Journal of Archaeological Science 35 (2008) 2813–2823 Contents lists available at ScienceDirect Journal of Archaeological Science journal homepage: http://www.elsevier.com/locate/jas Stone Age fishhooks – how were they dimensioned? Morphology, strength test, and breakage pattern of Neolithic bone fishhooks from Ajvide, Gotland, Sweden Carina Olson a, *, Karin Limburg b, Mikael Söderblom c a Osteoarchaeological Research Laboratory, Department of Archaeology and Classical Studies, Stockholm University, Lilla Frescati 7, SE-106 91 Stockholm, Sweden State University of New York, College of Environmental Science and Forestry, Syracuse, NY 13210, USA c Sadelgatan 279, SE-194 72 Upplands Väsby, Sweden b a r t i c l e i n f o a b s t r a c t Article history: Received 17 January 2008 Received in revised form 26 April 2008 Accepted 14 May 2008 The late Stone Age Pitted Ware site at Ajvide, Gotland, in the Baltic Sea, Sweden, has revealed a large deposit of fishbone and approximately 400 bone fishhooks, complete and incomplete. Cod (Gadus morhua), which is one of the most abundant fish species in the bone assemblage, was probably caught with hook and line fishing. To investigate the fishhooks’ field of application, a morphological and morphometric study was performed on 384 available hooks. Two sets of replicas made of four selected original fishhooks were submitted to a strength test. A breakage study of the incomplete hooks in comparison with the strength-tested hooks was carried out in order to distinguish fresh breaks from dry breaks. It seems that a certain morphology for fishhooks was preferred at Ajvide, indicating they were produced by skilled craftsmen for special usage. The strength test showed that the hooks had a weight bearing capacity more than the average size of cod caught at Ajvide. Using results of these tests, we predicted that the mean breaking strength of 46 intact Ajvide hooks was 96.6 26.1 (s.d.) Newtons (equivalent to 9.85 2.7 kg). The design of fishhooks changed somewhat over time, being slightly larger in the oldest layers of the site. The breakage patterns of the hooks show that the bow was the most common area of breakage. The design and weight bearing capacity of the hooks point to a specialized cod fishery from boats in deep water. Ó 2008 Elsevier Ltd. All rights reserved. Keywords: Neolithic bone fishhooks Hook morphology Hook sizes Fish sizes Material strength test Breakage pattern 1. Introduction In the middle Neolithic period, c. 3300–2300 BC (e.g. Berglund, 1999; Welinder et al., 1998), a coastal complex defined as the Pitted Ware culture appeared in Sweden, Norway, Denmark and Åland/ Finland. In Sweden, the Pitted Ware sites have been found in the coastal areas from Scania in the south (e.g. Burenhult, 1999; Malmer, 2002; Welinder et al., 1998;) up to Ångermanland in northern central Sweden (Björck, 1997; Färjare and Olsson, 2000), as well as on the islands of the Baltic Sea (Janzon, 1974; Stenbäck, 2003; Storå, 2001; Österholm, 1989). The bone refuse from all these sites indicates a similar coastal economy, based on fishing and seal hunting (e.g. Edenmo et al., 1997; Ericson, 1989; Storå, 2001, 2002) and a roughly similar diet (Eriksson, 2003; Lidén, 1995). Most Pitted Ware site refuse faunas are dominated by seal bones, and where preservation conditions are good, large amounts of fishbone are also found, e.g. on the island of Gotland in the Baltic Sea, with its calcareous soil and bedrock. Wild boar bones are also present in rather large quantities at many Pitted Ware sites (Aaris-Sørensen, * Corresponding author. Tel.: þ46 8161393; fax: þ46 8164476. E-mail address: carina.olson@ofl.su.se (C. Olson). 0305-4403/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.jas.2008.05.009 1978; Ekman, 1974; Rowley-Conwy and Storå, 1997; Hedell, unpublished data; Lepiksaar, 1974). Sites with Neolithic finds of bone fishhooks on Gotland are Gullrum (Lithberg, 1914), Visby, Fridtorp (Nihlén, 1927), Västerbjers (Stenberger, 1943), Ire (Janzon, 1974; Ekman, 1974), and Ajvide (Burenhult, 1997; Österholm, 1989). Ajvide on the southwestern coast of Gotland is exceptionally rich in fishbone, and has also revealed the largest amount of bone fishhooks among the Swedish Stone Age sites to date. The large deposits of fishbone from marine species at Ajvide indicate the importance of fish in the diet of the Ajvide people. A heavy dependence on marine proteins is supported by stable isotope studies of human remains from Ajvide (Lidén, 1995; Lindqvist and Possnert, 1997). The fish fauna from Ajvide consists of fresh-water, marine and migratory species. Herring (Clupea harengus) and cod (Gadus morhua) are the dominant species making 65% and 30%, respectively, of the identified fish remains based on NISP. Other fish species found at the site include pike (Esox lucius), perch (Perca fluviatilis), flatfish (Pleuronectiformes), salmonids (Salmonidae), cyprinids (Cyprinidae), whitefish (Coregonidae), and eel (Anguilla anguilla) (Olson and Walther, 2007). Archaeological finds show that hooks, nets and leisters (pronged spears) have been used as fishing implements at Ajvide. The fisheries of Ajvide included species that stayed in or passed through the 2814 C. Olson et al. / Journal of Archaeological Science 35 (2008) 2813–2823 Fig. 1. Location of the study. Map showing Gotland in the Baltic Sea, and the location of Ajvide on the southwestern coast of Gotland, latitude 57 140 N, longitude 18 80 E. The map of Gotland shows the shoreline from the end of the Middle Neolithic period. Modified from Österholm (1989). deep waters of the open sea during migration, as well as species that visited or resided in the littoral waters and in streams. Flatfish and cyprinids confirm fishing during the warmer part of the year, while the other species presented fishing possibilities more or less on a year round basis (Olson and Walther, 2007). The main aim of this study was to examine the fishhook design to evaluate if they were manufactured according to specific morphological criteria, to maximize catch efficiency for particular fishing strategies or species. In particular, we wished to test the hypothesis that the hooks were designed for cod fishing on local stocks. In this case, the hook morphology and strength should match the size of the cod caught at Ajvide, for which we here provide weight estimates from fish bones. By using modern replicas in a strength test, we also investigated the hooks’ weight bearing capacity. The strength test would shed light on the breakage patterns of the fishhooks and thus facilitate comparisons with the breakage patterns of the original hooks in the Ajvide assemblage. The breakage and fracture patterns of the archaeological hooks were studied in detail to be able to distinguish fresh breaks (i.e. breaks from the time of active use) from dry breaks (breaks that occurred from the time of deposition until discovery) and also to estimate the effect of post-depositional destruction. Preservation issues may have affected the possibilities of evaluating the morphology of the hooks, and thus must be checked. 2. Material transgression layer, separates the lower older layers 4 and 5 from the upper younger layers 1 and 2. The layers correspond to 10-cm thick technical (excavation) layers. The hooks investigated in the present study derive both from graves and from layers 1 to 5 of the settlement area. 2.2. The Ajvide fishhook assemblage This is the largest assemblage of bone fishhooks from the Stone Age in Sweden. Out of 384 fishhooks, 354 derive from the settlement area, and the remaining 30 originate from 22 different grave contexts. The fishhook assemblage comprises of complete (n ¼ 54), incomplete (n ¼ 327), and unfinished hooks (n ¼ 3). Nearly all fishhooks from Ajvide are made of long bones, like the tibia of wild boar (Fig. 2). A few are made of wild boar tusks, and possibly are some also made of metapods of elk or moose. No composite or stone hooks have been recovered at Ajvide, but there is one doublepointed gorge of bone with centrally placed grooves for attaching a line. 2.3. Fishhooks selected for the strength test Four original, intact bone fishhooks (Fig. 3) from the Ajvide fishhook collection were selected to be copied and used in a breaking strength test. The four original fishhooks seem to be cut out of the tibia of wild boar (Sus scrofa). Wild boar is the most common terrestrial mammal in the faunal material at Ajvide. 2.1. The site Ajvide is located in Eksta parish on the southwest coast of the island of Gotland in the Baltic Sea (Fig. 1). Comprehensive studies and yearly excavations have been carried out at Ajvide since the early 1980s (Burenhult, 1997a,b, 1999, 2002; Österholm, 1989). The excavations have revealed an interesting complex of both settlement areas and a large burial ground. Considerable amounts of animal bones, more than 7000 artifacts, and 2.5 tons of pottery have been recovered from Ajvide so far. The main period of use was between 3200 and 2300 BC (Burenhult, 1997). A transgression that took place at ca. 2900 BC cal. submerged the site for a short period of time and divided the cultural layers, which is of importance for the stratigraphy and the interpretation of the site. Layer 3, the Fig. 2. Ajvide fishhook made from a wild boar tibia. Measure from the proximal end of the tibia to the bow of the fishhook ¼ 83.8 mm. C. Olson et al. / Journal of Archaeological Science 35 (2008) 2813–2823 2815 Fig. 3. Original intact Neolithic fishhooks from Ajvide, Gotland, copied for strength test. From left to right: H1, H2, H3, and H4. Squares: 30 30 mm. However, since there are occasional imported finds of both moose (Alces alces) and elk (Cervus elaphus) at Ajvide, bone of these species cannot be disregarded as raw-material. The long bones of the most common mammal at Ajvide, the harp seal (Phoca groenlandica), were not suitable for fishhook manufacturing due to their morphology, but they were often used for the production of awls. Our criteria for the selection were (1) that the hooks, in size and morphology, should be representative for the Ajvide fishhook assemblage, (2) that hooks with holes and grooves for line attachment should be tested, and (3) that a comparison of raw-material should be carried out, using bone tissue of two mammal species (local and imported) that we knew were available at Ajvide during the Neolithic. The original fishhooks H1 and H4 are equipped with a hole and a groove for attaching the fishing line. On both these hooks the hole diameter decreases towards the center of the shank, showing that it was drilled from both sides. H2 and H3 lack holes, but are equipped with grooves for line attachment. H2 has two grooves and H3 has one groove on each side of the head of the shank. The upper ends of the shanks of H2 and H3 include cancellous bone, while H1 and H4 are made solely of compact bone. The raw-material for the replicas comes from a captive population of wild boar from the Åland Islands in the northern part of the Baltic Sea, and from moose from the mainland of central Sweden. Two sets of four replicas each were made. One set of replica hooks was made from wild boar bone (designated with ‘‘-wb’’) and the other set was made of moose bone (designated with ‘‘-m’’) (Fig. 4). 3.1. Manufacturing the replicas The fishhook manufacturing began with preparing the rawmaterial. The selected limbs were stripped off and cleaned of skin, sinews, marrow, and the periosteum. This was done while the material was fresh, to avoid the fat affecting the bones. Boiling the bones, which may be a simpler way to clean them, is more time consuming, and could also impair the special qualities of the raw bone, such as strength and elasticity. After the preparation and cleaning were carried out, the bone element was cut in two, and the 3. Methods All 384 available hooks from the Ajvide fishhook assemblage from graves and three chronological levels of the cultural layer were morphologically analyzed. We measured eight different parameters (to the nearest 0.01 mm) including overall length, bow height and width, point length, type of line attachment, gape, and gripping angle (Fig. 5a and b). We identified and classified type of fracture for the incomplete hooks. Three of the incomplete hooks were not measured since they are in an ‘‘unfinished’’ state, meaning, not separated from the raw-material. Fig. 4. Replicate fishhooks constructed of moose (m) and wild boar (wb) bone showing fractures. Scale same as in Fig. 2. 2816 C. Olson et al. / Journal of Archaeological Science 35 (2008) 2813–2823 Fig. 5. (a,b) Fishhook morphology and measured parameters. (a) A ¼ fishhook head incl. line attachment area, B ¼ shank, C ¼ bow, D ¼ point. (b) Fishhook parameters that were measured: OL ¼ overall length, MBH ¼ maximum bow height, MBW ¼ maximum bow width, PL ¼ point length, GAPE ¼ hook gape, GA ¼ gripping angle. when the hook is just penetrating the skin of a fish, while loading at the bottom of the bow would rather emulate the case when the fish is already hooked. Since the hooks are much thicker at the bow than elsewhere, this indicates that rupture often occurred in this area, and that is likely due to bending moments from forces close to the tip. Of course, under actual usage conditions, the load is distributed over the hook and the force on different parts of the hook varies during the catch. Considering this, it was assumed that loading close to the tip, avoiding point forces, would be adequate. Thus a small metal cylinder (nut) was threaded over the hook’s tip and the external force was applied to that cylinder some millimeters from the very tip. The hooks were fixed at their opposite ends to the testing machine, using parts of a cord. Loading equipment. The tests were run in a universal servo-hydraulic material testing machine, MTS load frame 50/100 KIP, model 309.03, and INSTRON 8500þ digital control. A HBM load cell, model U2B 0.5 kN #H01092 was used to measure the applied load. The piston position was indicated by an integral linear variable differential transformer (LVDT). Load and displacement were stored in a connected PC. 3.3. Breakage pattern of Neolithic hooks part intended as raw-material for the replica was selected. The hook manufacturing consisted of a combination of fine cutting, scraping, and grinding. As the project did not include the use of authentic tools, steel-edged tools and grinding cloths were used, but the procedure would have been similar if flint-edged tools and sandstone had been used. The holes in the upper part of the shank of hook H1r-m, H1r-wb, H4r-m and H4r-wb were drilled from both sides with a bow-drill. The drill was made of round steel. The pointed end of the round steel was cold forged into a cutting edge. The copies are almost identical to the originals. There are only diminutive measurement discrepancies which are considered to not affect the strength test in any significant way (Table 1). 3.2. Material strength test In the purpose to find out the weight bearing capacity of the selected fishhooks from Ajvide, two sets of replicas, consisting of four hooks each, were subjected to tensile strength testing at the Department of Solid Mechanics, KTH Royal Institute of Technology in Stockholm. 3.2.1. Load application The position of the applied loads is of main importance. Loading at the very tip would result in lower breaking force than loading at the bottom of the bow. Loading at the tip would emulate the case To gain more insight into the hook and line fishing technology at Ajvide, we studied the fracture areas of the incomplete hooks in order to detect the fresh breaks, and to learn how the fishhooks had ruptured during fishing. Three criteria of fracture were used for identifying and distinguishing fresh breaks from post-depositional breakage of dry bones (Johnson, 1985; Outram, 2001, 2002). A fracture is identified and classified from (1) the fracture outline (shape), (2) fracture angle, and (3) fracture texture (smooth or rough). Our observations of the Ajvide fractures corresponded well with the classification of fresh and dry breaks made by Johnson and Outram, even if those did not include bone fishhooks. For the Ajvide hooks, we define fresh breaks as having a smooth fracture texture and mostly a sloped fracture area with rather sharp edges, while the dry breaks often have a more straight fracture angle, rough fracture texture, and rather blunt edges. 3.4. Cod sizes and weights at Ajvide Methods for size estimations of archaeological remains of prehistoric fish have been established by a number of researchers (Bødker-Enghoff, 1983, 1994, 1995; Casteel, 1976; Lepiksaar and Heinrich, 1977; Morales and Rosenlund, 1979; Wheeler and Jones, 1989). Size estimates (maximum length) of cod at Ajvide have been presented earlier based on measured otoliths (ear stones) and Table 1 Parameter values of original and replicate fishhooks selected for the strength test ID Context Hook no. OL (mm) MBH (mm) MBW (mm) PL (mm) Hole/groove HI Ø (mm) Gape (mm) GA Weight (g) Ultimate load 30826 30826 30826 CL CL CL H1 H1r-m H1r-wb 51.29 51.35 51.55 10.87 11.27 11.05 12.88 12.95 12.88 24.27 24.28 24.79 Hole Hole Hole 3.74 4.11 3.65 10.20 10.22 9.59 19 19 15 2.94 3.4 3.7 Original 109 N 106 N 30447 30447 30447 G6b G6b G6b H2 H2r-m H2r-wb 51.41 51.49 51.01 7.55 7.55 7.91 13.34 13.78 13.79 25.04 25.88 25.93 Groove Groove Groove 11.01 10.60 10.48 20 17 20 2.68 2.97 3.01 Original 86 N 106 N 30511 30511 30511 G9 G9 G9 H3 H3r-m H3r-wb 62.28 62.28 62.01 9.08 9.16 9.04 11.24 11.24 11.22 29.46 29.38 29.39 Groove Groove Groove 11.82 11.83 11.74 20 20 17 2.13 2.45 2.62 Original 57 N 60 N 34544 34544 34544 sf G21 sf G21 sf G21 H4 H4r-m H4r-wb 52.04 51.54 52.09 10.13 10.17 9.38 14.45 14.57 14.12 31.5 31.58 31.51 Hole Hole Hole 11.12 11.50 10.06 14 18 13 2.98 3.38 3.03 Original 69 N 82 N 3.81 3.81 3.66 CL ¼ cultural layer, G ¼ grave, sf ¼ sieve find, OL ¼ overall length, MBH ¼ maximum bow height, MBW ¼ maximum bow width, PL ¼ point length, GAPE ¼ hook gape, GA ¼ gripping angle. C. Olson et al. / Journal of Archaeological Science 35 (2008) 2813–2823 vertebrae (Olson et al., 2002; Olson and Walther, 2007). In the present study, the weight (g) of codfish caught at Ajvide was backcalculated from regression relationships developed between body length and sizes of preserved hard parts (otoliths and vertebrae). We extrapolated weights of cod from a modern-day length–weight relationship derived from Baltic Sea cod (Marine Research Laboratory, Lysekil, unpublished data), using the following formula: weight ¼ 0.0039621(TL)3.2375, N ¼ 40, R2 ¼ 0.99. TL (total length) is the maximum length of the cod with mouth closed and tail fins squeezed together (Casteel, 1976). The calculated weights supplied information about approximate weights of the cod captured at Ajvide, and were compared with the loads used in the strength test. 3.5. Statistics Oneway ANOVAs were calculated on all fishhook parameters with a known layer affiliation (layers were the categorical variable) to test whether there were any significant changes in design over time. Post hoc tests (Fisher’s Least Significant Difference) were made, and an alpha of 0.05 was used as the level of significance. In addition to this, the correlation (Pearson’s r) between all combinations of the fishhook parameters was calculated to see if they correlated proportionally within the hook assemblage regardless of hook size. These analyses were designed to test whether or not a homogeneity of design, indicating a specialized fishing technology was statistically detectable. To help understand the factors that promoted increased loadbearing strength, parameters from the eight replica hooks were examined by regressing load (newtons) on individual parameters, and a principal components analysis (PCA) confirmed which combination of variables had strong effect on load. From this, a multiple linear regression was developed and used to predict potential loads that could have been borne by original hooks; this last analysis was performed on intact hooks from layers 1 to 5. 4. Results 4.1. Fishhook morphology The manufacture of the hook replicas required a detailed examination of the original fishhooks. The four original fishhooks had been carved out of the flat side of the bone and followed the natural curvature of the bone surface. All hooks were carefully worked and have finely polished surfaces, although on close examination it is possible to detect parts of the hooks that carry traces of harsher working. Grooves for attaching the fishing line are more common than holes in general, but the frequency of holes decreases over time, from 45% in layer 4 þ 5 to 20% in layer 1 þ 2. Most hooks have barbed points (82 of 91 intact points). All bows of the hooks are more or less strengthened, which means the bone material in the bow is thicker than elsewhere in the hook. The hook morphology of all sizes of hooks is similar, except for a few among the smallest Table 2 Mean parameter values, Ajvide fishhooks Fishhook Complete þ incomplete originals Strength-tested replicas Parameter Number Mean s.d. Number Mean s.d. OL MBH MBW PL Gape GAa 134 211 120 91 61 53 53 9 13 28 10 21 8.5 2.1 2.4 5.7 1.8 8.7 8 8 8 8 8 8 54 9 13 28 11 17 4.9 1.3 1.3 3.0 0.8 2.5 Parameter mean values for complete þ incomplete, and strength-tested hooks. OL ¼ overall length, MBH ¼ maximum bow height, MBW ¼ maximum bow width, PL ¼ point length, GAPE ¼ hook gape, GA ¼ gripping angle. a One hook holding a GA of 10 is not included. 2817 Table 3 P-values for Ajvide fishhook measure categories from three chronological levels (i) Level (j) Level OL MBH MBW PL GAPE GA Layer 1 þ 2 Layer 3 Layer 4 þ 5 Layer 4 þ 5 0.043* 0.046* 0.949 2.720 117 119 0.880 0.429 0.352 0.471 182 184 0.717 0.570 0.777 0.168 99 101 0.532 0.091 0.209 1.497 74 76 0.869 0.277 0.290 0.716 46 48 0.322 0.358 0.910 0.453 40 42 Layer 3 F-value df n Statistic method ¼ Oneway ANOVA, post hoc Fischer LSD. * ¼ Significant at alphalevel 0.05. hooks, that have a design that differs from the rest of the hooks in the assemblage. Some of these are proportionally short, while others have shanks (perhaps reworked) that are almost as short as the point length. The overall lengths (OL) of 134 measurable (complete and incomplete) hooks range from 30 to 79 mm, with a mean value of 53 mm (s.d. ¼ 8.5). Point lengths greater than 50% of the overall length (OL) are most common. The point lengths of 91 measurable hooks range from 11 to 40 mm. The mean value is 28 mm (s.d. ¼ 5.7). The maximum bow heights (MBH) of 211 measurable hooks range from 6 to 20 mm. The mean value is 9 mm (s.d. ¼ 2.1), making 13–20% of the OL, and 69% of the maximum bow width. The maximum bow widths (MBW) of 120 measurable hooks range between 6 and 20 mm. The mean value is 13 mm (s.d. 2.4). The gapes, (i.e. the horizontal distance between the top of the point and the shank) of 61 measurable hooks range between 5 and 15 mm, with a mean value of 10 mm (s.d. 1.8). Except for one hook of a very special design, showing a gripping angle (GA) of 10 , the gripping angles of 53 measurable (intact) hooks vary between 10 and 63 . The mean value is 21 (s.d. 8.7). The mean values for the hooks selected for the strength test are – overall length: 50 mm, maximum bow height: 9 mm, maximum bow width: 13 mm, point length: 27 mm, hook gape: 10 mm, and gripping angle: 21 (Table 2). Statistical analyses revealed that only one design parameter changed over time. The overall length (OL) showed a significant difference (p ¼ 0.043) between layer 4 þ 5 and layer 1 þ 2, and also between layer 3 and layer 1 þ 2. (Table 3). The hooks were larger in layers 3–5, than in layer 1 þ 2. Moreover, there are tendencies that point length, maximum bow width, gape and gripping angle were slightly larger in the upper, younger layers, even if these changes are not statistically significant. The maximum bow height proved to be the most static measure, and showed only very small differences among the layers (Table 4). The correlation between the different measure categories of the fishhooks, calculated from the complete hooks in the hook assemblage, was statistically significant for most of the parameters, only MBH-gape, MBH-GA, MBW-GA and PL-GA did not exhibit significant correlations (Table 5). That the MBH did not correlate with the gape and gripping angle seems logical, since the value of the bow height cannot affect the value of the gape and gripping angle, while the lack of correlation between the MBW-gripping angle and the point length-gripping angle is more difficult to Table 4 Parameter mean values of fishhooks from three chronological levels Unit Parameter Layer 4 þ 5 Mean s.d. Mean s.d. Mean s.d. mm mm mm mm mm Degrees OL MBH MBW PL GAPE GA 54.55 9.46 12.61 26.12 9.24 15 8.35 2.39 2.57 7.37 2.01 12.94 54.43 9.1 12.78 28.33 9.98 16 7.85 1.86 2.33 5.69 1.6 6.7 50.72 9.15 12.98 29.3 10.08 20 8.53 1.98 2.32 5.01 2.03 10.59 Calculated from 577 measures. Layer 3 Layer 1 þ 2 Total number 120 185 102 77 49 42 2818 C. Olson et al. / Journal of Archaeological Science 35 (2008) 2813–2823 Table 5 Ajvide complete fishhooks Parameter r p Significance OL–PL OL–MBH OL–MBW OL–gape OL–GA MBH–MBW MBH–PL MBH–gape MBH–GA MBW–PL MBW–gape MBW–GA PL–gape PL–GA Gape–GA 0.662 0.571 0.323 0.366 0.275 0.402 0.578 0.244 0.128 0.658 0.472 0.140 0.553 0.023 0.535 0.000 0.000 0.017 0.006 0.044 0.003 0.000 0.076 0.356 0.000 0.000 0.314 0.000 0.871 0.000 <0.01 <0.01 <0.05 <0.01 <0.05 <0.01 <0.01 <0.01 <0.01 <0.01 Fig. 6. Predicted weight bearing loads (newtons, means þ s.e.) of 46 intact fishhooks from Ajvide, by chronological layer. N for each layer is written on the column. Eight intact hooks from unknown time horizons are not included in the figure. <0.01 The correlation (Pearson’s r) between different parameter categories (n ¼ 54). explain, since these measures are closely related and ought to affect one another. Although there is a variation in the size between the hooks, the rather strong correlation between most of the parameters indicates that the hooks were morphologically proportional. 4.2. Strength test and breakage of hook replicas The test-hooks ruptured either in the bow or close to that (Fig. 4). The ultimate loads range from 57 to 109 N (Table 1). The strength of the hooks is enough to bear 6–10 kg dead weight in air. The ultimate load is due to three groups of factors: the geometry, the material and the loading. The possible geometry may be restricted by factors that influence the fishing efficiency. Bone is an anisotropic material with a strength that varies with the loading direction. The precise loading position is unknown in advance. Thus it is impossible for the maker of the hooks to choose the optimal geometry for best strength in all individual cases. A compromise based on experience is necessary. The design of the tested hooks did not seem to affect the rupture areas as much as the difference in raw-material. All the hooks made of moose bone ruptured in the center of the bow, while only one made of wild boar had a center bow fracture, and the remaining three ruptured in connection to the bow area. The raw-material comparison also showed that three of the four fishhook types (H2, H3 and H4) had higher Newton values when made of wild boar bone. Only hook type H1 manufactured of moose bone withstood somewhat more load than H1 made of wild boar bone (Table 1). 4.3. Predicted loads of hooks from Ajvide Regression and PCA revealed that three parameters correlated strongly with load-bearing strength in the replica hooks. Two of these parameters, hook gape and point length, were inversely related to load, whereas hook weight was positively correlated. A multiple regression with load as the dependent variable and these three parameters as independent variables yielded the following predictive equation: 38.8–173.8 N (3.9–17.7 kg). ANOVA tests of predicted load by burial layer showed similarity of loads across time, with the possible exception of layer 5 which had somewhat higher mean loads (Fig. 6). However, given that only two intact hooks were found in layer 5, it is difficult to draw conclusions. 4.4. Fracture pattern of hooks from Ajvide Only 14% of the Ajvide fishhooks we studied are complete (intact), and the remaining 86% (incomplete hooks) have one or two fractures. There are 10 locations of fractures registered: four single fracture locations (head, shank, bow and point) and six double fracture locations (head þ bow, head þ shank, head þ point, shank þ bow, shank þ point, and bow þ point). The surface of the rupture areas of the replica hooks and the originals is rather similar, but the original hooks tend to show a more sloped or angled fracture area. Single fractures are most common and make 77% of the total amount of fractures. Shank fractures predominate (33%) somewhat over bow fractures (32%), both in settlement and grave contexts. Point fractures made 10% and head fractures only 2% of the total number of fractures. Double fractures were registered on 23% of the hooks, of which there is a slight dominance of the combination of shank þ point fracture (Table 6). Fresh breaks were noted on 39% and dry breaks on 20% of the incomplete hooks. The remaining fractures (indeterminate 41%) were difficult to assess, either due to the state of preservation or due to the ambiguity of the ruptured surface (Table 6). The fracture texture of the fresh breaks proved to be rather smooth, like those of the tested hooks. The edges of the fresh breaks were sharp. The fracture surfaces of the dry breaks were rough and the edges in general not as sharp as the fresh breaks. Like the tested hooks, bow fractures dominate among the original hooks with fresh breaks. Shank fractures are the most common type of fracture in the group of dry breaks, and more than 50% of these fractures were caused by post-depositional processes. 4.5. Cod sizes The reconstructed weights, obtained from 486 otoliths and vertebrae of the Stone Age cod from Ajvide (Fig. 7) show that 40% of Load ðNÞ ¼ 276:8636 3:0823 ðpoint lengthÞ 12:5851 ðgapeÞ þ 9:3323 ðweightÞ Table 6 Number of fractures per breakage category from all contexts Breakage HF HF&BF HF&SF HF&PF SF R2 ¼ 0:90; F3:4 ¼ 12:33; p < 0:05: We used this equation to predict the loads potentially borne by 46 intact hooks found at Ajvide. Predicted loads had a normal distribution with a mean of 96.6 N (9.85 kg) 26.1 (s.d.), range Fresh Dry Indet Sum % 2 1 5 8 2 3 3 1 2 6 8 2 1 6 7 2 SF&BF SF&PF BF 36 26 4 46 16 108 20 33 6 5 23 28 9 BF&PF PF Sum % 75 6 9 21 3 102 12 31 4 13 138 42 10 52 16 8 137 42 31 327 10 100 HF ¼ head fracture, SF ¼ shank fracture, BF ¼ bow fracture, PF ¼ point fracture. C. Olson et al. / Journal of Archaeological Science 35 (2008) 2813–2823 2819 taken place from boats in deep waters which large fish occupy. They also noted a change in design where the older hooks from the Mesolithic usually were U-shaped, while those of later periods were more V-shaped and had longer shanks and a differently designed upper shank where the line was attached. In Norway, Stone Age fishhooks confirm a variety of fisheries by the many different sizes and forms of hooks found there. A characteristic of the Norwegian hooks is, that the design of is often of such a distinctive feature that the hooks can be connected to a certain site (Bakkevig, 2003). Tegermark (1997) made a study of the distribution of fishhooks within the Ajvide site area (n ¼ 261 available at the time). He concluded that 9% of the hooks derived from grave contexts and 91% from the cultural layer. These figures remain the same after our recent study. He also made a morphological survey of the hooks and concluded from their size and form that they would have been suitable for line fishing for cod. However, all previous studies lack a detailed investigation of the morphology and use of the fishhooks. 5.1. Morphology Fig. 7. Cod weights predicted from vertebrae and otoliths from Ajvide cod (n ¼ 486). the captured cod weighed less than 2 kg, 52% weighed 2–3.9 kg, and 8% weighed 4–7 kg. Consequently, the fishhooks from Ajvide were well dimensioned to carry the sizes of the captured fishes, when one considers that live weight puts more load on a hook than dead weight. No statistically significant chronological change in cod size was observed. The largest and the smallest individuals were found in the oldest level. However, even if not statistically observable, there is a tendency of an increase of cod mean size over time. The sizes of the fishhooks show that both the largest and the smallest ones were found at the transgression layer (layer 3), but, the highest hook mean size was found at the oldest level where also the largest cod were found (Table 7). Interestingly, a strong correlation between mean, minimum, and maximum hook and cod lengths is apparent, with cod lengths being approximately 10 times that of hook overall lengths (cod length ¼ 10.37 cm, hook length ¼ 0.59 cm; R2 ¼ 0.93). Frequency distributions of all cod and hook lengths are strikingly similar (Fig. 8), offset by a factor of 10, suggesting a correspondence. 5. Discussion When comparing the Ajvide fishhooks with contemporaneous fishhooks from other Pitted Ware sites on Gotland (e.g. Janzon, 1974; Schnittger and Rydh, 1940; Stenberger, 1943), and Öland in the Baltic Sea (Schulze, 1973), the resemblance in design is obvious. Stenberger (1943) and Janzon (1974) examined Gotlandic fishhooks and their use. Janzon also classified hooks based on the morphology of the line attachment area and the point. Lekholm (1951) studied prehistoric hooks from Scania in southern Sweden and discussed their field of application by the way they were constructed. Hernek and Jonsson (2003) dealt with fishhooks from the southwestern coast of Sweden. They concluded that hook and line fishing had The similarities in measurement values for both complete and incomplete hooks of the fishhook assemblage from Ajvide support their homogeneity. The average Ajvide fishhook has a length of 53 mm, bow height of 9 mm, bow width of 13 mm, point length of 28 mm, gape of 10 mm and a gripping angle of 17. Grooves are more common than holes for attaching the line. Grooves dominate in all periods, and the occurrence increases over time while the frequency of holes decreases. Since head fractures are so few, the reason for the decrease is probably not related to stress, but perhaps to the manufacturing process. Grooves may have been less time consuming to make, but had apparently the same functional qualities as holes. The strengthened bows occur in all periods, probably due to their high exposure to stress, which is shown in the breakage study and the strength test. The design of a fishhook is of importance for its function, and the morphology can to a certain extent reveal its field of application. An interesting detail of the Ajvide fishhook design, from a manufacturing view, is that the inner bow of many hooks from Ajvide including hook H2, seems to have been bored, instead of carved. Judging by the surface of the walls inside the holes, a high revolution drilling-tool must have been used, perhaps a bow-drill. This was a procedure that would have facilitated the manufacturing process considerably. This method, known within prehistoric technology (e.g. Chaussonet, 1995; Craig, 1967; Johansson, 1993), was also used on the Stone Age Scanian Fishhooks described by Lekholm (1951). 5.2. Fishing technology Considering the location of the Ajvide site, on the seashore, facing both a lagoon with shallow water and the open sea, various kinds of capturing methods must have been used for the different fish species inhabiting these waters. From the recovered fishbone material, it can be anticipated that fish species like herring, flatfish, pike, perch, salmonids and eel were mostly caught with stationary Table 7 Comparisons between cod and fishhook weights and cod total length (TL) and fishhook overall length (OL) Parameter Cod w (g) Hook w (g) Cod (cm) TL Hook (mm) OL Layer 4 þ 5 Layer 3 Layer 1 þ 2 Total number Min Mean Max Min Mean Max Min Mean Max 218.1 0.4 29.1 36.2 1952.9 2.0 54.7 54.6 6728.9 4.1 84.0 76.7 509.2 0.4 37.9 30.5 1892.5 2.6 55.1 54.4 4576.6 5.6 74.6 79.0 307.8 0.9 32.4 34.0 2201.9 2.2 56.6 51.0 5130.7 2.9 75.0 70.3 Minimum, mean and maximum values from three chronological levels. 366 46 486 120 2820 C. Olson et al. / Journal of Archaeological Science 35 (2008) 2813–2823 Fig. 8. Cod and hook size distributions (n ¼ 477 cod, 134 hooks). traps, nets and sometimes leisters. However, for cod, which seasonally stay in coastal waters, but not often in the immediate vicinity of the shore, methods such as hook and line fishing would be more efficient. The large amount of cod bones and fishhooks at Ajvide certainly point to the importance of this fishing method. The fact that the size of the hooks decreased slightly over time reflects some kind of change. Whether it was due to an improved fishing technology, a more specialized fishery or other causes, this indicates a change in the subsistence economy. A further indication of such a change is found in the mammal remains from the same time span at Ajvide. They show that seals decrease, while the amount of wild boar increases over time (Lindqvist, 1997; Rowley-Conwy and Storå, 1997; Storå, 2001). At Ajvide marine, diadromous (migratory) and fresh-water fish species were consumed, although cod and herring were the two dominant species. This makes Ajvide somewhat different from coastal sites having access only to marine species, which may to a certain extent have affected Ajvide’s fishhook material, in terms of size, form, and function. There is always the possibility that the hooks from Ajvide occasionally were used for species other than cod, such as cyprinids, perch, pike and salmon. However, using stationary fishing equipment for fishes active in shallow lagoon waters, creeks or streams were presumably more productive and energy saving than using a hook and line (e.g. McQuade and O’Donnell, 2007; Pedersen, 1995). Younger codfish may occasionally have been caught in fish traps or nets in shallower waters along the coastline, as described by Bødker-Enghoff (1986, 1989, 1994, 1995) in her studies of Danish Mesolithic fishbone materials. Pickard and Bonsall (2004) also studying Mesolithic fishing in Europe, only found evidence for deep-sea fishing in regions where deep water was found close to shore. They further comment that according to ethnographical data, offshore subsistence fisheries were practiced only where there was limited availability of terrestrial resources, a fact that agrees well with the restricted land mammal fauna on Gotland. Based on the size and form of the majority of the fishhooks from Ajvide, and the estimated sizes of the recovered cod from the site, we believe that the hooks were mainly intended for fishing gadids in deep waters some distance from the settlement. Our definition of hook and line fishing in this study is: a fishery with a hook attached to a hand line, hanging more or less vertically in the water down from the boat. We suggest the boats were some kind of dug out canoes, perhaps of a stable outrigger type like those used in the South Pacific (e.g. Österholm, 1997). We assume the fishery took place at depths of approximately 50 m or more, about 5–6 km away from the shore. The penetration power of the hook is closely related to its design. A hook with a short shank and a broad gape needs greater power to penetrate the mouth of the fish than a long shank with a small gape. For hook and line fishing bent hooks can penetrate the fish more efficiently than a straight hook (Broman, 1979). This fits in well with the design of the Ajvide hooks which are long, have rather small gapes, and mostly a point length area that is more or less bent in relation to the shank. Consequently, we conclude that the Ajvide hooks were designed for the use of low penetration power during hook and line fishing. There are no remains of the lines that were used together with the hooks from Ajvide. The strength, elasticity, and material of the fishing lines that once were attached to the fishhooks are unknown variables. However, prehistoric finds and ethnographic studies show that the manufacturing of lines, strings and ropes was common in low-technology communities (e.g. Andersen, 1985; Murdoch, 1892; Stewart, 1982). Lines could be produced both from animal and vegetable materials. The bast of linden-tree (Tilia spp.) has, for example, been used for making lines in temperate areas, like the uniquely preserved line from the Mesolithic site Tybrind Vig in Denmark (Andersen, 1985). In arctic and sub-arctic areas, with poor access to deciduous forests, long, braided sinews (sometimes more than 20 m) were used for purposes from retrieving lines for harpoons to reinforcements on bows (Callahan, 1991; Murdoch, 1892). When sinews are braided, they withstand the dissolving effect of water. Another vegetable material besides linden bast that could have been used at Scandinavian latitudes is nettle (Urtica dioica) fibre, which was frequently used by the indigenous fishers at the northwest coast of USA and Canada (e.g. Stewart, 1982). The different attachment areas for the lines are considered to have had little or no effect on the fishing techniques, while the strength of both the hook and the line was a determining factor. Salls (1989) tested the strength of fibre plant lines similar to those used in prehistoric California to see if they were strong enough to carry the sizes of different fish species found at excavations in southern California. His results revealed that the native fibre plant lines probably would have been limited to catching fish weighing 10–20 pounds (5–9 kg), which agreed well with the archaeological record showing that native fishermen mainly utilized these fish sizes in the nearshore habitats. Unaware of what lines were used at Ajvide, one could argue that the lines were the weak point, and the cause of the lack of large cod. However, the finds of harpoons at Ajvide indicate they must have had lines strong enough for harpooning all the seals found at the site. Therefore, it is not likely they lacked the knowledge how to make fishing lines sufficiently strong for their purposes, but, despite this observation, we cannot totally exclude the possibility that the lines sometimes broke during fishing. Our study demonstrates that fishhooks from Ajvide had the capacity to hold very large fishes with line fishing. Even if there is no evidence yet of cod body lengths over 900 mm (comparable to a weight of 6–7 kg) in the osteological material from Ajvide, it is still possible that these sizes of cod were caught at times. One cannot disregard that the fishhooks were, in their ultimate strength, made to handle the occasional large codfish. One must also consider that a fishhook in everyday use is exposed to a variety of stress factors that cannot be simulated in a test-bench in air. Such stress could be caused by the movements of the fish in relation to the boat’s vertical and horizontal movements, the elasticity of the line, the way the point of the hook penetrates the mouth or body of the fish, and the handling of the fish when taken out of the water. Therefore, it needs to be considered that the hooks may have ruptured at lower loads when exposed to higher stress in water. We assume the fishers at Ajvide were well aware of all the stress affecting factors, and had them in mind when they manufactured the hooks, which could explain the high weight bearing capacity proven by the strength test. The somewhat stronger fishhooks made of wild boar bone compared to those made of moose bone, may indicate a difference C. Olson et al. / Journal of Archaeological Science 35 (2008) 2813–2823 in bone strength between the hooks, even if not statistically demonstrated. The hooks made of moose bone, all consisted of compact bone, while those made of wild boar bone included some cancellous bone since the porous upper part of the wild boar tibia shaft had to be used. The compact area of the tibia was not large enough for cutting out a complete hook. This however, did not affect the outcome of the strength test as all hooks broke in or close to the bow area. We certainly believe that the Stone Age craftsmen were well aware of how different bone tissue affected the function of their tools and that the manufacturing followed well developed strategies. The availability of raw-material at Ajvide was probably good, and it is not likely that fishhooks were made from bone of dubious quality. The use of fracture analysis methods for unworked bone elements (Johnson, 1985; Outram, 2001, 2002) proved to be applicable to bone artifacts, in this case with bone fishhooks. The fracture analyses showed that the fresh fractures of the original fishhooks were more sloped than the fresh fractures of the replicas, presumably owing to the difference between real fishing and static load stress. Both replicas and originals showed a smooth texture of the fracture surfaces. The breakage pattern gives an indication about a number of weak parts of the hooks. The bow (n ¼ 75) and shank (n ¼ 36) hold the highest record of fresh fractures while the point (n ¼ 13) and head or line attachment area (n ¼ 2) were not as often fractured. The variety of fractures displayed on the tested wild boar hooks agrees well with the fractures occurring on the original fishhooks from Ajvide, which were also primarily made of wild boar bone. The fact that the rupture areas are mostly in or close to the bow, explains why this exposed area of the hook was almost always strengthened in the manufacturing process. Although the bows were strengthened, they show the highest amount of fresh breaks, indicating this was still the weakest part of the hook. However, to further strengthen the bow would perhaps have affected the functional qualities of the hook. Allen (1996) discussed functional and stylistic traits on fishhook assemblages from a number of island sites in the South Pacific, where she also concluded the bend was the weakest part of one-piece hooks. In the South Pacific, large one-piece hooks decline while two-piece hooks increase in abundance over time. The explanation to this may be the functional superiority of the two-piece hook, where ‘‘the weak bend region is effectively ‘broken’ and then reinforced with a flexible lashing’’ (Kirch ref. to in Allen, 1996). As earlier indicated there are no traces of two-piece hooks at Ajvide, leaving an eventual similar development in fishing technology unknown. There is a dominance of bow fractures below the transgression layer (layer 3). Within the transgression layer shank fractures and bow fractures show almost equal amounts, while shank fractures clearly dominate above the transgression layer. It is also obvious that the incidence of both shank and point fractures increases over time (Fig. 9). A high post-depositional destruction is indicated by the increased occurrence of dry and indeterminate breaks within 2821 Fig. 10. Breakage category per chronological level (n ¼ 279). the youngest layers (Fig. 10). A similar indication of fracturing is described for wild boar (Outram, 2001, 2002) and bird bones (Mannermaa and Storå, 2006) from the fauna at Ajvide, where dry fractures increase and fresh fractures decrease towards the upper layers of the stratigraphy. It is obvious that post-depositional breakage (dry breaks) may mask the original fracture patterns of the hooks. However, since we, in this case, carried out the strength test of the replicas and the breakage study, we could distinguish the fresh breaks from the dry breaks, and reveal the bow as the weakest part of the fishhook, in spite of the higher occurrence of shank fractures among the archaeological hooks. 6. Conclusions The Stone Age site of Ajvide on Gotland, Sweden yielded a remarkable collection of fishhooks indicating the importance of fishing and fishhook production. The homogeneity in size and form of the fishhooks points to a planned manufacturing. Hook morphology remained more or less constant over time, although hooks with grooves rather than holes for attaching the lines became more frequent, while the overall length of the hooks decreased somewhat. The strength test of the replicas proved no significant difference in strength between the two raw-materials, moose and wild boar bone. However, the ultimate loads at rupture were in three cases of four somewhat higher for the hooks made of wild boar bone. The fracture areas and surface textures of the replicas were similar to those of the original incomplete hooks with fresh breaks. The breakage study of the strength-tested replicas and original incomplete hooks made it possible to distinguish fresh breaks from dry breaks. The breakage pattern showed that bow fractures were the most common among fresh breaks. Shank fractures were most frequent, but were dominated by dry breaks, and therefore more related to post-depositional processes. The strength test and the morphology study supported our hypothesis that line fishing for cod, in deep waters, from boats, some distance from the site, was the field of application for the majority of the fishhooks from Ajvide. Hook sizes and cod sizes appear to be related, also suggesting that cod was the main target of the Ajvide hook and line fishery. Acknowledgements Fig. 9. Percentage single fractures (fresh and dry) per chronological level (n ¼ 189). HF ¼ head fracture, SF ¼ shank fracture, BF ¼ bow fracture, PF ¼ point fracture. We thank Hans Öberg, Senior Research Engineer at the Department of Solid Mechanics, KTH Royal Institute of Technology in 2822 C. Olson et al. / Journal of Archaeological Science 35 (2008) 2813–2823 Stockholm, who carried out the strength test and documented the results from this test. Without his kind help and interest in Stone Age technology, we doubt this test could have taken place. We are also most grateful to Dr. Inger Österholm, now sadly deceased, Professor Göran Burenhult and Johan Norderäng at the University of Gotland for access to the fantastic fishhook assemblage from Ajvide and generous permission to borrow original fishhooks for the manufacturing of replicas for this study. We thank Dr. Jan Storå for critical comments, fruitful discussions and for continuously commenting on the manuscript. Thanks also to Dr. Sabine Sten at the University of Gotland and fisherman John Nordberg from Fårö, Gotland, who made a fishing experiment possible by using a replica from the Ajvide assemblage for fishing cod (and caught one!). Carina Olson also wishes to acknowledge the support of her senior supervisor professor Ebba During who was highly involved in the fishhook study, but sadly passed away before it was completed. Finally, we thank the Helge Ax:son-Johnson Foundation for financing the manufacture of the replicas and the strength test, the Berit Wallenberg Foundation, the Royal Gustav Adolf Academy and the Hildebrand Foundation for financing the study of the Ajvide fishhook assemblage. KL was supported by National Science Foundation grant DEB-0238121. References Aaris-Sørensen, K., 1978. Knoglematerialet fra den mellemneolitiske boplads ved Korsnäs. With English Summary. Riksantikvarieämbetet och Statens Historiska Museer Rapport 1978, 8. Stockholm. Allen, M.S., 1996. Style and function in East Polynesian fish-hooks. Antiquity 70, 97– 116. Andersen, S.H., 1985. Tybrind vig. A preliminary report on a submerged ertebølle settlement on the west coast of Fyn. Journal of Danish Archaeology 4, 52–69. Bakkevig, S., 2003. På fisketur med steinalderkroker. Fra haug ok heidni. Tidskrift for Rogalands Arkeologiske forening, no 1–2, pp. 9–19. Berglund, B.E., 1999. Odlingslandskapets framväxt i Norden. In: Burenhult, G. (Ed.), Arkeologi i Norden 1. Natur och Kultur, Sweden, pp. 250–255. Björck, N., 1997. New perspectives on the Pitted Ware culture in Northern Sweden. In: Burström, M., Carlsson, A. (Eds.), Current Swedish Archaeology, vol. 5, pp. 19–39. Broman, A., 1979. Modernt mete 1. Förlagsaktiebolaget Västra Sverige, Kungsbacka. Burenhult, G. (Ed.), 1997a. Ajvide och den moderna arkeologin. Natur och Kultur, Stockholm, Sweden. Burenhult, G. (Ed.), 1997b. Thesis and Papers in North-European Archaeology 13a. Remote sensing. Applied techniques for the study of cultural resources and localization, identification and documentation of sub-surface prehistoric remains in Swedish archaeology, vol. I. Department of Archaeology, Stockholm University, Sweden. Burenhult, G. (Ed.), 1999. Arkeologi i Norden 1. Natur och Kultur, Stockholm, Sweden. Burenhult, G. (Ed.), 2002. Thesis and Papers in North-European Archaeology 13b. Remote sensing. Applied techniques for the study of cultural resources and localization, identification and documentation of sub-surface prehistoric remains in Swedish archaeology, vol. II. Department of Archaeology, Stockholm University, Sweden. Bødker-Enghoff, I., 1983. Size distribution of cod (Gadus morhua L.) and whiting (Merlangus merlangus L) (Pisces Gadidae) from a Mesolithic Settlement av Vedbæk, North Zealand, Denmark. Videnskabelige Meddelelser fra dansk naturhistorisk forening 144, 83–97. Bødker-Enghoff, I., 1986. Freshwater fishing from a sea-coast settlement – the Ertebølle locus classicus Revisited. Journal of Danish Archaeology 5, 62–76. Bødker-Enghoff, I., 1989. Fishing from the stone age settlement norsminde. Journal of Danish Archaelogy 8, 41–50. Bødker-Enghoff, I., 1994. Fishing in Denmark during the Ertebølle period. International Journal of Osteoarchaeology 4, 65–96. Bødker-Enghoff, I., 1995. Fishing in Denmark during the Mesolithic period. In: Fischer, A. (Ed.), Man & Sea in the Mesolithic. Coastal Settlement Above and Below Sea Level. Oxbow Monograph, vol. 53, pp. 67–75. Callahan, E., 1991. Arctic archery. Bulletin of Primitive Technology 1 (2), 47–54. Casteel, R.W., 1976. Fish Remains in Archaeology and Paleo-environmental Studies. Academic Press, London, New York, San Fransisco. Chaussonet, V., 1995. Crossroads Alaska – Native Cultures of Alaska and Siberia. Smithsonian Institution Press, Washington. Craig, A.K., 1967. Some observations on the manufacture and utilization of fishhooks among Indians of North America. The Florida Anthropologist XX (1–2), 79–88. Edenmo, R., Larsson, M., Nordqvist, B., Olsson, E., 1997. Gropkeramikerna – fanns de? Materiell kultur och ideologisk förändring. In: Larsson, M., Olsson, E. (Eds.), Regionalt och interregionalt. Stenåldersundersökningar i Syd- och Mellansverige. Riksantikvarieämbetet arkeologiska undersökningar Skrifter, vol. 23, pp. 135–213. Ekman, J., 1974. Djurbensmaterialet från stenålderslokalen Ire, Hangvar sn, Gotland. In: Janzon, G.O. (Ed.), Gotlands mellanneolitiska gravar, Acta Universitatis Stockholmiensis. Studies in North European Archaeology, vol. 6. Stockholm University, Sweden, pp. 212–246. Ericson, P.G.P., 1989. Säl och säljakt i Östersjöområdet under stenåldern. In: Iregren, E., Liljekvist, R. (Eds.), Faunahistoriska studier tillägnade Johannes Lepiksaar. Symposium 26 maj 1988. Report Series No. 3. Institute of Archaeology, Lund University, Sweden, pp. 57–64. Eriksson, G., 2003. Norm and difference. Stone age dietary practice in the Baltic region. In: Thesis and Papers in Scientific Archaeology 5. Stockholm University, Sweden. Färjare, A., Olsson, E., 2000. Lill-Mosjön – boplatslämningen och fångstgropar från neolitikum, äldre järnålder och historiska tid. UV-mitt dokumentation av fältrbetsfasen 2000:5. Arkeologisk undersökning. Riksantikvarieämbetet, Stockholm, Sweden. Hedell, L., unpublished data. Osteological analysis of faunal remains from Åloppe. Antiquarian-Topographical Archives, Stockholm, Sweden. Hernek, R., Jonsson, L., 2003. Förhistoriskt fiske I Bohuslän – främst under stenålder och bronsålder. BOHUSLÄN Årsbok 2003, 83–102. Janzon, G.O., 1974. Gotlands mellanneolitiska gravar. In: Acta Universitatis Stockholmiensis. Studies in North European Archaeology, vol. 6. Stockholm University, Sweden. Johansson, T., 1993. Forntida teknik. ICA Förlaget, Västerås, Sweden. Johnson, E., 1985. Current developments in bone technology. In: Schiffer, M.B. (Ed.), Advances in Archaeological Method and Theory, vol. 8. Academic Press, New York, pp. 157–235. Lepiksaar, J., 1974. Djurrester från den mellanneolitiska (gropkeramiska) boplatsen vid Äs, Romfartuna sn, Västmanland. In: Löfstrand, L. (Ed.), Yngre Stenålderns Kustboplatser. Undersökningarna vid Äs och Studier i den Gropkeramiska Kulturens Kronologi och Ekologi. Aun, 1. Uppsala University, Sweden, pp. 140–156. Lepiksaar, J., Heinrich, D., 1977. Berichte über die Austrabungen in Haithabu. In: Untersuchen an Fischresten aus der fruhmittelalterlichen Siedlung Haithabu. Bericht, vol. 10. Karl Wachholtz Verlag, Neumünster, Germany. Lekholm, C.G., 1951. A technical study of some Scanian Bone Fish-hooks. Kungl. Humanistiska Vetenskapssamfundets i Lund Årsberättelse 1950–1951, IV, Meddelanden från Lunds Historiska Museium. Lund University, Sweden, pp. 245–265. Lidén, K., 1995. Prehistoric diet transitions. A dietary perspective on Swedish hunter-gatherer and Neolithic populations. An analysis of stable isotopes and trace elements. In: Thesis and Papers in Scientific Archaeology 1. Stockholm University, Sweden. Lindqvist, C., 1997. About the importance of fine mesh sieving, stratigraphical and spatial studies for the interpretation of the faunal remains at Ajvide, Eksta parish, and other Neolithic dwelling sites on Gotland. In: Burenhult, G. (Ed.), Remote Sensing, vol. I. Thesis and Papers in North-European Archaeology 13:2. Stockholm University, Sweden, pp. 91–112. Lindqvist, C., Possnert, G., 1997. The subsistence economy and diet at Jakobs/Ajvide and Stora Förvar. Eksta parish and other prehistoric dwelling and burial sites on Gotland in long-term perspective. In: Burenhult, G. (Ed.), Thesis and Papers in North-European Archaeology 13a. Remote sensing, vol. I. Stockholm University, Sweden, pp. 29–90. Lithberg, N., 1914. Gotlands stenålder. Jacob Bagges Söners AB, Stockholm, Sweden. Malmer, M.P., 2002. The Neolithic of South Sweden. TRB, GRK and STR. The Royal Swedish Academy of Letters History and Antiquities, Stockholm, Sweden. Mannermaa, K., Storå, J., 2006. Stone age exploitation of birds on the Island of Gotland, Baltic Sea: a taphonomic study of the Avifauna on the Neolithic Site of Ajvide. International Journal of Osteoarchaeology 16, 429–452. McQuade, M., O’Donnell, L., 2007. Late Mesolithic fish traps from the Liffey estuary, Dublin, Ireland. Antiquity 81, 569–584. Morales, A., Rosenlund, K., 1979. Fish Bone Measurements. An Attempt to Standardize the Measuring of Fish Bones from Archaeological Sites. Steenstrupa, Copenhagen, Denmark. Murdoch, J., 1892. Ethnological results of the point barrow expedition. In: 9th Annual Report of the Bureau of American Ethnology for the years 1887–1888, Washington, DC, USA. Nihlén, J., 1927. Gotlands stenåldersboplatser. KVHAA 36:3. Stockholm, Sweden. Olson, C., Walther, Y., 2007. Neolithic cod (Gadus morhua) and herring (Clupea harengus) fisheries in the Baltic Sea, in the light of fine-mesh sieving: a comparative study of subfossil fishbone from the late Stone Age sites at Ajvide, Gotland, Sweden and Jettböle, Åland, Finland. Environmental Archaeology 12, 175–185. Olson, C., Limburg, K., Patterson, W.P., Elfman, M., Kristiansson, P., Ehrensberg, S., 2002. Reconstruction of fisheries and marine environment. Preliminary studies of hard parts of Codfish (Gadus morhua) from Ajvide, Gotland, Sweden. In: Burenhult, G. (Ed.), Thesis and Papers in North-European Archaeology 13b. Remote sensing, vol. II. Stockholm University, Sweden, pp. 375–385. Österholm, I., 1989. Bosättningsmönstret på Gotland under stenåldern. In: Thesis and Papers in Archaeology 3. Stockholm University, Sweden. Österholm, S., 1997. Forntidens båtar – ett försök med expreimentell arkeologi. In: Burenhult, G. (Ed.), Ajvide och den moderna arkeologin. Natur och Kultur, Sweden, pp. 161–171. Outram, A.K., 2001. A new approach to identifying bone marrow and grease exploitation: why the indeterminate fragments should not be ignored. Journal of Archaeological Science 28, 401–410. Outram, A.K., 2002. Distinguishing bone fat exploitation from other taphonomic processes: what caused the high level of bone fragmentation at the middle Neolithic Site of Ajvide, Gotland? In: 9th ICAZ Conference, Durham, pp. 32–43. C. Olson et al. / Journal of Archaeological Science 35 (2008) 2813–2823 Pedersen, L., 1995. 7000 Years of fishing: stationary fishing structures in the mesolithic and afterwards. In: Fischer, A. (Ed.), Man & Sea in the Mesolithic Coastal Settlement Above and Below Present Sea Level. Oxbow Monograph, vol. 53, pp. 75–86. Pickard, C., Bonsall, C., 2004. Deep-sea fishing in the European Mesolithic: fact or fantasy? European Journal of Archaeology 7 (3), 273–290. Rowley-Conwy, P., Storå, J., 1997. Pitted Ware seals and pigs from Ajvide, Gotland: methods of study and first results. In: Burenhult, G. (Ed.), Thesis and Papers in North-European Archaeology 13a. Remote sensing, vol. I. Stockholm University, pp. 113–127. Salls, R.A., 1989. To catch a fish: some limitations in Southern California with special reference to native plant fiber fishing line. Journal of Ethnobiology 9 (2), 173– 199. Schnittger, B., Rydh, H., 1940. Grottan Stora Förvar på Stora Karlsö. Kungliga Vitterhets Historie och Antikvitets Akademien. W&W, Stockholm, Sweden. Stenberger, M., 1943. Das Grabfeld von Västerbjers auf Gotland. Kungl. Vitterhets Historie och Antikvitets Akademien, Stockholm, Sweden. 2823 Stenbäck, N., 2003. Människorna vid havet. Platser och keramik på ålandsöarna perioden 3500–2000 f.Kr. In: Stockholm Studies in Archaeology, vol. 28. Stockholm University, Sweden. Stewart, H., 1982. Indian Fishing. Early Methods on the Northwest Coast. University of Washington Press, Seattle, USA. Storå, J., 2001. Reading Bones. Stone Age Hunters and Seals in the Baltic. In: Stockholm Studies in Archaeology, vol. 21. Stockholm University, Sweden. Storå, J., 2002. Seal hunting on Ajvide. A taphonomic study of seal remains from a Pitted Ware culture site on Gotland. In: Burenhult, G. (Ed.), Thesis and Papers in North-European Archaeology 13b. Remote sensing, vol. II, pp. 387–428. Tegermark, L., 1997. Fiskekrokar. En studie av ett föremåls betydelse i ett mellanneolitiskt samhälle på Gotland. Seminar paper, Gotland University, Sweden. Welinder, S., Pedersen, E., Widgren, M., 1998. Det svenska jordbrukets historia. Jordbrukets första femtusen år. 4000 f.Kr.–1000 e.Kr. Natur och Kultur, Sweden. Wheeler, A., Jones, A.K.G., 1989. Fishes. Cambridge Manuals in Archaeology. Cambridge University Press.
© Copyright 2026 Paperzz