Italian leech-shaped glass fibula bow from the Hallstatt period, discovered in Poland

The article presents the results of a laboratory analysis of the glass of a leech-shaped fibula bow discovered in a Hallstatt C grave in Gorszewice, Greater Poland. The fibula comes from Italy and is dated sometime to the end of the 8 th or the 7 th century BC. Both of the inner sides with a yellowish colour and the external side that appeared dark blue, nearly black, were examined. Like low magnesium and medium potassium glass of glassy faience present in the Hallstatt C period, the glass of the Gorszewice fibula bow is characterized by medium contents of K 2 O, high Al 2 O 3 and Fe 2 O 3 , and a relatively high level of B 2 O 3 , TiO 2 and BaO. The inner and outer glass of the Gorszewice fibula bow have an identical, or nearly the same chemical composition. Higher levels of several oxides occurred in the dark glass, indicating that they must have been introduced with the colourant. The yellowish glass was not coloured intentionally; its colour is the result of the presence of iron compounds in the sand.


Introduction
The cemetery in Gorszewice (western Poland) is attributed to the Lusatian culture and is dated to Hallstatt C (about 750/700-600 BC) and the beginning of Hallstatt D (Pieczyński 1953;Narożna-Szamałek -Szamałek 2007). The graves excavated at the site featured a large number of objects imported from Italy and the eastern Alpine regions. A richly furnished grave (LV) from the Hallstatt C period yielded, among other items, a bronze fibula with a glass leech-shaped bow (Fig. 1).
Fibulae of this kind, Glasbügelfibeln in German (e.g . Haevernick 1959;Koch 2010; sometimes reffered to as brooch slider or brooch decoration in English), vary in terms of their shape, decoration and production technique. According to L. C. Koch (2010, 276), the Gorszewice fibula is a 1dA form in her classification, although by size it should actually be a 1cA form. Jewellery of this type was common in the Apennine Peninsula, mainly in the vicinity of Bologna and in Verucchio, around the end of the 8 th and in the 7 th century BC (Koch 2011, 103). The Gorszewice brooch is the only 1cA or 1dA fibula found outside Italy (Fig. 2). It must have come to Poland either from Bologna or from its close vicinity (Koch 2010, 194).
A few of the 200 known Glasbügelfibeln have had the chemical composition of the glass analyzed. C. Braun (1983, Tab. 20) examined a leech-shaped fibula bow of unknown provenance dated to Hallstatt D, and A.C. Towle examined four leech-shaped fibula bows from 800-600 BC, discovered in Italy (however, fully quantitative results were obtained for only five analyses of samples taken from three artefacts, Towle 2002, 315;Towle -Henderson 2007, 50 ff.). The glass of the Gorszewice bow was analyzed once before, a quarter of a century ago (Frána -Maštalka 1990, Tab. 2 and 3), but because of the imprecise determination of levels of potassium oxide among others (the K 2 O content was marked as "<2.5%"), this analysis was not fully successful. Glass chemical composition analyses were carried out (but not published) for two leech-shaped fibula bows from the cemetery of Magdalenska Gora in Slovenia (Towle -Henderson 2007, 57).
Glasbügelfibeln representing the form discovered in Poland are generally well-known in terms of provenance (Italy, most likely the vicinity of Bologna) and chronology (end of 8 th -7 th century BC). An examination of the fibula bow from Gorszewice is of considerable significance for indicating the place of production and dating of other objects found in central Europe and made of a similar kind of glass. The aims of this paper are thus: (i) to characterize the glass chemical composition of the fibula bow from Gorszewice; (ii) to point out similarities and differences in the chemical composition of the inner and outer glass layers; (iii) to compare the results with the outcome of analyses of leech-shaped fibula bows discovered in Italy; (iv) to make a comparison with the results of analyses of artefacts (beads, heads of pins) made of glassy material dating from the Hallstatt C-D discovered in Poland.

Samples
The matrix of the fibula bow from Gorszewice was comprised of two layers of glass: the inner (sample no. 23a/1), which was weakly translucent and yellowish, and the outer (sample no. 23a/2) not translucent, and dark blue, nearly black (black is the impression to the naked eye, while dark blue is visible under a microscope). The decoration was composed of 24 threads of opaque yellow glass applied to the matrix (at a right angle to the channel axis), subsequently scratched with a burin 12 times in alternating directions, to one side and then to the other (parallel to the channel), forming in effect a herringbone pattern in relief on the body (Purowski 2012, 82, 104, Fig. 30).

Analytical techniques
An ELAN 9000 (Perkin Elmer SCIEX, Canada) inductively coupled plasma mass spectrometer equipped with a laser ablation system LSX-213 (CETAC, USA) was used to obtain information about the elemental composition of the glass fibula bow. The laser ablation system combines a stable, environmentally sealed 213 nm UV laser (Nd-YAG, solid state) with a high sampling efficiency, variable 1 to 20 Hz pulse repetition rate and maximum energy up to 5 mJ/pulse. The experiment was performed using Ar as the carrier gas. The samples were placed inside the ablation cell with the Standard Reference Material NIST SRM 610 (http://www.nist.gov/). Three replicate single point ablations (φ=100 micrometers; 4.5 mJ of energy and repetition rate equal to 10 Hz) were carried out on each sample to register signal intensities for the selected isotopes with a 10 ms dwell time.
The fibula was too large to fit into the commercially available closed ablation cell, therefore a homemade open cell (Wagner et al. 2011) with an effective volume of 4.5 cm 3 was used within this work.
It was important to seal the ablation cell while mounted to the fibula for the entire measurement period. The fibula is characterized by an uneven rough surface, which cannot be damaged during the investigations. To protect the sample-cell configuration from unfavourable possible air leaks between the sample and cell, the fibula was wrapped tightly with elastic PARAFILM (Pechiney Plastic Packaging), and the open cell was attached directly to its surface prior to the measurements.
During the measurements, 39 major, minor and trace elements were determined in the inner and outer glass of the fibula bow (Tab. 1), with three replicate single point ablations taken from the glass surface. The NIST SRM 610 calibration material was measured twice at the beginning and twice at the end of each run to correct the instrumental drift using the algorithm proposed by Longerich et al. (1996). Signal intensities were recorded for the following isotopes: 7 Li, 11 B, 23 Na, 26 Bi,232 Th and 238 U. Transient signals were saved and the background corrected and integrated using the LAMTRACE program created by Jackson (2008). The results were recalculated to the content of the oxides with NIST 610 as the external standard and 29 Si as the internal standard. Sum normalization to 100 wt% was applied based on the corresponding oxide concentrations.
The accuracy of the measurements was established by comparing the results for archaeological reference glass Corning B examined as an unknown sample to the values recommended for this glass in the literature ) and which are given in Tab. 2.

Types of glassy materials known from Poland
Glassy materials from the Early Iron Age in Poland can be subdivided into glassy faience and true glass (Purowski et al. 2014). The physical structure, meaning the quantitative proportions between Archeologické rozhledy LXVIII-2016 111 glass and unreacted crystalline grains (most often quartz) or newly formed grains and crystals, is the classification criterion (cf. Santropadre -Verità 2000;Angelini et al. 2004). Based on potassium and magnesium oxides levels, the chemical composition of glassy faience can be divided into LMMK (low magnesium and medium potassium glass) and LMG GF (low magnesium glass of glassy faience), whereas true glass is represented by HMG (high magnesium glass) and LMG (low magnesium glass; Purowski 2012;2013;Purowski et al. 2012;2014).

Chemical composition of the glass from Gorszewice
The inner glass (sample no. 23a/1) and the outer glass (sample no. 23a/2) of the Gorszewice leechshaped fibula bow contain similar or identical amounts of Na 2 O, K 2 O, CaO, MgO and Al 2 O 3 (Tab. 1). The K 2 O content is average, MgO and CaO rather low, Al 2 O 3 high; B 2 O 3 , TiO 2 and BaO relatively high. The similar properties of the glass among glassy materials from the Early Iron Age in Poland have been noted for LMMK glass 1 (Figs. 3-5). Similar quantities of the said compounds were also recorded for some glass leech-shaped fibula bows from Italy, especially samples 177, 178 and 373 (marked in Figs. 3-4) examined by A.C. Towle. However, the glass from Italy is distinguished by a higher CaO content (>2.7 %); the B 2 O 3 content was not significant (Towle 2002, 315;Towle -Henderson 2007, Tab. 5).
The glass from Gorszewice was made of a two-component batch (sand and soda source). It cannot be ruled out that the glass from Italy with a higher CaO content received a third component, that is, a lime-bearing raw material (possibly sand with a higher content of calcium compounds).
The sand was also a source of other components besides silica, that is, compounds of aluminium, iron (the dark glass also contained another source of Fe 2 O 3 ; see below), titanium, barium and boron, which usually occurred in a higher concentration in LMMK glass compared to HMG, LMG and LMG GF glass (Purowski et al. 2014, 297). The high amount of aluminium compounds suggests the use of granitic sands, which are very rich in a feldspathic component (Arletti et al. 2010, 710). Relatively large amounts of barium compounds presumably entered the glass with the feldspathic component in sand (cf. Silvestri 2008Silvestri , 1498Panighello et al. 2012Panighello et al. , 2950. In the two glasses forming the fibula bow from Gorszewice, ZrO 2 is well correlated with Al 2 O 3 , TiO 2 , B 2 O 3 and SrO (Fig. 6). The strontium content in ancient glass is dependent on whether the sand used in production was inland or coastal. In the first case, with calcium carbonate derived from limestone, there is less SrO (<0.02%) and more ZrO 2 (>0.015%); in the second (if Mediterranean coastal sand is used) ZrO 2 (<0.01%) is typically low and SrO (>0.03%) high due to the aragonite in shells Table 1. Chemical composition of the glass of a brooch slider from Gorszewice (oxides wt%) obtained with LA-ICP-MS. Tab. 1. Chemické složení skla návleku lučíku spony z Gorszewice (údaje v hm.%) získané metodou LA-ICP-MS. that are part of beach sand (e.g., Freestone et al. 2003;Silvestri 2008;Panighello et al. 2012 and references therein). Relatively high ZrO 2 (0.019% and 0.02%) and low SrO (0.014% and 0.016%) in the Gorszewice glass indicates the use of inland sand in production. A similar situation was observed for most LMMK glass from Poland, whereas HMG from Hallstatt C from the Świbie site produced entirely different results (ongoing study).
Average amounts of K 2 O and low MgO in the fibula bow glass make it difficult to indicate the kinds of soda (whether natron or plant ash) used as a fluxing agent in the glass batch. Towle -Henderson (2007, 57) determined the glass from Italy to be low magnesia (natron) soda-lime-silica glass. In our opinion, it is more likely that they were melted using plant ash (Purowski et al. 2014, 297).

Colourants of the glass from Gorszewice
The inner (yellowish) glass and the outer (dark blue going into black) glass of the Gorszewice fibula bow differ in their content of CoO, NiO, CuO, Fe 2 O 3 , PbO and Sb 2 O 5 (Fig. 7), and to a lesser degree also SnO 2 , ZnO and As 2 O 5 . Higher levels of these oxides occur in the black glass, showing that they were introduced into the glass with the colourant. The yellowish glass was not coloured intentionally, hence its colouring is due to the presence of iron compounds in the sand used for its production.
The presence of cobalt oxide in sample no. 23a/1 indicates the intent to give the glass a blue colour. Even a small amount of CoO (> 0.005%) turns the glass blue (Bachtadze et al. 2002, Tab. 3), whereas the glass from Gorszewice contains a seldom seen amount of 0.88% CoO (for examples of such high content, see Purowski 2013, 58;Purowski et al. 2014, Fig. 16). This excessive amount may have been the reason the glass is nearly black (Santropadre -Verità 2000, 39;Towle 2002, 108). Alum, used commonly for colouring glasses during the New Kingdom in Egypt, was certainly not the colourant in the case of the Gorszewice fibula bow. In the Egyptian glass, the presence of cobalt is correlated strongly with higher levels of other elements, mainly oxides of aluminium, magnesium,  manganese, nickel and zinc (Shortland -Tite 2000, 145). Sample no. 23a/2 (coloured) demonstrates practically the same amounts of Al 2 O 3 , MgO and MnO as sample no. 23a/1 (not coloured).