LB I Ceramic Connections between Thera and Kos

These two wares found at the Seraglio on Kos seem to be locally manufactured. The Light-on-Dark ware is found in large quantities in the LB I layers of the settlement and has a rich repertoire of decorative motifs in several shapes of domestic pottery. There are some overfired examples. The distribution of this pottery at the settlement of Trianda on Rhodes is different. Comparatively few specimens have been found until now, but macroscopically the fabric looks very similar to that of the Koan examples.

Examples of these wares from Akrotiri and the Dodecanese have been analysed chemically by atomic absorption spectrometry. Their compositions seem to be consistent with manufacture in a single region. It is likely that one of the places of manufacture was on Kos.

1.       INTRODUCTION

This paper describes the class of LB I pottery, decorated in light-on-dark and dark-on-light, found at Akrotiri, Thera, that was probably imported from the south-east Aegean. The Akrotiri finds are presented, followed by the corresponding evidence, much of it from new excavation, from Seraglio on Kos. The examples of this class found at Trianda on Rhodes are also considered. In the third section of the paper, the hypothesis of southeast Aegean production is tested by chemical analysis of specimens of this pottery from Akrotiri, Seraglio and Trianda.

2.       THE THERA EVIDENCE

   South-east Aegean wares at Akrotiri:

Pottery from the last occupation period at Akrotiri, Thera, includes imports from many different sources, the most important of which have already been recognized. They came from Minoan Crete (Niemeier 1979, 19-22; 1980, 41-63), the Greek mainland (Marthari 1980, 182-207; 1988, 60; 1990), as well as other islands of the Cyclades (Marthari 1990). This paper focuses on a smaller class of imported pottery which differs in fabric and decoration from the above. Its diagnostic features are considered here, but a detailed typological analysis will be presented elsewhere.

The fabric is usually dark brown in colour, occasionally light brown, and includes flakes of gold mica. All the vases are of closed forms, mainly cut-away jugs (Fig. 1-4) and oval-mouthed amphorae (Fig. 5).

Two decoration techniques have been distinguished, light-on-dark (LOD) and dark-on-light (DOL), the latter being less common. On LOD vases the decoration is in matt and often chalky white on a washed surface (Fig. 5), or in some cases on top of broad bands of dark paint (Fig. 1). The decoration on DOL vessels is in matt black on a washed or slipped surface (Fig. 2). The slip is sometimes creamy and thick and flakes easily (Fig. 4). The coexistence of these two techniques has been noted on an eyed jug (Fig. 3); on this vase white motifs applied on dark bands are combined with dark motifs put directly on a washed surface.

The most common motifs are double wavy lines (Fig. 1-2), groups of horizontal bands (Fig. 3-5) and isolated large spirals (Fig. 5). Other, but less frequent, decorative motifs are running spirals, zig-zags, quirks (Fig. 3) and a floral motif that may represent barley (Fig. 4). Some of these motifs occur on both LOD and DOL.

The LOD and DOL wares described above seem to be very similar to a type of pottery found at many south-east Aegean sites, such as Miletus and Iasos on the coast of Asia Minor, Seraglio on Kos and Trianda on Rhodes, and considered to be local to this region (Morricone 1973, 296; Davis 1982, 1-4; Marketou below). In terms of fabric, form and decoration, the vases at Akrotiri find close parallels with the corresponding wares from the south-east Aegean, as described in detail by Morricone (1973, 139-396) and reviewed by Marketou below in the light of new evidence. It is most likely, therefore, that the Akrotiri vases are imports from the south-east Aegean. The excellent state of preservation of the Akrotiri vases, a feature common to so many of the finds at the site, can help give a better understanding of the forms and decoration of this class of pottery which, Akrotiri aside, is usually fragmentary. Akrotiri is the second Cycladic settlement where the LOD and DOL wares have been detected, Ayia Irini on Kea being the first (Davis et al. 1983).

   Excavation data and chronology:

More than twenty intact vases of these wares have been found at Akrotiri. They are distributed throughout the settlement and usually come from rooms well known for their large concentration of imported pottery, such as the West House 6 and Complex Delta 9, 16 and 17 (Marthari 1980, 201). The wares are also represented in the sherd material from some buildings, indicating a higher frequency of the class than would be supposed from the number of whole vases. The buildings and their contents belong to the final occupation of the town (Last Period, phase B), assigned in Minoan terms to a late stage of LM IA (Marthari 1990). The chronological evidence from Akrotiri seems to accord with that from the sites yielding this pottery in the Dodecanese and Asia Minor (Davis 1982, 1-3; Marketou 1990 and below) and at Ayia Irini on Kea (Davis et al. 1983, 363). However, the presence of this pottery in the deposits of the Last Period, phase A (Marthari 1990), contemporary with an early stage of LM IA, cannot be excluded because these deposits have not been studied in detail.

   A commercial route connecting Thera and the south-east Aegean:

The south-east Aegean origin of the Akrotiri vases has been argued above. Nevertheless, a number of questions remain, one of which is how and why these vases reached Thera. The evidence from the recent excavations at Seraglio on Kos strongly suggests that this island could have been the production centre of the LOD and DOL south-east Aegean wares, as Marketou proposes below. This view is also supported by the chemical data, the compositions of the Akrotiri vases and the examples from Kos showing close resemblance, as shown below by Jones. Consequently, Kos could be the source, or at least one of the sources, of those vases imported to Akrotiri.

The frequency of the south-east Aegean class at Akrotiri, its distribution pattern, and the proximity of Thera to the Dodecanese and Asia Minor are consistent with a direct trade route, rather than the more occasional, less direct route (via Crete) envisaged by Davis et al. (1983) in the case of the Ayia Irini imported vases. Furthermore, it is clear from the map that Thera, the southernmost island of the Cyclades, and Kos lie on an imaginary line crossing the Aegean from West to East with intermediary stops at Anaphi and Astypalaea. This line could correspond to a commercial route connecting the Southern Cyclades with the Dodecanese at this time. The absence of Late Bronze Age sites on Anaphi and Astypalaea is probably more apparent than real, and can be attributed to incomplete investigation. For example, recent surface surveys (by the writer) on Ios discovered settlements at Skarkos and Maganari that were occupied during the early Late Bronze Age. The same picture could well apply to Anaphi and Astypalaea.

The south-east Aegean vases at Akrotiri are all closed and made in a coarse clay, and consequently were most likely imported to Thera as containers rather than for their aesthetic ceramic value. The local Theran decorated pottery is of a higher quality but could not nevertheless compete with the fine lustrous LM IA and LH I wares. For this reason, the inhabitants of Thera and other Cycladic islands imported this decorated pottery, mainly semi-globular and Vapheio cups, from Crete and the Greek mainland. The vases from the south-east Aegean, on the other hand, were more important for their possible comestible contents.

Considering the various commercial routes involving Thera in early LBA, it can be suggested that, apart from the well-established route linking Thera to the Greek mainland and the Western Cyclades to the north and Crete to the south, the so-called Western String (Davis 1979; Cherry and Davis 1982; Schofield 1982), there was another route connecting Thera to the south-east Aegean via Kos.

Akrotiri was an important port and urban centre of the early LBA. Its position in the centre of the Aegean encouraged trade connections with ports and trade centres in many directions. One of these could be Kos, as the pottery evidence has demonstrated.

M. MARTHARI

3.       THE EVIDENCE FROM KOS AND RHODES

   Seraglio:

The light-on-dark pottery ('con ornati bianchi') found in the deepest strata of the Italian excavations at Seraglio on Kos appears in large amounts and is decorated with a rich repertoire of motifs. Morricone (1973, 296-326), the excavator of Seraglio, examined the LOD and DOL pottery with great care. Although he associated it with a domestic class of pottery of MM III character, because of the white-painted motifs, as well as some of the vase forms which seemed to be Minoan, he emphasized its local, Koan, origin. He also observed that the LOD and DOL pottery was widely distributed in Asia Minor and the surrounding area (Morricone 1973, 296, n. 1).

According to the recent evidence from Seraglio and Trianda, it is now possible to review the character of the LOD and DOL pottery since these sites have yielded much more material of this class in stratified deposits. It is also possible to reconsider the more general problem of its distribution in the east and south-east Aegean and to give a new classification of both forms and decorative motifs. To the shapes identified by Monaco can now be added five new ones. There are more decorative elements such as different kinds of linear motifs, and several reed and plant patterns including the barley motif (Fig. 6). A group has been identified bearing some LOD motifs which were exactly copied in a DOL version (Fig. 7), suggesting that both may have been products of the same workshop.

The aim of this section is to present a description of the LOD and DOL pottery, and an assessment of its chronology and quantitative distribution in different parts of the Aegean, based on the finds primarily from Seraglio. A detailed analysis and classification of this pottery will be published elsewhere.

   Fabric:

The fabric of LOD and DOL looks similar to the domestic pottery, which should be local. The colour of the fired clay ranges from light brown-red to dark brown-red (in Munsell colour terms: 10R/6/6 - 6/8 to 2.5YR/6/6 - 7/6), and is sometimes reddish-yellow. The clay includes white and black inclusions, but the basic characteristic is the presence of gold mica, also observable in the earlier pottery at Seraglio (although in smaller size), as well as in the soil of Kos. The mica flakes are visible to the naked eye, although at times the surface of the vase is covered with slip. Light decoration motifs are made in a white paint (10YR/8.2) which sometimes wears away. The same white paint is usually used as a slip on the undecorated part of the vase. The vases are usually washed. The main decorative motifs are double wavy lines on a red-grey painted band (Fig. 8) on the upper part of the vase.

   Evidence for local manufacture:

The discovery of three pottery kilns (Vasiliou plot: ADelt Chronika 1981, forthcoming; Thalassinou plot: ADelt Chronika 1984, forthcoming) in the recent excavations at Seraglio lends very strong support to the notion of locally manufactured pottery. The tradition of pottery-making on the island, of course, continues into the historic period (Sherwin-White 1978, 14) up to recent times (Psaropoulou 1986, 63-77). The three Seraglio kilns are representative of three successive periods of occupation at the site. The first, very fragmentary kiln was of the EBA III period, the second belongs to the MBA and is related to the production of carinated cups, while the third is dated to LBA I - LM IA late Seraglio period (Marketou 1990) and was possibly constructed for the production of conical cups. Among the sherds and vases found close to the last two kilns were overfired and distorted pieces (Fig. 9 and 10), one of which was analysed chemically (K3, see below).

   Chronology:

The first excavations by the Italians at Seraglio did not reach virgin soil (Morricone 1973, 385). In the light of the subsequent Greek excavations, Morricone's 'prima citta' (Morricone 1973, 389-391) belongs to the third period of the Seraglio settlement, there being two earlier occupation periods underneath the LM IA pavements. The first is dated to the end of the Early Bronze Age and the second to the immediately succeeding Middle Bronze Age (Marketou 1990). Closer study of the latter period will undoubtedly shed more light on the beginning of the Late Bronze Age but cannot be discussed at this stage.

All clear LM IA late strata revealed abundant LOD pottery together with LM IA imports and the local conical cups as well. It is the presence of LOD in contexts overlying LM IA late pavements and immediately underneath the tephra layer that allows a precise chronology for LOD to be made.

The same applies to DOL which, as already mentioned, is contemporary with LOD and probably produced in the same workshop. As an example, a DOL sherd decorated with a couple of wavy bands was found exactly beneath the tephra layer and over the LM IA late street surface at Koukounas plot (ADelt Chronika 1987, forthcoming).

   Trianda:

The excavations at Trianda, which took place at about the same time as those at Seraglio, have also yielded some specimens of LOD and DOL pottery which belong to the LM IA late period according to the new stratigraphy (Marketou 1990, Table 1). They are, however, small in number and form a fraction of the imported pottery. It is clear, therefore, that the percentage of LOD at Trianda, as at other sites in Asia Minor and the Aegean where this pottery is found, is much lower than that at Seraglio. Stratigraphically, the Trianda and Seraglio material is confidently contemporaneous. Again at both sites, there are some instances of LOD and DOL pottery found with LM IB Marine Style. It seems, then, that the former pottery, together with conical cups, continued to be made on Kos after the Thera eruption and the tephra fall until at least the end of LM IB.

   The distribution of LOD in the south-east Aegean:

The LOD (and DOL) from Seraglio and Trianda share common characteristics which are easily distinguished from the imitations that appear at Trianda. The same observation can be applied to examples from Vathy Cave on Kalymnos (Maiuri 1928, 117), Miletus (Weickert 1940, 325-332, Pl. 24.1; Weickert 1957, 117-119, Pl. 28-1a, 3, Pl. 31 a, b, c; Weickert et al. 1960, 27, Pl. 8, 45, Pl. 34, 64, Pl. 42b; Schiering 1984, 187-189), Akbuk Teichiussa near Miletus (Voigtländer 1988, 605, 608, Fig. 39,2), Iasos (Laviosa 1978, 1093-99, Pl. 350. Fig. 13; Laviosa 1984, 187), Knidos (Mellink 1978, 321; pers. comm. I.C. Love), as well as further afield at Akrotiri (see Marthari above), Ayia Irini, Kea (Davis 1982, 33-41; Davis et al. 1983) and perhaps Tigani on Samos (Heidenreich 1936, 173, Pl. 49-1, 2).

Vases found at these sites and decorated with LOD and DOL motifs used to be recognized as MM II - III Kamares ware (Laviosa 1978; Laviosa 1984, 187; Maiuri 1928, 117; Mee 1975, 384; 1982, 80). This pottery was thus taken as evidence of Minoan trade and thalassocracy in the Aegean. Similar pottery found at Trianda is also very important because it was found in stratified deposits. The LOD and DOL found there by Monaco (1941, 92-93 Fig. 39, A; Furumark 1950, 154, 169, 184) belongs to Furumark's Trianda IIA, dated to LM IA end-LM IB, not LM IB/LH IIA (Davis 1982, 34). It is therefore Furumark who proposed a date later than MM III for this kind of pottery. We note here Davis's discussion of this matter and his assertion of an LB I date (Davis 1982, 31-44; Davis et al. 1983).

The writer has seen some of the LOD sherds and vases from Kalymnos, Miletus and Iasos, as well as Akrotiri, and ascertained their visual correspondence to the Seraglio group. It is interesting to note that while LOD and DOL pottery occurs also in many open and small-sized examples at Seraglio, the corresponding finds elsewhere belong only to closed shapes of large vases such as amphorae and jugs which were surely container vessels (see, for example, the vases from Akrotiri above). These vessels may have carried products from Kos itself, since the island is fertile and, moreover, its wine industry (and amphorae) in later times is well documented (Sherwin-White 1978, 224-235).

The LBA I - LM I settlement of Kos had many urban characteristics such as a wide street leading from a harbour to the Akropolis. It has similarities in this respect with Trianda (Marketou 1990).

The continuous occupation of the Seraglio hill from the end of EBA III, past the last phase of LH IIIC and up to the early Christian period, was due to its natural and safe harbour (Morricone 1950, 62 n. 36; Marketou 1990, 169), over which the town of Kos was built. From the port sailings must have taken place to the coast of Asia Minor, Rhodes and westwards to Thera. Its success as a port surely led to the later development of Koan trade.

To conclude this section, the following points may be outlined:

1. There are indications of pottery workshops at Seraglio which produced a large amount of LOD and DOL closed and open-shaped pottery, decorated in a rich repertoire of motifs.
2. There is evidence of a long tradition of pottery manufacture before the LBA I - LM I period.
3. According to the Seraglio and Trianda stratigraphies, which are equivalent to each other, LOD and DOL pottery is dated to the LM IA late - LM IB period.
4. LOD and DOL pottery is widely distributed in the eastern Aegean and the Asia Minor coast; it extends to Thera and Kea. Visually, at least some of it resembles the Seraglio group.
5. LOD and DOL pottery was exported, apparently as container vessels, through the important port at Seraglio.

T. MARKETOU

4.       CHEMICAL CHARACTERIZATION OF SOUTH-AEGEAN LOD AND DOL POTTERY

1.    Background to the analyses:

At the last Thera Congress, several papers were presented on the physico-chemical characterization of pottery from Akrotiri and the identification of possible imports. These contributions, together with Mrs J. Pittinger's unpublished petrographic study of Cycladic pottery which included material from Akrotiri, and the petrographic study by D.F. Williams (Davis et al. 1983) of some 'Eastern Aegean' LB I - II pottery from Ayia Irini, Kea, that relates very closely to the wares under consideration in this paper, have been reviewed by Jones (1986, 273-280). During the last few years, more systematic, broadly-based analytical work on prehistoric pottery from the Cyclades has got under way. Part of V. Kilikoglou's doctoral thesis (1988) concerned the neutron activation analysis of pottery of Middle and Late Cycladic date from Akrotiri and Phylakopi. Concurrently, the recognition that much Cycladic pottery is very amenable to petrographic analysis has stimulated a new study by S.J. Vaughan, who reports on some of her results in this volume.

The efforts of laboratory-based investigation have also extended to the Dodecanese and the south-west coast of Turkey. To the chemical and petrographic data base for prehistoric and later pottery from these islands described by Jones (1986, 290-299) should be added the data obtained by Whitbread (1986) and Watson (1985) who have respectively studied petrographically and chemically (with XRF) Hellenistic amphorae from Rhodes and Kos. Regarding sites on the Turkish coast, Goedecken (1984; 1988) has reported the chemical analysis (by neutron activation) of more than one thousand ceramic and clay samples (some of them also analysed petrographically) from Miletus and Didyma, the pottery ranging in date from LB I (and including some likely examples of LOD ware) to Archaic. The Lyon Ceramology Laboratory also has chemical (XRF) data for Miletus, as well as the Datça peninsula (Coja and Dupont 1979; Empereur and Picon 1988). Jones (1986, 665-666) has isolated a small chemical group of Archaic pottery of probable Milesian origin. There are petrographic and chemical data sets for Knidian amphorae (Whitbread 1986; Watson 1985).

2.    Analysis:

Into this framework, the present analytical study can be placed. Chemical analysis by atomic absorption spectrometry (AAS) was selected for two practical reasons: first, the examples of suspected LB I LOD and DOL ware at Akrotiri were whole (restored) vases, for which sampling had perforce to be in the form of a drilling. Second, much of the necessary comparative data was available: the Fitch Laboratory's data bank includes chemical compositions (obtained by AAS and/or optical emission spectroscopy (OES)) for prehistoric pottery from Rhodes and Kos, as well as many Cycladic Islands, much of the data being described by Jones (1986, 258f., 290f.).

The purpose of the analyses was two-fold: a. to establish whether the LOD and DOL ware found at Akrotiri was comparable in composition with that from Kos and Rhodes; b. trearing the Koan examples as local products, owing to their association with the kiln context (see Marketou above), to determine whether the compositions of the Akrotiri examples were consistent with production on Kos and/or elsewhere.

3.    Material analysed:

A.    LB I LOD and DOL South-East Aegean wares from Akrotiri

B.    LB I LOD ware from Seraglio on Kos

C.    LB I LOD ware from Trianda on Rhodes

D.    LC I large bird jugs from Akrotiri in the local fabric

4.    Analytical procedure:

All the Akrotiri samples were obtained by taking multiple drillings (with a tungsten carbide drill-head) into the base of the vase. The resulting powder was usually not less than 100 mg in weight. For A13-16 it was possible to take a duplicate sample in the form of a fragment which was broken off the sherd with pliers and the outer weathered surface removed before grinding in an agate mortar to give several hundred mg of powder. The same method was applied to the Seraglio and Trianda pieces which were in sherd form, the latter being heavily encrusted with a calcareous soil.

The Fitch Laboratory's analytical procedure, which has recently been described by Liddy (1989), comprised heat treatment of the samples at 120° C for three hours, followed by their preparation for analysis by AAS using the lithium metaborate fusion method. The samples were analysed in two consecutive batches. Calibrations were set up with two types of standard: 'Potmix' and Greek (and other) ceramic standards, leading to the determination of the concentration of eleven elements in their oxide form. The precision of the element determinations is monitored by the Laboratory at intervals by analysing series of replicate samples (usually fifteen in number). As expected, the value of the coefficient of variation is dependent on the nature of the element, manganese and chromium being the best and worst determined elements respectively. The ranges of coefficient of variation over a six-year period are: Al: 5-10%, Ca: 3-7%, Mg: 5-8%, Fe: 6-8%, Ti: 4-8%, Na: 8-12%,K: 6-10%,Mn: 1-2%, Cr: 10-15%, Ni: 5-10% and Si: 2.6-8%.

5.    Results and discussion:

The composition for all elements but Si are set out in Table 1. It is at once clear that there is a good measure of uniformity among the compositions of all but one of the LOD and DOL ware samples from Akrotiri and the Seraglio and Trianda samples. This observation seems to accord well with the similarity in the macroscopic appearance of this ceramic class, as noted by Marthari and Marketou above.

There are two features of the compositions: (a) they are typical of a non-calcareous clay, and (b) those of Akrotiri (rather less so those of Kos and Trianda) exhibit rather wide dispersions in a number of elements, notably Al, Na, Mn and K. This last observation may be associated with a sampling problem, because the compositions of two Akrotiri samples (14 and 15) obtained by the two sampling methods reveal some discrepancies in these elements, particularly Mn and K.

In any case, the compositions of the LB I LOD and DOL ware are easily differentiated from those of the local pottery of Akrotiri, A17-20, and from the northern half of Rhodes; the former are of calcareous type (Jones 1978, Table 1; Jones 1986, 277f.), as are the latter which also have a distinctive serpentinitic character (Jones 1986, 297f.). This picture is presented graphically in Fig. 12, the dendrogram resulting from Ward's method cluster analysis of all samples in groups A-D, together with a small group of Hellenistic fine ware pottery from Rhodes Town, acting as a 'control' for the northern part of that island, analysed by the same technique, AAS. The DOL and LOD ware samples form one broad cluster, reflecting the variation in composition just mentioned, which is quite separate from the Akrotiri and Rhodes clusters. A8, which is excluded from this cluster analysis, differs significantly in composition from the DOL and LOD ware owing to its high calcium content (see Table 1); its origin is currently being examined in the light of data obtained in a study (in collaboration with M. Marthari) on contemporary Cycladic imports to Akrotiri.

K1-4, and indeed the DOL and LOD ware samples generally, compare closely with previously obtained, admittedly limited, reference data for the Seraglio site (Table 2). For this reason, it is proposed that the LOD and DOL from Akrotiri and Trianda was produced at one or more centres, one of which was surely at Seraglio. Their production, however, at centres on the Bodrum and Datça peninsulas cannot be excluded because, on the basis of the available chemical data for Hellenistic amphorae, the clays there have compositions similar to or overlapping with those on Kos. This finding concurs with the conclusion of Davis et al. (1983) that the two examples from Ayia Irini, Kea, which both contained volcanic glass, were consistent with a source at the south-east end of the volcanic arc which includes Kos and the Bodrum region. On the other hand, the LOD and DOL ware does not match the few compositions available in the Fitch Laboratory's data bank for prehistoric pottery from Miletus; there are clear differences in Mg, Cr and Ni (Table 2).

In conclusion, the results of chemical analysis are reasonably satisfactory. They support the hypothesis based on the archaeological evidence and the macroscopic appearance of the fabric that one of the production centres of LB I DOL and LaD found at Akrotiri and Trianda was on Kos, probably at Seraglio. Looking to the future, further laboratory-based study of the LOD and DOL ware needs to be based on a larger corpus of data from a wider range of sites in the region than is presently the case. Thus, work which is already in progress on the chemical analysis of more LOD and DOL ware from the three sites considered thus far, should be extended to include examples from Miletus, Kalymnos and Iasos. However, progress would also benefit greatly from a combination of both chemical and petrographic characterization. Finally, valuable supplementary information would be obtained from identification of the paintings on this class of pottery, as well as possible organic residues remaining within the fabric.

R. E. JONES

--------------------------------------------

----------------------------------------