These included several fragments of hazelnut shell as well as seeds of hawthorn (Crataegus monogyna Jacq.), dwarf cornel (Cornus suecica L.), blackberry, sedge, heath-grass, ribwort plantain (Plantago lanceolata L.), buttercup, clover, knotweed and sheep's sorrel (Rumex acetosella L.), and a fragment of wild radish (Raphanus raphanistrum L.) mericarp.

The majority of the samples from House 10 came from sections through the house's narrow wall ditch (Figure 3). The find density was higher than in the postholes but still relatively low, at 1.48 identifiable macrofossils per litre. The finds were concentrated towards the centre of the house and mostly comprised indicators of grazing such as grasses, ribwort plantain, buttercups including creeping buttercup (Ranunculus repens L.), clover and violets (Viola sp.). Seeds of arable weeds such as hemp-nettle (Galeopsis sp.), knotweed, common knotgrass (Polygonum aviculare L.) and corn spurrey were also found, but the main find from this feature was 13 fragments of large legumes (cf. Vicia/Lathyrus / Pisum) spread across four samples. The location of these finds can be seen in Figure 3 – note that the distribution does not precisely follow that of taxa identified as grazing indicators. As can be seen in Figure 4, these are poorly preserved, lacking their seed coat and with the hilum partially or wholly absent or obscured. Their identification is therefore somewhat uncertain: the seeds are generally a little smaller than published archaeological finds of Vicia faba L. and Pisum sativum L. (e.g., Kirleis, 2019; Neef et al., 2012; Treasure and Church, 2017) but they are significantly larger (ca. 3–4 mm in diameter) than finds of wild Vicia/Lathyrus known from the region (for further detail see discussion). The smaller size may be related to the shorter growing season offered by the West Norwegian climate. A direct radiocarbon date on several of these remains from sample 166 returned a date of 579–645 cal AD (see Figure 2 and Table 2). A single seed of flax (Linum usitatissimum L.) was also found in a sample from the wall ditch.

The samples from outside of House 10 were taken from a cooking pit and layers thought to be connected to the use of the house. The layers produced only a few macrofossils, identified as grasses including heath-grass. The cooking pit, on the other hand, contained cereal grains including barley along with arable weeds such as fat-hen, hemp-nettle and knotweeds, and grazing indicators such as segdes, heath-grass, ribwort plantain and clover.

Grave 2190

Feature 2190 was a cremation grave located in the central part of the site, about 25 m east of House 1. As well as burnt bone it contained an amber bead, a bone comb, a spindle whorl, and fragmented remains of iron and copper alloy objects (Demuth et al., 2024, p. 73). Barley grains from the grave fill were dated to the Merovingian period, ca. 595–649 cal. AD (Figure 2, Table 2). Two samples were analysed from the upper part of the grave fill – the same layer from which the finds were recovered. Some of the barley grains were identified as hulled barley. Cereals were present in large quantities when compared with the assemblages from the houses, in all 34 barley grains, 28 indeterminate cereal grains, and 104 grain fragments. The grave fill also contained hazelnut shell fragments and seeds of wild strawberry (Fragaria vesca L.), as well as seeds of grazing indicators such as sedges, grasses, buttercup, clover and violets, but the assemblage was dominated by arable weeds. Fat-hen, hemp-nettle and knotweed were recovered in particularly large quantities, along with black-bindweed [ Fallopia convolvulus (L.) Á.Löve] and chickweed [ Stellaria media (L.) Vill.]. The two samples averaged just under 198 identified seeds per litre.

The Merovingian features at Harestad in comparison with other periods

As mentioned above, the complete results of the analysis of plant macrofossil remains from Harestadkrysset are available online (Demuth et al., 2024), and are not presented in detail here. However, it can be useful to highlight some general trends in the macrofossil assemblage from the Merovingian houses compared to other periods at the site. The Merovingian period at Harestadkrysset shows in general a lower density of macrofossil finds: the samples from Merovingian house contexts averaged 1.48 identified macrofossils per litre, compared to 12.07 for all samples from the site and 28.78 for samples from Pre-Roman Iron Age houses (Demuth et al., 2024). The largest quantity of grain from a Merovingian house at Harestad was 13 grains in House 10, which is less than a quarter of the size of the smallest assemblage dated to the Pre-Roman Iron Age, and is dwarfed by the find of over 1,000 barley grains in a Late Neolithic/Early Bronze Age house at the site (Demuth et al., 2024) (Figure 5A). It can be seen in Figure 5B that the difference in number of crop macrofossil finds from the Late Neolithic/Early Bronze Age compared to other periods only becomes more stark when presented as finds per litre of sediment sampled. The difference in crop finds per litre between the Pre-Roman Iron Age and the Merovingian period is less significant, but viewed alongside Figure 6 it is clear that there is a decline in the proportion of the assemblage that relates to arable agriculture, as the Pre-Roman Iron Age assemblage is dominated by arable weeds (although see also discussion below regarding the edibility of common arable weeds). It is likely that this indicates a decline in arable production as part of a regional trend: there is significant evidence of surplus production of grain in the Pre-Roman Iron Age at Tjora, ca. 11 km south-southwest from Harestadkrysset (Fyllingen et al., 2023), but grain from later periods is often found in smaller quantities.

At the same time there also seems to have been a diversification of subsistence practice. This is visible in the archaeobotanical record in the form of increased presence of hazelnut shell and seeds of berries, as well as a change in crops: (hulled) barley remained common, but oats comprised a greater part of the assemblage than in other periods (Figure 7). Wheat was rare across the whole site, but was also identified in the Merovingian material, and likely cultivated legumes are only found in House 10. Flax was also found only in House 10 and House 7. The latter is of uncertain date (see section 3.1) but its low find density and the presence of woodland and heathland plants along with arable weeds and grazing indicators are most comparable with the macrofossil assemblages from the Merovingian Houses 1 and 10. The change in overall composition of the archaeobotanical assemblage is illustrated in Figure 6: note the increase in both forest and heathland plants (including foraged foods such as hazelnuts and berries) and grazing indicators.

The samples from Grave 2190 are not included in Figures 5–7, as their ritual context precludes direct comparison with remains from houses. However, these samples also suggest that barley was the dominant arable crop in the Merovingian period. The uniformity of the dates (Figure 2) strongly suggests that the grave is contemporary with Houses 1 and 10, but neither oats, legumes nor flax were present in the grave fill. The dominance of seeds of arable weeds often engenders an interpretation of such an assemblage as waste from crop processing (cf. Fyllingen et al., 2023), but its clear ritual/religious context here encourages other interpretation. Fat-hen, which accounts for 58% of the 1,343 identified seeds in Grave 2190, is also edible: its leaves and seeds been traditionally consumed in times of famine in Norway (Høeg, 1976, p. 104). Fat-hen seeds were also found in the stomach contents of the Tollund Man bog body (Nielsen et al., 2021) along with other “weed” taxa such as hemp-nettle and knotweed. This reminds us that the division between “weed” and “crop” is in many cases a modern construct, and that for foods such as porridge or gruel, people in the past were likely less concerned about the “purity” of their grain (Westling, 2025; Høeg, 1976, p. 104). There is no reason not to interpret the macrobotanical remains from Grave 2190 as a food offering, and this stands in sharp contrast to the paucity of seeds from other Merovingian contexts at Harestadkrysset.

As seen above, the Merovingian samples from houses at Harestadkrysset are characterised by low find density and increased diversity – including potential legume cultivation. It cannot be ruled out that the legume finds result from cultivation for grazing or fodder, but the taxa traditionally used for this purpose in Norway [e.g., clover, Vicia cracca L. (tufted vetch)] produce much smaller seeds (Høeg, 1976). The only native legumes producing seeds of a similar size are Lathyrus japonicus Willd. (sea pea), and Vicia pisiformis L. (pea vetch) (Mossberg and Stenberg, 2018). The Harestadkrysset fragments compare poorly to the former, which grows on sandy and stony beaches, while the latter is only rarely found in eastern Norway and not native to Rogaland. If the Harestadkrysset legumes belonged to either of these taxa, they would almost certainly have to result from cultivation. Unfortunately, the poor preservation of the Harestadkrysset finds, including the lack of preservation of the hilum, precludes their further identification. Furthermore, although it has been demonstrated that geometric morphometric analysis can be helpful in the identification of uncertain legume finds (cf. Tarongi et al., 2024), the fragmented nature of the Harestadkrysset finds means that this method is not applicable in this case.

Despite the uncertainty surrounding the precise taxonomic identification of the Harestadkrysset legumes, for the purposes of this discussion they are treated as domesticated legumes, such as Pisum sp. or Vicia faba L., cultivated for human consumption. The interpretation of the Harestadkrysset legumes as cultivated is based on their size compared to other legume finds from Rogaland (see below, Section 4.1.1), their quantity compared to both legume finds elsewhere in Rogaland and cereal finds in Merovingian houses, and their distribution compared to that of taxa interpreted as grazing indicators. Figure 3 shows that while legumes are found in the same features as grazing indicators, such as the wall ditch, they are never found together with more than four seeds of grazing indicators, which are found in much larger quantities elsewhere. Although the material in the ditch is likely to be mixed, this slightly different distribution suggests possible different preservation pathways for legumes and grazing indicators, although both could derive from accidental (or indeed deliberate) burning of e.g., hay collected as fodder, or from the use of animal dung as fuel (for detail on the routes of entry of plant remains into charred assemblages see e.g., Berihuete-Azorín, 2016; Van Der Veen, 2007). Material from the postholes and wall ditches is likely to represent floor material, including ash spread on the floors to combat damp and odours (cf. Milek, 2012) and waste from in and around the house, and the legumes may have entered these assemblages through accidental burning during cooking or drying. While 14 legume finds may seem like a small number, it is significant compared to the general paucity of cultivated plant remains in Norwegian Merovingian period houses as discussed above, and equals the number of finds of cereal taxa from House 10.

The following discussion focuses on diversification, especially the evidence for potential legume cultivation at Harestadkrysset. I focus on the legumes because of their rarity in the archaeological record – cereal cultivation in Iron Age Rogaland has been extensively addressed elsewhere (Prøsch-Danielsen and Soltvedt, 2011; Westling, 2025). I first present a brief background to the cultivation history of legumes and evidence for their cultivation in Scandinavia, before situating the Harestadkrysset legume finds, and archaeobotanical evidence for Merovingian subsistence at the site, in the context of environmental change during this time period in southwest Norway.

Cultivation history of legumes

Legumes have been valued in agricultural systems since the origins of agriculture (Zohary et al., 2012). They are also an excellent source of protein, and taxa such as lentils (Lens sp.), chickpeas (Cicer sp.), peas (Pisum sp.) and broad beans (Vicia faba L.) are known as food crops from the earliest Neolithic onwards (Zohary et al., 2012). Legumes also fix atmospheric nitrogen in the soil and thus improve soil fertility. While smaller-seeded legumes such as vetches (Vicia sp.) and vetchlings (Lathyrus sp.) have also been important elements of human diet, they are also valuable for soil improvement along with taxa such as clover (Trifolium sp.) (Mikić, 2016). They may also have been used as fodder or forage, but this is difficult to demonstrate archaeologically. Legumes cultivated for soil improvement could be grown either as part of a crop rotation system, or through intercropping, in which legumes and cereals are planted on the same plot of land in the same season.

Cereal-legume intercropping is widespread in non-mechanised agricultural systems, and is considered to improve yields as well as being beneficial for disease/pest resistance, weed control, soil health, and biodiversity (Malleson, 2016; Rahman, 2021). However, in many regions this practice died out as farming was mechanised: it is quicker and more cost-effective to harvest fields by machine, but this in turn requires monoculture fields that can be harvested completely in one go, rather than only harvesting certain plants. Legumes appear very early in the archaeological record alongside cereals (Zohary et al., 2012), and although this is not necessarily evidence of intercropping it demonstrates their long history as a key element of human subsistence. It has been proposed that leguminous weeds were tolerated in cereal crops, and possibly seen as beneficial, in ancient Egypt (Malleson, 2016). Cereal-legume intercropping has also recently been proposed as a way to increase protein production while reducing fertiliser use in modern northwestern Europe (Landschoot et al., 2024), however without reference to any potential historical basis for the practice.

This is in part because when legumes are presented in archaeobotanical studies, it is usually the fact of their presence that is discussed, rather than how they are cultivated (e.g., Fuller and Harvey, 2006; Ljuština and Mikić, 2010; Treasure and Church, 2016). This is unsurprising given that legumes in general are rare in the archaeobotanical record, probably because they are generally less likely to come into contact with fire during processing, and are therefore less likely to be accidentally charred and preserved than cereal crops (Treasure and Church, 2016). Extensive meta-studies of archaeobotanical data does however suggest that legumes should be considered as an important element in agricultural systems in regions as diverse as Britain (Treasure and Church, 2016), India (Fuller and Harvey, 2006), and southern Europe (Mikić, 2016).

The Harestadkrysset legume finds in their regional context

Like elsewhere, finds of large legumes are rare in Scandinavia. It is most common for smaller seeds of Vicia and/or Lathyrus to be recovered as individual instances of likely wild taxa within assemblages of weed seeds, and species such as Lathyrus vernus (L.) Bernh., L. pratensis L., Vicia cracca L., V. sylvatica L. and V. sepium L. are widespread across the region (Mossberg and Stenberg, 2018). In Sweden, much archaeobotanical data is systematised and openly available through the Strategic Environmental Archaeology Database (SEAD, Buckland et al., 2018). A search for finds of Vicia and Pisum reveals 50 sites, of which 39 produced five or fewer legume finds (SEAD, 2025). The finds are concentrated in southern Sweden, and the largest is from Svarteborg in Bohuslän, dated to the Late Bronze Age or Pre-Roman Iron Age. This find comprised 126 legumes and fragments thereof, including both Vicia faba L. and Pisum sativum L (SEAD, 2025).

There are by contrast no openly available collections of archaeobotanical data from Norway and Denmark, so there is no comprehensive data on how often small numbers of legumes are present in archaeobotanical analyses conducted in the course of development-led archaeology. However, from experience and discussions with colleagues it is clear that seeds of small Vicia and Lathyrus species occur as individual finds with relative frequency, and legume pollen is not uncommon in palynological analyses. Large-seeded legumes, on the other hand, are less common. The earliest published finds of Vicia faba L. in Denmark are also from the Pre-Roman Iron Age in northern Jutland (Robinson et al., 2009), and finds have also been reported from Merovingian contexts at Ribe (Jensen, 2022). Elsewhere, finds of Vicia and similar Fabaceae tend to be smaller and interpreted as likely to represent either fodder plants or agricultural weeds. The latter may be the case at the 13th-14th century AD castle of Bjørnkær in northern Jutland, where Fabaceae seeds were common amongst the large assemblage of arable weed seeds interpreted as crop processing waste (Jensen and Larsen, 2022).

Despite the presence of cultivated legumes elsewhere in Scandinavia, archaeological studies often state that it is first in the Medieval period that legumes become part of agricultural systems in western Norway (e.g., Øye, 2004) as it is in this period that pollen and macrofossil evidence becomes more common in Medieval towns such as Bergen (Øye, 2002). In Stavanger, where significantly fewer archaeological investigations of Medieval town remains have been conducted, a major archaeobotanical find from near the city's harbour comprises peas (Pisum sativum L.) mixed with burnt grain, and finds of legume pollen suggest nearby garden cultivation in this period (Bakkevig and Time, 2007, p. 15). However, the oldest find of cultivated peas from Rogaland comes from the site of Stokka in Sandnes, around 15 km south-southeast of Harestad. Here, an impression of a pea, larger than native wild legumes but a little smaller than modern cultivated peas, was found in a ceramic sherd from a grave cairn dated to the Pre-Roman Iron Age (Bakkevig and Time, 2007, p. 14). Recently, another possible cultivated legume find has been made at the site of Sør Braut, also dated to the Merovingian period, around 30 km south of Harestadkrysset (Sara Westling, personal communication, April 2026).

There is therefore a small but growing body of evidence for past legume cultivation in Rogaland, and although archaeological legume finds are still uncommon, historical sources tell us of their importance in the past. The Norwegian law code Magnus Lagabøtes landslov, dated to AD 1274, includes punishments to be given for the theft of fruits and vegetables from gardens, including peas and beans (Rindal, 2024). A few years later, in AD 1277, peas in Norway were lagt tiende, meaning that they were included in the farming produce on which tithes to be paid to the church were calculated. Later again, in AD 1400, there is a record in Diplomatarium Norvegicum of a request for peas and beans to be sent for sowing in Borgund, Sunnmøre, indicating that legumes were also cultivated further north in Norway (Bakkevig and Time, 2007, p. 9).

Despite this, peas and beans are seldom mentioned in ethnobotanical sources in Norway. In Ove Arbo Høeg's Planter og Tradisjon (1976), widely regarded as the authoritative source on Norwegian traditional plant names and use, peas are allocated less than half a page and Vicia faba L. is not mentioned at all. Høeg spent 45 years, from 1925 to 1973, travelling around Norway and recording traditions regarding cultivated and wild plant use from all regions of the country. However, Rogaland is one of the most poorly-represented regions in this work: in most counties, respondents are recorded from more than 50% of municipalities, and around half of the counties are represented by respondents from more than 65% of municipalities. Rogaland, on the other hand, is represented by respondents from 44% of municipalities – only ahead of the northern counties of Troms and Finnmark (Høeg, 1976, p. 3). In Høeg's presentation, wild species of Vicia and Lathyrus are collected together under V. cracca L., which is reckoned as excellent grazing (Høeg, 1976, p. 676). It is recorded that two species of Pisum were cultivated in Norway: “hageert” (P. sativum L.) and “åkerert” (P. arvense, now regarded as a subspecies of P. sativum), but no further detail on their cultivation is given (Høeg, 1976, p. 504).

Peas were eaten both dried and fresh in Norway, likely from at least the Medieval period onwards (Øye, 2002, p. 324), but are rarely mentioned in accounts of agricultural practice. However, in Jæren, the fertile coastal plain that dominates western Rogaland, there still exists a local pea variety called jærerter (“Jæren peas”). Their cultivation is widely known in oral histories, but is mentioned in very few written sources (Bakkevig and Time, 2007, p. 17). This hardy taxon was cultivated together with oats and farmers believed that although the peas “used the oats as a climbing frame” and created some shade, this form of intercropping resulted in a good crop of oats, and peas as well (Bakkevig, 2020, p. 17). As with all legumes, the peas also helped fertilise the fields, but the intercropping practice died out from the 1950s onwards, with the introduction of mechanised harvesting and cheap artificial fertilisers (Bakkevig, 2020, p. 18). Even then, intercropping of peas and oats seems to have been a very localised practice: while oats were commonly grown across Jæren into the 20th century, jærerter seem to have only been grown alongside them in certain areas (Bakkevig and Time, 2007, p. 25). Sverre Bakkevig (2020, p. 19) has proposed that the pea impression found at Stokka, mentioned above, can be of jærerter, pushing their cultivation history back 2000 years. This is of course speculation, but this existence of a local pea variety and a clear tradition of intercropping suggests that we may be tracing a lasting practice that for various reasons has remained elusive in the archaeological and documentary records.

Merovingian subsistence at Harestadkrysset – an adaptation to a climate “crisis”?

There is no direct evidence to suggest that the legumes from Harestadkrysset were cultivated through intercropping – even if cereal and legume pollen were to be found in the same cultivation layer, this could indicate crop rotation or even the presence of arable weeds in the Fabaceae family. However, given the local history of this practice and its suitability for poor soils, it is possible. Whether grown as part of a rotation system or through intercropping, the large size of the legume finds from Harestadkrysset and their relative dominance in the archaeobotanical material is strongly indicative of cultivation. This fits into an established pattern of crop and foodway diversification in the Late Iron Age of Rogaland: it is at this time that there is a significant shift towards outfield resources, especially fishing and marine resources (Mooney et al., 2022a). However, while this shift has been studied in detail (e.g., Mooney et al., 2022b; Indrelid et al., 2015), agricultural diversification and shifts in diet around this period are less well-explored.

Agricultural adaptation in Rogaland

The key study of agricultural changes, and corresponding changes in plant macrofossil assemblages, around the time of the 6th century crisis in Norway, is Sara Westling's doctoral research (Westling, 2025). Focusing on Rogaland, she argues that the influence of climatic downturn is visible in the archaeobotanical material in the form of changes in preferred cereal crops, a changing ratio between arable agriculture and animal husbandry, and experimentation with different cereal crops (Westling, 2025). In a synthesis of sites from across the region, Westling demonstrated that sites where barley accounted for more than 90% of the cereal grain finds were significantly more likely to survive into the Viking Age, while those where oats comprised over 40% of the assemblage were most likely to be abandoned in the Merovingian period (Westling, 2025, p. 505). This may be because sites where oats were an important crop were also more vulnerable to climatic change: the mechanism is unclear. It is however clear that successful sites showed an increased reliance on animal husbandry, as demonstrated by changing wild flora assemblages (Westling, 2025, p. 506). This is also supported by evidence from elsewhere in western Norway: Dahle (2025, p. 145) has observed that the 6th century crisis is not particularly visible in archaeological evidence for shieling activity (summer dwellings on seasonal pastures, often in upland areas, linked to transhumance practices) in Møre og Romsdal, which again indicates a continuing need for infrastructure related to animal husbandry.

One of the more surprising finds in Westling's corpus is the large quantity of charred rye grains found at the site of Hove-Sørbø, about 21 km south of Harestadkrysset. This find is interpreted as a cache of locally-cultivated rye, and provides the earliest known evidence of rye cultivation in Norway (Westling and Jensen, 2020, p. 97). This may be evidence for experimentation with a new crop in response to the pressures of climate change: although uncommon in Norway, rye was grown in Denmark and southern Sweden in the Merovingian period, and would have been available through trade contacts (Grabowski, 2011; Robinson et al., 2009; Westling and Jensen, 2020). The question of diversification and experimentation is an important one as it relates to self-sufficiency: if the crops that are usually grown are failing, at what point does one take a chance on something new or uncommon? Westling argues that the settlement at Forsand, Norway's largest prehistoric settlement, was abandoned in the Merovingian period because it was not self-sufficient in grain, relying also significantly on trade (Dahl, 2022, p. 93; Westling et al., 2022, p. 55). In settlements like Hove-Sørbø and Harestadkrysset, where the occupation period extends into and beyond the Merovingian period, one can see the unusual finds of rye and legumes as evidence of attempted adaptation to changing conditions.

However, we cannot say for certain that these practices were not more widespread than they appear. The Stokka ceramic sherd with its pea impression (Bakkevig and Time, 2007, p. 14), resulting from the use of plant material as temper (cf. Stilborg, 2001), suggests that cultivated peas were at least present in Rogaland in the Pre-Roman Iron Age, although again it is not possible to say with certainty that these were cultivated locally. Furthermore, new development-led excavations are steadily turning up new evidence for “minority” crops – often from the Merovigian period. As well as the possible pea find at Sør Braut mentioned above (Sara Westling, personal communication, April, 2026), there is both pollen and plant macrofossil evidence for the cultivation of rye in the Merovingian period and Viking Age at the site of Svadberg, ca. 35 km northeast of Harestadkrysset (Demuth et al., in preparation). Add to this the tendency for increased finds of wild fruits and berries in the Merovingian period and beyond, visible at Harestadkrysset and other sites in the region (Demuth et al., 2024; Westling, 2025), and we begin to glimpse what may be a trend towards more varied plant food consumption in response to climate pressure.

This paper has presented evidence for the presence of legumes (cf. Vicia/Lathyrus/Pisum) at the site of Harestadkrysset, southwest Norway, in the Merovingian period. The finds themselves are too poorly preserved to allow for precise identification, but their size and frequency compared to other sites in the region strongly suggests cultivation of domesticated legumes. Archaeobotanical evidence from Harestadkrysset also supports trends of reduced surplus grain production and diversification of diet in the Late Iron Age, which I argue is likely to result from climatic and demographic downturn in the 6th century AD. The legume finds are also presented in relation to the local pea variety jærerter, which was traditionally intercropped with cereals such as oats.

We cannot be certain that this pea variety, or the tradition of intercropping, can be traced back to the Merovingian period or earlier. We can, however, look at the archaeological finds of legumes from Harestadkrysset and other sites, the local tradition for intercropping, and the various environmental challenges posed by the 6th century crisis, and begin to tell a story of local agricultural innovation. We know that agriculture and diet in Rogaland changed significantly during and after the Merovingian period. The establishment of hulled barley and oats as the staple crops of the region, the increased reliance on wild resources, and the end of the large grain surpluses known from earlier periods form a pattern that continues into the Viking Age and Medieval Period. The extent to which this pattern holds true for the region (and further afield) can only be established by further investigation. I hope therefore that this paper will raise the profile of the existence of potentially cultivated legume finds in Iron Age western Norway, and provide a resource for others investigating non-cereal crop cultivation in the wider Nordic region.