The Soros Hill source was first reported towards the end of the 19th century as part of an early archaeological investigation of the Cyclades (Bent, 1884: 52). It represents one of the four major obsidian sources of the Aegean (Figs. 1 and 2), alongside Sta Nychia and Dhemenegaki on nearby Melos, plus Giali to the east in the Dodecanese islands. Renfrew et al. (1965) provided the source's first detailed report as part of a larger study of eastern Mediterranean obsidian, and analyzed three geological samples as part of the Old World's first major obsidian characterisation project. Using a bivariate Barium vs. Zirconium contents plot, they were able to discriminate the Antiparos obsidian from the products of other Aegean sources, together with those of the western Mediterranean, Carpathians and central Anatolia. The Soros Hill obsidian comprised the sole constituent of their source Group 3b (Renfrew et al., 1965: Figure 4, Table 1). Antiparos obsidian has subsequently been characterised by a number of other studies using various analytical techniques, though no project involved more than seven source samples (Table 1). It should be noted that with the exception of portable EDXRF, each of the techniques employed was destructive. Antiparos obsidian has also been analyzed for its thermoluminescent properties; the results show that it had a relatively low natural saturation peak of 350 °C (low compared to Melian and western Mediterranean obsidian), with a ten times brighter TL intensity (Göksu et al., 1988). TL has yet to be applied to Aegean obsidians for characterisation purposes, unlike K-Ar and fission track dating (Arias et al., 2006 and Durrani et al., 1971).

Archaeologically, it has long been appreciated that Soros Hill obsidian was fairly insignificant, arguably because while the material fractures conchoidally, the raw nodules were mainly in the form of “small lumps” less than 5 cm in length (Renfrew et al., 1965: 232). Instead, Sta Nychia and Dhemenegaki on Melos represented the primary Aegean obsidian sources (Carter, 2009). The preference for Melian obsidian by prehistoric tool makers is perhaps most clearly evidenced at the Late Neolithic (5th millennium cal BC) site of Saliagos (Fig. 1). While today Saliagos is situated on a small islet between Antiparos and Paros, these three landmasses were originally part of a single island, whereby the Soros Hill source was only a few kilometres to the south across dry land, yet the vast majority of the community's chipped stone assemblage was made from Melian obsidian (Cann et al., 1968).

In August 2010, we procured 44 samples from three collection areas on Soros Hill (Fig. 2). The hill, rising approximately 200 m from the Aegean at the southern end of Antiparos, is dissected by multiple shallow drainages and covered with scrub vegetation (Fig. 3); obsidian is available both in colluvium superficially mantling the slopes and exposed in eroding deposits of tuff with clasts of perlite and small obsidian nodules (Fig. 4). Obsidian is sparsely but relatively consistently available over an area of roughly 50 ha, but rarely were nodules larger than 4 cm in their long dimension observed (Fig. 5). The three collection areas were spaced approximately 500 m apart in this area, spanning an elevational range of 0–110 masl in order to capture potential variability between exposures of distinct flows. Each collection area was described in the field, documented with a recreational GPS (Garmin 60csx with an accuracy of ±6 m), and photographed. Sampling focused on in situ rather than colluvial obsidian so that the specific geospatial provenience of each sample could be confidently recorded.

Antiparos I – 36.9714N, 25.0528 E and 36.9719N, 25.0509 E. The first sample location was on the western slope of Soros Hill, where obsidian was both available in the colluvium mantling the slope and in inclusions of perlite and obsidian within exposures of tuff (Fig. 4). Inclusions of fragments of banded flows of rhyolite, perlite, and obsidian are also visible, but only up to 30 cm in their long axis and clearly reworked within the tuff deposit. The largest intact nodules of obsidian reached only 4 cm on their long axes. Eleven nodules were collected, weighing between 9 – 123 g (mean 38 g) and 1.9–4.3 cm long.

Antiparos II – 36.9662N, 25.0537 E. Approximately 0.5 km south of Antiparos I. Sixteen nodules were collected, weighing between 7–138 g (mean 19 g) and 1.3–6.8 cm long.

Antiparos III – 36.9661N, 25.0588 E. Approximately 1 km to the south-east of Antiparos I, in a modern roadcut at the foot of Soros Hill; another exposure of what appears to be the same deposit of tuff, likely the second exposure reported by Renfrew et al. (1965: 232). Obsidian nodules eroding out of this tuff matrix are predominantly less than 3 cm, but occasionally larger (one reaching 7 cm). Sixteen nodules were collected, weighing between 4–13 g (mean 8 g) and 1.4–2.8 cm long.

In each instance, the obsidian is recovered in the form of small lustrous jet-black nodules (Fig. 4). The raw material has a good conchoidal fracture habit and tends to be very homogenous, but most pieces are under 3 cm long. Moreover, we saw precious little evidence for the material having been knapped at the source, with the exception of one blade-like piece (not collected, being an archaeological artefact), that neatly illustrates the obsidian's grey hue and translucency when flaked (Fig. 6). Noteworthy is the fact that even the exterior surface is shiny, i.e. it lacks the coarser matte cortical surface that one associates with many obsidian source materials, as for example on Melos. Renfrew et al., 1965 (1965: 232) report that similarly lustrous nodules are to be found on some of the Slovakian nodules (Zemplin Hills, “Carpathian 1”), while we have seen similar small lustrous pieces from Erzincan in north-eastern Anatolia.

The homogeneity of the elemental composition of the various geological samples indicates clearly that we are, perhaps unsurprisingly, dealing with the products of a single flow (Table 2). In turn, through a bivariate Sr vs. Zr contents plot, the obsidian of Soros Hill can clearly be discriminated from the products of the other major sources of the Aegean, central Anatolia and the Carpathians (Fig. 7), i.e. Dhemenegaki and Sta Nychia on Melos (Cyclades), Giali (Dodecanese), Göllü Dağ and Nenezi Dağ (Cappadocia), Zemplin Hills/Carpathian 1 (Slovakia) and Tokaj Mountains/Carpathian 2 (Hungary). We aimed to discriminate these raw materials in particular as each has been attested in chipped stone assemblages of the prehistoric Aegean (Bellot-Gurlet et al., 2008, Carter and Kilikoglou, 2007 and Kilikoglou et al., 1996). Notable characteristics of the Antiparos obsidian are its enriched Th (47–63 ppm) and Rb counts (333–414 ppm), and depleted levels of Sr (8–16 ppm) and Ba (< 86 ppm, most below 41 ppm (Table 2)); these have been noted as defining features of Antiparos rhyolites more generally, along with high levels of Cs, U and Nb (Clapsopoulos, 1998).

This article contributes to eastern Mediterranean obsidian characterisation studies by producing a robust elemental data-set for the Soros Hill source on Antiparos. It confirms that the products of this source: •

are relatively homogenous in terms of their chemistry, size, knapping quality, colour, texture and translucency;