Introduction

Several researchers have demonstrated the usefulness of technological studies set within the chaîne opératoire (CO) framework for analysing Mesolithic bone industries. Different technical procedures generate specific types of production debris. Bone debitage, in combination with experimental replication, facilitate the identification of the different steps of the manufacture process, or chaîne opératoires and distinguish between different production concepts which are prerequisites for archaeological interpretations (David, 1999; 2007). The ultimate goals of CO-analysis are to explore the implications of technology for the social organization and reproduction of crafts in prehistoric societies. By employing the CO-approach systematically, we gain insight into the technical practice conducted at the site level. From this, we can gradually build knowledge of the regional and chronological distribution of raw materials and techniques (Bergsvik and David, 2015).

Bone debitage has been essential for identifying fishhook-manufacture and other forms of bone-tool production in archaeological assemblages from various geographical regions. The properties of bone as well as their morphological shape, size and thickness guides the transformation of a bone element into artefact (David, 2007, p.39). Bone fishhooks have previously been replicated from metapodials of red deer (Cervus elaphus) and elk (Alces alces) (Bakkevig, 2003; Mansrud 2017; Clausen, 2018). The majority of hooks in the NE Skagerrak region are however small (c.3cm long). Taking the size of the fishhooks as our point of departure, we decided to conduct an experiment using a bone element that corresponded with the size of the hook; the metapodials of roe deer (Capreolus capreolus).

Preliminary results from experiments with the manufacture of bone fishhooks made from roe deer metapodials will be presented and discussed in this paper. The main objectives of the experiments were to (i) attain hands-on understanding of the properties of roe-deer bones and the production process, (ii) to test whether it was possible to manufacture small fishhooks from roe deer metapodials using Middle Mesolithic lithic tools, and (iii) to achieve knowledge of the time and skills needed for the task.

Fishhooks of the NE-Skagerrak area: types and manufacture technique

The Middle Mesolithic fishhooks vary slightly in size, but are overall considered small, measuring approximately 3 cm long on average (See Figures 3, 4). The hooks were barbless and often have several small notches at the shaft, presumably for fastening the line. As shown on Figure 3, few hooks were complete. Most of the hooks were fragmented, consisting either of the shaft or the bend. Most of the hooks from sites in south-eastern Norway were also burnt (Mansrud, 2017; Mansrud and Persson, 2017).

A large number of similar but better-preserved fishhooks, dated within the same period, have been retrieved from well preserved osteoarchaeological assemblages excavated from caves and rock shelters in Western Norway (Bergsvik and David, 2015; Bergsvik, 2016). These hooks have a rounded bend, a pointed shank with notches on the external side and a pointed, barbless tip. They were often asymmetrical in form – the tip measured approximately half the length of the shank. The length of the shank was commonly 1-3 cm, but individual hooks measured up to 6-7 cm (Lund, 1951, p.27). The barbless fishhook is sometimes referred to as the “Viste-type”, named after a find from the Viste cave in Rogaland (Brøgger, 1908, p.546).

How were these tiny hooks manufactured?

Recent investigations based on the CO-approach have identified several different manufacture techniques for Mesolithic fishhooks in Scandinavia. These differences involved the whole process from blank production to finished hook (David, 1999; 2017a; Bergsvik and David, 2015; Clausen, 2018). In this paper we focused on the manufacture of the hooks.

Bergsvik and David (2015) suggest that the fishhook-manufacture of Viste-hooks in Western Norway involved a combination of fracturing and abrasive techniques, which leaves little debitage to be studied. The reconstructed CO suggests a method for serial manufacture of four hooks. The first step consists in drilling four holes symmetrically in the bone blank. Two holes opposite each-other were widened, and the inner part of the hooks shaped by scraping, to make two fishhooks. The bone blanks were then segmented by sawing the tip and shank cut off, and the finished hook polished. It has been further suggested that the notches were made by sawing with a “plate knife made of schist (Bergsvik and David, 2015, pp.207-208). Traces of the drilling were identified on some of the Viste-hooks, by the use of a stereomicroscope. In most cases, these traces were removed during the final polishing (Bergsvik and David, 2015, p.199). Experimental replication of this CO has so far been unpublished.

The Viste-hooks have been replicated on several occasions by Kutschera (See Figures 6, 7, 8) using a different technique than the one suggested above (Mansrud, 2017). Metapodials from different species of deer were utilized. These bones have an anatomy suited for making a range of different bone tools (David, 2007, p.37). The metapodial in the foreleg is termed metacarpal and the metapodial in the hind leg is the metatarsal (See Figure 7). The metatarsal may be longer than the metacarpal in cervids. When the epiphyses are fused and the long bone fully grown, the metacarpal is D-shaped at the proximal aspect along the shaft to the distal end whereas the metatarsal is somewhat rounded proximally and along the shaft.

The first step in making a fishhook consisted of boring (by hand) or drilling (by bow drill) a hole through one end of the blank. Thereafter, two longitudinal and divergent grooves were incised towards the hole, to shape the inner curve of the hook. The removed bone resulted in a triangular piece of debitage (Jonsson, 1996, p.41; Mansrud and Persson, 2017, p.146, fig 5). In some cases, the hole remained visible at the bend of the hook, occasionally in tandem with a small barb-like protuberance on the inner side of the tip (as shown on the hook depicted in Figure 4). In most cases, the hole has subsequently been removed by later grinding and polishing, to make a U-shaped hook with a rounded bend (See Figures 5 and 3, upper left). Replication of Viste-fishhooks was made by Bakkevig (2003) using elk metapodials. Bakkevig suggested a yet another CO in which the whole hook, including the notches, was completed while still attached to the bone.

Based on fishhooks and debitage from early Mesolithic Maglemose sites (dated between c. 8200-6000 cal. BC) in Zealand, Denmark, a third CO has been documented (Clausen, 2018, pp.6-13). In the initial stage the bone blank was punctured and left with a perforation at the end. From the perforation, a furrow was opened by using longitudinal grooving. This permits the placing of the shank. Thereafter the groove was widened by inner scraping to shape the inner part of the bend. Finally, the fishhook was detached from the blank at one end with a prepared flexion break, through the sawing of two grooves from opposing sides of the blank. While at the other end the shank is cut. The experiments undertaken by Clausen shows that the procedure of inner longitudinal grooving and scraping was a very time-consuming procedure. Still, this technique has been constantly applied for approximately 1900 years. This suggests that Middle Mesolithic crafters in this region were following a strict operational schema and a tradition which was culturally, rather than functionally, prescribed (Clausen, 2018, p.13).  

Availability of roe deer during the Middle Mesolithic period 

The Middle Mesolithic sites with fishhooks in the NE-Skagerrak area were dated between 8300 and 6300 cal. BC. This corresponds to the Boreal (c. 8400-6700 cal. BC) and the first part of the Atlantic climate phase (6700-3900 cal. BC). During this time span, several ungulate species were accessible to the coastal groups in the area (See Table 1).

Red deer were the most common deer species identified in the faunal assemblages (See Table 1).  Elk and red deer appeared in natural and anthropogenic bone assemblages already by the Preboreal (9500-8400 cal. BC) (Grøndahl, et al., 2010; Jonsson, 1995, p.150). In eastern Norway, the oldest finds of roe-deer are dated to c. 6300 cal. BC (Table 1, see also Hufthammer and Aaris-Sørensen, 1998). Roe deer have, however, been identified amongst the faunal remains from the Almeö site, which dated to the Preboreal (Mansrud and Persson, 2017, p.136), as well as from the Preboreal/early Boreal transition (c. 8000-7600 cal. BC) at Huseby Klev (Boethius, 2017, p.6; Jonsson, 2005). In Southern Scandinavia, roe deer was a common species throughout the Boreal and the Atlantic chronozone, appearing on inland as well as on coastal sites (Boethius, 2016, pp.160-162). The roe deer prefers mixed woodlands but can adjust to a variety of habitats. They are tolerant of climatic extremes, from hot and dry Mediterranean habitats to the cold boreal forests with deep snow (Sommer, et al., 2009; Tufto, et al., 1996). The lack of roe-deer in the faunal assemblages in South-Eastern Norway before the Atlantic climate optimum is therefore surprising. Since roe deer was targeted by coastal groups in Western Sweden, we find it probable that roe-deer was also present in South-Eastern Norway during the Boreal, but this is unproven by zooarchaeological assemblages. Overall, the number bones identified as roe deer in the faunal remains were very low. The low number of fragments from the cervids in general is possibly also related to the fact that these bone elements were heavily utilized for raw material for making of tools. Zooarchaeological reports rarely take into account the intense human utilization of cervid bone and antler during the Mesolithic, which have been demonstrated by current research into bone-technologies (David, 2017b; Elliott, 2012; Gummeson, 2018).

Table 1. Middle Mesolithic sites in the NE-Skagerrak area with fishhooks, fishhook-debitage, and type of deer species identified.

Experimental replication with roe deer metapodials

For the experiment, two metacarpals and two metatarsals of roe deer were utilized (See Figures 9, 10). The roe deer metapodials are considerably shorter and thinner than the metapodials of other deer species (See Figure 10a). One roe deer metapodial provided material for four blanks, which were then worked into fishhooks. The metapodial bones had been stored dry and were initially soaked in water for about five minutes in order to make the bone softer and easier to pierce. The first step in the manufacture sequence was the removal of the articular ends. The distal articular end was first removed. This was achieved by transversal sawing around the circumference of the bone, using a large flint blade (See Figure 9b) and followed by a flexion break through the use of a stone hammer. This initial procedure resulted in a characteristic piece of debitage: an articular end with transversal grooves and a small piece of bone still attached (See Figure 9b). The proximal epiphysis was then removed in the same way. If a younger individual with an unfused proximal epiphysis had been used, the splitting of the bone could have been performed without the initial removal of the proximal articular end.

Secondly, the metapodial was split in two, to make two rough outs for blanks (See Figure 9c). To split the bone, a deep groove was made on each side of the metapodial, following the natural longitudal groove on the diaphysis (sulkus dorsalis). Flint implements were then selected from a varied assemblage of blades, bladelets and microblades. The groove was incised using the corners of small, regular (intentionally or accidentally broken) bladelets, and during this procedure the bone was regularly soaked in water for 15-30 seconds. The splitting was performed by placing a sturdy blade of flint as a wedge at the top of the proximal end and hitting it gently with a stone hammer. During this stage of the production process, the roe deer metapodial tended to break or splinter inwards, rather than following the natural groove (See Figure 9d). The initial grooving, removal of articular ends and splitting of the bone into blanks took about 30 minutes.

The third step was to remove the excess bone on inside of the split metapodial, to make the blank pre-forms. We tried different ways to do this: twisting the excess bone out with a flint knife, removing it by a hard hammer blow, and by using a piece of flint as a wedge. However, as the roe deer metatarsus is thinner and much more fragile than bones of larger cervids, the bone tends to break or splinter. Thus, we found that the simplest way to proceed without breaking the bone, was by careful scraping/whittling with a flint blade, and then by grinding. A large grinding slab made from sandstone was used for this purpose (See Figure 9b-f). Water was mixed with crushed flint into powder on top of the slab to augment the grinding effect. The grinding method eventually formed the bone into an oval blank with a slightly U-shaped ventral side, formed by the medullar cavity. The medullar cavity was sometimes visible on the inner part of the bend of the archaeological fishhooks (See Figure 3, upper left).

The fourth step was the manufacture of the hook itself. The shape of the hook was etched on to the blank. Initially a hole was drilled. The hole will eventually become the base of the inner part of the hook. The hole was drilled by using a flint drill bit. A morphologically shaped/retouched drill bit (borer) turned out to be best suited for this purpose, but a naturally shaped pointed blade also worked well. The blade eventually is shaped into the drill bit. The hole may be placed symmetrically in the middle or a little bit to one side. An asymmetrical placement of the hole makes it easier to cut out the shank. When the hole was placed, the shape of the hook was carved out. We found it more convenient to make the hole and groove out the V-formed piece first, and then shape the outer part of the bend into desired form by grinding it on the slab (See Figure 9e). This part of the production stage was very time-consuming, but it had an advantage in that the bend may be shaped into the desired thickness and shape. The bend was identified as a weak point, and often broke during manufacture (See Figure 10e). Such fragmented hooks were also observed within the archaeological assemblages. Presumably, this inherent weakness was the main reason why the bend of the Mesolithic hooks was usually rather broad (See Figure 3).

Finally, the inner part of the hook was carved out by the use of a small, sharp bladelet or microblade. This last stage of the process was very time-consuming, and the bladelet had to be renewed often, as it gets blunt. New, sharp bladelets were regularly made by placing the blade on the edge of the grinding slab and snapping the worn section with a blow. This constant renewal of the working edge resulted in many small sections of bladelets and microblades.

As noted, removal of the inner part of the hook resulted in a triangular piece of debitage with grooved sides. Also observed were perpendicular traces of grooving in the distal end and traces of a controlled flexion break in the proximal end (See Figure 9f, and Figure 10f, Figure 11). Similar pieces of debitage have been distinguished in the well-preserved archaeological assemblages such as Dammen (See Figure 11). Based on analogy, similar pieces have also been recognized among burnt bone fragments at other sites (Mansrud, 2017), thus supporting a conclusion that fishhook manufacture was undertaken at the sites.

With no articular ends preserved, determination of which species and which bone element was used for making the Middle Mesolithic hooks remains challenging. The morphological characteristics and morphometric differences as well as the thickness of the bone cortex makes it relatively easy to distinguish between different species of cervids when the bones are well-preserved. Conversely, when the faunal remains were fragmented, worked and/or worn, or burnt, as is the case with most archaeological assemblages in the NE-Skagerrak area (See Figure 2), species and element identification remain largely impossible. The size and thickness of the metapodial walls may give an indication, however size and thickness are affected by age as well as size dimporhism between males and females. Resolving this issue would require a large comparative study of cervid metapodials of different ages and sexes.

Results, discussion, and implications for future research

The experimental replications presented here have added some new “flesh on the bone” in the study of Mesolithic fishhook manufacture. We have shown that small fishhooks could have been made from roe deer metapodials, but these bone elements are rather fragile and were inclined to break and splinter throughout several stages of the manufacture process: during the initial splitting, when the removal of excess bone took place, and during the final stages of manufacture. At this point, it is difficult to determine whether this fragility is inherent to roe-deer bones or mainly caused by the fact that bones used in our experiments were dry. For further exploration of these issues, more experiments utilizing fresh bone is required. Well-preserved assemblages in Northern and Southern Scandinavia roe deer body parts were less frequently utilized compared to red deer and elk during the Boreal period (David, 1999, pp.123-124; David, 2007, p.37; Leduc, 2012, p.75; David and Sørensen, 2016). From this it can be concluded that in terms of properties, roe deer metapodials may not be the preeminent osseous material for bone manufacture.

Since one of the present authors have extensive experience with bone tool replication and the other was a beginner, the experiment provided some insight into the level of skill and know-how needed for crafting these bone implements. Small fishhooks made from roe-deer metapodials were challenging to accomplish for a novice. As shown on Figure 10g, the fishhooks made by the experienced crafter resemble the archaeological fishhooks, whereas the ones made by the novice were more diverse, and many of them broke during manufacturing. The lack of experience presumably also contributed to the high degree of breaking. Whether fishhooks were made by individuals with different levels of skill during the Middle Mesolithic is not possible to confirm based on the fragmented archaeological material presented in this study. Distinguishing between fishhooks broken during use and during production might be possible by combining traceological and morphometric analysis of the archaeological specimen in future studies (Olson, et al., 2008). Above all, the fishhooks were very time consuming to make. Even for the experienced practitioner, the production manufacture of each hook from blank production to a finished tool took 3-4 hours. If several small fishhooks were used in longline fishing, as previously suggested (Åhrberg, 2007) a considerable amount of time must have gone into producing them.

Experimental replication of osseous technology furthers experiential knowledge of the lithic tools and technologies that were used when the fishhooks were made, because these two technical systems were fundamentally entangled (David and Sørensen, 2016). As stressed by Bergsvik and David (2015), grinding and abrasion techniques characterized the fashioning of fishhooks at the examined sites in Western Norway. An informational experience from our experiment is the importance of the grinding slab. By means of the slab, several stages of the manufacture process were made with less effort. Larger grinding slabs appear in South-Eastern Norway alongside ground axes from c. 8000 BC and it is often assumed that these slabs were used for grinding axes (Eymundsson, et al., 2017).  As shown here, grinding slabs were similarly functional for grinding and polishing bone. The utilization of these implements should be more thoroughly investigated by experiments. While performing this experiment we had no access to smaller grinding stones, i.e. ‘plate knives’ as were suggested by Bergsvik and David (2015, p.207) as a multi-purpose tool for manufacturing fishhooks in Western Norway. Hence it remains to be tested how such tools would have performed. The use of ‘plate knives’ has so far not been documented outside Norway and is suggested to represent a local adjustment (Bergsvik and David, 2015, p.215). The use of these tools is dated back to 7000 cal. BC in Western Norway. Similar items are termed ‘sandstone knives’ and appear in south-eastern Norway in the late Middle Mesolithic, around 7000 cal. BC (Reitan. 2016, pp.33-34). However, ‘plate knives’ never enter the Middle Mesolithic lithic repertoire in Western Sweden (Kindgren and Åhrberg, 1999; Nordqvist, 1999, p.242). This supports the conclusion that small bone fishhooks in this region of Skagerrak were manufactured with flint tools.

The experimental replications further shed light on the lithic implements that were used for manufacturing the blanks and the hooks. The lithic debitage which accumulated during the manufacture of the hooks corresponded well with the sectioned bladelets of flint that commonly occur at the archaeological Middle Mesolithic sites in the NE-Skagerrak area (Nordqvist, 1999; Mansrud, 2013). These bladelets were associated with an overarching technological concept that were introduced to the NE-Skagerrak from Russia and the Baltic around 8300 cal. BC (Sørensen, et al., 2013). Bladelets could have functioned as lithic inserts in composite tools, but also were used for working bones at the coastal sites (Mansrud, 2013, p.245). As demonstrated here, unhafted bladelets and microblades were ideal working tools for fashioning small fishhooks. Systematic use-wear analysis on archaeological assemblages of bladelets to verify this interpretation is an important task for future research.

As initially stated, the CO-approach is a bottom-up approach, aiming to study technology from the microscale of singular sites and gradually build empirical knowledge of larger regional and inter-regional scales. The ultimate objective is however to explore the implications of technology for the social organization in prehistoric societies – and essentially contribute to understanding how culture works and reproduces, as pointed out by Kjel Knutsson (2009) in the introductory quote. Traditional technologies are guided by ideology – decision making, ecological knowledge, and transmission of knowledge are entwined in tool production and social traditions, values, and worldviews, are communicated and mediated through everyday craft production (Lemonnier, 2013). Fishhooks are first and foremost functional tools. Their design needs to be adjusted to the size and feeding habits of the fish and their manufacture is further governed by raw material properties. But as demonstrated in this paper, there were many ways to manufacture a bone fishhook and their manufacture is also guided by traditions; cultural conceptions and expectations (i.e. David, 2009). Different techniques have been employed and persistently reproduced by Mesolithic crafters in Scandinavia. These preliminary results indicated at least three regional traditions during the period 8300-6300 cal. BC: Western Norway, the NE Skagerrak area, and Zealand. As such, technology is a process mediating mind, society, and the material environment. If we accept that technical processes are also socially meaningful (Lemonnier, 2013), actualistic experimental replications provide a fruitful road for future exploration of the interface between individual and society, ideology and nature in the Mesolithic.

Acknowledgements

The authors wish to thank prof. emeritus Kjel Knutsson (University of Uppsala) and Dr. Helena Knutsson for their contributions to Scandinavian Mesolithic archaeology in general and for their friendship continued support and help.