Et monumentalt midtsulehus ved Nørre Holsted i Sydjylland


  • Lars Grundvad
  • Martin Egelund Poulsen
  • Marianne Høyem Andreasen



langhus, midtsulehus, Nørre Holsted


A large two-aisled house at Nørre Holsted in southern Jutland – Analysis of a longhouse from Early Bronze Age period I

In 2011 and 2012, Sønderskov Museum investigated an area of 65,000 m2 at Nørre Holsted, between Esbjerg and Vejen. The investigation revealed a multitude of features and structures dating from several periods, including extensive settlement remains from the Late Neolithic and Bronze Age. Excavations have also been carried out in this area previously, resulting in rich finds assemblages. This paper focuses on the site’s largest and best preserved two-aisled house, K30, which is dated to Early Bronze Age period I (1700-1500 BC). This longhouse therefore represents the final generation of houses of two-aisled construction. It also contained charred plant remains, which provide information on arable agriculture of the time and the internal organisation of the building at a point just prior to three-aisled construction becoming universal. The remains indicate continuity in both agriculture and in internal organisation between the late two-aisled and early three-aisled longhouses. The two-aisled house at Nørre Holsted can therefore make a significant contribution to the long-running debate about this architectural change, which has often focussed on developments in farming: The increased importance of cattle husbandry is said to have been the main reason for breaking with the tradition of two-aisled construction.

The Nørre Holsted locality comprises the top of a sandy plateau that forms a ridge running north-south. The slightly sloping plateau lies 38-42 m above sea level and the ridge is surrounded by damp, low-lying terrain that, prior to the agricultural drainage of recent times, was partly aquiferous. The site occupies a central position in the southern part of Holsted Bakkeø, a “hill island” that is primarily characterised by sandy moraine. People preferred to live on well-drained ridges with sandy subsoil throughout large parts of prehistory and this was also true in the Late Neolithic and Bronze Age. On the area uncovered at Nørre Holsted, remains were found of 16 two-aisled houses, of which three had sunken floors. Ten of these houses are dated to the Late Neolithic and three are assigned to the first period of the Bronze Age. During Early Bronze Age periods II and III, a total of 14 three-aisled longhouses stood on the sandy plateau. As can be seen from figure 2, the houses from the Late Neolithic and Early Bronze Age lie more or less evenly distributed across the area. However, the buildings from the Late Neolithic/Early Bronze Age period I form a distinct cluster in the eastern part, while a western distribution is evident for the houses from Early Bronze Age periods II-III. The western part of the site lies highest in the terrain and a movement upwards in the landscape was therefore associated with the introduction of the three-aisled building tradition. Tripartition of the dimensions can be observed in both the two- and the three-aisled houses, with this being most pronounced in the latter category. The three-aisled Bronze Age houses from periods II and III, which represent the typical form with rounded gables and possibly plank-built walls, show great morphological and architectonic uniformity. Conversely, the two-aisled house remains are characterised by wider variation. The small and medium-sized examples, with or without a partly-sunken floor, represent some very common house types in Jutland. Conversely, the largest longhouse, K30, represents a variant that is more familiar from areas further to the east in southern Scandinavia.

The largest two-aisled house at Nørre Holsted was located on the eastern part of the sandy plateau, where this slopes down towards a former wetland area (fig. 3). The east-west-oriented longhouse had a fall of 1.5 m along its length, with the eastern end being the lowest part at c. 38 m above sea level. Its orientation towards the wet meadow and bog to the east is striking, and it stood a maximum of 50 m from the potential grazing area. A peat bog lay a further 100 m to the east and in prehistory this was probably a small lake. Sekær Bæk flows 600 m to the north and, prior to realignment, this watercourse was both deeper and wider where it met the former lake area. Access to fresh water was therefore optimal and opportunities for transport and communication by way of local water routes must similarly have been favourable. It should be added that the watercourse Holsted Å flows only 1 km to the south of the locality.

House K30 had a length of 32 m and a width of 6.5-7 m, with the western part apparently being the broadest, giving a floor area of more than 200 m2. The eastern gable was slightly rounded, while that to the west was of a straighter and more open character. The wall posts were preserved along most of the two sides of the building and the internal (roof-) supporting posts were positioned just inside the walls. Two transverse partition walls divided the longhouse, with its ten central posts, into three main rooms (fig. 5). These posts were the building’s sturdiest and most deeply-founded examples. Charcoal-rich post-pipes could be observed in section, and these revealed that the posts consisted of cloven timber with a cross-section of c. 25 cm. The central posts were regularly spaced about 3 m apart, except at the eastern and western ends, where the spacing was 4 m (fig. 5). The posts along the inside of the walls were less robust and not set as deeply as the central posts. There were probably internal wall or support posts along the entire length of the walls. These were positioned only 0.5 m inside the walls and must therefore have functioned together with these. Based on the position of these posts, the possibility that they were directly linked to the central posts can be dismissed. It seems much more likely that they were linked together by transverse beams running across the house – a roof-supporting feature that, a few generations later, moved further in towards the central axis to become the permanent roof-bearing construction. The actual wall posts or outer wall constituted the least robust constructional element of the longhouse.

Remains of the walls were best preserved in the eastern part, and the wall posts here were spaced 1.5 m apart in the eastern gable and 2 m apart in the side wall (fig. 5). The wall posts had disappeared in several places, particularly in the central part of the building. Entrances could not be identified in the side walls, possibly as a consequence of the fragmentary preservation of the post traces. Two transverse partition walls, each consisting of three posts, were present in the western and eastern parts, with the latter example being integrated into a recessed pair of posts. The western room had an area of 59 m2 and contained two pits, while the eastern part was filled with charred plant material, consisting largely of acorns. The actual living quarters may have been located here, even though the larger central room, with an area of c. 85 m2, could just as well represent the dwelling area with its large, deep cooking pit (fig. 5). The eastern room had an area of 60 m2 and therefore did not differ significantly in area from that to the west.

The entire fill from features that could be related to longhouse K30 was sieved. The objective was to retrieve small finds in the form of micro flakes and pottery fragments that are normally overlooked in conventional shovel excavation. The associated aims included ascertaining whether the flint assemblage could reveal the production of particular tools or weapons in the building. Unfortunately, not a single piece of pottery or any other datable artefacts were recovered. Only a few small flint flakes, which simply show that the finds from house K30 conform to the typical picture of a general reduction in the production of flint tools at the transition from Late Neolithic to Early Bronze Age. The 11 flint flakes from the longhouse merely reflect the simple manufacturing of cutting tools. Consequently, no bifacial flint-knapping activities took place within the building, and there is a lack of evidence for specialised craftsmen. The great paucity of finds is typical of houses from the Late Neolithic and Early Bronze Age which do not have a sunken floor. It is therefore important to look more closely at the charred plant material (plant macro-remains) concealed in the fills of the postholes and pits. In the case of house K30, the soil samples have provided a range of information, providing greater knowledge of what actually took place in a large house in southern Jutland at the beginning of the Bronze Age.

The scientific dating of house K30 is based on barley grains from two roof posts and from a wall post in the eastern part. The three AMS radiocarbon dates assign the longhouse to Early Bronze Age period I, with a centre of gravity in period Ib (fig. 6). Plant macro-remains have previously been analysed from monumental three-aisled Bronze Age houses in southern Jutland. It is therefore relevant to take a look inside a large longhouse representing the final generation of the two-aisled building tradition. Do the results of the analyses indicate continuity in the internal organisation of these large houses or did significant changes occur in their functional organisation with the introduction of the three-aisled tradition?

During the excavation of longhouse K30, soil samples were taken from all postholes and associated features for flotation and subsequent analysis of the plant macro-remains recovered. An assessment of the samples’ content of plant macro-remains and charcoal revealed that those from two central postholes and a pit contained large quantities of plant material (fig. 7), whereas the other samples contained few or no plant remains. It was therefore obvious to investigate whether there was a pattern in the distribution of the plant macro-remains that could provide an insight into the internal organisation of the house and the occupants’ exploitation of plant resources. The plant macro-remains can be used to investigate the organisation of the house because the house site lay undisturbed. The remains can therefore be presumed to date from the building’s active period of use. The plant remains lay on the floor of the house and they became incorporated into the fill of the postholes possibly as the posts were pulled up when the house was abandoned or when the posts subsequently rotted or were destroyed by fire. The plant macro-remains therefore reflect activities that have taken place in the immediate vicinity of the posthole in question.

Only barley, in its naked form, can be said to have been definitely used by the house’s occupants, as this cereal type dominates, making up 80% of the identified grains (fig. 8). It is also likely, however, that emmer and/or spelt were cultivated too as evidence from other localities shows that a range of cereal crops was usually grown in the Early Bronze Age. This strategy was probably adopted to mitigate against the negative consequences of a possible failed harvest and also in an attempt to secure a surplus. Virtually no seeds of arable weeds were found in the grain-rich samples from the postholes where the central posts had stood; just a few seeds of persicaria and a single grass caryopsis were identified. This indicates that the crops, in the form of naked barley, and possibly also emmer/spelt, must have been thoroughly cleaned and processed. In contrast, the sample from pit A2500, in the western part of the house, contains virtually no cereal grains but does have a large number of charred acorn fragments (fig. 9). The question is, how should this pit be interpreted? If it was a storage pit, then the many acorns should not be charred, unless the pit and the remnants of its contents were subsequently burnt, perhaps as part of a cleansing or sterilisation process. It could also be a refuse pit, used to dispose of acorns that had become burnt by accident. In which case this must have been a temporary function as permanent refuse pits are unlikely to have been an internal feature of the house’s living quarters. Finally, it is possible that this could have been a so-called function-related pit that was used in connection with drying the acorns, during which some of the them became charred.

From the plant macro-remain data it is clear that the occupants of longhouse K30 practised agriculture while, at the same time, gathering and exploiting natural plant resources. It should be added that they probably also kept livestock etc., but these resources have not left any traces in the site’s archaeological record – probably due to poor conditions for the preservation of bones. A closer examination of the distribution of plant macro-remains in house K30 reveals a very clear pattern (fig. 9), thereby providing an insight into the internal organisation of the building. All traces of cereals are found in the eastern half of the house and, in particular, the two easternmost roof postholes contain relatively large quantities, while the other postholes in this part of the building have few or no charred grains. This could suggest that there was a grain store (i.e. granary) in the vicinity of the penultimate roof-bearing post to the east, while the other cereal grains in the area could result from activities associated with spillage from this store, which contained processed and cleaned naked barley. No plant macro-remains were observed in the posthole samples from the opposite end of the building. The plant remains in this part of the house all originate from the aforementioned pit A2500, which contained a large quantity of acorns, together with a few arable weed seeds. The pit should possibly be interpreted as an acorn store or a functional pit associated with roasting activities or refuse disposal.

The distribution of the plant macro-remains provides no secure indication of the location of the hearth or, in turn, of the living quarters. However, if the distribution of the charcoal in the house is examined (fig. 10), it is clear that there was charcoal everywhere inside house K30. This indicates that the longhouse was either burned down while still occupied or, perhaps more likely, in connection with its abandonment. A more detailed evaluation of the charcoal found in the various postholes and other features reveals the highest concentrations in the central room, suggesting that the hearth was located here, and with it the living quarters. This is consistent with the presence of a large cooking pit, found in the eastern part of this room. Perhaps this explains the presence of open pit A2500 in the western part of the house, which constitutes direct evidence against the presence of living quarters here. Another explanation for the highest charcoal concentrations being in the central room could also have been the entrance area, where there would be a tendency for such material to accumulate.

Plant macro-remains have previously been analysed from large Bronze Age houses in the region, namely at the sites of Brødrene Gram and Kongehøj II, and plant remains from a somewhat smaller Late Neolithic house at Brødrene Gram were also examined. In many ways, K30 corresponds to the houses at Brødrene Gram (houses IV and V) and Kongehøj II (house K1). There is continuity with respect to the cereals represented in the Late Neolithic house at Brødrene Gram and the three-aisled Early Bronze Age houses at Brødrene Gram and Kongehøj II; naked barley and emmer/spelt are the dominant cereal types. There is, however, some variation in the cereal types present in the three-aisled Bronze Age houses, as hulled barley also occurs as a probable cultivated cereal here. It therefore seems that, with time, an even broader range of crops came to be cultivated when houses began to have a three-aisled construction. Another marked difference evident in the composition of the plant macro-remains is that the grain stores in the two-aisled houses contain only very few weed seeds, while those in the later houses are contaminated to a much greater extent with these remains. This could be due to several factors. One possible explanation is that the grain was cleaned more thoroughly before it was stored at the time of the two-aisled houses. Another explanation could be that there were, quite simply, fewer weeds growing in the arable fields in earlier periods, possibly because these fields were exploited for a shorter time and less intensively. This would mean that the field weeds were not able to become established to the same degree as later and fewer weeds were harvested with the cereal crop. As a consequence, the stored grain would contain fewer weed seeds relative to later periods. If the latter situation is true, the increase in field weeds could mark a change in the use of the arable fields, whereby each individual field was exploited for a somewhat longer period than previously.

A common feature seen in all the houses is that they had grain stores in the eastern part of the building and storage was therefore one of the functions of this part. No secure evidence was however found of any of the houses having been fitted out as a byre. The three-aisled house IV at Brødrene Gram apparently also had a grain store at its western end – where K30 had its acorn-rich pit. However, while the western end of the Brødrene Gram house, and that of the other houses, is interpreted as a dwelling area, this room apparently had another function in K30, where the living quarters appear to have been located in the central room, as indicated by the cooking pit and the marked concentration of charcoal.

Longhouse K30 differs from the later houses at Brødrene Gram and Kongehøj II in that these two three-aisled houses contain large quantities of chaff (spikelet forks) of wheat, possibly employed as floor covering, while no such material was observed in K30. However, it is unclear whether this is due to differences in the internal organisation of the buildings or to preservation conditions. Conversely, the use of possible function-related pits, like the one containing acorn remains in house K30, appears to have continued throughout the subsequent periods, as the Bronze Age house at Brødrene Gram also contains similar pits, the more precise function of which remains, however, unresolved. A high degree of continuity can thereby be traced, both in the crops grown and the internal organisation of the two- and three-aisled longhouses in southern Jutland. There was, however, some development towards the cultivation of a wider range of crops.

In turn, this suggests that, in terms of arable agriculture and internal building organisation, there was no marked difference between the late two-aisled and early three-aisled houses – or, more correctly, between the large houses of Bronze Age periods I and II in southern Jutland. More secure conclusions with respect to continuity and change in the internal organisation of the buildings would, however, require a significantly larger number of similar analyses, encompassing several house types of different dimensions from a longer period of time and across a larger geographic area. Nevertheless, let us address the problem by including house sites in other regions, because this should enable us to gain an impression of the degree to which the picture outlined above for southern Jutland is representative of larger parts of southern Scandinavia.

In several cases, both in the large two-aisled longhouses from Late Neolithic period II to Early Bronze Age period I and the large three-aisled longhouses from Early Bronze Age periods II-III, we see an internal division of the building into three main rooms. This tripartite division does, however, become clearer and more standardised with the advent of the three-aisled building tradition, which is a special characteristic of the longhouses of southern Jutland. Food stores were apparently often kept in the eastern parts of these houses. This is shown by the concentrations of charred grain found in these areas, and in some cases the larders must have been positioned immediately inside the eastern gable. Over time, traces of grain stores have been recorded from sunken areas in a number of house sites in Jutland. As a rule, these sunken floors constituted the eastern part of two-aisled houses of Myrhøj type, which were particularly common, especially in Jutland, during the Late Neolithic and the first period of the Bronze Age. One reason for lowering the house floor in this way was possibly a requirement for more space to store grain. It has been pointed out that a sunken floor gives greater head clearance in a room which, in turn, optimises the possibility of keeping the grain dry. In some cases, these sunken floors were almost totally covered by charred barley and wheat grains; surely the result of stored grain having fallen from an open loft during a house fire.

In the Late Neolithic, arable agriculture apparently increased in importance as it became more intensive and diverse, with a wider range of crops now being cultivated. Agriculture in the Early Bronze Age was simply a continuation of the agricultural intensification evident in Late Neolithic arable agriculture. There was a possible difference in that fields were probably more commonly manured in the Early Bronze Age, though the first secure evidence for manuring dates from the Late Bronze Age. The plant macro-remains from the Early Bronze Age include significantly greater numbers of weeds, suggesting that individual arable fields had a longer period of use. Moreover, nutrient-demanding hulled barley came on to the scene as a cultivated crop. This has been demonstrated for example in the aforementioned longhouses at Brødrene Gram and Kongehøj II, both of which date from the Early Bronze Age period II. However, a large component of hulled barley has actually been demonstrated in remains from a Late Neolithic sunken house site at Hestehaven, near Skanderborg.

Most Late Neolithic and Early Bronze Age farms in what is now Denmark were located on nutrient-poor sandy soils, and this was also the case at Nørre Holsted. In itself, location on these soils suggests that soil-improvement measures were employed. Indirectly, it can also tell us something of the significance of livestock, if it is assumed that cattle supplied a major proportion of the material used to manure the arable fields. Domestic livestock is, however, virtually invisible in the Late Neolithic settlement record, compared with that from the three-aisled contexts of the Bronze Age. There are records from Jutland of about 15 longhouses with clearly evident stall dividers, but this total seems very modest relative to a total number of Bronze Age house sites of around 1000. It has long been maintained in settlement archaeology that the three-aisled building tradition was better suited to the installation of a byre. On the face of it, this seems plausible for animals tethered in stalls. But the byre situation is, however, unlikely to have been a direct cause of the change in roof-bearing construction, as highlighted by recently expressed doubts in this respect. Neither are there grounds to dismiss the possibility that byres were installed in two-aisled longhouses. There is an example from Hesel in Ostfriesland, northwest Germany, where a large two-aisled house, measuring 35 x 5-6 m, contained stall dividers in its eastern half. An example from Zealand can also be mentioned in this respect: At Stuvehøj Mark near Ballerup there was a two-aisled longhouse, measuring 47 x 6 m, with possible post-built stall dividers in its eastern half. It stood on a headland surrounded by wetland areas and, like longhouse K30 at Nørre Holsted, it had a marked fall from the west to east gable.

Preserved stall dividers in Bronze Age houses are, therefore, still a rare phenomenon and phosphate analysis of soil has yet to produce convincing results in this respect. There must be another explanation for the change in building architecture. It is possible that the massive monumentalisation process of Early Bronze Age period II played a crucial role in this respect. As described in the introduction, the first three-aisled houses were built higher up in the terrain. A position on the highest points of the landscape is a recurring feature at many other localities with longhouses from Early Bronze Age periods II-III. This visualisation process involved consistent use of the timber-demanding plank-built walls and took place primarily in southern, central and western Jutland. Here, forests had to yield to the huge resource consumption involved in constructing three-aisled houses because it was here that the tradition of plank-built walls was strongest. This situation must be seen in conjunction with barrow building, where there was a corresponding and coeval culmination in the construction of large turf-built burial mounds. Was the three-aisled tradition introduced quite simply because it became possible to build both wider and higher? Period II has the largest longhouses found in Scandinavia to date and these could reach dimensions of 50 x 10 m. The buildings became much wider and the earth-set posts for the plank walls were in some cases founded just as deep as the roof-bearing post pairs, which could extend 50-70 cm down into the subsoil. This could, in turn, suggest that some longhouses had more than one storey. It should also be pointed out that the large-scale construction of longhouses and barrows came to a halt at the same time – in the course of period III, i.e. shortly before 1200 BC. It therefore seems likely that the three-aisled building tradition was introduced as an important step in the actual monumentalisation process rather than as a result of a need to adjust to new requirements for internal organisation. At the end of the Early Bronze Age and throughout the Late Bronze Age, the dimensions of three-aisled houses were reduced and the houses adopted a much less robust character. There was no longer a need for monumental construction. The significance and symbolism by the large buildings constructed in the Early Bronze Age period II and the first part of period III is though a longer and more complex story and it should not be studied in isolation from the barrow-building phenomenon of the time.

Lars Grundvad
Museet på Sønderskov

Martin Egelund Poulsen
Museet på Sønderskov

Marianne Høyem Andreasen
Moesgaard Museum






Grundvad, L., Poulsen, M. E., & Andreasen, M. H. (2015). Et monumentalt midtsulehus ved Nørre Holsted i Sydjylland. Kuml, 64(64), 49–75.