Lipari (Aeolian Islands) Obsidian in the Late Neolithic. Artifacts, Supply and Function

This study focuses on the Neolithic, particularly on the emergence and development of the Diana Culture in the Aeolian Islands. Since the 1950s, the archaeological excavations unearthed parts of a settlement in a plain near the sea, contrada Diana in Lipari. We discuss the technological and typometric study of obsidian from trenches XVII, XXI, and XXXVI. A series of pXRF analyses on obsidian were carried out to identify their sources. A selection of retouched and non-retouched artifacts was examined, showing the higher variability in forms than at importing sites. This significance of this workshop area on prehistoric trade is assessed.

Identifying possible prehistoric quarries (Buchner, 1949;Tykot, 2019;Tykot et al., 2006) will always be subject to objective limitations, due to the considerable transformation of the north-eastern slopes of the island of Lipari. The beginning of a stable population in the Aeolian Islands occurred in the Middle Neolithic with the facies of Stentinello. A group of people came from Sicily or Calabria to live on the islands of Lipari (settlement of Castellaro: Cavalier, 1979;Bernabò Brea & Cavalier, 1957;Nomi & Speciale, 2017) and Salina (settlement of Rinicedda: Bernabò Brea & Cavalier, 1995). The pottery was impressed (Stentinello style) and painted in red bands. There is one radiocarbon date from a piece of charred Erica cf. arborea from the Neolithic site of Rinicedda on Salina: 6325 ± 45 BP (Lab-code LTL4329A), 5390-5210 cal BC with 90.4% probability and 5470-5440 cal BC with 5% probability, using 2σ calibration with software OxCal 3.10. This date places the arrival of Neolithic peoples in the Aeolian Islands in the later 6 th millennium BC, raising the chronology of Aeolian prehistory (Martinelli, 2016).

Intensive Exploitation of the Obsidian at Lipari
The period of greatest production of obsidian artifacts in Lipari is the Late Neolithic. In the Aeolian Islands there was a population increase with the presence of many settlements dated to the Diana culture, Brought to you by | University of South Florida Tampa Campus Library Authenticated Download Date | 4/24/19 3:27 PM characterized by red surface pottery. At Contrada Diana (Fig. 2.1), today at the center of the town of Lipari, are the remains of a great settlement probably organized as groups of huts. During the Greek and Roman periods, the plain of Contrada Diana was the site of a large necropolis that damaged and partially destroyed the Neolithic layers. Some of these layers have preserved remains of hearths and pottery and a huge amount of obsidian that made it possible to interpret the Neolithic settlement as the main workshop station of the archipelago (Bernabò Brea & Cavalier, 1960).

Typological and Technological Study of a Sample of Obsidian Artifacts from the Contrada Diana Settlement
The systematic study started on obsidian knapping waste and artifacts from the workshop in Contrada Diana proved to be rather difficult given the number of products (flakes, blades, cores, etc.) collected in the various excavation trenches investigated since 1950 (Bernabò Brea & Cavalier, 1960). A typological Brought to you by | University of South Florida Tampa Campus Library Authenticated Download Date | 4/24/19 3:27 PM and technological study will be necessary to understand the collection methods and knapping technique to identify economic and social behavior (Iovino & Martinelli, 2008;Negrino & Radi, 2006). Based on macroscopic observation, confirmed by mineralogical analysis (Bullock, et al., 2017), we can distinguish three main obsidian types: black, black with phenocrysts, and gray. The first type is predominant, the others are present with amounts equal to approximately 5%. It appears that the ancient inhabitants of Lipari performed a selection of the raw material.
The partial results of the typological study show an obsidian lithic assemblage with very specialized characteristics. We were able to record all the stages of the chaîne opératoire required to produce blades and bladelets. In the first stage, the block of natural obsidian was cut and prepared at the extraction source as attested by two quarry points discovered by Buchner in 1949 inside the Gabellotto Valley. The resulting pieces were then brought into the settlement to be decorticated and cleaned from surface impurities, mostly caused by pumice. The knapping proceeded through the second stage by detaching surfaces to obtain blades and bladelets of medium, small and tiny dimensions, and some as flakes. The work in the village produced a lot of knapping waste that cluttered the floor.
At present, all the artifacts coming from trench XXXVI (Martinelli, 1994, pp. 257-269) were analyzed while the obsidian lithic industry coming from trench XVII (square A-H) is being studied (Fig. 2.2). This trench preserves integral the Neolithic layer in which four fireplaces were discovered. In this paper, we present the data obtained from the matrix squares G cut 1-2, square F cut 4 and square E cut 3. The study highlights the massive production of blades. In two histograms, it is possible to note that 9194 artifacts from the production workshop, including about 7 kg (100 pieces) of raw material, have been rejected. Histogram 1 (Fig. 3) and table 1 include the products of the second stage of debitage (preparation of the core) with 6299 pieces which represent the production of flakes of varying dimensions (debris) from knapping the core with the percussion technique. Histogram 2 (Fig. 4) and table 2 include the quantity of artifacts reaching the third stage of knapping, the final product obtained from the core (blades). We have counted 2895 artifacts for which we could distinguish and classify the blades as not retouched (Galiberti, 1990). Histogram 2 includes cores, bladelets and retouched tools (Laplace, 1964) to show the different quantities between knapping products and finished products. The point of impact of percussion on the artifacts is flat and chipped with proximal abrasions. The length of the blades is between 25 and 90 mm. The bladelets are thin and transparent with thickness between 1 and 3 mm, typical width 15 mm, length 40-50 mm. The cores of pyramidal (15%) and prismatic (85%) shape have various sizes: their height ranges between 18 and 70 mm. The cores were exploited until they crumbled or cracked. They account for a low percentage compared to the total of the lithic assemblage. In the third stage we could recognize also some examples of pressure technique.    Histogram 3 (Fig. 5) and table 3 include the synthesis of 2895 obsidian artifacts from trench XVII. The obsidian tools represent a low percentage compared to the number of by-products from knapping. The manufacture of tools for local use was probably secondary to the production of blades. Instead, flint tools are found in greater numbers than the cores and other products from knapping the same raw material. The flint tools are typologically very rich, as shows the study of lithic groups from Contrada Diana trench XVII, XXI-XXIII, IX, XI, XII; XV, XVI (Martinelli, 2000), and include all types of instruments among which are distinguished backed blades and points, truncations, geometric rhomboids, tools with flat retouch and fragments of sickles. The sickles were not produced in obsidian at the Diana site. We have recognized 175 obsidian tools divided by type (Laplace typology from 1964) as shown in histogram 4 ( Fig. 6) and table 4. They are produced from thick blades. The burins (Fig. 7) are large with lengths between 50 and 120 mm, widths 20-40 mm, and average thickness between 8 and 20 mm. We have observed a specialization to prepare end-scrapers (Fig. 8), of long and short frontal types. Three endscrapers were retouched on distal and proximal sides allowing their use on both sides. Another type of instrument present is the backed point prepared from a blade with the distal area retouched and rounded for use as an awl. The substratum group of artifacts includes a significant amount of points, blades and flake scrapers (Fig. 9), as well as denticulates. The retouch is marginal and scaled. The flat retouch is invasive in scrapers with ogival shape and large size (about 11x50x81 mm). A particular tool is an axe from a thick blade (12x36x67 mm) with two crests on the dorsal surfaces and two grooves on the sides. We have observed on the distal and proximal sides the retouch typical of the Campignano technique (Fig. 10). A lot of flakes present abrasion on sides caused probably by use or by accidental contact with surfaces. It may be difficult to understand the difference between deliberate retouching and accidental retouching within our limited selection.     The Contrada Diana settlement at Lipari has returned a huge amount of obsidian proving an intense knapping activity on location. This study recognized débitage resulting from the production of bladelets (typical length: < 50 mm; thickness: < 3 mm) and middle sized blades (length: > 50 mm).
We can summarize the processing of obsidian that took place in three main stages: 1. Cut, collection and preparation of raw material at the source; 2. Preparation of cores and production of massive amounts of knapping waste in the settlement; 3. Exploitation of cores to produce blades in the settlement.
The percussion technique is attested during the first and second stages and the pressure technique during the third stage. The tools are retouched for local use (in tiny amounts) by selecting end-scrapers and backed blades. Obsidian is not used to produce sickles: flint was used exclusively for these tools. In the Aeolian production chain, the bladelets appear to be the finished product that was then exported. In fact, their production is massive and highly specialized already during the Late Neolithic (Diana culture).

Sourcing
A random selection of 150 obsidian artifacts, representing many different trenches and squares, were selected for sourcing analysis to confirm that they all came from geological outcrops on the island of Lipari. In the central Mediterranean, trace element analysis is a well-established method for distinguishing not only the different island sources of Lipari, Palmarola, Pantelleria, and Sardinia, but also sub sources within each island (Tykot, 2017). For Lipari, the ancient sources of Canneto Dentro, Gabellotto Gorge, and Monte Guardia have different trace element values, although obsidian from Monte Guardia does not seem to have been sufficient in size for producing tools, and the minor use of Canneto Dentro suggests it was quite limited in the size/quantity available (Tykot et al., 2006;Tykot, 2017Tykot, , 2019. Using a portable, hand-held Bruker III-SD model X-ray fluorescence (pXRF) spectrometer, analyses were conducted within the Lipari Museum complex using settings of 40 kV, 11 µA, and 90 second, with a filter that enhances the precision of trace elements Rb, Sr, Y, Zr and Nb. The data produced were calibrated against 40 known obsidian standards incorporated in a software program specifically for this instrument, and also directly compared with geological obsidian samples from Lipari analyzed with the same pXRF ( Table 5).
The data obtained for the trace elements by the pXRF are highly precise and accurate. The values have been compared with those from geological samples analyzed on the same instrument, and those obtained using different technologies. XRF has limits in producing accurate values when used non-destructively on surface analyses when compared with destructive methodologies because of the low penetration of the X-ray on the artifact, which makes it vulnerable to the accidental analysis of eventual extraneous elements on the surface, as well as alteration of the original values due to weathering, although this is rarely an issue on smooth-surfaced glassy obsidian. The values obtained from the pXRF were calibrated using Compton normalization, mitigating the effects of air intrusion and variable artifact geometry, and the calibration software used incorporates 40 obsidian standards analyzed by INAA, LA-ICP-MS, and XRF by MURR. It is also possible to verify ratios of key trace elements, which are consistent for XRF instruments in general. There is a substantial literature assessing the scientific value of pXRF data in consideration to its technological limits (see Kasztovszky et al., 2018 for a recent discussion, including the Lipari geological samples, by a team not connected to the authors).
The trace element values, along with ratios of Fe/Sr and Rb/Sr, were used to determine that 149/150 of the artifacts tested specifically came from the Gabellotto Gorge subsource on Lipari, while one artifact (#25351) turned out not to be obsidian. Two of the artifacts (#25366, #25386) have slightly higher Sr values than the others, suggesting they may have come from a different area within the Gabellotto Gorge.

Discussion
The massive production and exchange of Lipari obsidian started at the beginning of the local Middle Neolithic and continued into the early Bronze Age, as demonstrated by findings across Italy and in neighboring regions (Tykot, 2011;Tykot, Freund, & Vianello, 2013). During the middle Neolithic, it is likely that the island of Lipari was occupied seasonally (in summer) by people that benefited from calm seas for navigating and during those months the activity of obsidian tools production took place regularly (Robb & Farr, 2008, p. 40). During the Late Neolithic, the production was generally intense, but the original seasonality was nonetheless preserved given that most of the production was aimed for export, and export could only take place during late spring and summer. The impact that trade and exchange had on these communities is substantial, and should not be thought as a constant and febrile production to prepare artifacts that could be exported only for a short time during the year. We have suggested some preference for black obsidian, but even so, too much material, including raw material, was discarded, and this is better explained by seasonal cycles of production (e.g. fall arriving earlier than expected and scrapping the last shipment) than excessive precision in producing the finished products.
There is ample evidence that bladelets were the main tool produced (Tykot, 2017), and that they were used as far as the southern part of the Po Valley in Emilia Romagna, with much lower quantities further continued north. The Alps provided abundant alternative materials such as greenstone, limestone and schist and others, which significantly limited the need to import obsidian, but obsidian artifacts from Lipari still reached the Alps in mostly smaller quantities, reaching southern France, and also Croatia by crossing the Adriatic Sea (Tykot, 2014). There are very few variations in the typology, primarily the medium length blades are appreciated more west (in the Palermo area of Sicily) and north (Tuscany and Emilia Romagna). Such a gigantic distribution of a product whose origins are firmly based on a tiny island north of Sicily has effectively encircled other areas of production of obsidian tools. Pantelleria obsidian reached in significant quantity only western Sicily and Malta (Tykot, 2017(Tykot, , 2019Tykot et al., 2013). Palmarola obsidian is also found in much of central Italy, even reaching the Adriatic, but almost always in much lesser percentage than Lipari. In both cases, these are small islands geographically located even farther away from the coast than Lipari. Sardinian obsidian did not circulate in any quantity in southern Italy, and only one piece has been identified so far in Sicily. There is some evidence for a limited (mostly on the southeastern coast of Sicily and in the immediate coastlines facing Lipari) direct trade of blades that is indirectly proven by lower presence of cores or debitage from cores and a high similarity in the morphometrics of the finished product (Freund, Tykot, & Vianello, 2015). In the rest of Italy, prepared cores are definitely the product that was exported, and this accounts for the small typological differences. Knapping obsidian requires skills that are in part different from those used for knapping flint, mostly due to the fact that obsidian is a glass that can shatter into a myriad of very sharp flakes if hit badly, and is therefore a hazard for people unfamiliar with the material. However, producing blades from cores must have reduced considerably this hazard, and may explain why obsidian travelled so far and its use changed so little in such a culturally diverse and vast region.
The size of the region and limited variability created also the conditions for substantial standardization in the manufacturing process, which is recognizable at Contrada Diana and elsewhere in Lipari. Most finished tools were bladelets and flakes, probably with a number of cores prepared to obtain the very same types, all were made in black obsidian and at least all those analyzed came from the same source, Gabellotto (Canneto Dentro accounts for 1% or less when detected, and the small size of the sample compared to the available evidence may have been insufficient to detect it).
The main form of distribution is through prepared cores, which were traded along main and secondary trade routes, and then worked (knapped) at regional centers, from where the blades were redistributed to individual sites. This type of trade is particularly evident in Calabria, where there are massive amounts of obsidian at Tyrrhenian sites facing the Aeolian Islands (e.g. Acconia Plain, Ammerman, 1985), with much smaller amounts and significant traces of cores on the Ionian coast (opposite). It is likely that people from both eastern Sicily and Tyrrhenian Calabria were involved in the exchanges, and whilst the activity may have been seasonal in Lipari due to the challenges presented by maritime crossings, in Calabria and elsewhere in peninsular Italy the trade may have been all year long.
The preparation of cores effectively prevented the production of other types of artifacts, and resulted in a standardization that encompasses the whole Italian peninsula, with minimal differences on the length of blades (they are longer in the north), but nothing substantial that would change the category of artifact or its function. This means that not only the type of artifact was the same, the function was the same as well, and consequently the needs that triggered its procurement were the same. For such a widespread use, across culturally different regions and throughout the whole Neolithic and Copper Age, we can only suggest a use inserted in a technological package associated with agriculture and farming. The local chronologies across the Italian peninsula may vary from the Aeolian one, and whilst the Aeolian chronology is one of the better studied and precise, at several sites obsidian is detected from the very beginning of the Neolithic period. Obsidian was not used as sickles or as an implement in threshing boards, but rather used as for tools aimed at cutting and slicing different materials. Working hides and cutting agricultural produce may have been an essential function.
The geographic distribution of obsidian matches the area covered by the "wave" of technological spread (Sargent, 1983;Brown, 1991;Skeates, 2000;Whitehouse, 2014) that arrived from Greece into the Tavoliere (Apulia) of southeastern Italy, and then spread northwards to reach the Po Plain, where a second wave from Central Europe arrived later (Isaac et al., 2010;Jones et al., 2012). This provides perhaps an insight into the Brought to you by | University of South Florida Tampa Campus Library Authenticated Download Date | 4/24/19 3:27 PM patterns of distribution of Neolithic technologies. From radiocarbon evidence Zilhão (2001) suggested that the expansion of farming in the west Mediterranean (Italy to Portugal) followed the process of maritime pioneer colonization, in a fast (up to six generations) and recognizable pattern of diffusion.
The standardization of Lipari obsidian suggests that the Neolithic package included only broad technologies, and perhaps a variable sample of plant and animal domesticates and tools. Choices depended on the local environment and local culture. Obsidian can therefore be interpreted as a local solution to produce a needed tool given its availability. Interpreting its pattern of distribution results therefore in high standardization until a natural frontier was encountered such as a mountain range or a significantly different environment or cultural practices. Obsidian consumption drops almost immediately past this frontier, in Italy located at the Po Plain, despite evidence that the exchange networks were able to carry obsidian much farther. In this perspective, obsidian may be considered an essential cutting tool to enable a certain type of production and consumption, which must have been similar to other solutions adopted elsewhere, and yet characteristic of the Italian peninsula. This combination of ability to spread technologies within a few generations while maintaining significant local traits in culture and socioeconomic organization, only partly dictated by the environment, is not fully understood. If it will be confirmed by future research, then Neolithic exchange networks were far more active and efficient than previously thought, and they acted as long distance information gateways, and probably as corridors for human mobility as well, in addition to moving products and materials. This would translate in the network being more sophisticated than as it appears from tracking just material culture.

Conclusions
The production and consumption of obsidian in Lipari is very significant not for the amounts of material worked locally, or their antiquity, or the typological diversity of the tools produced, but for its range of distribution and high levels of standardization, which is not immediately apparent considering only the sites within Lipari itself. Obsidian production was obviously a primary economic activity at sites in Lipari (Martinelli, 2016), but it is difficult to assess consumption and the likely link with farming and the Neolithic package. The sample studied, even if partial, allows us to hypothesize equivalence at the production stage between artifacts for local use and for export. In histogram of fig. 4 the quantity of instruments and bladelets are similar. This confirms the stability of the Diana settlement during the late Neolithic (Diana facies), a period in which the population extend to inhabit the other islands of the Aeolian archipelago.
Yet, the analyses of obsidian tools across Italy are revealing the importance of this trade, and how the distribution of obsidian matches both chronologically (Early Neolithic up to Middle Bronze Age) and geographically a whole wave that introduced and developed locally farming activities in the Italian peninsula and Sicily, and extended to Sardinia and beyond. This provides a unique opportunity to assess a specific category of artifacts, obsidian blades, together with plant and animal remains to track the effects of the adoption of farming across a vast area.
Obsidian tells us a story of rapid adoption, from the earliest Neolithic, which is confirmed by radiocarbon dates at farming sites. It also reveals how the march of penetration into the peninsula used both maritime and land routes, and how the mechanisms of redistribution saw, in the Neolithic, artisans operate at short distance, with a redistributive system able to replicate tool production at major sites on these routes thanks to prepared cores, and subsequently irradiate the tools at smaller sites in the surrounding area. It is an extremely efficient system, which is still in use today, for example by logistics companies. This made possible for Lipari obsidian to be known and regularly used 1,500 kilometers from the island itself, and probably in less than the six generations required to reach Portugal. On account of distance, as little as two generations passed to reach the farthest corners of its regular use area. More impressively, obsidian tools lasted many generations, in fact millennia, which is a testament to the resilience of the exchange networks. We still do not know exactly what obsidian was used for, except for a somewhat generic assertion that the tools built with obsidian blades were very efficient at cutting and slicing, with very few use wear studies having been done. It is actually possible that they were used like knives in a modern kitchen, and therefore employed to prepare foods for storage and Brought to you by | University of South Florida Tampa Campus Library Authenticated Download Date | 4/24/19 3:27 PM cooking, to help eating foods, and for other tasks that required their action (for example: wood and bone working, Iovino & Martinelli, 2008). There is nothing preventing obsidian blades to have been multi-purpose tools, with only the implements in which they were inserted limiting their use.
Indeed, the various proportions of domesticated plants and animals adopted by individual communities suggest that the obsidian tools we find today, without implements, are the materialization of a basic technological knowledge that spread far and fast. Their function may have been very different even at household level, with experimentation and adaptation to local needs always being a strong component. The sites on major routes received prepared cores and do not seem to have ever attempted to produce different tools, despite their varied lithic productions. Sites that received prepared blades were probably unable to request changes in the typology, but were probably free to experiment in obtaining different shapes from the available cores or use different implements. It is likely that the latter were more varied than we can detect, and the standardization we recognize today would have been lost to individual consumers in prehistoric times, in the same way that planting different seeds may look similar to us when we consider the technique, but actually people planted different species for different needs. We should therefore be cautious in over interpreting both standardization and range of use of tools from one site, which are significant to reveal some processes, but are still limiting in revealing how they affected daily life.
In Lipari obsidian was produced in massive quantities largely for export, and the only distinctive trait from the region in which Lipari obsidian is distributed is the larger number of types of lithic tools in obsidian found there. It is not possible to explain Lipari obsidian tools without considering the exchange network that carried them far away, because their function and typology was most likely determined by needs originating across the whole region of distribution. The farming needs in Lipari were likely so modest not to warrant any kind of selection or differentiation from the mainstream production of blades, and other tools were produced in the brittle and less efficient obsidian only because of familiarity in working obsidian and the easy availability of the raw product. Only accounting for the profound effect in the life in Lipari and the Aeolian Islands we can explain such a high number of blades, bladelets and sharp flakes in our sample, and the high degree of standardization in their production.