Hollow Comb Rivets Made from Stripdrawn Copper Wire and Two Possible Antler Draw Plates from 11Th–12Th C. Sigtuna, Sweden

Omslag 2019/2_Omslag 3/2004 (kopia) 2019-05-20 11:32 Sida 1

Innehåll F

57 Gardeła, L. et al. The spur goad from kort meddelande O R

Skegrie in Scania, Sweden. Evidence of 115 Scheglov, A. Omarbetning av Olaus Petris N elite interaction between Viking Age krönika: Ett politiskt vittnesbörd – och V Ä

Scandinavians and Western Slavs. heraldiskt? N N

75 Loftsgarden, K. The prime movers of iron E production in the Norwegian Viking and recensioner N Middle Ages. 118 Larsson, B.T. & Broström, S-G. Nämforsens 2 0 1 9 / 88 Wärmländer, S.K.T.S. & Söderberg, A. Hällristningar – Sveriges största och äldsta Hollow comb rivets made from strip- hällristningsområde med 2600 figurer FORN drawn copper wire and two possible ant- (The rock of Nämforsen). Anmälan av ler draw plates from 11th–12th c. Sigtuna, V. Mantere. Sweden. 120 Ahlström, C. The Viking Age. A time of many faces. Anmälan av L. Gardeła. debatt 123 Ljung, C. Under runristad häll: Tidig- 100 Kjellström, R. Stalofrågan – en personlig kristna gravmonument i 1000-talets slutkommentar. Sverige. Anmälan av P. Carelli. VÄNNEN 107 Lovén, C. Var Beowulf gute? JOURNAL OF SWEDISH ANTIQUARIAN RESEARCH

2019/2

issn 0015-7813 Omslag 2019/2_Omslag 3/2004 (kopia) 2019-05-20 11:32 Sida 2

Utgiven av Till författaren Kungl. Vitterhets Historie och Antikvitets Akademien i samarbete med Historiska museet. fornvännen välkomnar manuskript i nordisk arkeologi och äldre tiders konstvetenskap med an grän sande Fornvännen finns på webben i sin helhet från första årgången och publiceras löpande där ämnen. Bidrag kan vara avfattade på de skandinaviska språken samt engelska, tyska och franska. Abstracts och med ett halvårs fördröjning: fornvannen.se sammanfattningar skall vara på engelska, bildtexter på uppsatsens språk och engelska. Hela Fornvännens inne- håll publiceras fortlöpande både på papper och på internet, det senare med ett halvårs fördröjning. Kontakta Ansvarig utgivare och huvudredaktör gärna redaktionen inför och under skrivandet om frågor uppstår. Mats Roslund Vitterhetsakademien Tekniska och typografiska frågor Box 5622, 114 86 Stockholm Manuskript skall vara kompletta och noga korrekturlästa. De levereras med all text i en enda fil i RTF-format [email protected] eller annat vanligt ordbehandlingsformat. Filen skickas helst per e-post, annars på CD. Inget pappersmanu- skript behövs. Om invecklad typografi som tabeller, matematiska formler eller runtranskriptioner förekommer Redaktionssekreterare och mottagare av manuskript kan dock en utskrift eller PDF-fil vara bra. Peter Carelli Vitterhetsakademien Uppsatser Box 5622, 114 86 Stockholm Tät och kortfattad framställning är en dygd. En uppsats får inte överstiga 16 trycksidor à 4500 ned slag inkl. [email protected] abstract, illustrationer och tabeller. Uppsatser levereras med abstract (högst 1500 nedslag), summary (2000–4500 nedslag) och bildtexter, alla Redaktörer dessa på engelska. Herman Bengtsson, [email protected] Abstract inleds med författarnamn, årtal, uppsatsens titel med engelsk titel inom parentes och avslutas med Christina Fredengren, [email protected] författarens namn, institutionsadress och e-postadress. Åsa M Larsson, [email protected] Referenser Dessa utformas enligt Harvardsystemet med författarnamn, tryckår och sida inom parentes i texten. Under Teknisk redaktör rubriken Referenser förtecknas refererade eller citerade arbeten i bokstavsordning efter författarens efternamn Kerstin Öström följt av initialer, tryckår, titel, namn på eventuell tidskrift samt nummer, utgivande institution och förlagsort. Grävlingsvägen 50 Titlar på monografier och tidskrifter kursiveras. Inga sidnummer anges i referenslistan. Se nedan. 167 56 Bromma [email protected] Forsklund, F., 1954. Skäggmode under järnåldern. Esoteriska sällskapets årsskrift 26. Stockholm. Gendergren, G., 1993. Medeltida suffragetter i Burgund. Lund. Prenumeration Sviskonkvist, S., 1946. Priapos på Kullaberg. Adlerlöffel, X. & Ölkefjär, Y. (red.). Sydsvensk järnålderskult, Vitterhetsakademien ett symposium. Halmstad. Box 5622, 114 86 Stockholm e-post [email protected] Illustrationer Bankgiro 535-3552 Illustrationer skickas med minst 1500 pixlars bredd, per e-post, på CD eller i fysiskt original. Fornvännen pub- licerar endast illustrationer där rättighetsinnehavaren givit författaren tillstånd för publicering på internet. Årsprenumeration i Sverige (4 häften) 200 kronor, lösnummer 60 kronor Debatt och mötesreferat Journal of Swedish Antiquarian Research Debattinlägg och mötesreferat saknar abstract och summary, får ha referenser och illustrationer, överskrider published by The Royal Academy of Letters, History and Antiquities inte 5 trycksidor à 4500 nedslag inkl. illustrationer. Innan ett inlägg publiceras skickar redaktionen ofta slut- Subscription price outside Sweden (four issues) SEK 250:– redigerad version till berörd part för inhämtande av svar. Box 5622, SE-114 86 Stockholm, Sweden Korta meddelanden fornvännen började utges av Kungl. Vitterhets Historie och Antikvitets Akademien år 1906 och ersatte då Korta meddelanden får ha referenser och illustrationer, överskrider inte 3 trycksidor à 4500 nedslag inkl. Akademiens Månadsblad samt Svenska Fornminnesföreningens Tidskrift, som båda tillkommit under 1870-talets illustrationer. första år. Förutom i Sverige finns Fornvännen på drygt 350 bibliotek och vetenskapliga institutioner i mer än 40 länder. Recensioner Tidskriften är referentgranskad. Recensioner får inte ha illustrationer eller referenslista (kortfattade referenser läggs inom parentes i texten), omfattar högst 9000 nedslag inklusive mellanslag. Tonvikten skall ligga på vetenskaplig utvärde ring mer än på fornvännen (»The Antiquarian») has been published by the Royal Academy of Letters, His tory and referat av böckernas innehåll. Anti quities since 1906, when it replaced two older journals which had started in the early years of the 1870s. Outside Sweden Fornvännen is held by more than 350 libraries and scientific institutions in over 40 countries. Korrektur The journal is peer-reviewed. Författaren får efter textredigeringen den nya textversionen per e-post för granskning. Illustrerade bidrag skickas därtill, efter slutförd typografering och ombrytning, per e-post i form av PDF-fil för granskning. Inga issn 0015-7813 papperskorrektur förekommer. Printed in Sweden by AMO-tryck AB, Solna, 2019 Varje författare erhåller ett exemplar av det häfte i vilket bidraget blivit infört. Art. Wärmländer & Söderberg 88-99_Layout 1 2019-05-20 11:27 Sida 88

Hollow comb rivets made from strip-drawn copper wire and two possible antler draw plates from 11th–12th c. Sigtuna, Sweden

By Sebastian K. T. S Wärmländer & Anders Söderberg

Wärmländer, S. K. T. S. & Söderberg, A., 2019. Hollow comb rivets made from strip-drawn copper wire and two possible antler draw plates from 11th–12th c. Sig- tuna, Sweden. Fornvännen 114. Stockholm.

Modern metal wire is produced by drawing solid metal rods through a draw-plate. Scandinavian smiths used this technique already during the Viking Age, but little is known about earlier Scandinavian methods for making metal wire. It has previ- ously been suggested that the metal rivets in composite bone and antler combs may have been hollow and produced by strip-drawing, but no metallurgical studies have so far been carried out to investigate this possibility. Here, we used scanning electron microscopy (SEM) to investigate copper-alloy rivets in 11th – 12th c. composite combs from Sigtuna, the administrative centre for middle Sweden’s first Chris- tian kings during the early Medieval period. Our SEM images showed that while some rivets were made from solid circular wire, other rivets are hollow and proba- bly manufactured from strip-drawn wire. We also examined two perforated antler plates, likely dated to the 12th c. and excavated from a bronzesmith’s workshop in the block Trekanten in central Sigtuna. The copper and lead particles detected by SEM analysis around the plates’ holes indicate that the plates were used in metalworking activities. Because the holes are cylindrical and not conical, however, the plates would not be viable tools for drawing solid metal wire. In the strip-drawing technique, on the other hand, cylindrical holes might have been used to produce hollow metal wire. The holes in the studied antler plates have the same diameter – 2.0 mm – as many comb rivet holes, possibly suggesting a standardization for large- scale production. The bronzesmith’s location next to a combmakers’ workshop provides further support for a production connection between the two crafts. Tak- en together, our results indicate that the two perforated plates may have been tools for strip-drawing copper wire, used to make comb rivets.

Sebastian Wärmländer, Division of Biophysics, Stockholm University, SE-106 91 Stockholm, Sweden [email protected] Anders Söderberg, Sigtuna museum, Stora gatan 55, SE-193 30 Sigtuna, Sweden [email protected]

Introduction the desired size is obtained. This wire-drawing In modern society, metal wire is produced by technique was known in Scandinavia already at drawing solid metal rods through conical holes of the Viking Age, as demonstrated by the draw pla- increasingly smaller dimensions (Carroll 1972; tes encountered in Viking trading centres such as Newbury & Notis 2004; Oddy 1977). Each draw- Birka (Arrhenius 1968), Hedeby (Armbruster ing step reduces the diameter of the wire, until 2012), and Staraya Ladoga (Davidan 1982).

Fornvännen 114 (2019) Art. Wärmländer & Söderberg 88-99_Layout 1 2019-05-20 11:27 Sida 89

Hollow comb rivets… 89 Little is known about earlier Scandinavian portant centre also for craftspeople and artisans. techniques for crafting metal wire. In other cul- Many researchers have used Sigtuna’s rich ar- tures, a variety of techniques were used to pro- chaeological material to investigate Sweden’s duce metal wire before the invention of wire- Viking Age and early Medieval crafting techno- drawing (Muros et al. 2007; Newbury & Notis logy (Edberg 2013; Floderus 1928; Karlsson 2016; 2004; Oddy 1977; Scheel 1989), such as hammer- Pettersson 2007; Ros 1990; Sjöbeck 2016; Söder- ing, casting, block-twisting, strip-twisting (Car- berg 2006; 2011a; Söderberg & Gustafsson 2007). roll 1972; Thouvenin 1971), strip-drawing (Wil- Here, we investigate copper rivets from composite liams 1924), and roll-drawing (Özsen & Willer antler combs excavated from 11th–12th c. Sigtuna, 2016). Clarifying which wire-making techniques in order to characterize the manufacturing tech- were used where and when is important not only nology and chemical composition. We also in- from an archaeometallurgical point of view, but vestigate two perforated antler plates, excavated may also aid when determining the provenance in 1925 from a bronzesmith’s workshop in cent- and authenticity of ancient jewellery and other ral Sigtuna, and proposed by Floderus (1928) to objects containing metal wire (Carroll 1970; be draw plates used to produce metal wire. The Ogden 1983). objects are studied with scanning electron micro- A study of metal wires from the Viking Age scopy (SEM), a very useful technique for charac- town of Birka, Sweden, showed that most were terizing residue particles, corrosion, tool marks, produced by drawing, although occasional pieces and elemental compositions of archaeological ob- of jewellery contained block-twisted, strip-twisted, jects as well as modern replicas (Bartelink et al. and strip-drawn wire (Duczko 1985). As Birka was 2015; Boutrup et al. 2013; Scott et al. 2009; Smith an international trading centre (Ambrosiani 2012), et al. 2018; Smith et al. 2015). many objects were imported from faraway places (Wärmländer et al. 2015), and it is not clear which The strip-drawing, roll-drawing, and wire- of the wires studied by Duzcko (1985) were local- drawing techniques ly produced. Some researchers of Viking Age and Some of the various techniques that exist for pro- Medieval composite bone and antler combs in ducing metal wire involve drawing the wire ma- Norway (Hansen 2005), Sweden (Ros 1990), and terial trough a small hole. Wire produced with the British Isles (Galloway 1990) have noted that such methods, which involve strip-drawing, roll- many of the comb rivets appear hollow, and sug- drawing, and proper wire-drawing, but not block- gested that they might have been produced from twisting, strip-twisting, hammering or casting, metal wire manufactured with the strip-drawing will therefore display longitudinal striations along technique. Unfortunately, no metallurgical stu- the wire surface, created by the unevenness in the dies have so far been carried out to investigate holes of the draw plates. The strip-drawing and roll- this suggestion, which is in conflict with Duczko’s drawing techniques furthermore create a longitu- (1985) conclusion that by the Viking Age, the dinal seam on the wire surface, which might be technique of drawing solid wire had replaced ear- more pronounced than the striations. Yet, separat- lier Scandinavian techniques for wire producing these techniques based on visual inspection tion. Consequently, Duczko (1985) argues that of the wire surface is a very difficult task (Carroll objects containing non-solid wire should be pre- 1970; Newbury & Notis 2004; Oddy 1977; Og- Viking Age. In his study of comb rivets from Bir- den 1983; Özsen & Willer 2016). Instead, a cross- ka, Stjerna (1998) observed longitudinal stria- section analysis is much more informative. For tions on some rivets, indicating they were made proper wire-drawing, the transverse cross-sec- from drawn wire, but he did not carry out cross- tion is solid (fig. 1B), and the longitudinal cross- section analysis to determine if the rivets had section displays elongated grains (Oddy 1977). been made from solid drawn wire or from hollow For strip-drawing, the transverse cross-section is strip-drawn wire. hollow and slightly spiral-shaped (fig. 1A), while Sigtuna was middle Sweden’s administrative the grains in the longitudinal cross-section should centre for the early Medieval kings, and an im- not be significantly elongated. For roll-drawing

Fornvännen 114 (2019) Art. Wärmländer & Söderberg 88-99_Layout 1 2019-05-20 11:27 Sida 90

90 Sebastian Wärmländer & Anders Söderberg

Fig. 1. A) Strip-drawing, producing a hollow wire. B) Proper wire-drawing through a conical hole, which reduces the dimension of the solid wire. C) Roll- drawing is similar to strip-drawing, but the metal strip is here tightly rolled to produce a wire that is almost solid. Drawings by Sebastian Wärmländer based on images in Özsen & Willer (2016).

the transverse cross-section is distinctly spiral- semi-manufactures, mainly made from antler. shaped (Özsen & Willer 2016), possibly with a One common type is the Scandinavian early Me- small void in the centre (fig. 1C). dieval single-sided composite comb, which corre- In proper wire drawing, each drawing step re- sponds to Ambrosiani’s (1981) type B and Bro- duces the diameter of the solid wire and increa- berg & Hasselmo’s (1981) type 1 (fig. 2). Numer- ses its length. Thus, the corresponding draw pla- ous finds of components for this comb type, tes, which must be made of a material at least as including used tooth plates with remains of rivets, strong as the wire itself, typically display series of semi-manufactures for connecting plates, and conical holes of decreasing diameters (Newbury tooth plate semi-manufactures that appear to be & Notis 2004; Oddy 1977). The strip-drawing unused, were found in 1925 at an excavation in method, known from Ancient Egypt and the the block Trekanten conducted prior to the in- Roman Empire (Carroll 1972), involves less strain stallation of Sigtuna’s first water conduit system than proper wire drawing, due to the hollow inte- (Arbman & Floderus 2005; Edberg 2018). Four rior of the wire. Thus, strip-drawing does not sig- tooth plates excavated from the Trekanten nificantly increase the length of the wire, even if trench B (finds SF 407:b:1, SF 407:b:2, SF 453:f, multiple pulls are conducted to gradually reduce and SF 561:e) are shown in figs. 3–5. the wire diameter. It has been suggested that The trench at the block Trekanten revealed a draw plates for strip-drawing could have been bronzesmith’s and a combmaker’s workshop, situ- made from relatively soft materials such as wood ated close to each other. Due to the poor docu- or bone (Oddy 1977; Williams 1924), and coni- mentation of the early Sigtuna excavations, the cal holes are likely not required – cylindrical holes stratigraphic dating of these contexts is some- may suffice. Although these assumptions could what unclear. Much of the comb making debris be supported by future experimental evidence, appeared in trench sections B:V and B:VI, but a they will remain speculations until draw plates fair amount was together with the bronzesmith’s for strip-drawing have been identified in the ar- waste in sections B:II and B:III (Arbman & Flo- chaeological record, which to our knowledge has derus 2005, pp. 17; cf. Edberg 2018, p. 6). It is so far not happened. The roll-drawing method was therefore difficult to distinguish the two activi- only recently described (Özsen & Willer 2016), ties from each other, based on the excavated ma- and the material properties required of draw pla- terial, but this may not be due to poor documen- tes for roll-drawing are currently unknown. It can tation in the 1920’s. Instead, it might reflect a low be assumed that roll-drawing should require less level of crafting specialisation during the early force than drawing solid wire, but more force than Middle Ages. Different crafts could be housed un- strip-drawing hollow wire. der the same roof, and they could possibly also be performed by the same craftsmen (Hansen 2015; Sigtuna combs and comb fragments Ros 1990). Sigtuna’s archaeological record includes a multi- In the nearby block Trädgårdsmästaren, single- tude of combs and comb fragments of different sided composite combs with few but thick copper- types: 4 000 such finds are recorded in the Sigtu- alloy rivets (fig. 2), typically two per tooth plate, na Museum database. About one hundred are were common during the 12th and early 13th c. Iron intact combs, while the rest are fragments and rivets dominated in Scandinavian combs during

Fornvännen 114 (2019) Art. Wärmländer & Söderberg 88-99_Layout 1 2019-05-20 11:27 Sida 91

Hollow comb rivets… 91

Fig. 2. Two examples of the Scandinavian single-sided composite comb, both dated to the period 1125–1175 AD (cf. Söderberg 2011b, pp. 103). Top: Trädgårdsmästaren find No 15094. Bottom: Trädgårdsmästaren find No 26820. The white scale bar is 5 cm. Photo: Anders Söderberg.

the 11th c. (Smirnova 2005), which is clearly illus- kanten would be the late 11th or the 12th century. trated by the material from the Trädgårdsmäs- This dating correlates with the circumstances in taren block. Fragments of combs of a similar type, Trädgårdsmästaren, where antlerwork flourished with copper rivets and dated as early as to the sec- during the 12th c. (Pettersson 2007). The depth of ond half of the 11th century, were excavated in the finds from the bronzesmith’s workshop, i.e. 2014 at the Götes mack site. Being more recent- 1.0–1.2 meters below the 1925 street level (Flode- ly conducted, this excavation has proper strati- rus 1928), may roughly correlate to such a dating. graphic documentation. One fragment (181:3205: The dating is compatible also with the presence of 247) was found in connection with a bronze the simple double-sided comb among the finds workshop belonging to building phase V, i.e. from these layers, a type that started to appear in 1080–1150 AD, and another fragment (268:1: Sigtuna during the second half of the 11th c. (Wik- 251; fig. 6) was found in a waste dump belonging ström 2005) and became common during the to phase IV, i.e. 1060–1080 AD (Hed Jakobsson 12th c. 2017, pp. 41, cf. the Excel sheet “Matris med fasindelning” on the CD disc). An antler work- SEM analysis of the comb fragments shop involved in the manufacture of single-sided Scanning electron microscopy (SEM) images and composite combs and simple double-sided combs SEM-EDS spectra were recorded with a Hitachi was connected to the same chronological phase. TM-3000 table-top SEM unit, operating at 15 A direct contextual connection be-tween comb kV and equipped with a large sample chamber. fragment 268:1:251 and the workshop is howev- Thus, the entire objects were inserted into the er not certain. SEM unit and investigated without surface-coat- Overall, a fair dating of the workshops in Tre- ing or other preparations at low vacuum (i.e., in

Fornvännen 114 (2019) Art. Wärmländer & Söderberg 88-99_Layout 1 2019-05-20 11:27 Sida 92

92 Sebastian Wärmländer & Anders Söderberg charge reduction mode). SEM images were recorded in back-scatter mode where solid metal, metal residue, and metal corrosion appear brigh- ter than the surrounding bone or antler matrix. SEM-EDS analysis was used to map and calcu- late metal compositions. For tooth plate SF 561:e, the two rivets appear hollow upon visual inspection, and this impres- sion is confirmed by the SEM imaging (fig. 3). Both rivets appear to be around 1–1.5 mm in diameter, with a metal sheet thickness around 0.2– 0.3 mm and central voids with irregular shapes that have average diameters around 0.5 mm. Although both rivets are corroded, their central voids cannot reasonably have been created by corrosion processes, which typically are initiated at an object’s exterior surface. Instead, the most likely explanation is that the rivets were made from hollow wire, such as that produced by the strip-drawing technique (fig. 1). For tooth plate SF 407:b:1, the SEM analysis shows that both rivets are completely corroded: no solid metal is present (fig. 4). A close-up SEM image of the right rivet shows a series of concen- tric circles, remaining as structural features in the corroded material. The key feature appears to be two generally circular layers, each around 0.1 mm thick and rich in copper. These two layers encircle an oval boulder-shaped central feature, about 1 mm across and copper-rich (fig. 4). We interpret this as the corroded remains of a hollow rivet made from strip-drawn copper wire. The two circular layers in fig. 4 would then correspond to hollow wire made from two layers of copper strip, and the central boulder-shaped feature is corrosion (possibly mixed with soil/dirt) that has filled the central void. For comb plates SF 407:b:2 and SF 453:f, each plate has one missing rivet, allowing us to meas- ure the diameters of the rivet holes from the SEM images. Both were 2.0 mm (fig. 5). This observa- tion will be further discussed below, together with the analysis of the antler plates. For comb fragment 268:1:251 from the Götes Fig. 3. Top: An antler tooth plate (SF 561:e) mack site, the SEM analysis shows that the cop- excavated from the block Trekanten. The white per rivet consists of solid metal with small amounts scale bar is 2 cm. Photo: Anders Söderberg. Bottom: SEM images of the two corroded ri- of corrosion (fig. 6). The better preservation of vets in the tooth plate. Both rivets are clearly this object is likely related to it being more re- hollow. SEM images by Sebastian Wärmländer. cently excavated – in 2014 – than the comb frag

Fornvännen 114 (2019) Art. Wärmländer & Söderberg 88-99_Layout 1 2019-05-20 11:27 Sida 93

Hollow comb rivets… 93

Fig. 4. Top: An antler tooth plate (SF 407:b:1), excavated from the Trekanten site. The white scale bar is 5 cm. Photo: Anders Söderberg. Bottom: SEM image of one of the corroded rivets. The yellow circle indicates two circular layers of corroded copper, consistent with the rivet being made from strip-drawn wire. SEM image by Sebastian Wärmländer.

Fig. 5. Top: Two antler tooth plates (SF 407:b:2 and SF 453:f) for single composite combs, excavated from the block Trekanten in central Sigtuna. The white scale bar is 3 cm. Bottom: SEM image of the rivet hole from comb fragment SF 407:b:2, with a diameter of 2.0 mm. Images by Sebastian Wärmländer.

Fornvännen 114 (2019) Art. Wärmländer & Söderberg 88-99_Layout 1 2019-05-20 11:27 Sida 94

94 Sebastian Wärmländer & Anders Söderberg approximately 99.5% Cu and 0.5% Pb. The use of copper-zink alloys (i.e., brass) is not surprising, as brass was commonly used in the Viking and Medieval periods (Saage & Wärmländer 2018; Jouttijärvi 2002, p. 39) and traded from the Euro- pean continent as a standardized commodity (cf. Sindbæk 2001). The rather pure copper with a small amount of lead found in the rivet from comb fragment 181: 3205: 247, might originate from recycled brass: during each re-smelting cycle some zinc will evaporate, while any lead present will remain. This would not necessarily be a disadvantage, as it would result in a very Cu- rich alloy that would be softer and easier to draw.

The two antler plates The two studied antler plates, SF 697:h and SF 449:m (fig. 7), were found during the 1925 Tre- kanten excavation in direct connection with the bronzesmith’s workshop (Floderus 1928). With reference to the above discussion on the contexts and dating of the comb fragments, the antler plates can roughly be dated to the late 11th or the 12th c. Their dimensions are 5.0 x 4.1 x 0.5 cm (SF 697:h) and 4.1 x 3.7 x 0.3 cm (SF 449:m). Floderus (1928) described the material as elk antler, which Fig. 6. Top: Comb fragment 268:1:251 from the appears to be correct. SEM-EDS analysis revea- Götes mack site, with well-preserved rivets. The white scale bar is 3 cm. Photo: Anders Söderberg. led large amounts of calcium and phosphorus in Bottom: SEM image of the left rivet after cleaning the plates, which is expected from a bone/bioapa- the top surface with a Dremel rotary tool. This tite material. The microstructures of the plates rivet is made from solid circular wire. SEM image (figs. 7 and 8) are compatible with antler, and espe- by Sebastian Wärmländer. cially that of elk, where the spongy tissue is denser than in antler of e.g. red deer and reindeer (Ashby ments excavated from the Trekanten site in 1925. 2013; Karlsson 2016). The plate dimensions also The storage conditions for the Trekanten finds indicate a large animal such as elk. Determining were apparently not ideal, as often was the case species from morphological features is however for material from the early Sigtuna excavations. difficult, and biomolecular analysis of DNA or pro- The top view SEM image of the rivet effectively teins from the material can often provide a more shows it in transverse cross-section: it is clearly reliable species identification (Ashby 2013; Ashby circular and solid. Thus, this rivet is made from et al. 2015; Smirnova 2005; von Holstein et al. copper wire produced by a technique such as cast- 2014). ing, hammering or proper wire-drawing (Ros 1990). The scratches in the image originate from SEM analysis of the two plates the rivet ends being cleaned with a Dremel rotary SEM analysis identified metal particles consist- tool. The rivet composition is approximately 93.2% ing mainly of copper and lead around some of the Cu, 5.6% Zn, 0.6% Pb, and 0.6% Sn. The other holes in the plates, at both the front and back comb fragment from the Götes mack site, 181: sides (fig. 8). Because these particles are very small 3205: 247 (no image shown), displays a rivet with (around 1–10 micrometres) and also corroded a similar solid cross- section and a composition of (CuS and PbO appear to have formed), it is in

Fornvännen 114 (2019) Art. Wärmländer & Söderberg 88-99_Layout 1 2019-05-20 11:27 Sida 95

Hollow comb rivets… 95

Fig. 7. The two studied antler plates, SF 697:h and SF 449:m, excavated from the bronzesmith’s workshop in the block Trekanten in central Sigtuna. The white scale bars are 5 cm. Photos: Sebastian Wärmländer.

Fig 8. SEM images of holes in the studied antler plates. A) Hole from plate SF 449:m, front side. Note the metal residue in the bottom left corner. B) Hole from plate SF 697:h, front side. Note the false start of the hole (yellow rectangle), which proves that the hole was drilled. C) Hole from plate SF 697:h, back side. Note the large metal particle in the bottom right corner. D) Close-up view of the metal particle shown in (C). E and F) SEM-EDS maps showing the metal particle to contain large amounts of copper and smaller amounts of zinc – only slightly above the noise level. The diameters of the holes shown in (A-C) are all 2.0 mm. Images by Sebastian Wärmländer.

Fornvännen 114 (2019) Art. Wärmländer & Söderberg 88-99_Layout 1 2019-05-20 11:27 Sida 96

96 Sebastian Wärmländer & Anders Söderberg most cases not possible to elucidate from which rated antler plates might have been involved in original alloy compoitions the residue particles producing strip-drawn copper-alloy wire, possi- originate. One exception is a relatively large par- bly used to manufacture the hollow rivets found ticle (50 x 30 micrometres) next to a hole on plate in some of the studied comb fragments (figs. 3 SF 697:h (figs. 8C and 8D). This residue particle and 4). The antler plates’ holes all have diameters consists of brass with around 92% Cu and 8% Zn of 2 mm, and they display residue of copper and (figs. 8E and 8F). Some sulphur was also observed zinc (fig. 8). The comb rivets are made from cop- in the particle, indicating that it is at least partial- per and zinc, and the rivet holes are 2 mm across ly corroded. Hence, as zinc typically corrodes (fig. 5). The two antler plates were found together before copper, the original zinc content may have with various objects, including a strip of lead, fin- been somewhat above 8%. The holes are perfectly ished and unfinished antler combs, glass beads, circular with uniform diameters of 2.0 mm (figs. pottery sherds, and pieces of leather (Floderus 7 and 8). The false start shown in fig. 8B is evi- 1928). The presence of combs together with the dence that the holes were drilled. Because drills antler plates may be a coincidence, and the gen- during the Viking Age and Medieval Period typi- eral mixing of the waste from the bronzesmith’s cally were made from iron or steel (Ohlhaver 1939; and the combmaker’s workshops observed during Petersen 1951) – we are not aware of any evidence the excavation is perhaps not unexpected, given for the use of bronze or brass drills during these that the two workshops were located a mere 13 time periods – we conclude that the copper and meters apart (Floderus 1928). On the other hand, lead residue observed around the holes does not if the combmaker indeed obtained the rivets for his originate from when the holes were drilled. combs from the bronzesmith, it would make sense to have the two workshops situated next to each The antler plates as possible wire-making tools other (Croix et al. 2019). It might even have been The holes in the antler plates, and the observed the bronze smith’s task to attach the rivets to the metal residue around them, are likely related to combs, and some craftspeople could have worked how the plates were used. In his 1928 paper, in both workshops (cf. Hansen 2015). Floderus proposed that the plates were used for While these circumstances may appear con- wire-drawing. If so, the copper residue suggests vincing, they do not constitute proper proof. It is drawing of copper-alloy wire. The plates’ holes are quite possible that the antler plates had a totally however cylindrical and of uniform size, i.e., 2.0 different purpose. Or they might have been used mm wide (figs. 7 and 8 A–C). As discussed above, as expedient tools when making comb rivets, but this indicates that the plates were not used for not necessarily as draw plates. An alternative drawing solid wire, as plates used for that pur- function could be as support plates when head- pose should have conical holes, typically of diffe- ing the rivets. We do not know the steps involved rent sizes (Newbury & Notis 2004; Oddy 1977). when the rivets were attached and secured to the It is furthermore unclear if antler is a strong combs, but we might speculate that a piece of enough material to allow drawing of solid copper- copper wire was inserted into the rivet hole, and alloy wire. Antler plates may however be suffi- then both ends were hammered at to create the ciently strong to allow production of copper- rivet heads. In such a scenario, a thin support plate alloy wire by the strip-drawing method (Oddy with a hole of the same dimension as the comb’s 1977; Williams 1924). rivet hole would have been required when the Although the Saami people of northern Scan- first end was headed, to support the wire end pro- dinavia have a tradition of using plates of rein- truding from the other side of the comb (fig. 9). deer antler to draw tin wire (Zachrisson 1970), it Thus, the antler plates could have functioned simi- is generally uncommon to find tools of bone or lar to how nail headers are used in nail produc- antler in metal workshops (Grassi 2016; Mac- tion (Edberg 2013; Ryzewski & Gordon 2008; Gregor 1985, p. 171 fig. 89; Vuković-Bogdanović Wells 1998). Medieval smiths and metal-workers & Bogdanović 2016). Yet, there are a number of were clearly familiar with the concept of nail- circumstances to support the idea that the perfo- heading (and rivet-heading), as from the Viking

Fornvännen 114 (2019) Art. Wärmländer & Söderberg 88-99_Layout 1 2019-05-20 11:27 Sida 97

Hollow comb rivets… 97

Fig. 9. One possible use for the perforated antler plates. In a scenario where a rivet (light grey) is attached to a composite comb by inserting a piece of wire into the rivet hole and then heading it from both sides, a perforated plate will be needed to support the protruding wire end (at the bottom) when the oppo- site (upper) end is being headed. Image by Sebastian Wärmländer.

Age onwards iron nail headers were common posite combs were sometimes produced by hollow tools in Scandinavian smithies (Ohlhaver 1939; strip-drawn wire (figs. 3 and 4). This proves wrong Petersen 1951; Westphalen 2002) including those Duczko’s (1985) statement that objects contain- in Sigtuna (Edberg 2013). But this possible use of ing non-solid wire should be pre-Viking Age. Some the antler plates is just a speculation, as we do not comb rivets were however found to be made from know the procedures nor the tools used for attach- solid metal wire (fig. 6), which was produced either ing the rivets to the combs. by hammering, casting or solid wiredrawing (Ros The above discussion illustrates the problem 1990). Further research is required to clarify the of attributing a unique function to a tool that con- different wire-making techniques used in Medie- sists of a simple plate with seemingly randomly val Sigtuna, as well as when the strip-drawing me- distributed holes of equal size. It also illustrates thod eventually was discontinued, in Scandinavia the problem of identifying the tools used for strip- and elsewhere. drawing metal wire, as any perforated object like- The two perforated antler plates suggested by ly could be used. Overall, we believe that the best Floderus (1928) to be draw-plates were found to current interpretation of the two perforated ant- contain traces of copper next to their holes (fig. 8). ler plates is that they were used as strip-drawing Although they were not likely used to draw solid tools. Even though this tentative conclusion is wire, the plates may have been used to produce open for discussion, our results at least indicate hollow copper-alloy wire by strip-drawing. A pos- that tools made from soft materials such as bone, sible production connection between the antler antler, and wood might have been used when plates and the composite combs is supported by strip-drawing (and possibly roll-drawing) metal the hollow comb rivets arguably being made wire. Studies using experimental archaeology, from strip-drawn wire, by the close proximity of and ideally in combination with SEM (Özsen & the bronzesmith’s and the combmaker’s work- Willer 2016) or/and 3D analysis (Neiß et al. 2016; shops, and the identical dimensions of the rivet Sholts et al. 2012), should be able to investigate holes of the studied combs and the holes of the the material properties required of the drawing two antler plates (figs. 5, 7 and 8). However, other tools when producing wire of metals and alloys uses for the perforated plates are also possible, with different hardness and ductility (e.g., tin/ such as support plates when heading rivets (fig. 9). pewter, gold, silver, copper-alloys, and iron/ steel). Acknowledgments Conclusions We thank Kjell Jansson at Stockholm University Our analysis of 11th–12th c. composite antler combs for assistance with the SEM analysis, Sigtuna Mu- from Sigtuna confirms the suggestion by Hansen seum for help with accessing the studied objects, (2005), Galloway (1990), Ros (1990), and Stjerna and Björn Ambrosiani and an anonymous review- (1998) that copper rivets in early Medieval com- er for very helpful thoughts and comments.

Fornvännen 114 (2019) Art. Wärmländer & Söderberg 88-99_Layout 1 2019-05-20 11:27 Sida 98

98 Sebastian Wärmländer & Anders Söderberg References Floderus, E., 1928. Några brons- och silversmedfynd Ambrosiani, B., 2012. Birka. Brink S. & Price, N. (eds.). från det äldsta Sigtuna. Fornvännen 23. The Viking World. London. Galloway, P., 1990. The craft of comb making. Biddle, Ambrosiani, K., 1981. Viking age combs, comb making and M. (ed.). Artefacts from medieval Winchester, part II: comb makers: in the light of finds from Birka and Ribe. Object and economy in medieval Winchester. Oxford. Stockholm. Grassi, E., 2016. Bone anvils from the city of Sassari Arbman, H. & Floderus, E., 2005. Vattenledningsgräv- (16th–18th centuries AD). Vitezović, S. (ed.). Close ningen i Sigtuna 1925. Arkeologisk rapport, redigerad to the bone: current studies in bone technologies. Bel- och med en inledning av Rune Edberg. Sigtuna. grade. Armbruster, B., 2012. Wikingerzeitliches Goldschmiede- Hansen, G., 2005. Bergen c 800 – c 1170. The emergence of handwerk in Haithabu. Blankenfeldt, R. & Pesch, A a town. Bergen. (eds.). Goldsmith Mysteries: Archaeolocigal, Pictorial — 2015. Itinerant craftspeople in 12th century Bergen, and Documentary Evidence from the 1st Millennium Norway – aspects of their social identities. Hansen, AD in Northern Europe. Neumünster. G., Ashby, S.P., Baug, I. (eds.). Everyday Products Arrhenius, B., 1968. Ett tråddragningsinstrument från in the Middle Ages. Oxford. Birka. Fornvännen 63. Hed Jakobsson, A., 2017. I Sigtunas utkant. Slutunder- Ashby, S. P., 2013. Some comments on the identifica- sökningsrapport över gravar och bebyggelse vid Götes tion of species in worked antler. Choyke, A. & Mack. Arkeologikonsult rapport 2017:2696. Upp- O'Connor, S. (eds.). From These Bare Bones: Raw lands Väsby. materials and the study of worked osseous materials. Ox- Jouttijärvi, A., 2002. Metalhåndværk og håndværk- ford. spladser fra yngre germansk jernalder, vikingetid Ashby, S. P., Coutu, A. N. & Sindbæk, S. M., 2015. og tidlig middelalder. Henriksen, M. B. (ed.). Metal- Urban Networks and Arctic Outlands: Craft Spe- håndværk og håndværkspladser fra yngre germansk jern- cialists and Reindeer Antler in Viking Towns. alder, vikingetid og tidlig middelalder. Odense. European Journal of Archaeology 18. Karlsson, J., 2016. Spill – om djur, hantverk och nätverk i Bartelink, E. J., Sholts, S. B., Milligan, C. F., Van Deest, mälarområdet under vikingatid och medeltid. Stock- T. L. & Wärmländer, S. K. T. S., 2015. A Case of Con- holm. tested Cremains Analyzed Through Metric and Che- MacGregor, A., 1985. Bone, Antler, Ivory & Horn: The mical Comparison. Journal of Forensic Sciences 60. Technology of Skeletal Materials Since the Roman Period. Boutrup, J., Franzén, M.-L. & Wärmländer, S., 2013. London. Two 15th century openwork braids of silk andmet- Muros, V., Wärmländer, S. K. T. S., Scott, D. A. & Theile, al thread from Riddarholmen Church and Alvastra J. M., 2007. Characterization of 17th–19th centu- Abbey in Sweden. Fornvännen 108. ry metal threads from the colonial andes. Journal Broberg, B. & Hasselmo, M., 1981. Keramik, kammar of the American Institute for Conservation 46. och skor från 7 medeltida städer: fyndstudie. RAÄ & Neiß, M., Sholts, S. B. & Wärmländer, S. K. T. S., 2016. SHM rapport Medeltidsstaden 30. Stockholm. New applications of 3D modeling in artefact ana- Carroll, D. L., 1970. Drawn Wire and the Identifica- lysis: three case studies of Viking Age brooches. tion of Forgeries in Ancient Jewelry. American Jour- Archaeological and Anthropological Sciences 8. nal of Archaeology 74. Newbury, B. D. & Notis, M. R., 2004. The History and – 1972. Wire Drawing in Antiquity. American Journal Evolution of Wiredrawing Techniques. The Journal of Archaeology 76. of The Minerals, Metals & Materials Society 56. Croix, S., Neiß, M. & Sindbæk, S. M., 2019. The ré- Oddy, A., 1977. The production of gold wire in Antiq- seau opératoire of Urbanization: Craft Collabora- uity. Hand-making methods before the introduc- tions and Organization in an Early Medieval Work- tion of the draw-plate. Gold Bulletin 3. shop in Ribe, Denmark. Cambridge Archaeological Ogden, J., 1983. Jewelry of the Ancient World. New York. Journal. Ohlhaver, H., 1939. Der germanische Schmied und sein Davidan, O., 1982. Om hantverkets utveckling i Staraja Werkzeug. Leipzig. Ladoga. Fornvännen 77. Petersen, J., 1951. Vikingetidens redskaper. Oslo. Duczko, W., 1985. Birka V. Filigree & Granulation Work Pettersson, B., 2007. Kammakeriavfallets spridning på of the Viking Period: An Analysis of Materials from en tidigmedeltida stadsgård i Sigtuna. Situne Dei. Björkö. Stockholm. Ryzewski, K. & Gordon, R., 2008. Historical nail- Edberg, R., 2013. Ett nitjärn från Sigtunas vikingatid. making techniques revealed in metal structure. Situne Dei. Historical Metallurgy 42. – 2018. Fynden från vattenledningsgrävningen 1925: Ros, J. 1990. Horn- och benhantverk. Tesch, S. (ed.). katalog. Redigerad och med en inledning av Rune Makt och människor i kungens Sigtuna. Sigtuna- Edberg. Meddelanden och rapporter från Sigtuna grävningen 1988–90. Sigtuna. museum 61. Sigtuna. Saage, R. & Wärmländer, S. K. T. S., 2018. Metal resi-

Fornvännen 114 (2019) Art. Wärmländer & Söderberg 88-99_Layout 1 2019-05-20 11:27 Sida 99

Hollow comb rivets… 99

dues in 5th c. BCE–13th c. CE Estonian tools for ning i kv Trädgårdsmästaren 9 & 10 i Sigtuna 1988– non-ferrous metal casting. Journal of Archaeological 90. Meddelanden och rapporter från Sigtuna museum Science: Reports 19. 52. Sigtuna. Scheel, B., 1989. Egyptian Metalworking and Tools. Ayles- Söderberg, A. & Gustafsson, N.B., 2007. Från pres- bury. tigevarugjutning till myntning – Tidigmedeltida Scott, D. A., Wärmländer, S., Mazurek, J. & Quirke, S., metallurgi i kvarteret Trädgårdsmästaren, Sigtuna. 2009. Examination of some pigments, grounds Situne Dei. and media from Egyptian cartonnage fragments in Thouvenin, A., 1971. La Fabrication des Fils et des Fil- the Petrie Museum, University College London. igranes de Métaux Précieux chez les Anciens. Revue Journal of Archaeological Sciences 36. d’Histoire des Mines et de la Métallurgie 3. Sholts, S. B., Stanford, D. J., Flores, L. M. & Wärmlän- Wells, T., 1998. Nail Chronology: The Use of Techno- der, S. K. T .S., 2012. Flake scar patterns of Clovis logically Derived Features. Historical Archaeology points analyzed with a new digital morphometrics 32. approach: evidence for direct transmission of tech- Westphalen, P., 2002. Die Eisenfunde von Haithabu. nological knowledge across early North America. Neumünster. Journal of Archaeological Sciences 39. Wikström, A., 2005. Kammar. Wikström, A. (ed.). På Sindbæk, S., 2001. An object of exchange: Brass bars väg mot paradiset: Arkeologisk undersökning i kvarteret and the routinization of Viking Age long-distance Humlegården 3 i Sigtuna 2006. Meddelanden och exchange in the Baltic area. Offa 58. Neumünster. rapporter från Sigtuna museum 33. Sigtuna. Sjöbeck, A., 2016. Textilhantverk i det äldsta Sigtuna. Williams, C. R., 1924. Catalogue of Egyptian Antiquities: Situne Dei. Gold and Silver Jewelry and Related Objects. New Smirnova, L., 2005. Comb-Making in Medieval Novgorod York. (950-1450). BAR International Series 1369. Oxford. von Holstein, I. C. C., Ashby, S. P., van Doorn, N. L., Smith, K. N., Wärmländer, S. K. T. S., Vellanoweth, R. Sachs, S. M., Buckley, M., Meiri, M., Barnes, I., L., Smith, C. M. & Kendig, W. E., 2015. Residue Brundle, A. & Collins, M. J., 2014. Searching for analysis links sandstone abraders to on San Nico- Scandinavians in pre-Viking Scotland: molecular las Island, California shell fishhook production. fingerprinting of Early Medieval combs. Journal of Journal of Archaeological Sciences 54. Archaeological Sciences 41. Smith, K.N., Vellanoweth, R.L., Sholts, S.B. & Wärm- Vuković-Bogdanović, S. & Bogdanović, I., 2016. Late länder, S.K.T.S., 2018. Residue analysis, use-wear Roman bone anvils from Viminacium. Vitezović, patterns, and replicative studies indicate that sand- S. (ed.). Close to the bone: current studies in bone tech- stone tools were used as reamers when producing nologies. Institute of Archaeology. Belgrade. shell fishhooks on San Nicolas Island, California. Wärmländer, S. K. T. S., Wåhlander, L., Saage, R., Rezak- Journal of Archaeological Sciences: Reports 20. hani, K., Hassan, S. A. H. & Neiss, M., 2015. Analy- Stjerna, N., 1998. A short notice on the manufacture of sis and Interpretation of a Unique Arabic Finger copper-wire at Birka. Laborativ Arkeologi 10–11. Ring from the Viking Age Town of Birka, Sweden. Söderberg, A., 2006. Om två metallurgiska processer Scanning 37. knutna till vikingatidens betalningsväsende. Situne Zachrisson, S. (ed.), 1970. Sameslöjd – Sydsamernas horn-, Dei. trä- och tenntrådsslöjd. Arbetsteknik, mönster och histo- – 2011a. Eyvind Skáldaspillir’s silver – refining and ria. Umeå. standards in pre-monetary economies in the light Özsen, I. & Willer, F., 2016. Gezogener antiker Draht? of finds from Sigtuna and Gotland. Situne Dei. Zur Drahtproduktion des Kettenpanzers aus Zem- – 2011b. Fas-, tomt- och husbeskrivningar. Wikström, plín. Restaurierung und Archäologie 9. A. (ed.). Fem stadsgårdar – arkeologisk undersök-

Fornvännen 114 (2019)