Barshalder 3 Rojrhage in Grötlingbo a Multi-Component Neolithic Shore Site on Gotland

Barshalder 3 Rojrhage in Grötlingbo A multi-component Neolithic shore site on Gotland

Martin Rundkvist, Christian Lindqvist & Karl Thorsberg

Stockholm Archaeological Reports 41, 2004 Department of Archaeology, University of Stockholm

1 Acknowledgements Table of Contents This work has benefited from the comments of 1. Introduction ...... 3 Agneta Åkerlund, Fredrik Hallgren, Mats P. Mal- 2. Shore displacement, Neolithic mer, Inger Österholm, Ludvig Papmehl-Dufay topography and site size ...... 4 and Niklas Ytterberg. Remaining weaknesses are 3. Neolithic deposit preservation ...... 4 all the authors’ responsibility. 4. Stone Age surroundings ...... 6 Funding for radiocarbon analyses was kindly 5. Pottery ...... 9 provided by the National Heritage Board of Swe- 5.1. Introduction...... 9 den and the analyses were undertaken by Göran 5.2. Number and shapes of vessels...... 9 Possnert of the Ångström Laboratory in Upp- 5.3. Decoration, pottery styles and sala. chronology ...... 9 This publication has been produced and prin- 5.4. TRB pottery ...... 11 ted with the support of Stiftelsen Konung Gus- 5.5. GRK pottery...... 12 tav VI Adolfs fond för svensk kultur and Mag- 5.6. STR pottery...... 14 nus Bergvalls stiftelse. 5.7. Ware...... 14 The manuscript was completed in August 5.8. The pottery’s state of preservation ....15 2002. 6. Bone harpoon point ...... 16 7. Ground igneous rock axes...... 16 8. Flint...... 17 8.1 Knapping technique...... 18 8.2 Intended products...... 18 8.3 Raw material ...... 18 8.4 Find contexts ...... 18 8.5 Dating ...... 19 8.6 Iron Age flint ...... 19 9. Bones and radiometric analyses...... 20 9.1 Sampling ...... 20 9.2 Hunting and fishing ...... 20 9.3 Animal husbandry ...... 24 9.3.1 Cattle (Bos taurus)...... 25 9.3.2 Pig (Sus scrofa) ...... 27 9.3.3 Dog/fox (Canis sp.) ...... 29 10. Intra-site find distributions ...... 29 11. Phasing, site function and cultural interpretation ...... 31 12. References...... 32

© 2004 the authors ISBN 91-631-4947-8 Cover art: Samuel Rundkvist Messner Editing, maps, plans, photographs, graphic de- sign: Martin Rundkvist, [email protected] Printed by Intellecta Docusys, Sollentuna

2 1. Introduction stones from the graves. The field notes were ne- ver organised into a formal report. Rojrhage 1:1 is a small property in Grötlingbo In working with the Iron Age graves for my parish, southern Gotland, near the southern pa- doctorate and preparing a basic report of the rish border toward Fide. A house was moved to excavations, I, Martin Rundkvist, had to bring the site in 1925, and in 1955 the property was order also to the Stone Age finds. I found this to parcelled out from the southern part of the plot be an excellent opportunity to learn something allocated to the parish gravel pit, Uddvide Sam- about Neolithic matters. Knowing my limits, I fällt. Small archaeological rescue excavations took asked Karl Thorsberg to study the flint and Chris- place on the property in 1930-31, 1947 and 1951- tian Lindqvist the bones, and kept the pottery 52 due to the fact that it was now the garden of and sundries for myself. None of this work’s auth- a private house. Large-scale archaeological work ors took part in the excavations at Rojrhage 1:1. took place 1960-1967, first due to the widening After I had type-set this work in early 2003, of the road and subsequently as a research exca- but before funding had been secured for its pub- vation prompted by the rich finds. The trenches lication, Gunilla Eriksson published her docto- came to cover approximately 700 m2. Such were ral thesis at the University of Stockholm, Norm the horticultural consequences of the excavations and difference. Stone Age dietary practice in the Baltic that the Swedish National Heritage Board was for- region. It contains a paper on stable isotope analy- ced to buy the property in 1963 (KVHAA 1974:124- ses of bones from Västerbjärs in Gothem and 126), subsequently selling it to a tenant in 1975. Ire in Hangvar which has far-reaching implica- The property takes its name from the pas- tions for research into the Middle Neolithic of tureland of which it was part before the Laga Gotland. Four of Eriksson’s results are particu- Skifte land amalgamation reform in the early larly relevant to the present work but were not at 1890s. Rojrhage means ”pasture with cairns”. hand when it was written. These cairns are Iron Age (1st millennium AD) grave markers and the Rojrhage pasture formed 1. The marine reservoir effect for human bone the northern third of the great Barshalder ce- from GRK burials in Gotland is c. 70±40 metery (Rundkvist 1996a, 2003). The Rojrhage years. 1:1 property is part of the Registered Ancient 2. The people of the Pitted Ware Culture on Got- Monument area Grötlingbo Raä 54. land got their dietary protein almost exclusi- The excavations at the site were prompted by vely from seals. Any contribution of pork and rich graves from the Migration – Vendel Period fish to their diet was negligible in quantitative interface. However, it soon became evident that terms. the graves were superimposed upon Stone Age 3. Pig bones found at Pitted Ware sites on Got- deposits, and these are the subject of this work. land belong to animals with an entirely ter- At the excavations little attention was paid to restric diet. This means that the pigs did not the Stone Age deposits beyond the collection of have access to the rubbish of Pitted Ware flint and pottery. Documentation standards were habitation sites. This rubbish would, judging raised somewhat by Gustaf Trotzig in the 1963- from the isotope values of the human bones, 1967 campaign, when the culture layer was trea- have consisted mainly of bones and butchery ted separately from the Iron Age graves and a refuse from seals. The corollary of this is that 2.5 m sampling grid was used. However, Trotzig’s the pigs were not domesticated, but wild or excavations were cut short by insufficient fun- feral. ding and by his promotion, and much of the Sto- 4. The economy of the Pitted Ware Culture on ne Age deposits were excavated hurriedly when Gotland included neither agriculture nor ani- the trenches were being stripped of remaining mal husbandry on any appreciable scale.

3 2. Shore displacement, lines are 220 m apart at the site, and the 5 m a.s.l. Neolithic topography line corresponds to the shoreline about AD 1. Shore displacement between these levels is complicated and site size by the repeated transgressions of the Littorina Sea. The 15 m-above-sea-level (m a.s.l.) line on the In any case, the maximum conceivable width of economic map (sheet 5J6a FIDE) runs through the Neolithic deposits is far less than 200 m. The the excavated area at Rojrhage 1:1 between the Rojrhage deposit is located on the same level, slightly house and the road (fig. 1). The site is on the less than 15 m a.s.l., as the Neolithic sites of Stora NW slope of a ridge, close to the top. The 15 m Domerarve and Gullrum (see below). a.s.l. line corresponds to the Littorina maximum Turning to the phosphate-maps published by (c. 5000 cal BC) in this part of Gotland (Munthe Österholm (1989) and Samuelsson & Ytterberg 1910, re-printed in Österholm 1989 fig. 5 & 2002c (2002), they include five Neolithic shore sites that fig. 5), and formed a 700 m long and mainly less have been free to expand regardless of any limestone than 100 m wide NNE-SSW island or reef with escarpments toward the back and sides: Ajvide in the Rojrhage site just outside the northern end. Eksta, Västös in Hall, Gumbalde in Lau, Hemmor The area was located at the southern coast of in När and Gullrum in Näs & Hablingbo. All have ancient mainland Gotland from the time of the their greatest extent along the ancient shoreline. Littorina maximum until the mid-first millenni- Lengthwise, they vary from 350 to 600 m. Their um AD when the still ongoing land upheaval clo- widths vary from c. 25% to c. 50% of the length. It sed the straits of Fide and Öja. The Littorina should be noted that the phosphate distributions maximum provides a terminus post quem for all at these sites may extend much further than the archaeological deposits discussed in this work. culture layers (Hedemark et al. 2000:10). The edges of the Neolithic deposits have not As to the NE-SW length of the Rojrhage depo- been sought outside the trenches, but there are in- sits along the shoreline, there is only negative evi- dications that their landward limit to the SE may dence from nearby excavations of Iron Age graves have been within the excavated area. Peter Manne- where no Neolithic finds have turned up. The like’s excavations of 1960-61 between the house and mits are thus set by Manneke’s excavations of 1960 the road turned up neither a culture layer beneath (where seal bones, but no Neolithic artefacts, have the Iron Age graves nor many Neolithic stray finds been found in Bhr 1960:13 & 15) on Roes 1:36 pro- in their fills. Manneke, being known as a painsta- perty to the SW and archaeological rescue work king fieldworker, is unlikely to have missed much. 1899-1971 at the parish gravel pit to the NE, loca- Perhaps, then, the landward edge of the deposit ted about 200 m apart. The main trench at Rojrha- may have been at the 15 m a.s.l. line. ge was 35 m long with Neolithic finds from end to As to the seaward edge, there are no Neolithic end, and the Neolithic deposit is thus somewhere finds from the extreme seaward lines A and B of between 35 and 200 m long. Unless the Rojrhage the excavation grid (fig. 1, tables 10a-c), which would deposit has different spatial proportions than the place the NW limit of the deposit less than 20 m large sites or is discontinuous along the shoreline, from the SE one and above at least 13 m a.s.l. This then its width should be less than 100 m. interpretation should however be treated with some caution because the area excavated in line A and B 3. Neolithic deposit was located around the Late Roman Period grave preservation mound Bhr 1967:33, in the fill of which much Neo- lithic material, presumably taken from the surroun- The deposit is described as follows in Trotzig’s ding ground surface, had been re-deposited. The field notes. ”Thickness generally c. 10 cm. Pot- distribution of the Neolithic finds leaves an empty sherds, bones and a few flint fragments were zone all around the mound. found scattered through the layer, which was only The area NW of the trenches is still densely preserved in parts of the excavation surface. Most covered with Iron Age graves which have not pro- of the finds have been collected as stray finds duced any excavation records and would seem to during the excavation of the Iron Age graves. preclude phosphate mapping. The 15 and 5 m a.s.l. Here and there the layer seemed to thicken, and

4 Fig. 1. Plan of excavations 1961–1967 at Rojrhage 1:1. Neolithic sunken features and excavation grid for Neolithic deposits indicated. some of the hearth-like features discovered in gardless of whether artefacts where present or not. association with the graves should originate in Most if not all of the features should thus date the Stone Age layer although it has not been from the Neolithic. The finds from them were trea- possible to ascertain this.” ted inconsistently during fieldwork, in some cases There were 21 non-grave sunken features – in- kept separate and in others collected with the finds cluding Trotzig’s hearth-like features – in the exca- from the relevant grid square. Consequently, and as vated area (table 3a & fig. 1). 11 of them contained only one of them had enough structure to actually Neolithic artefacts, only one of them a probably be classified as a hearth during fieldwork, the finds intrusive Iron Age artefact as well as large amounts from the sunken features have been allocated to or of Neolithic pottery. Bones of the same animal divided among their respective grid squares in the species (determined by Petra Molnar, Åsa M. Lars- find distributions below (section 10). son and Christian Lindqvist) have been found re-

5 Table 3a. Non-grave sunken features. 1 Stora Domerarve 4:1, Stora Domerarve 1:3, Pottery Flint Medebys 1:2, Medebys 1:9, Stjups 1:27 & Feature Bone (g) (frags) Stjups 1:31. Hablingbo Raä 112. Site num- Bhr 1961:17 Ovicaprid 3- ber 62 & 52 in Österholm’s list. Coordinates (3/61) W-E: 1648340, S-N: 6341320. 10-15 m a.s.l. Bhr 1961:18 --gDo (4/61) Separate Late Mesolithic and Early Neolithic Bhr 1961:19a TRB deposits (Österholm 1989:126-140, -1- (5a/61) 2002b) with an Iron Age component. Bhr 1961:20 --Ovicapri d 2 Petsarve 1:38 & Härdarve 1:2 among others. (6/61) Bhr 1961:21a Eke Raä 20, 22, 86, 87, 92, 155. Not in Öster- --- (7a/61) holm’s list. Coordinates W-E: c. 1655850, S-N: Bhr 1961:22b 6-Ovicapri d 6339600. 15-20 m a.s.l. Late Mesolithic and Late (8b/61) Bhr 1961:22c Ovicaprid, dog, Neolithic deposit (Andersson 2000:98-99). -2 (8/61) bird 3 Brunns 1:21. Grötlingbo parish without Raä Bhr 1961:29 --- register number. Site number 56 in Österholm’s (17/61) list. Coordinates W-E: c. 1653500, S-N: Bhr 1961:30 --- (18/61) 6338250. 15-20 m a.s.l. Mesolithic deposit. Bhr 1961:32 4 Gullrum 1:20, Gullrum 1:11 & Stora Hajs- --- (20/61) lunds 1:4. Näs Raä 17 & Havdhem parish Bhr 1961:38 --- without Raä register number. Site number (30/61) Bhr 1961:39b 28 in Österholm’s list. Coordinates W-E: --Ovicapri d (32/61) 1649860, S-N: 6336850. 10-15 m a.s.l. Midd- Bhr 1961:39c --Ovicapri d le Neolithic GRK deposit (Hansson 1900). (33/61) Human SHM 8962, 9120, 10055, 10808, 12136, Bdhr 1967:27 I-n gri (intrusive?) 15274, 18311. B--lhr 1967:34 Sheep, sea 5 Suderkvie 5:1. Grötlingbo Raä 13. Site number 50 in Österholm’s list. Coordinates W- Bdhr 1967:35 I-dn gri In gri E: 1653470, S-N: 6335730. 10-15 m a.s.l. Bhr 1967:46 Pig, seal, bird, 2346 Early Neolithic TRB deposit and Bronze (feature A) fish Bhr 1967:47 Ovicaprid, seal, Age slab cist excavated in 1959 (Manneke 133 (Feature B) bird, fish 1961). SHM 27151. Bhr 1967:48 Mammal, bird, 5-1 6 Kattlunds 1:26, the farmstead museum. Gröt- (Feature C) fish Bhr 1967:49 lingbo Raä 55. Not in Österholm’s list. Coor- -1Mamma l (Feature D) dinates W-E: 1653505, S-N: 6334770. 10-15 Bhr 1967:50 320 Ovicapri d m a.s.l. Neolithic deposit discovered in 1971 (SA hearth 1) in a 3 x 0.5 m trench beneath the Medieval farmhouse at Kattlunds. I have not been able to locate the finds, which dated from histo- 4. Stone Age surroundings ric times except for a small collection of Neolithic pottery. The excavation report clas- Excavations and my own work in documenting sifies the pottery as Pitted Ware but con- private collections have so far turned up a total tains no descriptions or pictures of it (Falck of ten Stone Age sites in Österholm’s next-to- 1971). southernmost site cluster, Hablingbo-Näs-Gröt- 7 Grötlingbo parish gravel pit, Roes 1:28. lingbo (1989:168 fig. 83), on the ancient southern Grötlingbo Raä 45. Not in Österholm’s list. coast of mainland Gotland, as follow from north Coordinates W-E: 1652440, S-N: 6334155. 5- to south (fig. 2). Unexcavated find spots with ty- 10 m a.s.l. 50 m2 badly damaged deposit of pologically unspecific finds are not included. ringed seal bones (Pusa hispida, determined Österholm’s list refers to Österholm 1989:56-59 by Johannes Lepiksaar in 1967) and nine flint (also in Burenhult 1997b:XI, cf. Bägerfeldt 1992: fragments, excavated in 1966 (Rundkvist 44-46).

6 2001). Possibly Early Neolithic judging from 9 Rojrhage 1:1. Grötlingbo Raä 54. Site number a stray-found fragment of a thin-butted flint 8 in Österholm’s list. Coordinates W-E: axe. SHM 15609:3 & 32182. 1652340, S-N: 6333900. 10-15 m a.s.l. Multi- 8 Grötlingbo parish gravel pit, Norrkvie 1:16. component Neolithic deposit excavated 1930- Grötlingbo Raä 54. Not in Österholm’s list. 1967 (Rundkvist 2002). See table 3b for inven- Coordinates W-E: 1652343, S-N: 6334015. 5- tory numbers. 10 m a.s.l. Robbed LN slab cist excavated in 10 Poultry farm, Roes 1:36. Grötlingbo Raä 193. 1927 (Hansson 1927). Two inhumations, part Not in Österholm’s list. Coordinates W-E: of a polished shaft hole axe, two blunt bone 1652120, S-N: 6333760. 10-15 m a.s.l. Late Me- arrowheads and four burnished pottery frag- solithic aceramic deposit, surface-collected ments with wide incised lines. SHM 19055:08. by the landowner. Three pecked and gro- Part of a miniature flint dagger of similar date und greenstone axes and worked flint, kept was found in a Late Iron Age grave (Bhr 1957: in the Ohlsson collection (Rundkvist 1996b). 01, cf. Rundkvist 2003) less than 40 m from the cist.

1 2 3

4 5 6

7-10

Fig. 2. South Gotland. Stone Age sites in Hablingbo, Eke, Näs and Grötlingbo parishes. Three shorelines are shown: the present one of the Baltic Sea, the upper limit of the Littorina Sea (c. 5000 cal BC) and the upper limit of the Ancylus Lake (c. 8500 cal BC).

7 Table 3b. Neolithic finds from Rojrhage 1:1.

Igneous Context Neol. Bone Seal or fish Inv no. Flint rock (cf. Rundkvist 2003) pottery artefact bone artefact S7HM 19535 B-X---hr 1930:0

N0ot submitted B-X--.hr 1931:2 No ost. an - G)F C9582:1-23 BXX--hr 1947:02 (A/47 - G)F C9582:24-32 BXX--hr 1947:03 (B/47

G)F C9582:33-36 BXX-XXhr 1947:04 (C/47

G7F C9582:38-39,44 SXX--.tray finds 194 No ost. an - G1F C10176 B-X--hr 1951:0

S)HM pending alloc B----Xhr 1960:10a (2a/60 - S)HM pending alloc B-X--hr 1960:10b (2b/60

S)HM pending alloc B----Xhr 1960:11 (3/60

S)HM pending alloc B-XX-.hr 1961:15 (1/61 No ost. an - S)HM pending alloc BX---hr 1961:17 (3/61 - S)HM pending alloc BXX--hr 1961:17a (3a/61

S)HM pending alloc B-X--Xhr 1961:19 (5/61 - S)HM pending alloc B-X--hr 1961:19a (5a/61 - S)HM pending alloc B-X--hr 1961:21 (7/61 - S)HM pending alloc B-X--hr 1961:22 (8/61 - S)HM pending alloc BX---hr 1961:22b (8a/61

S)HM pending alloc B-X-XXhr 1961:33a (21a/61

S)HM pending alloc B----Xhr 1961:40 (34/61

SsHM 32623 SXX-XXtone Age feature

S1HM 32623 B----Xhr 1967:0

S3HM 32623 B----Xhr 1967:1

S0HM 32623 B----Xhr 1967:2

SaHM 32623 B----Xhr 1967:24

SbHM 32623 B----Xhr 1967:24

SdHM 32623 B----Xhr 1967:25ab

ScHM 32623 B----Xhr 1967:25

S6HM 32623 B----Xhr 1967:2

S3HM 32623 BXX--Xhr 1967:3

S4HM 32623 B----Xhr 1967:3

S6HM 32623 B----Xhr 1967:3

S7HM 32623 SXXX-Xtray finds 1963-6

8 5. Pottery ferent vessels including the flask or cup. This number disregards the fact that there are not 5.1. Introduction necessarily any rim fragments preserved from Neolithic pottery has been recovered at the site every single vessel at the site, and the actual num- through rescue excavations of Iron Age graves ber must be higher as demonstrated below. in 1947, 1951, 1952, 1960 and 1961; and through What kind of vessels were these, disregarding excavation of the Neolithic deposit 1963-1967. the flask or cup? The 14 measurable rim diame- The finds from 1947-1952 amount to only 17 ters vary between 12 and >35 cm with the medi- fragments and have not been included in the pre- an at 21 cm. Five bottom fragments have been sent study as they are kept in the Gotlands Forn- identified, all of them rounded. A vessel shape sal county museum, Visby. The finds from 1960- with a turned-out rim, a short neck, a lightly 1967 amount to 4170 g Neolithic pottery and do marked shoulder, a steeply inclined belly and a not quite fill two storage cases at the SHM in rounded bottom seems to have been the rule. Stockholm. One fragment (the largest one) includes rim, neck The vessels are described in a typological sys- and shoulder; two rim and neck; and three neck tem with four main parameters: ware, diameter, and shoulder. By far the most of the fragments vessel shape and decoration. To qualify for typo- cannot be placed precisely along the vessel’s pro- logical study of the co-variation of independent file. attributes (Malmer 1962:47-57), a fragment must be determinable in at least two of these dimen- 5.3. Decoration, pottery styles sions. and chronology All fragments can be studied as to ware, but 34 elements of decoration have been defined and this parameter does not lend itself to rigorous 27 of these combine on the fragments to form definitions. All fragments that allow measurement three distinct groups (table 5a & fig. 3). There is of diameter are identifiable vessel parts. This only one instance of each pairwise element com- means that, in order to be used, each fragment bination except 13+31 and 36+53 of which the- must either be a clearly identifiable part of a ves- re are each two. This low number of instances is sel, exhibit decoration, or both. 1300 g (31% of not astonishing given the small size of the samp- the total) of the material satisfies these demands, le. The largest group combines 20 elements in- amounting to 179 fragments, with an average cluding all the pit types and represents Pitted Ware weight per determinable fragment of 7.3 g. Of (GRK). The diagram also includes the dummy these fragments, only 21 lack decoration. This element no. 10 (damaged pits of indeterminable indicates that we are very likely dealing exclusive- type) to underline (redundantly) the integrity of ly with decorated vessels. the group. The two remaining groups, with four All determinable fragments but three can be and three elements respectively, represent TRB ascribed to beakers or larger pots. Of the three, and Corded Ware (STR). The pottery from Rojr- two fit together and belong to an undecorated hage is, thus, a multi-component assemblage of pottery disc (diam 9 cm) and one is a rim-frag- Early and Middle Neolithic pottery styles. GRK ment from a decorated pottery flask or miniatu- predominates but TRB and STR are also in evi- re cup (fig. 4D, diam 2.5 cm). dence. Österholm (1989:171, 175) was thus only partly right in classifying the site as a TRB settle- 5.2. Number and shapes of vessels ment. The first question to answer is how many pot- Adding vessels without preserved rims iden- tery vessels are represented by the finds. To de- tified among the TRB and Corded Ware vessels, termine a minimum number I have studied the a total of 41 individual vessels can be discerned rim fragments, which display the greatest typo- among the finds. This is still a minimum value, logical variation among the vessel parts. There and closer study of the Pitted Ware would no are 44 rim fragments representing at least 36 dif- doubt increase it.

9 Table 5a. Pottery decoration elements. References where applicable to the Ajvide GRK decoration classification system of Österholm (1989:103-109, 1997).

No. of Comb. Ö'holm's Nno. Definitio frags group code

1e0 P4it, uncertain typ 3KG-R

1l1 P2it, conica 1KG2R

1n2 P3it, U-shaped cross-sectio 1KG1R

1n3 P9it, orthogonal cross-sectio 1KG1R

1b4 P3Kit, central bottom kno G-R

1d5 P1Kit, 8-shape G-R

2e0 I7--mpression, uncertain typ

2s1 I1--mpression, square, triangular + rectangular c-

2d2 I1--mpression, pairwise shallow rhomboi

Impression, oblong, vertical, repeated horizontally, L 4-5 mm, 23 5KG-R 1.5-5 mm apart

Impression, oblong end wood, vertical, repeated horizontally, 25 3KG-R L 7-13 mm, 5-6 mm apart

2e7 I3--mpression, shallow oblique cylinder edg

Impression, oblong, vertical, repeated horizontally, L 4-12 28 5BT-R mm, >5 mm apart

3n0 L4ine, uncertain patter 1- -

3y1 L8Kine, acute angle pointing horizontall G9R

3t2 L4Kine, oblique cross-hatching, 3-5 mm apar G1R 1

3t3 L6Bine, vertical, across sherd, 4.5-8.5 mm apar T-R

Line, oblique, repeated horizontally, switching direction after 34 2KG8R >=4 (…////\\\\…)

3t5 L2Kine, horizontal, repeated, 3 mm apar G?R 29

3y6 L3ine, obtuse zigzag horizontall 1KG?R 73

Line, obtuse Vs alternately up and down repeated 37 1- - horizontally

4t1 C7Rord, horizontal, repeated, >=3.5 mm apar S-T

4t2 C3Bord, horizontal, repeated, <3.5 mm apar T-R

4t3 C1Bord stamp, >=3 horizontal, L 18 mm, 3.5-6 mm apar T-R

4e4 C1Rord stamp, vertical, side by side forming continuous surfac S6T 6

4l5 C2-?ord-twined twig stamp, horizontal and vertica 68

4t6 C1-5ord stamp, vertical, L >11 mm, repeated, 5-8 mm apar 6

5t1 R5Rim, line, transversal, 1-2 mm apar S-T

5g2 R1Kim, line, acute zigza G-R

5e3 R7Kim, line, obliqu G1R 7

5t4 R3Kim, line, transversal, >3.5 mm apar G8R 2 Fig. 3. Combination diagram of Neolithic

5t5 R2Kim, line, oblique cross-hatching, 4.5 mm apar G8R 3 pottery decoration elements from Rojrhage, cf. table 5a. Circles denote 5t6 R3Kim, impression, oblong end wood, transversal, 4.5 mm apar G-R different types of pits. Each connecting 5t8 R1Kim, denticulate stamp, straight, transversal, 4 mm apar G-R line represents one potsherd. Double lines denote double instances of an 5t9 R1Kim, double line, transversal, 2.5-5 mm apar G-R element combination. 6m1 D1Kenticulate stamp, straight, oblique, L >10 m G?R 72

7t1 O2Kuter rim edge, impression, oblique, 4-6 mm apar G-R

10 5.4. TRB pottery Compared to the eastern Danish TRB sequen- The pottery of the TRB combination group (fig. ce, the two least incomplete vessels appear to 4) comprises nine fragments from five excava- belong to Koch’s type III and the Svaleklint deco- tion contexts, probably representing five indivi- ration style. Type III differs from type IV mainly dual vessels. As only one of the fragments has a through its lack of sub-shoulder line decoration. preserved rim, four must be added to the num- Koch dates type III to the first part of the Early ber of identifiable vessels at the site based exclu- Neolithic, EN I, with an absolute dating of 3800- sively on rims. The decoration elements of this 3500 cal BC (Koch 1998:91-94). pottery are widely spaced vertical oblong impres- The cord-twined twig stamp (no. 45) and obli- sions (no. 28), widely spaced vertical lines (no. que cylinder edge impression (no. 27) lack combi- 33), closely spaced horizontal cord (no. 42) and nations in the Rojrhage pottery, but should belong vertical stacks of horizontal cord stamps (no. 43). with the TRB pottery judging from the Siretorp The typological determination of this pottery finds (Bagge & Kjellmark 1939, Pl. 61:2, 65:6, 67:10 is hampered by the lack of determinable belly & Pl. 62:1-5, 68:1-2, 68:10, 69:5) and the ware (see fragments. However, judging from the ceramic below). The fragment of a pottery flask or cup is wares as shown below, there is a strong positive also made from a ware most similar to the TRB indication that the TRB vessels at Rojrhage had pottery at the site. Its diameter is much smaller than rounded bottoms and no decoration below the that of the miniature cups found in Gotlandic GRK shoulder. contexts, and none of them are decorated (Sten- From a Scanian perspective, the vessels be- berger 1943:100, Taf. 35:1-3, Taf. 22 fig. 41:2; Jan- long to the Mossby group of the later EN I (Lars- zon 1974:104-105). The vessel is thus more likely to son 1994:217-223, 1997:95 & fig. 4:9), c. 3800- have been a collared flask than a miniature cup and 3650 cal BC. The Mossby group concentrates up- should be counted with the site’s TRB component. on southern and eastern Scania and has contem- The pottery disc, too, would belong here due to the porary parallels in Bornholm (Becker 1947: type’s scarcity in GRK contexts and absence from 161-169; 1990:41-44, 160-163), which is particu- STR contexts.At a diameter of only 9 cm, it seems larly interesting in the Gotlandic context. too small to be useful for bread baking.

Fig. 4A. TRB pottery from Rojrhage 1:1. SHM 32623. Grid square F1. Decoration nos 28 + 43. 4A

11 4C

Fig. 4. TRB pottery from Rojrhage 1:1. SHM 32623. 4B. Grid square H0-I0. Decoration nos 28 + 33. 4C. Stray find 1963-1967. Decoration nos 33 + 42. 4D. Flask or miniature cup, rim. Sunken feature B.

5.5. GRK pottery 1997a:30-33, Lindqvist 1997a:71-73, Österholm The pottery of the GRK combination group (117 2002a). fragments) is characterised by pits, oblique line In the 1980s it was believed that all the GRK patterns, vertical and horizontal zigzag lines, and burials of Ajvide dated to a late phase contempo- oblique lines on the rim (fig. 5). The low frequency rary with the settlement at Ajvide D Lower. The of denticulate stamp decoration (two fragments absolute dates of the Ajvide finds have since been plus two kept in the Gotlands Fornsal county clarified through an extensive programme of ra- museum) is remarkable. In the terminology of diocarbon analyses (Lindqvist & Possnert 1997a: Nihlén (1927) and Schnittger & Rydh (1940), this 55-57; Possnert 2002). Burial began already in the decoration corresponds to the Hemmor-Gull- middle phase concurrently with post-transgres- rum style. sional settlement at Ajvide D Upper. With cor- Compared to the pottery decoration of the rection for the reservoir effect, Lindqvist & Poss- stratified GRK site at Ajvide in Eksta parish, 22.5 nert date post-transgressional settlement at Aj- km to the NNW (Österholm 1989:103-115, vide D Upper and most graves (nos. 1, 2, 13, 28, 1997), the combination of decoration elements 29, 36, 41, 53, 1/94: cf. Burenhult 2002) within at Rojrhage is most similar to that at Ajvide D the interval c. 2900-2600 cal BC. The graves con- Upper, levels 1-2. This can be exemplified by the temporary with settlement at Ajvide D Lower ratio between fragments decorated only with (nos 6, 19 & 30), in their turn, are dated to the conical (no. 11) and cylindrical (nos. 12-13) pits interval c. 2450-2100 cal BC. respectively. The ratio is 0.08 at Ajvide D Upper Both the early and middle GRK phases at Aj- level 1, 0.13 at Rojrhage, and 0.19 at Ajvide D vide belong to the Hemmor-Gullrum pottery Upper level 2. The chronological dichotomy be- phase (cf. Lindqvist & Possnert 1997a:33). In tween conical and cylindrical pits also shows up current terms of southern Scandinavian relative in the combination diagram for Rojrhage’s GRK chronology, this translates to pre-Corded-Ware component (fig. 3), where they are found at op- Middle Neolithic, phase MN A. Thus, even with- posite ends of the combination cluster. Ajvide out the radiocarbon dates from Ajvide, Ajvide’s D Upper levels 1-2 represent the middle phase middle GRK phase and the GRK component at of Ajvide’s three-step GRK sequence, post-da- Rojrhage may be assigned to a late part of MN ting the Littorina transgression deposit and pre- A, corresponding to the interval c. 3000-2700 cal dating area D Lower (cf. Lindqvist & Possnert BC (Edenmo et al. 1997:136).

12 5G

5F 6A

Fig. 5. GRK pottery from Rojrhage 1:1. SHM7 32623. 5A. Grid square G1. Dec. nos 11 + 36 + 59. 5B. Stray find 1963-1967. Dec. nos 12 + 25 + 56. 5C. Stray find 1963-1967. Dec. nos 13 + 34 + 52. 5C 5D. Grid square F4. Dec. nos 23 + 53. 5E. Grid square F3-F4. Dec. nos 13 + 31. 5F. Stray find 1963-1967. Dec. nos 32 + 54. 5G. Stray find 1963-1967. Dec. nos 36 + 55.

Fig. 6. STR pottery from Rojrhage 1:1. SHM 32623. 6A. Grid square F2. Dec. nos 41 + 51. 6B. Grid square F4. Dec. nos 41 + 44.

Fig. 7. Bone harpoon point. Sunken feature A. SHM 32623. 5D

13 Charcoal, hazelnut shells and pottery crusts wise, it may be an example of contemporaneous from the eponymous Hemmor site have, confu- local traditions. The same argument applies to singly, mainly given much later radiocarbon da- the dearth of denticulate stamp decoration at tes, clustering in the interval 2764-2258 cal BC Rojrhage. (Wallin & Martinsson-Wallin 1996:14, Hedemark et al 2000:11). Only two samples (Beta-135646, 5.6. STR pottery 4730±110 BP, charcoal; Ua-17465, 4535±70 BP, The pottery of the STR combination group (fig. human bone, grave 1) from higher altitudes at 6) amounts to seven fragments from three exca- the site have given dates in the interval 2900-3800 vation contexts, probably representing one indi- cal BC. The site is thus largely contemporary not vidual vessel per context. One of these vessels with the early and middle GRK phases at Ajvide has no preserved rim. The decoration elements with Hemmor-Gullrum pottery, but with the late of this pottery are widely spaced horizontal cord GRK phase at Ajvide, with Visby and Ire-Väst- (no. 41), vertical cord stamp side by side forming erbjers pottery. The typology of the Hemmor a continuous surface (no. 44) and closely spaced site’s pottery finds is outside the scope of this transversal lines on the rim (no. 51). Compared work, but it seems warranted to check anew to the Corded Ware sequence of mainland Swe- whether the site’s pottery is really mainly an ex- den, the fragments from Rojrhage fit the defini- ponent of the Hemmor-Gullrum style. tions of Malmer’s groups A-C (Malmer 1962:8, Decoration at Rojrhage is concentrated on the 12, 13). However, rim decoration such as that upper part of the GRK vessels, the criterion used known from Rojrhage occurs only in group C by Malmer to separate the early (Fagervik III) (Malmer 1962 Abb. 5:6), which would place the phase from the late (Fagervik IV) phase at the pottery in group C. This group, known to Mal- Västerbjers cemetery (1962:736-737). Malmer mer only from settlement deposits and mainly as notes that the genesis of the Fagervik III type, single fragments, dates to period 6 (Malmer 1962: thus defined, may pre-date the cemetery. The 89), that is, the first phase of the Late Neolithic, Ajvide stratigraphy, where already the earliest after 2350 cal BC (Holm et al. 1997:215). GRK phase (D Upper levels 3-4) pre-dating the STR pottery is very uncommon on Gotland, graves is characterised by Fagervik III pottery, with only two previously published instances: indicates that this is actually the case. However, Stora Förvar on Stora Karlsö (Schnittger & Rydh Malmer’s (1962:732) suggestion that the begin- 1940 Pl. 24:9, Malmer 1962 Abb. 16:3 & Tab. 2) ning of the Gotlandic GRK cemetery horizon and Ardags in Ekeby parish (Åhlén 1972:21, 8, post-dated the appearance of the Battle Axe Cul- fig. 3.2). However, the presence of surface-cove- ture (that is, the beginning of MN B, 2700 cal ring vertical cord-stamp mainly in the final pha- BC according to Edenmo et al. 1997:136) on the se of the Ajvide pottery (element no. 66 in Ös- Swedish mainland is not borne out by the radio- terholm 1989:103-109, 1997) indicates that many carbon dates from Ajvide as reported by Lind- of Gotland’s GRK sites may conceal small qvist & Possnert. Rather, they seem contempo- amounts of late STR pottery. This seems to be raneous at the available dating resolution. the situation in the Late MN B coastal zone of Although the correspondence between the the Lake Mälaren area and Östergötland (Ols- pottery styles of Ajvide D Upper 1-2 and Rojr- son & Edenmo 1997:186-188). hage is clear, it should be noted that only half of the Rojrhage GRK decoration elements have 5.7. Ware clear counterparts at Ajvide. Most importantly, The following discussion of the ceramic wares horizontal lines of vertical oblong impressions treats the 133 fragments that could be allocated (nos. 23 & 25, n=8) are absent from the Ajvide to the three decoration groups, TRB, GRK and pottery. This may be due to chronology, in which STR. It is based on four variables recorded for case corresponding pottery might be sought each fragment: temper size (class 1-3), porous among the finds from excavations during the surface (yes / no), pink surface in fluorescent 1990s between areas D Upper and D Lower at tube lighting (yes / no) and minimum thickness Ajvide (cf. Burenhult 2002 colour plate A). Other- (mm).

14 Table 5b. Ceramic wares and pottery decoration groups. thick but neither porous nor pink, which indicates that most of them belong to the TRB ves- Unad- TKRB GRR ST sels. orned Small amounts of organic residue suitable for M8ean temper (!) 27. 10. 34. 2. radiometric analyses have been observed on the insides of a few coarsely tempered and thick P%orous 1%1 5%6 0%5 pottery sherds; none of them is, however, deco- P%ink 1%1 2%2 8%6 0 rated. M7ean thickness (mm) 71. 76. 64. 8. 5.8. The pottery’s state of preservation Of these, all except the thickness are impres- 4170 g of recovered pottery and 41 identifiable sionistic and the results are thus of highly ques- vessels entail a mean recovered weight per vessel tionable value if any at all outside the context of of <100 g. Large and sturdy pots of the types in this investigation. There is even a risk that my question can weigh several kilograms each, and impressionistic criteria may have varied during the finds are thus highly incomplete. This is hardly the recording, but as the material is small and surprising given the probable functional nature was studied during a short time span I assume of the deposit as refuse, the repeated use of the them to have been constant. This is all sloppy site into recent times, the limited extent of the method with many lamentable precedents in the excavations and the priority given during field- annals of archaeology, and I can offer only two work to the superimposed Iron Age graves. excuses. To quantify temper I would have had to The fact that so much Neolithic pottery (149 thin-section the sherds. I cannot think of any g) was found in the earliest Iron Age structure at workable stringent definition of a porous surfa- the site, the Late Roman Period grave mound ce. As for the colour variable, I have no excuse, Bhr 1967:33, gives some indication as to the post- as I might have used a standard colour chart. depositional hardships endured by the Neolithic The minimum thickness of each determinable deposits. Although the later Iron Age graves were fragment with both inner and outer surfaces pre- flatter, less fill-consuming, and seem mainly to served (n=165) has been measured and varies have been placed on top of the ground surface, from 4.5 to 13.4 mm with an arithmetic mean of it seems that very little of the Neolithic deposits 7.8 mm. The distribution is continuous from 5.5 could have survived the Iron Age untouched. to 9.8 mm. The mean fragment thickness of all Also, due to its multi-component nature, the de- determinable pottery exceeds that of all three posit constituted a palimpsest of repeated uses decoration groups. This is because most of the of the site all through the Neolithic. It had thus 19 unadorned fragments determinable as to ves- very likely already been thoroughly churned when sel part are above mean thickness, reflecting the it started to settle into place before the Bronze fact that the decoration concentrates on the up- Age hiatus at the site. This is borne out by the per, thin-walled parts of the vessels. observations that the fragments are small, that The sample sizes for TRB (n=9) and STR few fragments can be assigned to the same ves- (n=7) are small but should still permit an im- sel, and that the TRB and GRK pottery mixed pressionistic characterisation. As shown in table indiscriminately across the total pottery distribu- 5b, the ceramic ware of the three decoration tion. Interestingly, the latest pottery type, STR, groups differs dramatically. The TRB ware is coar- was also the only one with a focussed spatial dist- sely tempered and thick. The GRK ware is finely ribution in squares F2 and F4. tempered and mostly has a porous surface. The Most sherd breaks are covered with dirt but not STR ware is coarsely tempered, mostly pink, and visibly abraded. Only very few fragments show ab- thin. rasion possibly due to wave action on a seashore, The unadorned fragments determinable as to and, due to their scarcity, they were probably trans- vessel part are coarsely tempered and extremely ported to the site in their abraded state.

15 6. Bone harpoon point shown that the morphology of the igneous rock axes (as opposed to the flint axes) varies greatly, Three bone artefacts have been identified among exhibits very weak typological clustering, and does the finds, only one of which can be determined not conform to the TRB and STR norms of the as to function and type: the tip of a bone har- Swedish mainland. Petrén could divide the ig- poon point from sunken feature A (fig. 7). Only neous rock axes into only two vague types: an one barb remains, finely sculpted. Since so little EN-MN type with bulging broad sides and roun- is preserved of the point it is hard to determine ded neck, and an MN-LN type with flatter broad as to type. However, with the short barb and the sides and straight neck. The Rojrhage specimen, long acutely tapered tip it has no good parallels though lacking its neck, has flat broad sides and among the finds from MN graves and settlements thus belongs to the latter type. However, none published by Stenberger et al. (1943), Janzon of either Petrén’s 173 igneous rock axes nor his (1974) and Österholm (1989). Good matches are, 117 flint axes shows any close similarity to the however, found in harpoon points from excava- Rojrhage axe’s dimensions. His sample includes tion unit G6, Stora Förvar in Eksta parish the finds from MN graves and settlements pub- (Schnittger & Rydh 1940, pl. I, fig. 4) and Hem- lished by Stenberger et al. (1943), Janzon (1974) mor in När parish (Wallin & Martinsson Wallin and Österholm (1989). 1996:22 fig. 11, fourth harpoon point from the From a mainland Swedish perspective, the axe left). These points taper slowly from a wide flat shows little similarity with the Late Mesolithic base to an acute tip, with closely set equidistant sub-rectangular cross-section axes of the Lake barbs along one side formed by finely sculpted Mälaren area (Lindgren & Nordqvist 1997:59-61 shallow oblique notches, and ending in a tip long- & fig. 2:4). This supports a Neolithic date. Ac- er than the distance between two of the notch- cording to Malmer’s definition (1962:560), it is a es. cross-edged, thin-bladed igneous rock axe, a type Given the complete dominance of Hemmor- with only three specimens known to Malmer Gullrum pottery at Stora Förvar G6 and the lack from STR graves. The width and thickness of of parallels to the Rojrhage harpoon point in the the Rojrhage axe are both within the intervals Gotlandic MN B cemetery horizon, this type characteristic for the type. However, it is much seems to be contemporary with the GRK pot- longer in absolute and relative terms than all of tery from Rojrhage and earlier than the ceme- Malmer’s thin-bladed axes, both flint and igneous tery horizon. rock ones. Its relation to the grave finds of the Battle Axe Culture is thus uncertain. Igneous rock 7. Ground igneous rock axes axes were a late (late MN B and early LN) feature of the northern limits of the Battle Axe Cul- Two ground igneous rock axes have been identi- ture’s area in Sweden, the Lake Mälaren area and fied among the finds: one of them nearly com- Östergötland, where they took the place in the plete and collected as a stray find, and the other assemblage occupied by flint axes in more south- represented only by a flake found in an Iron Age erly areas (Malmer 1962:431). grave. Little can be said about the flake except As the axe differs markedly from the modes that the axe had a rounded cross-section. of common axe types, it is best interpreted as a The nearly complete axe (fig. 8) is made from local greenstone variation on the thin-blade the- greenstone. It is missing the point of its tapered me (personal communication, Mats P. Malmer butt. Otherwise it is in good condition. Except 2000). The Rojrhage axe would fit well in the esta- for the break, its surface is polished all over. Its blished view of MN Gotland as located outside cross-section is roundedly rectangular. The edge the area of the canonical Battle Axe Culture, but is distinctly off-centre but not hollow. The ex- in close contact with it, as evidenced e.g. by the tant maximum dimensions are 206.5 x 42.5 x 22.0 Scanian and Jutish battle-axes found in graves at mm. The thickness 2/3 of the probable original Västerbjers (Stenberger et al. 1943). length (c. 220 mm) from the edge is 19.4 mm. A fine specimen like that from Rojrhage would Michael Petrén’s (1992) eminent work with probably originate from a votive deposit or a grave. rectangular cross-section axes from Gotland has The edge, however, exhibits use-damage. This tal- 16 lies well with Österholm’s observation (1989:169 & fig. 84) that Gotlandic greenstone axes of all types from the Late Meso- lithic through the Early Bronze Age tend to an edge angle of c. 80º, while sharper-edged axes like the Rojrhage specimen (edge c. 30º) are often damaged. Indeed, it seems obvious that an axe like this would break when used, and it was thus very likely made primarily for display or burial. The Rojrhage axe lacks a detailed find context and belongs to a typologically unspecific artefact class. Therefore it cannot be dated any closer than Pet- rén’s MN-LN.

8. Flint In December 1999, 282 flint fragments were known from the site. This number will most likely be augmented in the future as further Iron Age cremations are analysed by osteologists. Neolithic artefacts have been mixed secondarily into the Iron Age contexts. Among the materials at hand are 21 fragments from a total of 18 Iron Age cremations analysed by osteologists, all of them excavated 1960-1961. No flint ap- peared when inhumati- Fig. 8. Cross-edged, thin-bladed igneous rock axe from ons were analysed. To this number should be ad- Rojrhage 1:1. Stray find 1963-1967. SHM 32623. ded 19 fragments collected already during field- Ventral broad side shown left. work from 10 graves, only one of which was an inhumation. No flint fragments were registered

17 from the graves excavated 1963-1967. This is paration. Finally, flakes with a V-shaped cross- probably due to the separation at these excava- section were produced by connecting inside the tions of Iron Age finds and finds from, or con- vertical scar of a previous flake. jecturally belonging to, the underlying Neolithic The strong presence of bipolar technique is deposits. The excavator alludes to such a practi- characteristic of Gotlandic flint knapping ce in the field notes quoted above. through millennia. Local flint nodules are simply In all, 57 cremation graves have been excava- generally too small to permit production of plat- ted at the site. The total number of flint frag- form cores (Thorsberg 1997:50-51). Many of the ments from these once they have all been sub- morphologically unspecific fragments from Rojr- ject to osteological analysis may be predicted at hage display sharp-edged high speed fractures. 57 x (21/18) ≈ 67. Of these, 40 have already been This indicates the use of an anvil, probably during collected, to which must be added an unknown bipolar work judging from the finds. number for 43 graves excavated 1963-1967. All No attempts were made to limit the width of this assumes a roughly homogeneous spread of the flakes, as shown e.g. by the practice of con- flint across the area prior to the graves’ construc- necting inside flake scars. This reflects an ideal tion. opposite to that of blade technique. Wide, thin Of the 282 known flint fragments, one was and straight flakes were the intended products. discarded at excavations in 1931 and 13 are kept in the Gotlands Fornsal county museum, Visby. 8.2 Intended products The remaining 268 fragments (95% of the known Nothing at the site indicates production of com- ones and >86% of the predicted total number) plex flint tools like axes or daggers, nor of simp- are kept at the SHM. They were determined as le tools like scrapers. The flakes made were cut- to type by Karl Thorsberg on 28 May 1999 (ta- ting tools that could be used on their own or ble 8a). fitted as edges onto composite tools. The best products, wide flakes with cutting 8.1 Knapping technique edges, seem to have been selected and removed, Despite the multiple Neolithic components dis- as probably also many of the bipolar flakes. How- played by the pottery from the site, the flint finds ever, no flake re-fitting to reconstruct cores has permit the reconstruction of a single coherent been attempted. The finds seem mainly to con- reduction process. A flint-saving method has sist of knapping debris. The actual sites of tool been practiced: work has begun with multifacial use may have been anywhere. cores from which flakes have been knapped. The- se flakes have been made as large and wide as 8.3 Raw material possible without platform preparation, that is, The raw material is hard to determine because unnecessary waste has been avoided. When the Gotlandic flint is so variable, but there is no cer- multifacial cores have become too small, reduc- tain evidence for the use of any imported flint. tion has continued on an anvil, since all collected The flint from grave 5/61 might possibly be cores are bipolar ones which have been used to South Scandinavian, but the greasy surface of the limit. The bipolar flakes are consistently smal- the fragments may equally well be due to the flint ler than the multifacial core flakes. The only ex- having been knapped during the Iron Age leading ception from this chaine opératoire is a bipolar to comparatively less time for patination. core made from a multifacial core flake. This core Beside the small nodules of local beach flint, is also the only intact one. the knappers have used at least one ground flint The multifacial core flakes demonstrate at least tool as raw material. It has not, however, been three ways of knapping flakes from a multifacial worked differently than the common nodules. core. Flakes with pseudo-facets on the impact surface were produced by connecting at the bot- 8.4 Find contexts tom of horizontal flake scars. Flakes with a level Most of the flint finds derive from the Neolithic platform were produced by connecting far from deposits. There are no significant differences in the edge of the core without initial platform pre- the flint from different parts of them. However,

18 Table 8a. Flint fragments. At the Neolithic site of Mölner in Väte parish, Gotland, with a main TRB component, there T.ype No is knapping debris from complex flint tools and S2plinter from a multifacial core a large component of South Scandinavian flint (Thorsberg 1997:54-55). If Mölner can be taken F3lake from a multifacial core 7 as representative of Gotlandic EN sites, then an B6ipolar core or fragment thereof 2 EN date can be excluded for the Rojrhage flint. Early Bronze Age flint would most likely in- B9ipolar flake or fragment thereof 5 clude debris from surface pressure flaking and F1lake from ground tool platform preparation. The same may probably be said for the LN, although the Gotlandic re- F3lake knapped with metal hammer cord is weak on this point. These dates may thus C3hunk also, provisionally, be excluded. On the basis of these deliberations, one may U1nspecific 10 thus most probably date the Rojrhage flint finds to the MN. Given the coherent reduction process perceived in the flint finds, it would seem S8um 26 that they represent only one of the site’s chronological components, viz the late MN A. To there are differences between the flint from the explain the apparent lack of earlier and later lithics Neolithic deposits and the Iron Age graves. The in the material, a comparison with the pottery flint from the Iron Age graves is more patinated, may be instructive. Of the pottery fragments that but most fragments are not visibly burnt with a could be ascribed to one of the three decoration cracked surface. element groups, only 12% belonged to TRB and Fragments with a greatest measurement <10 STR. 12% of the determinable flint fragments mm are very uncommon but do occur among amounts to only 20, which is of course not artefacts found during osteological analysis of the enough material to reconstruct a single lithic in- bones from the Iron Age graves. This indicates dustry, let alone two. This argument presuppo- that although the deposits were, judging from the ses a constant ratio through time between the minimum size of the collected pottery fragments, amount of pottery and flint discarded at the site. screened, the smallest flint fragments were not collected during fieldwork. 8.6 Iron Age flint As noted above, the flint found in the graves 8.5 Dating exhibits deviant patination. Of the fragments On typological grounds, the flint finds may not from Neolithic contexts, only about 1% are be dated closely except for the observation that burnt, while the ratio among the finds from the the reduction process at Rojrhage is the same Iron Age graves is 12.5%. This is probably due one as identified at a number of Gotlandic GRK to the cremation pyres and indicates that at least sites (Thorsberg 1997:53). However, a few argu- some of the cremations were performed on-site. ments ex silentio may be suggested, fully acknow- More remarkably, three of the fragments ledging their weaknesses. found in the graves have a wavy surface indica- Mesolithic lithics would most likely contain a ting that they were knapped with a metal ham- much larger component of igneous rock. One mer. None of the fragments found in the Neo- would also expect more blade technique with lithic deposits do. This hints at an Iron Age flint oblong flakes and prepared knapping platforms. industry, possibly linked to mortuary ritual.

19 9. Bones and radiometric dual in the Iron Age context. The bones of hum- analyses ans and dogs, on the other hand, were common in the graves and very rare in the substratum. 9.1 Sampling The horses, bears and lynxes of the Iron Age Bone samples for dating were chosen on the fol- graves are entirely unknown among the Neolithic lowing grounds. The selection was limited to spe- finds. Ovicaprids are common in both contexts. cies-determined bones from seemingly unadul- terated Neolithic contexts. One sample per spe- 9.2 Hunting and fishing cies was selected, and two extra human samples Three seal species have been identified among were added, bringing the total up to ten samples the bones from Neolithic contexts at Rojrhage: (table 9a, no dating was achieved for cod verteb- grey seal (Halichoerus grypus, Sw. gråsäl), ringed seal rae). It was hoped that the human bones might (Pusa hispida, Sw. vikaresäl) and harp seal (Phoca provide the first occasion on Gotland to compare groenlandica, Sw. grönlandssäl), see table 9b. The- EN human d13C values with seal, fish and do- re are remains of at least two very young ringed mestic animal values from the same context. Un- seals in the Rojrhage material, probably indica- fortunately, none of the two human cranial bo- ting breeding not far away and severe ice condi- nes retained sufficient collagen for radiometric tions. The abundance of grey seal at Rojrhage is analysis, although a human tooth gave a date in unusual, as harp seal predominates among the the 1st millennium cal AD (Ua-16504). Instead seal bones from most other Neolithic sites on an extra sample each of cattle and ovicaprid were Gotland. Usually, grey seal predominates only at selected. These were preferred over the wild spe- Early and Middle Mesolithic habitation sites, and cies and pig because we considered them more is very rare at Late Mesolithic and Early-Middle likely to date from the EN and thus be relevant Neolithic sites. Nevertheless, two grey seal samp- to the study of the neolithisation of Gotland. les from Rojrhage have been dated to the Neo- MN seal-hunting, fishing and pig-herding is al- lithic, proving that the grey seal bones are actual- ready well attested on Gotland. No EN dates, ly approximately contemporaneous with the ring- however, resulted from the analyses. One of the ed seal and Neolithic domestic animal bones (ta- cattle bones produced a date in the MN, which is ble 9a). rather uncommon. The apparent difference in radiocarbon age, A comparison of the species frequencies for 251 (140-345) years, between the marine influ- the Neolithic deposits and the Iron Age graves enced seal dates and the domestic animal dates, at the site reveals interesting differences. Bones is probably mainly due to the reservoir effect, of cattle, pig, seals, birds and fish are common not an actual difference in age. The fairly low among the Neolithic finds but very rare among d13C values of the seal bones, which is not un- the Iron Age finds. They may actually all be resi- usual in the Baltic, especially in harp seal bones,

Table 9a. Radiocarbon dates. Go, Grötlingbo psh, Rojrhage 1:1, Raä 54, SHM 32623. Samp- Age Std Corrected date, Corrected Lab no. Context d13C Date cal 1 s le no. BP dev RE = -250 years BP period

U1a-16500 P8ig, dens, E 3-522. 4043 7C3-300-2920 cal B MN A

U3a-16501 O5vicaprid, dens, G 1-022. 1523 6D7-10-890 cal A I.A. VII-VII I

U4a-16502 G7rey seal, b. tymp., G 1-023. 3081 1C1 2C460-2040 cal B 2I040-1740 cal B LN I-I

U6a-16503 H3arp seal, b. tymp., F 4-019. 4065 1C0 3C630-3570 cal B 3A320-2900 cal B MN

U8a-16504 H0uman, dens, F 1-022. 1073 1D0 1-40-430 cal A I.A. IV-V I

U0a-16505 10O2vicaprid, tibia, H -023. 1032 8D6-40-810 cal A I.A. VI I

U1a-17545 12C0attle, dens, F1- -523. 4532 6C3-020-2880 cal B MN A

U3a-17546 15R0inged seal, dens, G -017. 4067 6C3C520-3360 cal B 3A310-2920 cal B MN

U4a-17547 11G0rey seal, dens, H -517. 4557 7C3C500-3100 cal B 3A090-2870 cal B MN

20 Table 9b. Skeletal element abundance (number of may perhaps indicate that the calculated reser- identified fragments/specimens, NISp.) in the seal (grey voir effect should be somewhat lesser and hence seal, ringed seal and harp seal) bone material from that there may in fact be a small difference in age Rojrhage. between the seal bones and the domestic animal No. of bones, the date of which is 3349-2707 cal BC (all Grey Ringed Harp Skeletal element frags, seal seal seal calibrations according to Stuiver & Pearson NISp. 1993:1-23). However, compared to the calcula- T7emporale 1111 ted reservoir effect at Ajvide D upper (c. 305 years O0ccipitale for seal bones, c. 215 years for human bones; cf. table 9c and Lindqvist & Possnert 1997a: 55-57, C9ranium-fr. 1997b:73-74) and at ”Grottan” (280 years, cf. ta- N0asale ble 9d), the age difference appears reasonable M0axilla when interpreted as due to the reservoir effect. The latest grey seal date from Rojrhage is in- M0andibula 1+2?1 teresting, provided that it is correct (its quite ter- D831entes restrial d13C value is obviously incorrect), since it V0ertebra 1 may indicate that the grey seal multiplied again C6osta when the harp seal declined in the Baltic Basin toward the end of the Neolithic. S1ternebra Another exception to the predominance of S4capula the harp seal is found in the Neolithic levels (G.7- H3+umerus 5) in the cave of Stora Förvar on the island of Stora Karlsö in Eksta parish, where the harp seal R1adius is only the third in abundance after ringed seal U1lna and grey seal. The Neolithic ringed seals caught C2arpi (7 sp.) at Stora Karlsö were about 7 months old and the harp seals about 9-10 months old, indicating seal M6etacarpus, M.c. I-V hunting during the late summer-autumn-early Phalanx anterior, Ph. 5 winter (Lindqvist & Possnert 1997a). Although 1, I-V some grey seal samples from the Neolithic levels P1h. 2, I-V in Stora Förvar have been dated to the Mesolith- P0h. 3, I-V ic, indicating residual material from lower levels C8oxae (G.11-8), there are also grey seal samples that have been dated to the Neolithic (table 9e). B1aculum The reason for the shift from grey seal to harp F0emur seal predominance, indicated by most late Meso- T2ibia lithic and Neolithic habitation site materials, may be a combination of two factors. On one hand, F4ibula competition as pelagic species in the same eco- T0arsi (7 sp.) logical niche; on the other, a harsher winter cli- M2etatarsus, M.t. I-V mate, that is a greater abundance of permanent ice in the Baltic Basin, caused by the Sub-boreal Phalanx posterior, 1 Ph. 1, I-V, climatic shift. This probably meant very severe ice conditions in the Arctic, causing a southward P3h. 2, I-V relocation of the harp seal and the establishment P0h. 3, I-V of a population in the Baltic Basin enduring as a M6.p. relict for at least 1500 years. It should be noted that porpoise – although P2h. quite common in the final Mesolithic materials, and sometimes occurring also in the Early Neo- lithic ones – does not occur at these sites. Due 21 also to severe ice conditions in the Baltic Basin, Two species of fish have been identified in the grey seal may have been forced to the south- the Rojrhage material: cod (Gadus morhua) and west and even past the straits of Öresund and pike (Esox lucius), see table 9f. Of these the cod the Bälts. It is an interesting thought that the lar- predominates, as it does at most final Mesolithic ge grey seal canines in grave number 2 at the Aj- and Neolithic habitation sites on Gotland. The vide D upper Pitted Ware (Middle Neolithic) ce- cods were caught during the late summer or au- metery, Eksta parish, may have been traded from tumn at an age of at least 2-4 years and a length the south-west. However, the finds of grey seal of 45-83 cm (table 9g). However, the fish bones at Rojrhage and in the Neolithic levels of the are few, and herring is lacking, which is possibly cave of Stora Förvar seem to indicate that the due both to taphonomy and excavation techni- animal could still survive, perhaps seasonally, per- que. Where fine-mesh sieving has been conduc- haps to some extent breeding, around south- ted, herring is sometimes the predominant spe- western Gotland, whereas the grey seal was ab- cies, at least at sites from the middle of the Middle sent further to the north. Perhaps this is also an Neolithic (Ekman 1974; Lindqvist 1988 & 1997c). indication that the grey seal, which breeds during At ”Grottan”, Överstekvarn II, many fish bo- the autumn and early winter in colonies on is- nes, mainly of cod, were recovered from the Early lands and sheltered bays in the North Atlantic Neolithic levels. (as they probably did on the island of Stora Bird bones are likewise rare. They have not as Karlsö and on the shore terrace below the escarp- yet been identified as to species. Petra Molnar ment at Visborgs Kungsladugård, Visby parish, (1998) has identified goosander (Mergus mergan- during the early Mesolithic; see Lindqvist & Poss- ser, Sw. storskrake) among the bones from the nert 1997a; Lindqvist 1997c), finally began to Iron Age graves. adapt to ice-breeding during the late winter and Judging from the relative abundance based on spring in the Baltic Basin. Later, during the Bron- the number of bone fragments, hunting and fish- ze Age and early Iron Age, the grey seal bred ing do not seem to have been of greater impor- further north in the Åland archipelago. tance than animal husbandry at Rojrhage.

Table 9c. Neolithic AMS radiocarbon-dated animal bone and tooth samples from Jakobs/Ajvide, Eksta parish. (The apparent difference between the seal and pig dates, c. 250 (140-355) years, may well be due to the reservoir effect, and hence they are probably contemporaneous in true age.)

Skeletal 14C years d13C ‰ Lab. no. Csontext Specie element BP vs. PDB

Jlakobs/Ajvide B Harp sea Phoca groenlandica T0emporale, dx. 58020 ±6 -118. Ua-325

Aljvide D Harp sea Phoca groenlandica T5emporale 46465 ±6 -116. Ua-354

Aljvide D Harp sea Phoca groenlandica M5andibula, dx. 45425 ±6 -216. Ua-354

Glrave 29 Harp sea Phoca groenlandica T0emporale, dx. 43505 ±7 -117. Ua-1043

Glrave 19 Harp sea Phoca groenlandica T?emporale, dx. 43730 ±75 -616. Ua-1727

Glrave 41 Sea Phocidae D5ens C. 40460 ±7 -816. Ua-1727

Glrave 2 Grey sea Halichoerus grypus D5ens C. 41450 ±6 -815. Ua-1042

Ggrave 2 Hedgeho Erinaceus europaeus M5andibula, sin. 44150 ±6 -921. Ua-1042

Ggrave 41 Hedgeho Erinaceus europaeus M5andibula, sin. 40210 ±6 -021. Ua-1728

Ggrave 41 Pi Sus scrofa/Castor D0ens C. 40285 ±8 -921. Ua-1727

Ggrave 29 Pi Sus scrofa D0ens C. 41195 ±7 -022. Ua-1043

22 Table 9d. Neolithic AMS Skeletal 14C d13C ‰ radiocarbon-dated Lsevel Specie Lab. no. element years BP vs. PDB animal bone and tooth samples from ”Grot- 9l-6 Harp sea Phoca groenlandica T5emporale, sin. 48980 ±6 -317. Ua-354 tan” at Överstekvarn II, Lummelunda parish. 9e-6 Cattl Bos taurus M0etacarpus 45700 ±8 -820. Ua-265 (The apparent difference between the seal and cattle dates – 280 years – may to a large extent be due to the reservoir effect, and hence they are probably roughly contemporaneous in true age.)

Table 9e. Neolithic AMS Skeletal 14C d13C ‰ Lsevel Specie Lab. no. radiocarbon-dated element years BP vs. PDB animal bone and tooth samples from the cave Gl.5 Grey sea Halichoerus grypus M5andibula, dx. 46410 ±7 -514. Ua-294 of Stora Förvar on the Gl.4 Grey sea Halichoerus grypus T0emporale, sin. 45240 ±7 -915. Ua-1717 island of Stora Karlsö, Eksta parish. (There are Gl.7 Ringed sea Phoca hispida M0andibula, dx. 44460 ±8 -916. Ua-293 a couple of apparently Gl.6 Ringed sea Phoca hispida M5andibula, dx. 40115 ±7 -818. Ua-1717 earlier marine animal dates from G.7 which Gl.6 Harp sea Phoca groenlandica M5andibula, dx. 42575 ±9 -216. Ua-294 may be contempora- Gl.5 Harp sea Phoca groenlandica M5andibula, sin. 40120 ±6 -717. Ua-1718 neous with the Early Neolithic domestic Gl.4 Harp sea Phoca groenlandica M0andibula, sin. 40165 ±7 -616. Ua-1717 animal dates. If so, the Gl.7 Harp sea Phoca groenlandica M5andibula, sin. 50205 ±8 -418. Ua-1717 calculated reservoir effect was c. 210 (135- Ge.7 Porpois Phocoena phocoena V0ertebra 59225 ±6 -615. Ua-324 290) years during the Early Neolithic.) Gp.7 Shee Ovis aries U5lna, sin. 59070 ±7 -221. Ua-495 Ge.7 Cattl Bos taurus R5adius, sin. 43935 ±7 -821. Ua-324

Gg.7 Pi Sus scrofa T0ibia, sin. 48800 ±6 -720. Ua-324

Gg.10 Pi Sus scrofa C5osta II, dx. 47325 ±8 -021. Ua-495

Gg.8 Pi Sus scrofa N5asale, sin. 45095 ±8 -420. Ua-293

No. of Table 9f (left). Verte- Age L Sum- L. sum. / Aut- L. aut. Skeletal element frags, Number of bone bra (yrs) (cm) mer E. aut. umn NISp. fragments/specimens (NISp.) of 13661 V1omer fish (Pisces), 243811? P3remaxillare mainly of cod (Gadus morhua) – 3-35411 M4axillare one bone from 44-251 O1perculum Rojrhage is of pike (Esox lucius). 55- 1 1 B3ranchiostegale 64--1? C1leithrum (of pike) Table 9g (right). Approximate 74- 1 1 P1ostcleithrum age, length and 83- 1 1 catching season P3arasphenoid 92- 1 1 for cod (Gadus V1ertebra atlas/cervicale 1 morhua) based 1400 711? on vertebrae V8ertebra precaudale 1341 51 from Rojrhage. V0ertebra caudale 1 1-2 2841 ?

V1e. fr. 1 1253 611?

C1osta fr. 13-4 1

P2isces indet. 12-5

S0um 6 S22um 18?

23 9.3 Animal husbandry in levels 9-6 at ”Grottan”, Överstekvarn II in During the Mesolithic all larger terrestrial game Lummelunda parish and in levels G.8-7 in the species were absent from Gotland due to zoogeo- cave of Stora Förvar on the Island of Stora graphical isolation. For example, aurochs, wild Karlsö, and probably at Stora Domerarve II, boar and wild ovicaprids were absent, and thus Hablingbo parish. While the date of the domes- had to be introduced by humans. This was ac- tic animal bones from G.7 in the cave of Stora complished through voyages by boat across the Förvar is 4030–3581 cal BC and that of the catt- Littorina Sea, possibly also involving cargoes of le bone from ”Grottan” at Överstekvarn II is leaf and hay fodder and cereals. Pollen of graz- 3647-3149 cal BC, that is, clearly Early Neolithic ing indicators such as narrow-leafed plantain dates; the date of the cattle bone from Stora (Plantago lanceolata) and later also cereal pollen have Mafrids is 2923-2790 cal BC and that of the Rojr- been identified in lakebed cores of the early Sub- hage domestic animal bones is 3349-2707 cal BC, boreal period from main Gotland and Stora clearly Middle Neolithic dates, although earlier Karlsö. than most dates from Pitted Ware Culture sites All the important domestic species are repre- (table 9h). sented in the Rojrhage material: cattle (Bos tau- Table 9h. Early dates for domestic animal bones from rus), pig (Sus scrofa) and sheep (Ovis aries) Gotland. or goat (Capra hircus). Of these, cattle Radiocarbon Age cal Age cal Ssite Specie and ovicaprids are markedly more age BP BC 1 s BC 2 s abundant than pigs, which seems to be Sptora Förvar G.7 S5hee 59070 ±7 38964-377 4030-369 typical for the Early Neolithic Funnel Beaker culture subsistence economy, Setora Förvar G.7 C5attl 47935 ±7 31788-364 3941-354 whereas pig dominates in the Middle Sgtora Förvar G.7 P0i 41800 ±6 38647-358 3699-337 Neolithic Pitted Ware culture subsisten- "eGrottan" 8-6 C0attl 43700 ±8 39626-336 3647-314 ce. This is an Early Neolithic pattern already registered at other habitation Setora Mafrids C5attl 45295 ±3 20916-288 2923-279 sites on Gotland. At least this is the case Rgojrhage P0i 42435 ±7 30301-292 3349-290

Reojrhage C0attl 45325 ±7 37022-288 3094-270 Table 9i. Periodical subdivision of the Sltora Domerarve II C0harcoa 40750 ±14 37663-336 3891-309 Neolithic on Gotland, with faunal changes seen in the bones finds from habitation Sltora Domerarve II C5harcoa 47425 ±13 35343-289 3502-269 sites. Impressionistic abundance scale 1-5.

Levels Por- Fish Period at Stora Psottery style Seal Seheep Cgattl Pei Hgors Do poise (cod) Förvar

Late -545----- Mesolithic

Er.N. Funnel Beake 213 3 41-- ?

G.6, G.7, Funnel Beaker / Early M.N. 313 2 33-- ? (G8?) Sät.-Fag. II Middle G.6, G.7, Sät.-Fag. III / 414 --4-2 ? M.N. (G8?) Hemmor-Gullrum Sät.-Fag. IVa / L6ate M.N. G.5, G. 414 1 13-2 Visby Late M.N. / Sät.-Fag. IVb / V G.4, G.5 414 1 1312 early L.N. / Ire-Västerbjers

L4.N. Ge. G1-132121allery grav

B4ronze Age Ge. Bronze Ag ---4 2131

24 The cattle dates from Stora Mafrids and Rojr- Several of the measurable bones and teeth of hage are unusually late ones for the Early Neo- cattle and pig from Rojrhage correspond in size lithic-Early Middle Neolithic period, since cattle to finds from other Early and Middle Neolithic bones are entirely lacking at most middle Middle habitation sites on Gotland and elsewhere in Neolithic/Pitted Ware ”Hemmor-Gullrum/Sä- Scandinavia. Hence, it is fairly unlikely that any ter III” cultural layers, e.g. at Ajvide D upper large portion of the unburnt bones and teeth of (Lindqvist & Possnert 1997a; Storå 1997 & 2001). these species found at Rojrhage should derive I, Christian Lindqvist, divide the Middle Neo- from the late Iron Age cemetery on top of the lithic of Gotland into at least three parts: early Neolithic deposits. The cattle and pigs of the MN (=late Funnel Beaker, with cattle, sheep and Iron Age were generally far smaller than their pig, as during the EN), middle MN (=Hemmor- Neolithic forebears. Gullrum/Säter III, with pig and dog, but cattle and sheep are lacking!) and late MN (=Visby- 9.3.1 Cattle (Bos taurus) Ire-Västerbjers/Säter IV-V or IVa-IVb, with pig, While the unfortunately extinct ”Gotland cow” dog, a few cattle and sheep and rare horse; see or ”Småland cow” breed that existed on Got- Lindqvist 1997c, Lindqvist & Possnert 1997a). land until the late 19th century was small (height (Table 9i.) at withers: 112 cm, weight: 350 kg; Hallander The fact that bones of domestic animals are 1989:138-149, 140, 144, fig. 69-70, 192/193 pl. more abundant at Rojrhage than those of seal 11), Early Neolithic cattle was rather large jud- and fish is of great interest. At ”Grottan”, Övers- ging from bone measurements. According to the tekvarn II the domestic animal bones are more distal width of a tibia and metacarpus from Rojr- abundant than the seal bones, although many fish hage, and the length of a calcaneus and distal width bones were found at the site. It is probable that of a metacarpus from ”Grottan” at Överstekvarn the subsistence economy at the inland habita- II, these cattle from Gotland seem to have occu- tion sites, such as Ardags in Ekeby parish, Gräne pied the same size interval as the cattle from the and Mölner in Väte parish, was entirely based on Alvastra pile-dwelling in Östergötland on the animal husbandry. Unfortunately, no Early Neo- mainland, as well as those from Troldebjerg and lithic radiocarbon dates exist from these locali- Bundsø in Denmark, all of which had a height at ties. the withers of 113-139 cm. They were, however, clearly smaller than the aurochs, the female of Table 9j. Measur ements of teeth of ca ttle ( Bos taurus) which according to reference skeletons reached from Gotland, with compar ative ma terial. *After During 1986:94 ta b. 34. ”St. Förvar*” inc ludes all levels, a height at the withers of 146-147 cm (according including IA/Medieval! to Johannes Lepiksaar c. 150 cm), while the male reached nearly 200 cm. This can be Length Width Shite Toot Period compared to the skeleton of a ”dwarf (mm) (mm) ox” (Sernander 1898:334-335) found R?ojrhage F.2 M2.2 247. 2N1. EN, M in a bog at Dammen, Fröjel parish, which has been interpreted as an abo- R.ojrhage F.1-2 M8. sup 266. 2N3. EN, M riginal wild dwarf ox subspecies on R.ojrhage F.1-2 P2. sup 189. 2N0. EN, M Gotland (Österholm 1989). However, its bone measurements correspond to "Grottan", M2.1 205.0; 30. 2N0.2; 18. E Överstekvarn II a height at the withers of 90-101 cm, that is, even smaller than the Gotland "2Grottan" M7. 206. 2N2.4; 18. E cow of recent centuries. Due to bin- ".Grottan" M7.3 inf 322. 1N2. E ding, stalling and insufficient fodder during the long winters, Swedish Me- S3tora Förvar* M0. 207.0-29. 1N5.0-20. dieval cows sometimes grew, as for ex- EN-early ample at Lödöse in western Sweden, A3lvastra pile dwelling* M0. 2)8.0-35. (20.0-22.3 MN no larger than 100 cm in withers A.urochs, female* M5.3 inf 405. 19. height (Lepiksaar 1977:83, 90). Hen- ce, it appears – when compared to 25 Table 9k. Measurements of Approx. height Skeletal Width Length bones of cattle (Bos taurus) Site at withers Period element (mm) (mm) from Gotland, with compari- (cm) sons (and approximate height- "sGrottan", Överstekv. II V2e. Atla >5124. 8N0. E at-withers calculations). Dsammen V9e. Atla 1?0 Medieval *After During 1986:101 tab. 43. Sstora Förvar G.7 H3umeru 6-5. EN-MN I ”St. Förvar*” includes all Sstora Förvar G.7 R8adiu 8-0. EN-MN I levels, including IA/Medieval! Dsammen R5adiu 213 1?0 Medieval #Height at withers after M.c. Drammen F0emu 208 9?Medieval and M.t.: Matolcsi 1970/ 71:113; von den Driesch & R.ojrhage G.1 T5ibia, dist 6-7. EN, M N Boessneck 1974:336 tab. 4; S.tora Förvar* T0ibia, dist 5-N5.0-59.

Sjøvold Ms. 1986; During A.lvastra pålb.* T0ibia, dist 5N5.8-68. EN-early M 1986:105 tab. 47. B.undsø* T0ibia, dist 5N7.0-70. M

Reojrhage C3arpi ulnar ?437. ?N24. EN, M

Keudruküla C5arpi ulnar 465. 4y4. Narva potter

R?ojrhage G.2 C9entrotarsale 585. 5N7. EN, M

A.lvastra pile dw.* C8.t 4N7.8-55. EN-early M

A.urochs, female* C0.t 65.

Sstora Förvar G.7 T8alu 470. 6I2. EN-MN

Sstora Förvar* T0alu 302.0-42. 53.0-65.

Aslvastra pile dw.* T9alu 451.3-50. 6N1.8-76. EN-early M

Bsundsø* T0alu 6N0.0-74. M

"?Grottan", Överstekv. II C5alcaneus/m 443. 1N30. E

Dsammen C0alcaneu 1?1 Medieval

Sstora Förvar* C0alcaneu 301.0-41. 1N11.0-145.

Aslvastra pile dw.* C)alcaneu (727.7-38.5 1N26.5-134. EN-early M

Asurochs, female* C0alcaneu 501. 162.

RVojrhage G.1 M3etacarpus III+I 5-8. EN, M N

".Grottan", Överstekv. II M0.c 5-5. E N

D.ammen M5.c 47116 1?0 Medieval

S.tora Förvar* M0.c 500.0-65. 1#68. 8)4.7-106.3 N (EN-IA

A.lvastra pile dw.* M1.c 504.2-67. 1#96.5-219. 1N14.6-138.6 EN-early M

T.roldebjerg* M0.c 503.0-73. 1#92.0-214. 1N13.3-135.5 Early. M

B.undsø* M0.c 506.8-71. 1#96.0-204. 1N14.1-129.1 M

A.urochs, female* M5.c 608. 2#38. 143.5-145.3

SVtora Förvar* M)etatarsus III+I 4)3.0 (juv. 1#71.0 (juv. 7N7.0-96.1 (juv.)

A.lvastra pile dw.* M0.t 407.8-67. 2#48.0-250. 1N29.8-140.5 EN-early M

B.undsø* M0.t 501.0-64. 2#24.0-237. 1N13.4-133.2 M

A.urochs, female* M5.t 605. 2#73. 145.5-147.0

"1Grottan", Överstekv. II P0halanx 207. 5N9. E

D1ammen P0h. 4?5. Medieval

"2Grottan" P0halanx 205. 3N9. E

"2Grottan" P0h. 255. 3N8. E

"2Grottan" P1h. 215. 3N8. E

D2ammen P0h. 2?3. Medieval

"3Grottan" P0halanx 552. 6N9. E

"3Grottan" P2h. 500. 6N7. E

"3Grottan" P0h. 560. 6N6. E

26 the large Early Neolithic cattle – probable that pig from Rojrhage indicates a height at the with- the ”dwarf ox” from Dammen was in fact a star- ers of c. 81 cm, which compares well with the ved Medieval cow. maximum of 83-84 cm in the material from the The bone measurements, e.g. the distal Metacar- cave of Stora Förvar, level G.7, and 81 cm in the pus width (45 mm) of the Dammen cow are com- Stora Mafrids material, although the pigs at Gull- parable to the measurements of cattle bones from rum seem to have reached 89-90 cm. The earli- the period after AD 1200. Vretemark has shown est pigs on Gotland were large: they were proba- that this measurement decreased from 50-53 mm bly recruited from the wild boar population on to 47-50 mm about AD 1200 in the town of Skara, the Swedish mainland and apparently not from Västergötland; and from the interval 50-53 mm in the smaller domestic pigs of Central Europe or the Viking Period (10th century) Birka material to Denmark. Neither were they recruited from the 47-49 mm in the High Medieval material from Stock- East Baltic area, where the wild boars were lar- holm (Helgeandsholmen, after c. AD 1270). This is ger than the Swedish ones (Jonsson 1986). Scan- interpreted as a result of a decreasing grazing and dinavian wild boars vary according to reference meadow area and fodder per animal, when cereal skeletons between 78 and 100 cm, living animals agriculture intensified (Vretemark 1997). vary between 85 and 102 cm, maximum 110 cm The difference in size between the small Dam- (Brehm 1931:477, Siivonen 1968/76:126, Curry- men cow, with a distal Metacarpus width of 45 mm, Lindahl 1975/82:263). and the larger ”Grottan” and Rojrhage specimens, A set of complete rear long bones of ”wild 53 and 57 mm, is probably due both to the more boar” from the bog of Libbenarve myr, Havd- favourable climate and less restricted grazing during hem parish (Sernander 1898:337), indicate a the Stone Age and to the fact that the earlier cattle height at the withers of 84-93 cm, that is, the had been domesticated for a much shorter time. It is furthermore quite Table 9l. Measurements of molar teeth of pig (Sus scrofa) from Gotland, possible that Neolithic cattle, as well with comparative material after During 1986:128 tab. 60. (Wild boar: 41-50 mm; domestic pig: 23-40 mm according to Clason 1967:63). as pigs and sheep, was herded loose throughout most of the year (table 9j Length Width Shite Toot Ndote Perio & 9k). Unfortunately, neither the Dam- (mm) (mm) men nor the Libbenarve bones (see R.ojrhage E.3 M7.3 sup.? inf >136. 1?7. DNom EN, M below) have been possible to locate for Alvastra EN - M5.3 sup. 356.1-42. 1d9.0-22. Dom/wil radiocarbon dating. pile-dw.* early MN T.roldebjerg* M7.3 sup 3m2.0-36. DNo Early M 9.3.2 Pig (Sus scrofa) All anatomic parts of the domestic B.undsø* M0.3 sup 209.0-38. 1m7.5-21. DNo M animals from Rojrhage, including the W.ild boar* M7.3 sup 304.5-44. 2d1.5-24. 'Dom'/wil pigs, are represented, which indicates that butchering has occurred at Table 9m. Measurements of tibiae of pig (Sus scrofa) from Gotland, with comparative material *after During 1986:130 tab. 62. ”St. Förvar*” the site and that transportation of includes all levels, including IA/Medieval! (Distal width of wild boar: 37- certain cuts to or from the site is un- 43 mm; domestic pig: 24-35 mm according to Clason 1967:63). likely. The pigs have been treated like Skeletal Width Site Ndote Perio the other domestic animals. Relativ- element (mm) ely many domestic animals have been slaughtered at an early age. Crush- S.tora Förvar G.7 T8ibia, dist 3m2. DNo E ing of the long bones to gain access S.tora Förvar G.4 T5ibia, dist 2m9. DAo LN-B to the nutritious marrow, and further S.tora Förvar* T0ibia, dist 3m4.0-36. DNo decay through trampling and other G.ullrum* T0ibia, dist 2d7.0-37. DNom/wil M taphonomic processes have made the bones fragmented, eroded and A.lvastra pile dw.* T7ibia, dist 3d2.1-38. DNom/wil EN-early M porous, and hence – with a few ex- B.undsø* T0ibia, dist 3m1.0-32. DNo M ceptions – difficult to identify and W.ild boar* T0ibia, dist 3d3.0-42. 'Dom'/wil measure. A 45.4 mm long talus of 27 Approx. Tables 9n:1 & 9n:2. Approximate calculations of withers Width Length height at Stite Skeletal elemen Ndote Perio (mm) (mm) withers (cm) height based on long extremity bones of pig (Sus scrofa)

S)tora Förvar G.7 R9adius (juv. 831. >m4 DAo EN-MN from Neolitic habitation sites on Gotland, with compa- S)tora Förvar G.6 R?adius (juv. 1?47 >m77 DAo MN rative material. Sstora Förvar G.4 R4adiu 1828. 6mDAo LN-Br *after During 1986:131 tab. 63. Sstora Förvar G.4 R8adiu 1523. 6mDAo LN-Br ”St. Förvar*” includes all levels, including IA/Medieval! Sstora Förvar G.4 R5adiu 831. 4mDAo LN-Br #Height at withers according to Talus and M.t. III & IV Sstora Förvar G.4 R9adiu 729. 4mDAo LN-Br after Teichert 1969 tab. 7; von den Driesch & Boessneck Sstora Förvar G.4 R9adiu 724mDAo LN-Br 1974:341 tab. 9; Sjøvold Ms. 1986; During 1986:133 tab. Sstora Förvar G.4 R9adiu 724mDAo LN-Br

S)tora Förvar G.6 U?lna (juv. 2411 >?8 WAild MN 66.

Lribbenarve myr F6emu 235 9?W?ild N

Laibbenarve myr T3ibi 241 8?W?ild N S)tora Förvar G.7 F5ibula (juv 1632. >m5 DAo EN-MN Skeletal Width Length Period Site Rsojrhage E.2 T4alu 246. 415. 8?WNild EN-M element (mm) (mm)

S.tora Mafrids T7alus, fr 233. >145. >?8 WNild EN-M S?tora Förvar G.6 P6halanx 1, III-IV 150. 2A7. MN Gsrottan, Överstekv. II T5alu 136.9; 18. 335. 6mD?o BA S.tora Förvar G.6 P4h. 1, post 167. 4A5. MN Sstora Förvar G.7 T1alu 436. 8?WAild EN-MN

Sstora Förvar G.7 T7alu 483. 7dDAom/wil EN-MN S1tora Förvar G.6 P1h. 127. 4A6. MN

Sstora Förvar G.7 T3alu 483. 7dDAom/wil EN-MN S1tora Förvar G.6 P7h. 154A3. MN Sstora Förvar G.7 T6alu 319. 7mDAo EN-MN S)tora Förvar G.5 P7h. 1 (ad. 184B5. MN Sstora Förvar G.6 T5alu 356. 6mDAo MN

Sstora Förvar G.6 T3alu 303. 6mDAo MN S)tora Förvar G.5 P2h. 1 (ad. 118. 4BMN

Sstora Förvar G.4 T3alu 231. 477dDAom/wil LN-Br S)tora Förvar G.5 P2h. 1 (juv. 176. 3B7. MN Sstora Förvar G.4 T8alu 129. 420. 7mDAo LN-Br S)tora Förvar G.5 P-2h. 1 (ad. 4B1. MN Sstora Förvar G.4 T6alu 147. 335. 6mDAo LN-Br

Sstora Förvar GH.3 T3alu 136. 361. 5mDAo Ir S)tora Förvar G.5 P-1h. 1 (juv. 3B7. MN

Sstora Förvar* T0alu 3#7.0-49. 6d6.2-87.7 DNom/wil S1tora Förvar G.5 P2h. 148. 3B5. MN Gsullrum* T0alu 4#2.5-50. (d76.1-89.5) DNom/wil M

Aslvastra pile dw.* T5alu 3#7.0-49. 8d3.2-88.6 DAom/wil EN-MN S1tora Förvar G.5 P6h. 150. 2B7. MN

Bsundsø* T0alu 4#1.0-46. 7d3.4-82.8 DNom/wil M S)tora Förvar G.4 P2h. 1 (ad.? 156. 3A8. LN-Br

Ksudruküla T4alu 278.4; 30. 532. 9d4. Wyil Narva potter S)tora Förvar G.4 P3h. 1 (juv. 183. 2A7. LN-Br Wsild boar* T0alu 4#3.7-56. (d78.2-100.2) Wil

R)ojrhage Anl. A C5alcaneus (juv., fr. 27>36 >m6 DNo EN-M S1tora Förvar* P0h. 37.0-43.

S)tora Förvar G.7 C7alcaneus/m? (ad. 849. 8?WAild EN-MN G1ullrum* P0h. 3N8.0-45. M

S)tora Förvar G.6 C6alcaneus (juv. 737. >m7 DAo MN A1lvastra pile dw.* P2h. 3A3.0-58. EN-MN S)tora Förvar G.6 C8alcaneus (juv. 615. >m6 DAo MN

S)tora Förvar G.6 C6alcaneus (juv., fr. >165. >m6 DAo MN RVojrhage Anl. A P1halanx 2, III-I 291. 2N6. EN-M

S)tora Förvar G.5 C7alcaneus (juv. 256. 805. >?8 DBom/wild MN S2tora Förvar G.7 P9halanx 2A6. EN-MN

S)tora Förvar G.5 C7alcaneus (juv. 254. 749. >m7 DBo MN S.tora Förvar G.6 P9h. 2, anter 186. 2A6. MN S)tora Förvar G.5 C6alcaneus (juv. 235. 749. >m7 DBo MN

S)tora Förvar G.5 C6alcaneus (juv. 274. 771. >m6 DBo MN S.tora Förvar G.6 P6h. 2, post 153A0. MN

S)tora Förvar G.5 C2alcaneus (juv. 272. 63>m6 DBo MN S2tora Förvar G.5 P4h. 195. 2B5. MN

S)tora Förvar G.5 C6alcaneus (juv. 159. >266. >m6 DBo MN S2tora Förvar G.4 P3h. 183. 2A7. LN-Br S)tora Förvar G.5 C2alcaneus (juv. 250. 604. >m6 DBo MN

S)tora Förvar G.5 C2alcaneus (juv. 222. >064. >m6 DBo MN S2tora Förvar* P0h. 22.0-27. S)tora Förvar G.5 C5alcaneus (juv. 296. 860. >m7 DBo MN G2ullrum* P0h. 2N2.0-29. M S)tora Förvar G.5 C5alcaneus (juv. 252. 659. >m6 DBo MN A2lvastra pile dw. * P0h. 2A6.8-29. EN-MN S)tora Förvar G.5 C1alcaneus (juv. 28692. >m5 DBo MN

S)tora Förvar G.5 C9alcaneus (juv. 17536. >m5 DBo MN RVojrhage Anl. A P5halanx 3, III-I 134. 3N0. EN-M S)tora Förvar G.5 C4alcaneus (juv., fr. >825. >872. >m6 DBo MN S3tora Förvar G.7 P9halanx 3A1. EN-MN S)tora Förvar G.4 C7alcaneus (juv. 263. 771. >m6 DAo LN-Br S3tora Förvar G.5 P1h. 3B4. MN S)tora Förvar G.4 C5alcaneus (juv. 108. 66>m5 DAo LN-Br

K)udruküla C1alcaneus/m? (juv. 289. 807. >d82. Wyil Narva potter S3tora Förvar G.5 P3h. 3B5. MN SItora Förvar G.6 M6etacarpus II 8A1. MN S3tora Förvar G.5 P7h. 3B1. MN SItora Förvar G.4 M2.c. II 91. 5A7. LN-Br S3tora Förvar G.5 P6h. 3B3. MN SVtora Förvar G.6 M1etacarpus I 8A4. MN

SItora Förvar G.7 M3etatarsus II 892. 7m6. DIo EN-MN G3ullrum* P0h. 2N8.0-33. M Lsibbenarve myr M0etatarsu 1?0 W?ild N S3tora Förvar* P0h. 31.0-33. SItora Förvar G.6 M3.t. II 931. 8d5. WAil MN A3lvastra pile dw.* P5h. 3A1.8-37. EN-MN S)tora Förvar G.4 M3.t. III (juv. 435. >?42. DAom LN-Br

SItora Förvar* M9.t. II 8#8d3.1-89.4 DNom/wil

AIlvastra pile dw.* M5.t. II 9#8d8.7-93.9 WAil EN-MN

SVtora Förvar G.6 M3etatarsus I 998. 8d6. WAil MN

SVtora Förvar G.5 M7.t. I 750. 6m2. DBo MN

SVtora Förvar* M0.t. I 9#4.0-99. 8d3.1-91.2 DNom/wil

AVlvastra pile dw.* M0.t. I 7#1.5-107. 6d3.2-96.5 DAom/wil EN-MN

28 Table 9o. The Neolithic faunal remains from Rojrhage Table 9p. Species list based on the osteological 1:1, osteological analysis by C. Lindqvist 1999. material (no. of fragments, NISp.) from Rojrhage, Grötlingbo parish, Gotland. Of birds, goosander Sspecies Fsrag Comment (Mergus merganser) has been identified among the Fragments bones from the Iron Age graves (Molnar 1998). probably mostly Mammalia representing No. fr., Mammals, unidentified 2313 Species indet. domestic animals, NISp. i.e. excluding seal bones. 1nHuma Homo sapiens 13

Human Homo sapiens 13 2eCattl Bos taurus 102 Excluding bones Phocidae Seals 107 identified as to 3tSheep or goa Ovis aries, poss. Capra hircus 79 indet. seal species. 4gPi Sus scrofa 34 Halichoerus Grey seal 7 grypus 5gDo Canis familiaris 1?

Ringed seal Pusa hispida 13 6xRed fo Vulpes vulpes 2?

Phoca Harp seal 2 7lGrey sea Halichoerus grypus 7 groenlandica 8lRinged sea Pusa hispida 13 1.29 Total seal bones

Vulpes 9lHarp sea Phoca groenlandica 3 Fox? 3? vulpes? Indet. seal Phocidae indet. 107 Cattle Bos taurus 101 1r1 Goosande Mergus merganser Pig Sus scrofa 35 Indet. bird Aves indet. 29 Ovis aries sive Sheep or goat 82 Capra hircus 1d2 Co Gadus morhua 57- Total domestic 218 1e3 Pik Esox lucius 1 animal bones.

Birds, unident. Aves indet. 2.9 Total bird bones lennia on Gotland at the introduction of agri- Mergus Goosander merganser culture. Of these canid species only traces have

Fish Pisces indet. 5.7 Total fish bones been found at Rojrhage. A couple of bone frag-

Gadus ments (a Phalanx 1, V, sin., a Scapula fragment Cod (>50) morhua and a Fibula? fragment) may derive from a very Pike Esox lucius (1) small dog or, more probably, a fox. As both species have also been identified in the Iron Age same size as the pigs of the Neolithic, e.g. from materials, these bones may be intrusive. Gullrum (table 9l, 9m & 9n). Perhaps this bog find indicates that Neolithic pigs were herded 10. Intra-site find distributions loose in the in the same way as on the Mediterra- nean Islands (Lewthwaite 1984, Lindqvist & Poss- All three main find categories – pottery, flint and nert 1997a:66 with refs.) and in Spain, Hungary bones – display the same general distribution and Russia (Brehm 1931:487-488, 490) in recent across the excavated area (tables 10a-c). The main centuries, as was the forest pig (”acorn pig”) in concentration forms an arc from squares I0 and Sweden and Norway (Brehm 1931:494, Lepiksaar H0 via E2 and E3 to G6 and H6. This distribu- 1977:91) into the 19th century and later. tion avoids only one of the Iron Age graves, the abovementioned mound, and should thus have 9.3.3 Dog/fox (Canis sp.) some significance for the interpretation of the Dogs (Canis familiaris), which judging from bone Neolithic finds. finds seem to have been large, may have been As mentioned previously, the TRB and GRK introduced on Gotland to serve as sheepdogs or pottery mix across the area, but the STR pottery watchdogs during the Neolithic. Foxes (Vulpes is focused near the middle of the arc. Neither vulpes), which judging from bone finds seem to the GRK pottery nor the flint finds exhibit ty- have been small, had already lived for two mil- pological variation within the distribution. Given

29 the homogeneity of the artefacts’ distribution, it a distribution in a multi-component deposit for- is rather surprising to find significant variation med by the refuse of centuries and then badly among the bones of the identified animal spe- damaged by later features can, of course, only cies. Cattle and sheep bones concentrated at the be a matter of speculation. 0-end, while bones of pig and seal were evenly distributed across it. The interpretation of such

Table 9q. Summary of the osteological material (number of fragments, NISp.) from Rojrhage 1:1.

Ovis/ Canis/ Halich- Phoca Phoci- Mamm. Av- Pisc- Context Hsomo Bo Sus Pusa Capra Vulpes oerus groenl. dae Indet. es es

C?.4 1-5-- -1--3 1

C.5 ------1-- 4

D.3-4 ------2

D.22 ------2--

E--11-.2 a 28--

E--2--.2 b 171--3

E-2-*.3 3?191313*9 26

E?.4 1---?2128311

F1-*.1 11- 61--1 5

F-61?.1-2 3-27 ? 6--

F-2*.2 5-- 21-5 9

F?.4 31-3-1* * 12101120

F-112-1.5 -14--

F-1---.6 ----

G-*.1 3*2 5--1? *3-184 ?1-*9 10

G-31--.2 -01--

G-2---.3-4 -02--

G.4 -----342 * 1--

G-?.5 312- 1 74-- 8

G-1---.6 1231--

G.7 ----- 123 ?2--

H-?.0 I.0 23-- 471- 2

H?.0 4-6--218 ?41-

H-33--*.1 1-24--

H.5 ------3-- 1

H.6 ------4--

I-?.0 9-4- 1 65-- 3

F--?eat. A 15- 1 6566 3

F-?eat. B 18-- 1 31515 9

F-?eat. C 1-?1- -321 8

Feat. D ------1--8 6

SA hearth --5-- -2--9 9 1

S32tray 333?1?0 -?1?1-19 21933

30 Table 10a. Distribution of Neolithic pottery (gram- 11. Phasing, site function and mes) from Neolithic contexts at Rojrhage. The numbers in tables 10a-c have been adjusted propor- cultural interpretation tionally upwards for find-yielding squares along the edges of the trench that were not excavated in their As shown by the artefact studies and radiocarbon entirety. dates, the Rojrhage deposit exhibits three tempo- 0 1234567 rally separate Neolithic components. The first one dates from the late EN I and includes only a small C ---03392-- amount of TRB pottery. No radiocarbon dates of this period have resulted from the dating program- D -0212120-- me. The second component dates from the late E -9190 128 330 10 000 MN A and includes a great deal of GRK pottery, flint debitage, a bone harpoon point and bones of -4414 306 39391 700 F pig, cattle, harp seal, ringed seal and grey seal. The G -2680 2028375101 third component dates from the early LN and includes a small amount of STR pottery and bones H 282 7000930 of grey seal. The LN date for a grey seal bone is 13 I 159 00000- - compromised by the fact that the d C value of the sample (Ua-16502) indicates a terrestric diet for the Table 10b. Distribution of flint (fragments) from animal. A greenstone axe belongs either to the Neolithic contexts at Rojrhage. MN or the LN component. Rojrhage breaks the pattern of Österholm’s sett- 0 1234567 lement model (1989:173-180) at two points. Firstly, C ---0862-- in the Barshalder area coastal activities continued through the EN at Rojrhage and possibly also at D -00110- - the parish gravel pit 270 m to the NNE. This is E -116 51 6000 evidence, not very surprisingly, of a sustained mari- time component also in the EN economy of Got- F -8402 7 670 land (as in the Lake Mälaren area, Segerberg 1999: 198-199). Secondly, the model of generally a single G -733310 111 GRK site in each settlement cluster (Gullrum in H 11 2000510 the case of Hablingbo-Näs-Grötlingbo) is contra- dicted by the GRK sites of Rojrhage and Kattlunds I 000000- - (Raä 55) in Grötlingbo, none of which has been delimited spatially but which seem rather small. The Table 10c. Distribution of bone (fragments) from second point is foreshadowed by Österholm’s ob- Neolithic contexts at Rojrhage. servations of multiple GRK sites in När-Lau, Vis- by and Ire-Lickershamn (1989:174, 176, 177). 0 1234567 Source quality at Rojrhage permits few detailed C ---0103 74-- conclusions about the cultural behaviour leading to the formation of the Neolithic deposit. The simp- -02110- - D lest interpretation seems to be that the deposit for- E -538177 264 4 000 med from refuse dumped during several episodes of inhabitation through centuries. The inhabitants F -3396 90162 4410 used and broke pottery, made simple flint tools and G -4310 61103267193 procured and consumed the meat of wild and tame animals. Chance preservation leaves us a single har- H 1691 97404134300 poon point as the only artefactual evidence of their economy, while the ecofactual evidence of the bo- I 114 00000- - nes and the coastal location speak more eloquently.

31 12. References Theses and papers in North-European archaeology 13b. Department of Archaeology, Åhlén, Marianne. 1972. Stenåldersboplatsen på Ar- University of Stockholm. dags i Ekeby, Gotland. BA thesis. Department Curry-Lindahl, Kai. 1975/82. Däggdjur i färg – alla of Archaeology, University of Stockholm. Europas arter. Färgserien. Stockholm. Åkerlund, Agneta; Bergh, Stefan; Nordbladh, Jarl von den Driesch, A. & Boessneck, J. 1974. Kri- & Taffinder, Jacqueline (eds). 1997. Till Gun- tische Anmerkungen zur Widerristhöhenbe- borg. Arkeologiska samtal. Stockholm Archaeo- rechnung aus Längen-massen vor- und früh- logical Reports 33. Museum of National Anti- geschichtlicher Tierknochen. Säugetierkundliche quities & Department of Archaeology, Uni- Mitteilungen 22. München. [26 pp] versity of Stockholm. During, Ebba. 1986. The fauna of Alvastra. An os- Andersson, Torgny. 2000. Eke sockens förhistoria. teological analysis of animal bones from a Neolithic Eke. [150 pp] pile dwelling. Ossa 12, suppl. 1. Osteological Re- Bägerfeldt, Lars. 1992. En studie av neolitikum på search Laboratory, University of Stockholm. Gotland. Problem och konsekvenser, utifrån under- Solna. sökningen av en dös och neolitiska lösfynd. Gamle- Edenmo, Roger; Larsson, Mats; Nordqvist, Bengt by. & Olsson, Eva. 1997. Gropkeramikerna – Bagge, Axel & Kjellmark, Knut. 1939. Stenålders- fanns de? In Larsson & Olsson 1997. boplatserna vid Siretorp i Blekinge. KVHAA Mo- Falck, Waldemar. 1971. Ang. undersökning av golv nografier 26. Stockholm. vid Kattlunds, Grötlingbo. Report dated 1971-12- Becker, Carl Johan. 1947. Mosefundne lerkar fra 20 at ATA (dnr 321-1907-2000), Stockholm, yngre stenalder. Aarbøger for nordisk oldkyndig- with a copy in the archives of Gotlands Forn- hed og historie 1947. Copenhagen. sal county museum, Visby. Becker, Carl Johan. 1990. Nørre Sandegård. Arkæo- Hallander, Håkan. 1989. Svenska lantraser – deras logiske undersøgelser på Bornholm 1948-1952. His- betydelse förr och nu. Veberöd. torisk-filosofiske skrifter 13. The Royal Da- Hansson, Hans. 1900. En stenåldersboplats på nish Academy of Science and Letters. Copen- Gotland. Svenska Fornminnesföreningens Tidskrift hagen. X. Stockholm. Brehm, Alfred Edmund. 1931. Apor, halvapor, Hansson, Harald. 1927. Excavation report dated uddatåiga och icke idisslande hovdjur. Djurens 1927-09-27 at ATA (dnr 4043/1927), Stockholm. liv, band 1. Malmö. Hedell, L. 1910. Benfynd från Hemmor, Got- Burenhult, Göran (ed.). 1997a. Remote sensing. land. Ms. at ATA. Applied techniques for the study of cultural resources Hedemark, Åse; Samuelsson, Christoffer & Ytter- and the localization, identification and documentation berg, Niklas. 2000. Stenåldersboplatsen vid Hem- of sub-surface prehistoric remains in Swedish archa- mor i ny belysning. Gotländskt Arkiv 72. Fören- eology. Vol. 1. Osteo-anthropological, economic, envi- ingen Gotlands fornvänner. Visby. ronmental and technical analyses. Theses and pa- Holm, Jenny; Olsson, Eva & Weiler, Eva. 1997. pers in North-European archaeology 13a. De- Kontinuitet och förändring i senneolitikum. partment of Archaeology, University of Stock- In Larsson & Olsson 1997. holm. Janzon, Gunborg O. 1974. Gotlands mellanneoliti- Burenhult, Göran. 1997b. Introduction. In Bu- ska gravar. Studies in North-European archaeo- renhult 1997a. logy 6. Department of Archaeology, Univer- Burenhult, Göran (ed.). 1997c. Ajvide och den mo- sity of Stockholm. derna arkeologin. Stockholm. Jonsson, L. 1986. From wild boar to domestic Burenhult, Göran (ed.). 2002. Remote sensing. App- pig – a reassessment of Neolithic swine of lied techniques for the study of cultural resources and Northwestern Europe. Königsson, L.-K (ed.). the localization, identification and documentation of Nordic Late Quaternary Biology and Ecology. Stri- sub-surface prehistoric remains in Swedish archaeolo- ae 24. Uppsala. [3 pp] gy. Vol. II. Archaeological investigations, remote sen- Jonsson, L. 1988. The Vertebrate Faunal Remains sing case studies and osteo-anthropological studies. from the Late Atlantic Settlement Skateholm

32 in Scania, South Sweden. Larsson, L. (ed.). The kott på Ajvide under gropkeramisk tid. In Bu- Skateholm Project I. Man and Environment. Inter- renhult 1997c. [3 pp] disciplinary Studies. Stockholm. [31 pp] Lindqvist, Christian. 1997b. Ansarve hage-dösen; Kjellberg, A.-S. 1988. Osteologisk analys av djur- Tvärvetenskapliga aspekter på kontext och den bensmaterial från Stora Domerarve I-II. Ms. neolitiska förändringen på Gotland. In Åker- Koch, Eva. 1998. Neolithic bog pots from Zealand, lund et al 1997. [18 pp] Møn, Lolland and Falster. Nordiske Fortidsmin- Lindqvist, Christian. 1997c. About the importan- der B:16. Det Kongelige Nordiske Oldskrift- ce of fine-mesh sieving, stratigraphical and selskab. Copenhagen. spatial studies for the interpretation of the KVHAA. 1974. Kungl. Vitterhets Historie och Anti- faunal remains from Ajvide, Eksta parish and kvitets Akademiens fastigheter av antikvarisk ka- other Stone Age dwelling sites on Gotland. raktär jämte övriga områden ställda under akademiens In Burenhult 1997a. [21 pp] eller riksantikvarieämbetets förvaltning. Andra ut- Lindqvist, Christian & Possnert, Göran. 1997a. ökade upplagan. Bilaga till KVHAA:s årsbok The subsistence economy and diet at Jakobs/ 1973. Stockholm. Ajvide and Stora Förvar, Eksta Parish and oth- Larsson, Mats. 1994. Lokale Gruppen des Früh- er prehistoric dwelling and burial sites on Got- neolithikums in Südschonen, Schweden. In land in long-term perspective. In Burenhult Hoika, Jürgen (ed). Beiträge zur frühneolithischen 1997a. [62 pp] Trichterbecherkultur im westlichen Ostseegebiet. 1. In- Lindqvist, Christian & Possnert, Göran. 1997b. ternationales Trichterbechersymposium in Schleswig Om reservoareffektproblemet. In Burenhult vom 4. bis 7. März 1985. Untersuchungen und 1997c. Materialien zur Steinzeit in Schleswig-Holstein Lindqvist, Christian & Possnert, Göran. 1999. aus dem Archäologischen Landesmuseum der The first seal hunter families on Gotland. On Christian-Albrechts-Universität 1. Neumünster. the Mesolithic occupation in the Stora För- Larsson, Mats. 1997. Skåne. In Kihlstedt, Britta; var Cave. Current Swedish Archaeology 7. The Larsson, Mats & Nordqvist, Bengt. Neolitiser- Swedish Archaeological Society. [23 pp] ingen i Syd-, Väst och Mellansverige – social Lindqvist, Christian & Possnert, Göran. subm. och ideologisk förändring. In Larsson & Ols- a. The Neolithic swine on Gotland: feral game son 1997. or loosely herded forest pigs? Journal of Osteo- Larsson, Mats & Olsson, Eva (eds). 1997. Region- archaeology. alt och interregionalt. Stenåldersundersökningar i Syd- Lindqvist, Christian & Possnert, Göran. subm. och Mellansverige. Skrifter 23. Riksantikvarieäm- b. Baltic stages, early marine adaptation and betet, Arkeologiska undersökningar. Stock- the Mesolithic-Neolithic transition on Got- holm. land, based on bone d13C and AMS 14C-data. Lepiksaar, Johannes. 1977. Människan och hus- Journal of Archaeological Science. djuren. Fauna och flora, svensk faunistik och floris- Lindqvist, Christian & Storå, Jan. 1997. Havet och tik 72. Stiftelsen för svensk faunistik och flo- sälarna. In Burenhult 1997c. [2 pp] ristik. Lund. [23 pp] Malmer, Mats P. 1962. Jungneolithische Studien. Acta Lewthwaite, James. The art of Corse herding. Ar- Archaeologica Lundensia, series in octavo 2. chaeological insights from recent pastoral University of Lund. practices on west Mediterranean islands. Clut- Manneke, Peter. 1961. Gotland, Grötlingbo, Suder- ton-Brock, Juliet & Grigson, Caroline (eds.). kvie 1:14. Report at ATA (dnr 1669/1963), Animals and archaeology 3. Early herders and their Stockholm. flocks. BAR International Series 202. Oxford. Matolcsi, J. 1970. Historische Erforschung der Lindgren, Christina & Nordqvist, Bengt. 1997. Körpergrösse des Rindes auf Grund vom un- Lihultyxor och trindyxor. Om yxor av basiska garischen Knochen-material. Zeitschrift für Tier- bergarter i östra och västra Sverige under me- züchtung und Züchtungsbiologie 87. Berlin. [49 pp] solitikum. In Larsson & Olsson 1997. Molnar, Petra. 1998. Osteologisk analys av skelett- Lindqvist, Christian. 1997a. Betydelsen av ben- materialen från Barshalder-gravfältet på Gotland. och tandfynd av rödräv, hund, svin och igel- Delrapport II. Ms. at ATA.

33 Munthe, Henrik Vilhelm. 1910. Studier öfver Gott- Rundkvist, Martin. 2002. Go, Grötlingbo sn, Rojrha- lands senkvartära historia. Sveriges geologiska ge 1:1, Raä 54. Barshalder cemetery. Rescue excava- undersökning Ser. Ca, Avhandlingar och upp- tion of 41 Late Iron Age graves and extensive Neo- satser 4. Stockholm. lithic deposits by Gustaf Trotzig 1963-1967. Rep- Nihlén, John. 1927. Gotlands stenåldersboplatser. ort dated 2001-11-22 at ATA (dnr 321-0612- KVHAA Handlingar 36:3. Stockholm. 2002), Stockholm, with a copy in the archives Olsson, Eva & Edenmo, Roger. 1997. Östra of Gotlands Fornsal county museum, Visby. Mellansverige. In Edenmo et al. 1997. Rundkvist, Martin. 2003. Barshalder 1. A cemetery Österholm, Inger. 1989. Bosättningsmönstret på in Grötlingbo and Fide parishes, Gotland, Sweden, c. Gotland under stenåldern. En analys av fysisk miljö, AD 1-1100. Excavations and finds 1826-1971. ekonomi och social struktur. Theses and papers Stockholm Archaeological Reports 40. Depar- in archaeology 3. Department of Archaeolo- tment of Archaeology, University of Stock- gy, University of Stockholm. holm. Österholm, Inger. 1997. An analysis of the cera- Samuelsson, Christoffer & Ytterberg, Niklas. 2002. mics from the Ajvide site. [Augmented trans- Arkeologisk undersökning. Hemmorboplatsen. Dalbo lation of Österholm 1989:97-117.] In Buren- 1:45, 1:44, Hemmor 1:4, 1:19, Bomunds i Burgen 1:16, hult 1997a. Raä 145, När sn, Gotland. Rapport, Länsmuseet Österholm, Inger. 2002a (1986). The Jakobs/Aj- på Gotland. Visby. vide site at Eksta, Gotland. In Burenhult 2002. Schnittger, Bror & Rydh, Hanna. 1940. Grottan Österholm, Inger. 2002b (1986). The St. Domer- Stora Förvar på Stora Karlsö, undersökt av Lars arve site At Hablingbo. In Burenhult 2002. Kolmodin och Hjalmar Stolpe. KVHAA Mono- Österholm, Inger. 2002c (1986). The pattern of grafier 5. Stockholm. Stone Age settlements on Gotland. In Buren- Segerberg, Ann. 1999. The Bälinge bogs. Neolithic hult 2002. coastal settlement in Uppland, eastern middle Swe- Petrén, Michael. 1992. Neolitiska yxor från Gotland. den. Bälinge mossar. Kustbor i Uppland under yngre Form och fördelning. BA thesis. Department of stenåldern. Aun 26. Department of Archaeology Archaeology and Ancient History, University and Ancient History, University of Uppsala. of Uppsala. Sernander, Rutger. 1898. Zur Kenntniss der qvar- Possnert, Göran. 2002. Stable and radiometric tären Säugethier-Fauna Schwedens. Bulletin of carbon results from Ajvide. In Burenhult 2002. the Geological institution of the university of Upsala Rundkvist, Martin. 1996a. Lägesrapport från 3 (1897). Barshalder. Aktuell Arkeologi V. In Renck, Siivonen, Lauri. 1968/76. Nordeuropas däggdjur. Anna Maria & Stensköld, Eva (eds). Stock- Andra omarbetade upplagan. Praktiska fälthand- holm Archaeological Reports 30. Department böcker i färg. Stockholm. of Archaeology, University of Stockholm. Sjøvold, Torstein. Ms. 1986. Unpublished manu- Rundkvist, Martin. 1996b. Go, Grötlingbo, Roes 1:36, script on the calculation of height at withers intill Raä 54. Gårdssamling: trindyxeboplats mm. in domestic animals. Report dated 1996-11-25 at ATA (dnr 439- Stenberger, Mårten; Dahr, Elias & Munthe, Hen- 6516-1996), Stockholm, with a copy in the rik. 1943. Das Grabfeld von Västerbjers auf Got- archives of Gotlands Fornsal county muse- land. KVHAA Monografier 30. Stockholm. um, Visby. Teichert, Manfred. 1969. Osteometrische Unter- Rundkvist, Martin. 2001. Go, Grötlingbo sn, Uddvi- suchungen zur Berechnung der Widerristhö- de Samfällt + Roes 1:28 + Roes 1:31, Raä 45. he bei vor- und frühgeschichtlichen Schwei- Barshalder gravel pit. Rescue excavation of 27 Late nen. Kühn-Archiv 83. Berlin. [56 pp] Viking Period graves and a small Stone Age deposit Thorsberg, Kalle. 1997. Den gropkeramiska kul- by Gustaf Trotzig 1966. Report dated 2001-11- turens o-väsen. In Åkerlund et al. 1997. 21 at ATA (dnr 321-3341-2001), Stockholm, with Wallin, Paul & Martinsson-Wallin, Helene. 1996. a copy in the archives of Gotlands Fornsal Åter till Hemmorboplatsen – utgrävningar, county museum, Visby. analyser och reflexioner. Gotländskt Arkiv 68. Föreningen Gotlands fornvänner. Visby.

34 Vretemark, Maria. 1997. Från ben till boskap. Kost- håll och djurhållning med utgångspunkt i medeltida benmaterial från Skara. Skrifter från Skaraborgs länsmuseum 25. Skara.