82 Journal of the North Atlantic Special Volume 2 *Department for Medieval and Renaissance Archaeology, University of Aarhus, Denmark; mogens.hoegsberg@gmail.com. Introduction The dwellings are one of the fundamental aspects of the archaeology of Norse Greenland. Since its publication in 1941, the primary work on the topic has been Aage Roussell’s thesis, “Farms and Churches in the Mediaeval Norse Settlements of Greenland,“ published in Meddelelser om Grønland (Roussell 1941). In this thesis, Roussell presented a theory about the development of the dwelling types of Norse Greenland, based on his comprehensive survey of all the available evidence. While later datings have since invalidated the central points of Roussell’s theory, namely that the various house types represented a chronological development where one type replaced the other, the typology itself has only been subject to minor revisions (e.g., Andreasen 1981). After circa 1950, relatively few houses have been completely excavated, and little work has been done on the topic of the Norse dwellings. The excavations of the two Western Settlement farms V54 (excavated during the Inuit-Norse Project in 1976–77) and The Farm Beneath the Sand (GUS, excavated 1991–96) were the fi rst to seriously undermine Roussell’s theory of a chronological development of house types. According to the theory, the centralized house type was the last to develop, but at both sites such houses existed contemporaneously with other types of houses (Andreasen 1981:182, Arneborg 1998:80–81). Discussions about the actual typology have also been rare. In his 1981 article, “Langhus-ganghuscentraliseret gård,” Claus Andreasen offered a slight revision of Roussell’s typology (Andreasen 1981). Svend Erik Albrethsen has also written on the topic, and in the article, “Træk af den norrøne gårds udvikling på Grønland,” he presents a number of possible houses from the early phase of the settlement in Greenland which can be deduced from the excavation plans of the pre-war excavations (Albrethsen 1982). He also offered an interpretation of the buildings phases of one of the large centralized complexes in the Western Settlement, V53d (Albrethsen 1982:271). In this project, Albrethsen tackled a major obstacle in working with evidence from the old excavations. Up until circa 1960, excavations of Norse sites in Greenland were not carried out stratigraphically. The excavators located the wall lines, after which the individual rooms were excavated while the walls were left standing. This method resulted in very little insight about the way the Greenlandic houses were constructed and how they developed over time. In many cases, old excavation plans therefore show what in reality were multiple phases—the initial structure, and additions and rooms that may have existed in one phase of the building’s history but went out of use in a later time. This uncertainty in building phase chronology obviously presents us with serious problems, and should be kept in mind at all times when reading the following. This article does not propose to discuss all of the issues surrounding the various dwelling types of Norse Greenland. First and foremost, it aims to discuss the typology which Roussell developed, focusing on the terminology. No criticism of Roussell is implied—his work was state of the art when it was carried out, and his thesis has been the basis of all the work which has been done on Greenlandic houses since then. Roussell’s Typology Roussell divided the Greenlandic dwellings into three types: the long house, the passage house, and the centralized house. The long house was defi ned simply as a house with one or more rooms. The passage house was defi ned as a house with rooms in rows behind one another and connected by one or more passages. Finally, the centralized house was defi ned as a house where dwelling and animal stalls are in the same block, which is usually of the passage-house type. Morphologically, then, we are only dealing with two types since the passage house and the centralized house were distinguished from each other only by the absence or presence of rooms for animals (Roussell 1941:137). Roussell gave a number of examples of the different types. The long house was exemplifi ed by, Continuity and Change: The Dwellings of the Greenland Norse Mogens Skaaning Høegsberg* Abstract - In his 1941 thesis, Aage Roussell established a typology for the dwellings of the Greenland Norse. While some of his conclusions have since been dismissed, Roussell’s typology itself has not been subject to a fundamental review, although there are problems with it. This paper reviews Roussell’s typology for the dwellings of the Greenland Norse and suggests a revised typology and terminology. It argues that the medieval architecture of the Greenlanders may have been more inspired by contemporary Scandinavian architecture than Roussell imagined, and that the so-called passage house is very sparsely represented in Greenland. 2009 Special Volume 2:82–101 Norse Greenland: Selected Papers from the Hvalsey Conference 2008 Journal of the North Atlantic 2009 M.S. Høegsberg 83 e.g., the earliest phase at Ø83 (Hvalsey) and V51 (Sandnes), the passage house by, e.g., Ø2 and Ø29 ruin 18, and fi nally the centralized house was exemplifi ed by, e.g., Ø52a and Ø16 (Roussell 1941:138) (Figs. 1, 2, 3). Among the long houses, Roussell defi ned a secondary type, what he called “the fully developed long house,” exemplifi ed by the dwelling at Ø51 (Fig. 4). It should be noted that Roussell did not classify this as a separate type as such, but as as sub-type of the long house (Roussell 1941:149, 206). Roussell’s central theory about the different house types was that they represented a chronological development where one type replaced the other: the passage house replaced the long house and was itself later replaced by the centralized house. Roussell principally regarded the development of the passage house and the centralized house as practical and adaptive measures. The move from the long house to the fully developed long house was supposedly a result of the “unsatisfactory nature” of the ordinary long house, but Roussell did not offer a more detailed explanation of the reasons behind this shift (Roussell 1941:203, 206). The fully developed long house was seen as the immediate precursor to the passage house, due to its central room functioning as a sort of a lobby, giving access to the other rooms in the building (Roussell Figure 1. The dwelling at ruin group Ø83, usually taken to be the farm Hvalsey, mentioned in written sources. Aage Roussell, who excavated the dwelling here, believed room IX—the middle room in the upper row—to be the oldest part of the house. From Roussell (1941:141). 84 Journal of the North Atlantic Special Volume 2 1941:206–207). Roussell explained the move from the fully developed long house to the passage house in terms of adaptation: the passage house being better suited to the colder Greenlandic climate (Roussell 1941:212). The same argument accounted for the incorporation of animal stables into the passage house, creating the centralized house. Terminology and Typology The terminological problems connected with Roussell’s typology relate to the two fi rst types, the long house and the passage house, and the core of the problem lies in the implicit assumptions underlying the usage of the two terms. Regarding the long house, it is obvious that Roussell viewed the Greenlandic long house simply as a local variation on the Scandinavian house type. In the case of the passage house, it is equally obvious that Roussell viewed the Greenlandic passage house as essentially the same type of house which was known from Iceland, although in Roussell’s time only post-medieval examples were known. The long house The long house as a term is not easily defi ned, and it is important to note that the term is not exclusively used about Scandinavian or even European buildings. There is no doubt that Roussell used the term with reference to Scandinavian long houses, but even in a Scandinavian context, the term is not used unambiguously. Some defi ne long houses as buildings incorporating both dwelling and animal premises, while others use the term to include buildings which only contain a dwelling. As such, a “long house” has also been used to describe dwellings of the Icelandic Viking age type which are sometimes referred to as a “hall” or “skáli,” and there can be no doubt that Roussell’s use of the term referenced this type. In my opinion, this association is problematic, as there are major differences between Roussell’s Greenlandic long houses and buildings of the hall type. When Roussell wrote his thesis, there were no buildings in Greenland which were directly comparable to Icelandic hall-type houses, but a number of examples have been discovered since. One of these is the earliest phase of the so-called landnáma farm in Narsaq (ruin group Ø17a) which had straight long walls. Another example of the same type of house is the oldest phase of the Farm Beneath the Sand (Figs. 5, 6). Two examples with curved long walls were found in Qassiarsuk (ruin group Ø29a) in the 1960s and 1970s (Figs. 7, 8). One of these had been partially documented during Poul Nørlund and Mårten Stenberger’s excavations there in 1932, but it was not recognized at the time for what it was (Albrethsen and Ólafsson 1998:19, Krogh 1978:421, Nørlund and Stenberger 1934:61, Vebæk 1993:14). The two Qassiarsuk buildings have not been fully Figure 2. The dwelling at ruin group Ø2, excavated by Daniel Bruun in 1894. To date, this house remains the best Greenlandic example of a passage house in the Icelandic sense in that it does have a passage which runs the depth of the building. Nevertheless internal traffi c also seems to have fl owed from room to room as exemplifi ed by the doorways between rooms III and V and rooms IV and VI. From Bruun (1896:216). 2009 M.S. Høegsberg 85 Figure 3. The dwelling at ruin group V52a which was used by Aage Roussell as an example of the centralized house, containing both dwelling and livestock premises. As with most buildings of its type, the dwelling here seems to represent multiple phases, and remains of earlier buildings were recognized by Roussell in rooms I, II, IV, VIII, and IX. Note how internal traffi c is not dominated by any central passage. From Roussell (1941:161). 86 Journal of the North Atlantic Special Volume 2 excavated, while both the landnáma farm and The Farm Beneath the Sand are completely excavated. Artifact typology and carbon dates place the landnáma farm in the early 11th century, while the carbon dates from the hall at The Farm Beneath the Sand are more ambiguous, ranging from the 11th and into Figure 5. The dwelling at ruin group Ø17a, the so-called landnáma farm in the modern town of Narsaq. Room I was determined as the oldest phase of the house, making the house one of the few excavated Greenlandic examples of a long house or hall. From Vebæk (1993:13). Figure 4. The dwelling at ruin group V51, usually interpreted as Sandnes of the written sources. From Roussell (1936:31). 2009 M.S. Høegsberg 87 the 13th century, with the majority in the 12th century (Arneborg et. al. 1998:27–29). It is the opinion of the excavators that the hall at The Farm Beneath the Sand was originally constructed in the 11th century (Albrethsen and Ólafsson 1998:25). The house had a central hearth and benches along the walls, and one end of the house may have been partitioned off from the rest by a wall of lighter materials (Albrethsen and Ólafsson 1998:22). We know considerably less about the landnáma farm in Narsaq, but here, too, there was a central hearth, and it must be considered likely that there were benches along the walls (Vebæk 1993:14). Whether the oldest phase of the landnáma farm had interior subdivisions or not, we do not know, and the same is true for the two Qassiarsuk-houses since these are not completely excavated. Nevertheless, all four houses can readily be seen as “long houses” in the way Roussell thought of the term. They seem to refl ect the contemporary building tradition in Iceland and, to a certain extent, also in the rest of Scandinavia. The row-house The houses that Roussell actually used as examples of the Greenlandic long house seem to belong to a different tradition. While they are certainly elongated, they differ from the above mentioned houses in that the interior of the dwelling is clearly and markedly subdivided—not by walls of lighter materials, but by solid turf and stone walls. They are usually characterized by a row of rooms which may be supplemented by one or more rooms at the back. Houses of this type may be called row-houses, and a number of houses can be singled out as examples of this type of building. Examples include the later phase of the landnáma farm in Narsaq, but also the dwellings at Ø20, V51, and Ø47 (Figs. 4, 9, 10). Rather than seeing these simply as Greenlandic variations of the hall-type houses discussed above, I believe they should be regarded as a new development. In stressing the subdivision of the dwelling interior, these houses essentially follow a development which happened in Europe during the late Viking age and early middle ages. Here, dwellings became more complex than earlier styles, with more rooms and probably more specialized room functions (Roesdahl and Scholkmann 2007:159). This trend can also be traced in Iceland where dwellings of the so-called Þjórsárdalur-type, with Stöng as the best known example, appear to refl ect Figure 6. The exacavated part of the oldest phase of the Farm Beneath the Sand. As the oldest phase at Ø17a, it was a long house or hall with straight, long walls. The house measures about 12 x 5 m on the inside. North is up. From Albrethsen and Ólafsson (1998:21). 88 Journal of the North Atlantic Special Volume 2 Figure 7. The dwelling at ruin group Ø29a, Brattahlíð. On the original plan from Mårten Stenberger and Poul Nørlund’s excavation in 1932 are marked the older remains of a long house which were recorded in 1932, but not realized for what they were. From Albrethsen (1982:275). 2009 M.S. Høegsberg 89 A very similar house has been found in Greenland at Ø71 North (Fig. 12). At fi rst glance, it does not look much like Stöng, but closer inspection reveals that it follows much the same scheme: there is a living room (III) to the west of the main room (I+II) of the dwelling, and at the back of the house there are two small outshots, mirroring the same elements at Stöng. Ø71 is not a carbon copy of Stöng, and it has further rooms placed lengthwise to the main room. Nevertheless, the similarities outweigh the differences, and one of the interesting similarities is also the way in which the front wall of the living room in both instances has been pulled back from the front wall of the hall proper, accentuating both elements as different parts of the building. Other interesting examples of the row-house are the dwellings at Ø20 and V51, which are the houses that Roussell called the fully developed Greenlandic long house (Fig. 4 and 9). He used this term because he viewed houses such as these as the immediate precursors to the passage house. The two houses have remarkably similar layout schemes. Both have three rooms where the middle room functions as the main conductor of internal traffi c. The main difference between the two houses is that at V51, the central room (room II) is quite narrow and works more as a sort of lobby than is the case at Ø20, where the middle room is somewhat larger and clearly a functional living space. Another similarity is the smaller rooms at the back of both houses. Daniel Bruun, who excavated Ø20, identifi ed a doorway from the middle room to the room at the back, while Aage Roussell was not able to fi nd a doorway from the middle room, or lobby, into the room at the back of V51 (Roussell 1936:33). Nevertheless, a door might well have existed in the back wall of room II; in the 1920s and 1930s, excavators often had great diffi culty fi nding the doors between rooms, something which can also be seen in the plans of the landnáma farm in Narsaq and the dwelling at Ø47 (Figs. 5, 10). It must be considered likely that there was a door in the back wall of room II at V51, making the layout of these two houses even more similar. Ø20 and V51 are the only clear examples we have of this particular version of the row-house. None of them are closely dated, but at W51 the house seems Figure 8. Sketch plan of ruin 60 at ruin group Ø29a, Brattahlíð. The ruin was partially excavated by Knud J. Krogh in 1975. From Albrethsen (1982:276). the same development (Fig. 11). At Stöng, the hall was still the central part of the house, but Stöng also features a number of other rooms: a living room is placed lengthwise to the hall, and two smaller rooms adjoin the back side of the hall. Note that while I designate the individual units of the houses as “rooms,” this is only to avoid terminological confusion: in both Greenland and Iceland, the proper designation of many of the rooms would be “houses” as they had their own roof-bearing structures. Figure 9. The dwelling at ruin group Ø20, excavated by Daniel Bruun in 1894. The layout of the house is much akin to V51, although the middle room at this farm is clearly still a proper room and not just an entrance room as at V51. The house is about 23 m long. From Bruun (1896:267). 90 Journal of the North Atlantic Special Volume 2 Figure 10. The dwelling at ruin group Ø47, the bishop’s seat at Garðar. The plan shows multiple phases. Aage Roussell considered the southern part of the house to be a passage house due to the existence of the passage numbered X. From Nørlund (1930:83). 2009 M.S. Høegsberg 91 ogy, we see remarkably similar layout schemes as parts of these larger buildings. Parts of both V53c and Ø64a are characterized by two rooms at either side of a central lobby and with rooms at the back (Figs. 13, 14). It seems likely that these houses were originally of the same type as Ø20 and V51, but as mentioned in the introduction, it must be remembered that our understanding of the development of these houses is extremely poor. to belong to the last phase of occupation at the site. If V51 was abandoned when the Western Settlement was depopulated in the mid-14th century or sometime in the second half of the century (which the latest 14C-datings suggest), that indicates a 14th-century date for that house (Arneborg et.al. 1999:161, and see below). Looking at the excavation plans of some of the houses classifi ed as centralized in Roussell’s typol- Figure 11. The house at Stöng in Þjórsárdalur, Iceland, excavated by Aage Roussell in 1939. The long house or hall, making up rooms I and II, has now been extended with several other rooms. From Roussell (1943:78). Figure 12. The dwelling at ruin group Ø71 North (the left part of the complex). Note the similarity of the layout to the one at Stöng. From Vebæk (1992:30). 92 Journal of the North Atlantic Special Volume 2 Houses with extremely similar layouts to Ø20 and V51 have also been found in Iceland, although they were unknown at Roussell’s time. The best example is the dwelling at the farm Gröf, in the Öræfi - district of Iceland, which was excavated in 1955–57 (Fig. 15). The farm is believed to have been destroyed by an eruption of Öræfajökull in 1362, which indicates a 14th-century date of the building, making it roughly contemporary to V51 in Greenland (Ágústsson 1982:258; Gestsson 1959:8, 84). The layout of Gröf is very similar to that of Ø20 and V51. At Gröf, there is a separate room at each end of the building, but the central part of the house retains the division into three parts with rooms at the back. Figure 13. The dwelling at ruin group V53c, a centralized house. Top: the original plan. Bottom: the possible original core of the building marked in black. Similarities with the dwellings at V51 and Ø20 may be seen in the rooms numbered II, III, IV, VII, and VIII. The main entrance gives access to a small room or lobby from which access is gained to rooms at either side as well as two smaller rooms on the back of the house. In the top plan, also note how no central passage dominates internal traffi c which is conducted through a number of smaller doorways from room to room. From Roussell (1941:172). 2009 M.S. Høegsberg 93 land, and despite Gröf-type houses being forerunners of the Icelandic passage house, Roussell’s claim that the Greenlandic houses of the V51-type were the precursors of a Greenlandic passage house is not necessarily correct. Roussell saw a close connection between Greenlandic and Icelandic houses, and even went so far as to suggest that the passage house could have been a Greenlandic development which At Gröf, the middle room must be classifi ed as a passage which could only function as a conductor of traffi c. There is little doubt that houses of the Gröftype can be seen as the immediate precursor to the Icelandic passage house, of which only few medieval examples are yet known. In spite of the great similiarities between houses such as Ø20 and V51 in Greenland and Gröf in Ice- Figure 14. The dwelling at ruin group E64a, a centralized house. Top: the original plan. Bottom: the possible original core of the building marked in black. Similarities with the dwellings at V51 and Ø20 may be seen in the rooms numbered III and VI—with an unnumbered passage between them, and the rooms IV and V at the back of the house reached through the passage. In the top plan, note how no central passage dominates internal traffi c which is conducted through a number of smaller doorways and passages from room to room. From Vebæk (1943:24). 94 Journal of the North Atlantic Special Volume 2 The passage house and the centralized house The primary reason for this lack of consideration of a Norwegian inspiration is the great found its way to Iceland (Roussell 1941:212). This scenario seems unlikely for a number of reasons. First of all, houses that might be characterized as passage houses in the Icelandic sense are very rare in Greenland (see below). Secondly, houses have been excavated in Scandinavia which may have been the inspiration for both Greenlandic houses with the V51-layout and Icelandic houses of the Gröf-type. The layout with two rooms at either side of a central room or lobby is known from Scandinavia from the 11th century onwards. Here the layout appears in urban environments, with examples from Trondheim, Oslo, and Lund, and later in the middle ages it is found in urban stone architecture (Fig. 16) (Christophersen 1994:184, Fett 1989:34, Kristensen 1999:79, Schmidt 1994:80). Although this hypothesis cannot be substantiated without further dating evidence from Greenland and Iceland, it is possible that the inspiration to adopt the layout with two rooms at either side of a central lobby was transplanted to Iceland and Greenland from Scandinavia. The fact that the Scandinavian houses existed in urban environments, obviously much different from the dispersed rural settlements of Greenland and Iceland, would seem to speak against this theory. However, it should be remembered that most of the contact between Greenlanders and Norwegians probably came in the form of Norwegian town-based merchants. The possibility that Norwegian houses could be the inspiration for houses in Greenland was not considered by Roussell, because he had fewer buildings with which to compare his Greenlandic material and also no dating evidence. Another reason was probably that Roussell saw a closer connection between the architecture of Iceland and Greenland than the material really supports. Figure 16. Three examples of houses with three rooms from Trondheim. In all three instances, the entrance appears to have been in the middle room, from which access was gained to the rooms at either side. The lower house differs only in that the entrance room does not run the depth of the building, but the central idea is intact. The houses are from 1050–1150 (top), 1150–1175 (middle), and 1000–1050 (bottom). From Christophersen (1994:186). Figure 15. The dwelling at the farm Gröf in Iceland. Here the entrance room has been diminished to a passage proper, but apart from this and the added rooms at either end, the similarities with the Greenlandic houses at V51 and Ø20 are striking. From Ágústsson (1982:258). 2009 M.S. Høegsberg 95 Figure 17. An Icelandic passage house as they might look in the late 19th century when Valtýr Guðmundsson wrote Privatboligen paa Island i Sagatiden. This is the farm Laufás as it looked in 1876. The term passage house is apt, since the passage clearly guides internal traffi c in the building. From Ágústsson (1982:265). influence of Valtýr Guðmundsson’s 1889 book, Privatboligen paa Island i Sagatiden. Guðmundsson set up a typology of Icelandic buildings, based primarily on his interpretation of how buildings were described in the Icelandic sagas, but undoubtedly the typology was also informed by his knowledge of contemporary Icelandic architecture. At the time, the passage house was widespread in Iceland. They were large structures with a clearly conceived basic concept, where internal traffic was effectively conducted by a central passage running the depth of the building (Fig. 17). Although there are many variations on the type and although not all rooms of a late 19th-century passage house could necessarily be reached from the central passage, most of them could, and the term passage house—which, incidentally, was not used by Guðmundsson, but was adopted later—is perfectly suited for this type of building. Guðmundsson’s book was to have a profound infl uence on Greenlandic Norse archaeology. Thus, when Daniel Bruun conducted his famous exploratory journey in South Greenland in 1894, he knew the book, and undoubtedly attempted to fi nd similar houses in Greenland (Bruun 1896:178). Perhaps it does not come as a surprise that he did. It was Daniel Bruun who found and excavated the dwelling at ruin group Ø2, which remains to this day the only good Greenlandic example of a passage house in the Icelandic sense (Fig. 2). At Ø2, the house has a central passage running the depth of the building, distributing traffi c to the various rooms. However, in addition to the passage, there are doorways between several of the rooms, meaning that internal traffi c was conducted as much between the individual rooms as through the passage. The same is the case in most of the other Greenlandic examples of the passage house and centralized houses which, according to Roussell, were morphologically passage houses. The dwelling at Ø29 was categorized by Roussell as a passage house, yet the passage (no. III) neither runs the depth of the building nor works as the main distributor of traffi c (Fig. 18). The same goes for the centralized houses at V53c and V53d, where most of the internal traffi c is conducted from room to room (Fig. 13, 19). In all these cases, it should be kept in mind that the excavation plans incorporate multiple phases; however, no matter how these houses looked at any given time, it does not change the fact that in most cases passage happened from room to room. Comparing the Greenlandic passage houses and centralized houses with a late medieval Icelandic passage house, the differences are striking. The farm Forna-Lá, which is dated to ca. 1450–1550, is clearly laid out as a direct descendant of the house type seen at Gröf and a precursor to later examples of this type (Fig. 20) (Ágústsson 1982:258; Eldjárn 1951:104, 108). The fi rst row of rooms in the house is laid out as at Gröf, with two rooms at either side of a passage, but at Forna-Lá the rooms on the back are considerably larger. The same scheme is seen at the farm Sandártunga, which is later, having been deserted after an eruption of Mount Hekla in 1693 (Fig. 21) (Ágústsson 1982:260; Eldjárn 1951:108, 111). In both cases, there is an extra room in the fi rst row of rooms, meaning that here, too, there is passage directly from room to room. However, this does not change the clear layout of both buildings, which are both planned around the central passage. 96 Journal of the North Atlantic Special Volume 2 Such a clear layout is not seen in any of the Greenlandic houses which have been termed passage houses or centralized houses. The Greenlandic buildings have a more “organic” feel to them, and appear to be less clearly organized than the late medieval and post medieval Icelandic passage houses. The Greenlandic houses give the impression of having grown in a more haphazard way. In fact, the differences are so marked, that the term passage house ought not to be used to describe the Greenlandic buildings. Using the term for Greenlandic houses implies a closer affi nity between the architecture of the two countries than the evidence supports. Compared with other Greenlandic houses, the passage houses and centralized houses differ in that they do not appear to have been organized with a primary row of rooms placed end to end—hence the impression of a more haphazardly organized building where rooms seem to “huddle” together with most traffi c conducted directly from room to room. Also, there appears to be a greater number of rooms in these complexes than in most row-houses. A major caveat is obviously still our poor understanding of the development of the houses. For instance, it seems highly unlikely that all the rooms in a large complex such as V53d (Fig. 19) were in function at the same time. I suggest that Roussell’s passage house and centralized house both be termed “conglomerate buildings” to acknowledge the shared morphological features of these buildings, no matter whether they contain stables and/ or byres or not. This terminology can be clarified by referring to “a conglomerate building without stable/byre” or “a conglomerate house with stable/byre.” The question of animal stables and byres still remains, because when looking at the evidence today, the number of conglomerate buildings without rooms for animals is quite small. Most of Roussell’s examples of the “passage house” in Greenland were centralized houses. He only had a few examples of the type without rooms for animals. The archaeological record thus begs the question whether or not conglomerate buildings without rooms for animals were ever a common type in Greenland. Roussell’s examples included the houses at Ø2, Ø29, Ø83, and the southern part of the dwelling at Ø47 (Figs. 1, 2, 10, 18). Of these, at least Ø47 ought to be discounted. There is a passage in the southern end of the house here, but most of the rooms are lined up in one row, and the passage is not a defi ning trait of the house nor of the way traffi c fl owed through it. Rather, the dwelling at Ø47 should be counted among the row-houses. The house at Ø83 is not as easy to discount, but as the plan also shows, Roussell had great diffi culties with the excavation, and there are many uncertainties about the building. It is possible that it was a conglomerate house without rooms for animals , but for the time being, it cannot be attributed to any grouping. The houses at Ø2 and Ø29 remain as possible examples of the conglomerate building without rooms for animals, and the same can be said for Ø66 (Fig. 22). All were excavated by Daniel Bruun, and in no instances did he fi nd evidence for a byre. One possible explanation for this could be that Bruun did Figure 18. The dwelling at ruin group Ø29, which was originally excavated by Daniel Bruun in 1894 and re-excavated by Poul Nørlund and Mårten Stenberger in 1932. Aage Roussell used this as one of his examples of a Greenlandic passage house, even though most internal traffi c fl owed from room to room. It is true that a passage (III) conducts traffi c from room I to IV, V, and VI, but it does not run the depth of the house. Also rooms I–II, VI–VII, and VI–VIII are directly connected and not dependant on the passage. From Nørlund and Stenberger (1934:73). Larger numbers added by the present author. 2009 M.S. Høegsberg 97 to have recognized them for what they were. In any case, until further buildings of the conglomerate type are excavated, the type without rooms for animals seems to be relatively rare. not expect to fi nd rooms for animals, and therefore did not identify them. However, byres and stables are usually characterized by either stall stones and/ or stone fl oors and as such one would expect Bruun Figure 19. The dwelling at ruin group V53d, the largest centralized house excavated in Greenland. While we do not know which rooms are contemporaneous, it remains clear that internal traffi c was from room to room and not via a central passage. From Roussell (1941:180). Figure 20. The dwelling at the farm Forna-Lá in Iceland, an early example of a true passage house. The outshots at the back are now larger rooms, and even though a small room in the front row is reached through another room, the passage clearly dominates internal traffi c. From Ágústsson (1982:259). 98 Journal of the North Atlantic Special Volume 2 heim, is dated to a phase which ends circa 1025. The house at V51 obviously predates the abandonment of the Western Settlement, which today is assumed to have happened in the mid- to late 14th century. The latest carbon dates from Sandnes are cal. 1413 A.D. (1393–1432 σ1 standard deviation) and cal. 1390 A.D. (1323–1412 σ1 standard deviation), suggesting a late 14th-century abandonment of Sandnes (Arneborg et.al. 1999:161). What this means for the dating of the dwelling is another question, but welltended turf houses might have stood for between 60 and 100 years (Gestsson 1982:168). Thus, at least the type must have been known since the fi rst half of the 14th century, but even this may be a conservative estimate. Conglomerate buildings are known from both settlements, but the largest appear in the Western Settlement. When they appeared cannot be determined on the basis of the available evidence, but they obviously predate the abandonment of the Western Settlement. With reference to the conglomerate building at V54, Claus Andreasen (1981) notes that while there were traces of earlier occupation phases, no traces of freestanding byres could be found in the area—suggesting that V54 may have been a conglomerate building with rooms for animals from its inception in the early 11th century. Similarly, it has been suggested that The Farm Beneath the Sand may have been a conglomerate Dating Presently, we only have a very vague idea of the dates of the different types of Greenlandic houses. The long house at Ø17a is dated to the early period of Norse occupation in Greenland, and the same may be true of the long house at The Farm Beneath the Sand. Judging from the morphology of the two long houses at Ø29, an early date also seems likely for them (Albrethsen and Ólafsson 1998:25; Vebæk 1993:30, 73). The appearance of the row-house cannot be closely dated, but the type would appear to have had an extended period of use. In most cases, the exacavators of the 1920s and 1930s only investigated the latest phases of the dwellings, and judging from the presence of the type at a site such as Ø47 (Garðar), it must have been in use until the abandonment of the Eastern Settlement sometime in the mid- 15th century. If it is correct that the row-house was a Greenlandic variation on European developments in the late Viking and early medieval periods, it would seem likely that the type evolved at the latest in the early 12th century, but quite likely earlier. The appearance of houses with the layout seen at Ø20 and V51, with a central lobby, cannot be closely dated either. If it was inspired by Scandinavian houses with similar layout, it could date from any time after the mid-11th century, since the oldest currently known example of the type, from Trond- Figure 21. The dwelling at the farm Sandártunga in Iceland. The layout is strongly reminescent of Forna-Lá. From Ágústsson (1982:260). 2009 M.S. Høegsberg 99 Conclusion It hardly needs stating that the reassessment of the house typology of Norse Greenland presented here can only be considered as provisional. Further excavated examples of Norse houses and more solid dating evidence is needed to further elucidate the variety of the house types of Norse Greenland. When the excavation at The Farm Beneath the Sand is published, it may give answers to crucial questions—most importantly how a conglomerate house developed over time. Based on the extant evidence, it seems clear that the architecture of Norse Greenland, and its development, was more complicated than Aage Roussell was able to ascertain. This new interpretation is now possible, partly because we have more excavated buildings at our disposal today than Roussell did, some from Greenland and more from both Iceland and Scandinavia. This additional data allows us to make new inferences about the typology and also suggests that we need to be careful in our choice of terminology. Scandinavian-type long houses were unknown in Greenland when Roussell wrote his thesis, but later investigations have shown that this type of house was also in use in Norse Greenland. It is reasonable to use the term long house to describe houses of this type. Roussell’s long houses, however, appear to represent another type of building. While elongated, they are not long houses in the Scandinavian sense. Their heightened focus on room divisions as well as their spatial organization— a row of rooms with or without further rooms at the back, sets them apart building from a very early point (Andreasen 1981:182, Arneborg 1998:81). Along with renewed investigations into the different house types of Norse Greenland, a systematic campaign of dating is necessary in order to chronologically defi ne the appearance and prevalence of the various types. Figure 22. The dwelling, churchyard, and church at ruin group Ø66, excavated by Daniel Bruun in 1894. Along with the dwelling at Ø2, this could be a passage house in the true sense of the word since most rooms appear to have been reached through a central passage running the depth of the building. The scale at the bottom totals 100 feet. From Bruun (1896:377). 100 Journal of the North Atlantic Special Volume 2 no longer be substantiated that any of the Greenlandic house types were prerequisites for another to develop. Later dating evidence has demonstrated that row-houses and conglomerate houses existed contemporaneously in the Norse settlements in Greenland, suggesting a far more varied architectural landscape where continuity and change are not mutually exclusive. This muddied picture is a natural consequence of the larger amount of data available to us today, new dating evidence, and advances in excavation methods which force us to question the validity of older excavation results. It also means that this article does not pretend to present a replacement of Roussell’s typology, merely a temporary adjustment which should be further amended as new evidence is brought to light. Literature Cited Ágústsson, H. 1982. Den islandske bondegårds udvikling fra landnamstiden indtil det 20. århundrede. Pp. 255–268, In B. Myhre, B. Stoklund, and P. Gjærder (Eds.). Vestnordisk byggeskikk gjennom to tusen år—Tradisjon og forandring fra romertid til det 19. århundre (AmS-skrifter 7). Arkeologisk Museum i Stavanger, Stavanger, Norway. 287 pp. Albrethsen, S.E. 1982. Træk af den norrøne gårds udvikling på Grønland. Pp. 269–287, In B. Myhre, B. Stoklund, and P. Gjærder (Eds.). Vestnordisk byggeskikk gjennom to tusen år—Tradisjon og forandring fra romertid til det 19. århundre (AmS-skrifter 7). Arkeologisk Museum i Stavanger, Stavanger, Norway. 287 pp. Albrethsen, S.E., and G. Ólafsson. 1998. A Viking age hall. Pp. 19–26, In J. Arneborg and H.C. Gulløv (Eds.). Man, Culture, and Environment in Ancient Greenland: Report on a Research Programme. The Danish National Museum and Danish Polar Center, Copenhagen, Denmark. 212 pp. Andreasen, C. 1981. Langhus-ganghus-centraliseret gård. Hikuin 7:179–184. Arneborg, J. 1998. The Farm beneath the Sand: Summary. Pp. 80–82, In J. Arneborg and H.C. Gulløv (Eds.). Man, Culture, and Environment in Ancient Greenland: Report on a Research Programme. The Danish National Museum and Danish Polar Center, Copenhagen, Denmark. 212 pp. Arneborg, J., J. Heinemeier, N. Rud, and Á. Sveinbjörnsdóttir. 1998. AMS Dates from the hall (XVII). Pp. 27–30, In J. Arneborg and H.C. Gulløv (Eds.). Man, Culture, and Environment in Ancient Greenland: Report on a Research Programme. The Danish National Museum and Danish Polar Center, Copenhagen, Denmark. 212 pp. Arneborg, J., J. Heinemeier, N. Lynnerup, H.L. Nielsen, N. Rud, and Á.E. Sveinbjörnsdóttir. 1999. Change of diet of the Greenland Vikings determined from stable carbon isotope analysis and 14C dating of their bones. Radiocarbon 41(2):157–168. Bruun, D. 1896. Arkæologiske Undersøgelser i Julianehaabs Distrikt. Meddelelser om Grønland XVI:173– 461. from the Scandinavian-style long houses. Therefore a new term, row-houses, is suggested for them. As regards Roussell’s passage house and centralized house, it is suggested they be seen as one type, since they appear to be morphologically similar, albeit with many variations. This type is highly ambiguous, particularly because of our poor knowledge of how such buildings developed. They appear to be made up of a number of agglomerated rooms, and compared with row-houses, they display a much greater variation in their spatial layout. As they appear to us today, most of these buildings are not organized around a passage functioning as the primary conductor of internal traffi c, and their resemblance to Icelandic passage houses does not seem as clear as Roussell’s term would suggest. Therefore, a new term, conglomerate buildings, is suggested. Once again, it must be stressed that most of the known examples of conglomerate houses were excavated non-stratigraphically, meaning that the plans we have today are composite, displaying multiple phases. Consequently the conglomerate building as a “type” is likely to be subject to major revisions as new evidence is produced. A few of the excavated examples of row-houses, V51 and Ø20, appear with a layout that recalls similar buildings from Iceland, such as Gröf, and similar buildings seem to have been the core of later conglomerate houses. This layout, with two rooms at either side of a middle room, functioning as a lobby of sorts, and with one or more rooms at the back side, is very similar to a number of urban houses which have been excavated in Scandinavia. Lack of dating evidence does not allow us to conclude that the Greenlandic and Icelandic buildings were inspired by Scandinavian urban houses, but the possibility should be kept in mind, and perhaps new dating evidence will be able to throw further light on this. If this should turn out to be the case, it suggests a closer association between Scandinavia and Greenland than Roussell imagined. It also suggests a need to look further than Iceland to fi nd possible parallels to the Greenlandic material—something the early 20th-century pioneers of Norse Greenland archaeology did less than we do today, in part because there was less excavated material from Scandinavia at the time with which to compare the Greenlandic material, and partially because there was a tendency to see Greenland as a country which was, in the middle ages, very far away from Scandinavia and increasingly isolated. A certain reluctance to consider Norwegian material, due to the controversy between Denmark and Norway over Greenland, may also have played a part in this, although this factor should not be over-emphasized. Certainly, the picture presented here is less tidy than the one Roussell painted in his thesis, and it can 2009 M.S. Høegsberg 101 Christophersen, A. 1994. Gård og grunn. Pp. 113–212, In A. Christophersen and S.W. Nordeide. Kaupangen ved Nidelva (Riksantikvarens skrifter nr. 7). Riksantikvaren, Trondheim, Norway. 324 pp. Eldjárn, K. 1951. Tvenna bæjarrústir frá seinni öldum. Árbók hins íslenzka fornleifafélags 1949–1950:102–119. Fett, T.M. 1989. Bygninger og bygningsdetaljer. Pp. 15– 92, In E. Schia (Ed.). De arkeologiske utgravninger i Gamlebyen, Oslo bind 6—Hus og gjerder. Alvheim and Eidi—Akademisk Forlag, Øvre Ervik, Norway. 180 pp. Gestsson, G. 1959. Gröf í Öræfum. Árbók hins íslenzka fornleifafélags 1959:5–87. Gestsson, G. 1982. Brugen af sten og tørv i de islandske huse fra landnamstid til nyere tid. Pp. 162–172, In B. Myhre, B. Stoklund, and P. Gjærder (Eds.). Vestnordisk byggeskikk gjennom to tusen år—Tradisjon og forandring fra romertid til det 19. århundre (AmS-skrifter 7). Arkeologisk Museum i Stavanger, Stavanger, Norway. 287 pp. Guðmundsson, V. 1889. Privatboligen paa Island i Sagatiden samt delvis i det øvrige Norden. Andr. Fred. Høst and Søns Forlag, Copenhagen, Denmark. 270 pp. Kristensen, H.K. 1999. Land, by og bygninger.54–81, In E. Roesdahl (Ed.). Dagligliv i Danmarks middelalder— En Arkæologisk Kulturhistorie. Gyldendal, Copenhagen, Denmark. 420 pp. Krogh, K.J. 1978. Brattahlið. Pp. 417–421, In J. Hoops (Ed.). Reallexicon der Germanischen Altertumskunde Band III. Walter de Gruyter. Berlin, Germany. Nørlund, P. 1930. Norse Ruins at Gardar: The Episcopal Seat of Mediaeval Greenland. Meddelelser om Grønland LXXVI:3–170. Nørlund, P., and M. Stenberger. 1934. Brattahlid. Meddelelser om Grønland 88(1):5–161. Roesdahl, E., and B. Scholkmann. 2007. Housing Culture. Pp. 154–180, In J. Graham-Campbell and M. Valor (Eds.). The Archaeology of Medieval Europe— Volume 1: Eighth to Twelfth Centuries AD. Aarhus University Press, Århus, Denmark. 479 pp. Roussell, Aa. 1936. Sandnes and the neighbouring farms. Meddelelser om Grønland 88(2):3–219. Roussell, Aa. 1941. Farms and churches in the Mediaeval Norse settlements of Greenland. Meddelelser om Grønland 89:3–342. Roussell, Aa. 1943. Stöng, Þjórsárdalur. Pp. 72–97, In M. Stenberger (Ed.). Forntida Gårdar i Island. Ejnar Munksgaard, Copenhagen, Denmark. 332 pp. Schmidt, H. 1994. Buildings Customs in Viking Age Denmark. Bergiafonden, Nivaa, Denmark. 178 pp. Vebæk, C.L. 1943. Inland farms in the Norse East Settlement. Meddelelser om Grønland 90(1):3–110. Vebæk, C.L. 1992. Vatnahverfi : An inland district of the Eastern Settlement in Greenland. Meddelelser om Grønland – Man and Society 17:1–132. Vebæk, C.L. 1993. Narsaq: A Norse landnáma farm. Meddelelser om Grønland – Man and Society 18:1–85.