KONISPOL CAVE, ALBANIA:
A Preliminary Report on Excavations and Related Studies,
Chronology and Chronometry I: Radiocarbon Dates
Chronology and Chronometry II: Archaeomagnetic Dating
Investigations at Xarë, 1992
The Konispol Cave has produced a modest but typical selection of small finds indicating a variety of crafts and economic pursuits in the prehistoric period. Notable are terracotta spindle whorls typical of the Neolithic and Early Bronze Ages of the southern Balkans and a variety of bone awls and polishers. One hollowed and pointed longbone (found in a mixed Late Neolithic-Eneolithic unit in Trench VIII in 1992) is of a type known at Starçevo sites as a fishing tool. Only one groundstone tool has been excavated from the site, a small gray celt from an Early-Middle Neolithic unit in Trench XXI. Notable is a fragment of an obsidian blade; on visual inspection its source appears to have been the island of Melos in the Cycladic Islands. It should be noted here that obsidian has been found only very rarely on Albanian sites.
In 1993 we recovered a buff-colored sandstone pillow-shaped object with nine grooves incised in one face; it was perhaps an object used for grinding or dulling (Fig. 12). The site has produced a number of fragments of clay "anchors" (Fig. 10:5) which are known over much of Macedonia, Thessaly, and southern Greece (Forsén 1992:221-32). Their function is unknown, but they are horizon markers of the Early Bronze Age. Finally, we would note the recovery of a few pieces of what appear to be the feet of Danilo "coal scuttles"; these are small theriomorphic vessels attested widely in the Neolithic Balkans. Again, their use is unknown, but they are classified as rhyta by Albanian archaeologists and are believed to have had religious or cultic function.
A regime of water-sieving was instituted in field season 1993. A fair amount of trial and error was necessary before we settled on a routine that worked. The amount of water available restricted us to a sampling of the earth from the trenches: from every excavation unit in 1993 (in all trenches) we water-sieved between 6 and 12 liters of earth.
The deposits from the cave proved difficult to break down on account of their heavy clay content, thus slowing the processing. Owing to these limitations, our water-sieving operation was something less than state-of-the-art; nonetheless, we were able to make some progress in documenting the natural and human-influenced vegetation during the Stone Age. Many dozens of samples of carbonized wood were taken to the U.S. for microscopic examination. Most of the samples examined thus far are hardwood species; a few pieces from pre-Neolithic levels appear to be softwoods (i.e., conifers).
Among the carbonized seeds identified thus far are the following: One grain of emmer wheat from an Early or Middle Neolithic level in Trench XXI; one fragmentary vetch from the same excavation unit; and one lentil from pre-Neolithic levels in Trench XXI, together with several seeds that may be wild grape. Domesticated grape seeds occur in the upper levels from Trench VIII, but several of these are uncarbonized, and hence probably of recent deposition.
Only the most preliminary statement can be made at this point about the paleoethnobotany of Konispol Cave. Certainly the emmer wheat and vetch indicate a typical Early Neolithic agricultural assemblage for the eastern Mediterranean. The presence of lentil in pre-Neolithic levels is also common in the eastern Mediterranean (e.g., at Franchthi Cave in southern Greece), as well as in Sicily (Grotta dell'Ouzzo) and southern France (L'Aberadeur). No conclusions can be drawn yet from the wood charcoal. It will be necessary to establish at least the common genera that were exploited before environmental implications can be considered.
In the 1993 season, we began a survey of the modern vegetation in the vicinity of the Konispol Cave, from the top of the Saraqint Ridge to the alluvial plain below.
The Konispol Cave excavation is the first archaeological project in Albania to save all faunal remains for specialist analysis. The 1993 season was begun by examining the bone recovered during the 1992 season. The entire faunal sequence from Trench VIII in both seasons was analyzed, spanning the occupation of the cave from the preneolithic period to the present day. This single excavated sequence provides the basis for the following preliminary report.
The remains consisted of 4,141 pieces of bone, of which 1,280 (30.9%) were identified to at least the genus level (or sub-family, in the case of sheep/goat). The bone is generally well preserved, although a few pieces show signs of being worn, perhaps by running water in the cave. A striking exception occurs in Unit 22 and to a lesser extent Unit 23, where most of the bone is very poorly preserved, appearing as if it had spent considerable time near the surface in a loose matrix. This may represent a hiatus in deposition--and hence perhaps occupation--that will be investigated stratigraphically and with respect to other materials recovered. Another, probably shorter, such break may have occurred around Unit 12, where there is relatively little bone but a noticeable amount of lime concretion that appears to have been formed as a result of dripping water.
For the purposes of this preliminary report, our analysis of the sheep and goat are combined, since in many cases they cannot be distinguished. The fauna are tabulated in terms of diagnostic zones (Watson 1979), which provides a good system of quantification for the description of the balance of species.
Not surprisingly for a site in the Mediterranean zone, sheep and goat predominate through much of the sequence, forming at least 50% of the identified fauna in all excavation units from 21 up. In some cases they constitute more than 90% of the fauna. Cattle (nearly all of which appear to be domestic) are of relatively little importance, in only one case rising as high as 20% of the fauna. This is despite the fact that taphonomic biases would be expected to favor the preservation of the larger cattle bones over those of sheep and goats. Pigs (mostly domestic) play a similarly small role in the later units, but wild pigs are often a quite significant portion of the fauna of the lower units. Similarly, red and roe deer are variable in abundance, but generally are not of major significance in the later levels. They tend to be much more frequent, however, in the earlier units.
One question of obvious significance to be addressed through the fauna is at what point in the sequence domesticated animals appear. We are as yet unable to give a definitive answer at this time, although we can make a beginning. Pottery disappears after Unit 18. Domesticates are definitely present as deep as Unit 21, however, as attested by unmistakable sheep and goat horn cores. (That these were animals raised at the cave rather than obtained by hunters through trading with--or raiding--contemporary farmers, is suggested by the presence of a shed deciduous sheep/goat tooth.) Small and probably domesticated pigs make their earliest appearance in Unit 22. All cattle from Unit 23 and below appear to be wild. The complication arises from the presence of small bovids more or less throughout the sequence. (The lowest units have very little faunal material at all, making their characterization more difficult.) Some of these are definitely ibex, and some may be chamois. Due to inadequate comparative material and the intrinsic difficulties of separating the congeneric goats and ibex, we have not as yet been able to establish whether all the small bovids in the lower levels can be assigned to these two species. Contamination from higher levels either through postdepositional processes or the occasional bone dropping out of a scarp from above (particularly a consideration for Unit 32, the first unit excavated in 1993 after the trench stood open over the winter) is also a possibility. We suspect that the lowest levels (ca. Unit 28 and below; Unit 24 is the lowest with substantial amounts of fauna, so perhaps the distinction should be made somewhat higher) may originally have contained no faunal materials at all. These units have only a few small pieces of bone that might well have been brought down by rodents or other postdepositional processes. Such movement would obviously have happened above as well, and is particularly likely to bring down the bones of smaller animals such as sheep and goat.
As a working hypothesis, we propose that the layer of heavily weathered bone seen in Units 22 and 23 marks a break in occupation between the pre-Neolithic and Neolithic levels. There is no unequivocal evidence for domesticated fauna below this layer. Small bovids drop dramatically in proportion at about this point, and domesticated cattle and pigs disappear. Furthermore, we would suggest that the earliest levels represent not an occupation per se, but rather the use of the cave as a comfortable, protected spot to carry out primary reduction of local stone materials by people who lived and ate elsewhere. The implication of this scenario, sadly, is that the cave might have relatively little to tell us about either the putatively Upper Palelolithic occupation of the region or the nature of the Mesolithic/Neolithic transition. On the plus side, there is the possibility that we might be able to document an aceramic Neolithic phase, and of the Mesolithic/Epipaleolithic occupation that preceded it by an unknown separation in time. Moreover, although there is some evidence of carnivore modification of the bone assemblage, it appears that humans and their dogs were responsible for the vast majority of the fauna as recovered. There is no indication that the cave being occupied by bears or other wild carnivores.
It is particularly striking that the sheep/goat remains are throughout heavily weighted toward very young ages. As can be seen in Fig. 13, this exceeds even the rather extreme milk maximization kill-off strategy documented by Payne (1973) for Anatolia, wherein most males are eliminated shortly after birth. This is despite the fact that postdepositional processes tend to eliminate differentially the bones of younger animals. Many of these animals appear to have been very young indeed (fetal, neonatal, or under three months). While further analysis is indicated, the pattern thus far suggests that we are recovering only a part of the seasonal round. It would seem that the cave was occupied for only a part of the year, at least from the Neolithic period on. It may have been functionally restricted as well, e.g., occupied by shepherds with their flocks. Such a functional restriction is likely to be reflected in the artifact assemblage, although it would be useful if we had a broader range of sites from the region--including open-air Neolithic settlements--for comparison. The fact that roasting rather than boiling appears to have been the predominant cooking method could be a result of such functional restriction; again, it would be helpful to have some other local faunal assemblages for comparison.
These comments are indeed preliminary. In particular, little allowance has been made for postdepositional alteration of the assemblage. The study of the metrical data, use of better comparative material, seasonality analysis, and the examination of material from other trenches may well alter these results and their interpretation.
Finally, we note that the water sieve yielded a few dozen vertebrae and ribs of small fish in an excavation unit in one trench only (XXI; ceramic context mixed, but mostly Eneolithic). In addition, several dozen scales of sturgeon were recovered from Unit 9 (Late Neolithic) in Trench XII. The finding of only a dozen or so marine mollusc shells from our two seasons of excavation in the cave--Ostrea, Monodonta, and Patella--suggest only passing interest in the exploitation of the sea. These items were perhaps picked up as curios, and do not suggest that the sea was a source of food for those who used the cave. The few valves of Mytilus might have come from the brackish lagoon just north of Butrint, which is in fact farmed for mussels today.
In the 1993 season, four locales were cored in an effort to document the Pleistocene and post-Pleistocene environment of southern Sarandë district:
(1) In the marshes of the alluvial plain near the village of Suk, bordering the saline lagoon near the Roman site of Butrint. A core 5.50 m long was retrieved. The deposits consist mostly of gray clay with inorganic layers and pieces of wood from a previous swamp forest; at 4.70 m marine influence is concluded from the presence of Cardium shell fragments.
During the 1993 season the Konispol Cave itself was also investigated. A core 1.94 m in length was retrieved from the bottom of Trench VIII, beginning 4.10 m beneath the modern floor of the cave. This material apparently predates the investigated human habitation of the cave. The deposit is a crumbly, calcareous material, red in color, which indicates oxidation. Black layers may indicate soil formation or hearths. The fine crumbling material likely fell from the ceiling of the cave. At ca. 1.80 m core depth (5.90 m depth from the current floor of the cave) the deposit is a fine orange loam that seems to have been formed under wet conditions. At 1.94 m gravel and stones made further coring impossible. The interior of the cave was disturbed throughout by recent visits, suggesting that samples taken from the floor would be contaminated. One sample, taken at the western end of the cave interior, contained pollen from small mosses growing on a ridge not far from the "window" above that part of the cave.
(2) In the alluvial plain southeast of the village of Vrinë and the Butrint canal. A core 12.25 m in length--the longest core thus far retrieved from Epirus--was extracted. At a depth of 7.50 m the sediment turns soft and there are fragments of marine shells.
(3) In a cultivated field adjacent to the village of Shkallë. A core cosisting of 4 m of soft clay containing wood remains was collected in a marshy spot southeast of the spring of the village before bedrock was encountered.
(4) Adjacent to an irrigation conduit between the villages of Çiflik and Shkallë, just northwest of the town of Konispol. A core 6.70 m in depth was extracted. It consisted mainly of a tough mottled clay, comparable to the clay collected in core no. 2, above.
The sample from Trench VIII, as it turns out, did not contain any pollen. The fossil contents of the core from Suk (no. 1 above) suggests a vegetation with central European traits under relatively moist conditions locally (alder swamp and a mixed deciduous forest at higher elevations). Significantly, the Albanian cores produced enough organic (peat) layers and wood remains to permit radiocarbon dating.
During two weekend field trips around Sarandë district, surface samples were collected and the local vegetations were described. Remnants of forest on the southeast of the plain near Vrinë and near Çiflik were not xerophytic Mediterranean, although the elevation was only 10-100 m above sea level. The vegetation differs from the Greek coastal area in displaying a less Mediterranean character. Celtis spp., Ulmus spp., and Quercus trojana were characteristic elements of the remnants. The forest vegetation of the (protected) classical site of Butrint is also of a composition indicating that the natural vegetation of the coastal lowlands of southern Albania is fairly humid compared with the Greek and Turkish coastal areas. The current vegetation in the vicinity of the cave itself was studied up to the crest of the mountain chain above it. The barren chalk of the ridge shows no sites with sediment formation suitable for palynological investigations. Up to the top of the ridge, surface samples were collected to study modern pollen precipitation in relation to the extant vegetation.
CHRONOLOGY AND CHRONOMETRY I: RADIOCARBON DATES
Albanian prehistorians generally avoid citing even the most approximate absolute dates in their publications, preferring instead to speak in terms of cross-datings (typically ceramic), especially when parallels can be drawn with materials in contiguous regions (e.g., Prendi 1990:425). Lack of contact in the past with foreign scholars and with radiocarbon laboratories, as well as financial exigencies, have been the cause of this. Under the circumstances, this was a prudent course of action.
Before our project began, only two radiocarbon dates had been available from all of Albania: one--unpublished--from the Neolithic site of Dunavec (personal communication, M. Korkuti), the other from a Copper Age level at Maliq (Guilaine and Prendi 1991). Both of these sites are in Korçe District, in the eastern part of the country. In 1992 we retrieved a number of samples of carbonized material from the Konispol Cave, and submitted them to Beta Analytic Laboratory in Miami, Florida, USA for analysis. The results are tabulated in
Fig. 14 (see Petruso et al. 1994 for the first report on dates from the 1992 season).
The dates are intriguing. They are internally consistent and of the right order of magnitude. We note in passing that the radiocarbon date obtained for Copper Age Maliq (Phase IIa), analyzed at the radiocarbon laboratory of the Université de Lyon I, is 5530 ± 110 B.P. (Ly-4975; Guilaine and Prendi 1991:575). The authors of that study have argued, incidentally (1991:577), that the Late Neolithic of Maliq (Phases Ia-Ib) should be placed within the period ca. 4000-3700 b.c. (uncalibrated radiocarbon years), which accords well with the single Late Neolithic date we obtained from Konispol Cave. Clearly the several dates for our site have the potential to provide a chronometric framework for Albania, and they will figure prominently in any subsequent discussion of the absolute chronology of the country.
CHRONOLOGY AND CHRONOMETRY II: ARCHAEOMAGNETIC DATING
The Konispol Cave preserves hearths throughout its prehistoric sequence. In 1992 we extracted several dozen samples from these hearths to test for their remanent magnetism. The hearths turned out to have strong signatures, and provided surprisingly precise archaeomagnetic absolute dates. In 1993 we retrieved over 600 hearth samples in a systematic effort to document the archaeomagnetic properties of the site.
Ellwood has defined two major objectives in his research in the cave; these objectives have ramifications for our understanding of the stratigraphy, tectonic history, and absolute chronology of the Konispol Cave, as well as for refining the absolute chronology of southern Europe. The objectives of this research are:
(1) To establish a new archaeomagnetic curve for southeastern Europe for the Stone Age. The unique aspect of these data is that the paleomagnetic results and radiocarbon dates will be obtained from the same site where a number of stratigraphically well- defined and well-preserved hearths associated with datable charcoal have been identified; and
(2) To facilitate ongoing archaeological investigations at the site, bulk magnetic susceptibility measurements will be used for within-site microstratigraphic correlations.
Archaeomagnetic Dating. Archaeomagnetic studies of materials from archaeological sites can provide a number of different kinds of useful data. Most such studies involve relative dating at localities where other methods do not work or where independent tests of anomalous dates are needed. To use archaeomagnetic dating techniques, it is necessary that standard curves for the region of interest be available. Where such data are poor, the utility of archaeomagnetic dating is diminished.
Ideal sites for archaeomagnetic dates are those with a sequence of isotopically datable hearths associated with rocks or underclays which exhibit a stable magnetism. The Konispol Cave, as it happens, is just such a site. The pilot sediments recovered in 1992 were from areas of normal sediment deposition and from a single hearth. The hearth has basal clays that were heated sufficiently to have recorded a remanent magnetic signature. This resulted from oxidation and the production of relatively stable magnetic minerals like hematite, and the acquisition of a thermal remanent moment as the basal sediment layers cooled. It was the purpose of the pilot study to sample the sediments and measure the magnetic properties of these materials, and then, through comparison with radiocarbon dates associated with these samples, to evaluate the potential of producing an archaeomagnetic curve for the Neolithic and Bronze Age in Albania. Results of the pilot study were compared to published archaeomagnetic dates from elsewhere in southeastern Europe.
In 1993 samples were obtained from recently-excavated surfaces by pushing cubic plastic paleomagnetic sample boxes into the loosely consolidated underclays. These boxes (8 cc. in volume) were then oriented using a Brunton compass. Pilot samples were returned to the U.S., and the remanent magnetization was measured in the paleomagnetism laboratory at the University of Texas at Arlington using a Molspin Minispin spinner magnetometer. Progressive alternating field demagnetization was carried out on all samples to magnetic induction levels of 80 mT.
A paleopole calculated for a hearth in excavation unit 3 of Trench XIII, without structural correction, has been compared with geomagnetic poles for the past 5,000 years, calculated from a southeastern European database (Kovacheva, 1980). That excavation unit contained Early Bronze Age pottery. An initial archaeomagnetic age range determined for the stratigraphic level is 2700-1900 BC. It appears that statistical precision for the data is excellent for the site, and we will continue to pursue these efforts.
The Konispol Cave has a large number of exposed hearths of varying ages which are ideal for archaeomagnetic measurements. Because radiocarbon ages of at least 11,000 B.P. have been determined for sediments in the cave, this site offers the potential for an archaeomagnetic curve from a single site going back at least into the Upper Paleolithic. Such a dataset would be unique, because most archaeomagnetic curves are developed from within a region, not from within a single site.
Magnetic Susceptibility Measurements. Bulk magnetic susceptibility is an indicator of iron mineral concentration within individual samples which can be affected by a number of natural processes, such as weathering and oxidation. Human activity can also alter the magnetic susceptibility of materials. For example, fires built to cook food and/or for heat can cause oxidation and alteration of iron-bearing minerals in specimens.
Magnetic susceptibility data have been applied in solving a number of widely differing types of problems. For example, these data have been used as paleoclimatic indicators, for stratigraphic correlation between lake sediment cores, in a broad range of environmental applications, and in granites to characterize where the magma originated within the earth. Bulk susceptibility can be measured quickly (less than one minute per measurement) using a susceptibility bridge.
It is clear from our experience that susceptibility measurements, especially during excavation, can be very useful for intra-site correlations. Stratigraphic ambiguities that arise can be resolved on-site, and data sets such as these can be used essentially as relative stratigraphic markers. If dates are available, the data provide intra-site relative dates. It is the purpose of part of this research to generate these readings during excavation at the site, and to use the measurements to resolve stratigraphic problems that arise during data analysis.
INVESTIGATIONS AT XARË, 1992
The site of Xarë (Fig. 15), which lies on nine relatively narrow plowed terraces on the south bank of the river, has been known to Albanian archaeologists for some years but has never been excavated. In order to determine the excavation potential of the site, we carried out an electrical resistivity survey in 1992. A 10 x 30 m grid was laid out on the broadest of these terraces and the one that has yielded the greatest number of stone tools (Fig. 16). Sediments over anomalies located in the resistivity survey were investigated subsequently by augering. The sediments on this terrace were found to have been sufficiently disturbed by plowing that culturally stratified materials are unlikely to be found, at least in the uppermost 2 m. The higher terraces, however, are planted with trees and not plowed, and might thus hold better potential for excavation.
The site is indeed rich. It has yielded flints of the Middle and Upper Paleolithic periods; significantly, there are no ceramic remains or other evidence of Neolithic or any later occupation. Earlier reports that there are both Middle and Upper Paleolithic components in the surface collection have been confirmed. The former is indicated by four Levallois flakes and several deliberately retouched tools; the latter by blades, bladelets, and prismatic cores (Fig. 17). The Middle Paleolithic tools are reminiscent of Mousterian assemblages from adjacent Greek Epirus (Runnels 1988; Huxtable et al. 1992). The bladelets suggest parallels with later Upper Paleolithic assemblages, again on analogy with better-known Greek assemblages (Bailey and Gamble 1990).
The geoarchaeological explorations at Xarë and in the Konispol Cave are the subject of Ellwood et al. 1993.
Proceed to Part 3