Chemical and Mineralogical Investigations of Sherds from the Akrotiri Excavations

These investigations reveal that six of these sherds were manufactured on Thera using altered rocks from this island as raw material (phyllitic and volcanic rocks).

The content of calcite in some of these sherds (carbonatic rock fragments and shells of Globigerina) indicate a firing temperature not higher than about 700^o C.

In contrast, one sherd ("imported ware") shows significant different features regarding temperature of firing, mineralogical and chemical composition. These point to quite another origin for this type of sherd.

INTRODUCTION

The Minoan ceramics of Thera pose a special problem concerning the origin of raw material as no clay deposits for a large scale production of ceramics are known on the island. Some sherds are called "imported" (possibly from Mycenae or Crete) by reason of their stylistic characteristics and characteristics of manufacture. But it is almost certain that a lot of ceramics were manufactured on Thera.

It is the aim of this short investigation to throw some light on the type of raw material and its possible origin on Thera. This was done by analysing the chemical and mineralogical composition of some sherds from the excavations near Akrotiri.

SAMPLES 

The sherds belong to the destruction level of the West House, room 5, with the exception of sample K 3. Different types of sherds were investigated to get a wide spectrum of material. They were classified in seven groups on the basis of colour and macroscopically visible content of different minerals and rock fragments (fig. 1 and tab. 1). The sherds are unpainted with the exception of sherd K 1 ("imported ware"), and belong probably to simple, daily used wares. Sample K 3 is a small fragment of a big pithos found in the Jar Room A1 (Pl. P.).

Furthermore, two samples of weathered phyllitic rocks (P 1 and P 2) from Thera were investigated for their chemical and mineralogical composition. Sample P 1 comes from the Sellada, sample P2 from Profitis Elias about 2 km west of Kamari.

TABLE I: Description of analysed sherds

ANALYTICAL PROCEDURES

Al, Mg, Ca, Na and K were analysed by atomic absorption spectrophotometry (AAS); Fe, P, Ti and Si were determined photometrically. Loss on ignition was determined by weight difference after heating the sample 1 hour at 1050° C.

Some trace elements were also determined, Ni with AAS, the other elements by instrumental neutron activation analysis (INAA).

Furthermore, the sherds were investigated microscopically, using thin sections made through the sherds, and some samples, in addition, by x-ray diffraction.

CLAYISH MATERIAL ON THERA

At present Thera is mainly covered by different volcanic rocks and to a subordinate extent by outcrops of phyllitic and carbonatic rocks. Local weathering and hydrothermal alteration has resulted in the formation of clay-containing material:

1. Fouqué (1879) cit. in Philipson (1899) mentioned some occurences of phyllitic rocks in deeply carved valleys in the region near Pirgos and Gonia which are altered partially to a yellowish clay material.
2. The phyllitic rocks of the Sellada are partially decomposed to clay-rich weathering products beneath a crust of carbonatic debris from Profitis Elias up to a thickness of one meter. This material consists mainly of illite, kaolinite, quartz, dolomite and a small amount of montmorillonite and felspar as shown by x-ray diffraction (sample P 1).
3. Small shear zones in the phyllitic rocks often show a strong decomposition. A larger occurrence of such material is exposed about 2 km west of Kamari on the north-slope of Profitis Elias. Here phyllites separated by a fault zone are underlying the marbles. Mechanical destruction of the fabric at this boundary and restrained penetration of water have altered the phyllites to clay-rich material with some bigger fragments of phyllite up to few cm in size. X-ray diffraction studies of this material (sample P 2) reveal that it consists of illite, calcite, felspar and quartz and a small amount of kaolinite. Until recently this material was used by the inhabitants of the near-by villages to wall up their ovens (Ch. Sigalas, pers. comm).
4. It is well known that fumarolic activity can result in the formation of kaolinite by altering volcanic rocks (Kesler 1970). On Thera two areas with fumarolic activity are known:

     a. Lavas on Nea Kameni are decomposed by recent fumaroles. In these rocks kaolinite was found (Schorin 1972).

    b. Dacitic tuff (Günther 1970) are exposed on the Akrotiri "peninsula" not far away from the excavations. These rocks are altered to various degrees by ancient fumarolic activities (Reck 1936). Even in those rocks available to the inhabitants of ancient Akrotiri kaolinite was found (Schorin 1972). It is not known whether this material can be used for the production of ceramics.

The above mentioned outcrops have a very limited extension. Nevertheless they show the possibility of using native material for ceramics. Furthermore larger extended areas of similar clayish material could have been available to the potters of Ancient Akrotiri.

CHEMISTRY OF SHERDS AND CLAYISH MATERIAL

The chemical composition of sherds and clayish material is given in tab. 2. Comparing the composition of the sherds one sees that they differ mainly in the contents of Fe, Mg and Ca. The content of Al is relatively high in all sherds.

In fig. 2 the composition of the investigated material is plotted in terms of SiO₂ - Al₂O₃ - CaO + MgO. From a chemical point of view part of the sherds could be termed as limestoneware lying in the triangle quartz-diopside-anorthite (sherds K 1, K 3, K 5, K 7 and K 9). The other sherds however show very low values of CaO + MgO. For these samples it seems improbable that marly clay or clay-rich marl, though often used in the ancient Aegean and Mesopotamian regions (Noll 1976) has been used as raw material.

Not only the amount of CaO + MgO but even the ratio CaO /MgO is very different from sample to sample. For example the unusually high value of MgO (9.7 %) in sherd K 1 with an equally high content of CaO could mean that dolomitic raw material was used in contrast to the CaO-rich (15.6 %) but MgO-poor (2.7 %) sherd K 3 for which calcite is confirmed as a component by x-ray diffaction.

Whereas eight sherds show an iron-content between 4.9 and 8.9 percent Fe₂O₃ sherd K 3 is almost free of iron (0.2 % Fe₂O₃). This points to a different kind of raw material.

Another point of interest concerns the loss on ignition. Usually the content of volatiles (H₂O, CO₂ etc.) in sherds is very low because of their high firing temperature. The sherds here investigated show a loss on ignition up to 15 weight percent. This is due mainly to some carbonate. Indeed the samples K 3, K 7 and K 9 have a high amount of calcite as shown by x-ray diffraction and microscopical investigations.

The chemical composition of the altered phyllitic rocks P 1 and P 2 in tab. 2 is in fairly good agreement with the range of the sherd's composition (though a more accurate comparision should be carried out on the basis of "dry" material, e.g. after correction for loss on ignition). The two samples plot slightly outside the range of the sherds in fig. 2. However a smaller content of dolomite (in sample P 1) and of quartz (in sample P 2) would lead to a good agreement with the composition of the sherds. Such variations of the mineralogical composition are not unusual in impure clay deposits. Indeed the unweathered phyllites of the Sellada are not homogeneous and show numerous mainly carbonatic layers.

In tab. 3 the trace elements are listed. They reveal a large range of concentration due to the different composition of the sherds. Some remark should be given upon the high content in Cr and the Ba content of sherd K 1 which differs significantly from the other sherds. Noll (pers. comm.) has found similar high Cr concentrations in some sherds from Knossos, as Harbottle et al. (1968) did in comparing trace elements of sherds from Knossos and Mycenae. The Ba content is the lowest of all sherds and significantly outside the range of the other sherds despite the high standard deviation.

It does not seem worth while interpreting these data in more detail because the number of investigated sherds is small and the material is very different. Nevertheless the data are given for reasons of completeness in the data collection of the ceramics in this region.

The chemical data show clearly that different types of raw material were used for the ancient production of ceramics in Thera. Either diverse clay deposits with different composition of lean clay were used, or different lean clays were added to more or less homogeneous clay, both resulting in a different chemical composition of the sherds.

MINERALOGICAL COMPOSITION (Pl. P)

With the exception of K 1 in all thin sections various minerals and rock fragments were found (tab. 4). They belong to different types of rocks: 

1. Metamorphic rocks: these are reddish brown to violet phyllitic fragments up to some mm. in size (fig. 3) and quartzitic rocks with hornblende and/or biotite (fig. 6). Some gneissic fragments were also found.
2. Magmatic rocks: most sherds contain a lot of minerals of volcanic origin (e.g. zoned, idiomorphic plagioclase, ortho - and clinopyroxene, greenish and brownish idiomorphic hornblende) and volcanic rock fragments (pumice and fragments with small crystals of plagioclase in a dark brown to black coloured glassy matrix) (fig. 4 & 5).
3. In some sherds non-metamorphosed carbonatic rocks were found. Three sherds contain microfossils (sherds K 3, K 7 and K 9). These are shells from foraminifera (Globigerina) of Upper Cretaceous to Lower Tertiary age (fig. 5).

DISCUSSION

As shown in tab. 4 most of the sherds contain two or three components of the above mentioned rock types and as such can be taken to indicate a mixture of clays from different sources. Furthermore some minerals and rock fragments indicate an origin on Thera. Therefore it is confirmed that native materials were used for the production of ceramics: weathered phyllitic rocks and altered material of the volcanic sequences (especially brownish and greenish hornblendes) are a characteristic constituent of the dacitic lavas and tuff of the Akrotiri region (Pichler & Kussmaul 1972) pointing to these rocks as a source of raw material.

However some types of rocks are not known from Thera or show a different feature. There is no doubt that the carbonate fragments are not similar to the marbles from Profitis Elias: those rocks have a triassic age in contrast to the Upper Cretaceous to Lower Tertiary age of the fossils in the sherds and show a weakly metamorphic fabric. Because the sherds in question have some pumice and volcanic rock fragments or minerals it is certain that they were produced on Thera.

Possibly some of the carbonatic layers in the phyllites contain such fossiliferous material. The age of the fossils and the Eocene age of the phyllites (Tartaris 1964) are compatible.

In some sherds the occurence of higher grade metamorphic rocks (gneiss, hornblende-biotite quartzite) and plutonic rocks (granite) points to the participation of the crystalline basement of the Aegean region though this is not known in exposures on Thera. The presence of volcanic material in these sherds also indicates their production on Thera. Their occurence could be due to :

1. Clay from the weathered crystalline basement as brought to Thera. The nearest island with such material is the island of Anaphi about 25 km east of Thera.
2. The rock fragments are components of partially greywacke-like layers in the phyllites.
3. Possibly the crystalline basement exists on Thera under the pumice layer and was available in earlier times.
4. It is known that some pyroclastic rocks and lavas on Thera contain xenoliths of gneiss and granitic rocks (Lacroix 1893, Liatsikas & Georgalas 1936). Similar inclusions were also found on the island of Christiana south of Thera (Puchelt et al. 1977). It is likely that the use of these rocks has resulted in the incorporation of metamorphic material in the sherds.

This question is open to further investigations.

While the result of the investigations on sherds K 3 to K 9 shows that Thera has been the place of production and source of the raw material of the sherds, no similar indications are given in sherd K 1. This sample shows some essentially different features:

1. Mineralogical composition: in thin section only quartz could be identified. Rock fragments are rare and could not be identified because of their fine grained composition. Certainly the sherd contains no volcanic material or metamorphic rocks similar to the above mentioned. The sherd consists mainly of quartz, plagioclase and diopside, as indicated  by x-ray diffraction.
2. The formation of diopside needs temperatures of about 900^o C (Peters & Jenni 1973). In contrast, at least the carbonate-containing sherds were fired at essentially lower temperatures because calcium carbonate desintegrates to CaO and CO₂ between 600^o C and 700^o C (Peters & Jenni). Rearrangement of CaO to CaCO₃ would have destroyed the fabric of the microfossil's shells. Therefore the firing temperature for these sherds hardly exeeded 700^o C. Also SEM-photographs of sherd K 1 show a high degree of sintering compared to the low sintered sherds of K 3 and K 9. 
3. Chemical composition: as shown above, dolomite seems to be a constituent in the raw material of K 1 in contrast to calcitic components in some other sherds. Cr and Ba show concentrations outside the range of these elements in the other sherds.

Because of these differing features it seems improbable that sherd K 1 is of native origin. Trace element investigations on a various number of this type of sherd from Thera compared with contemporaneous and similar sherds from Crete and Mycenae probably can lead to more precise results.

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

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