Akrotiri and its Neighbours to the South: Conical Cups Again

A specific item, the Minoan conical cup, found in Akrotiri in abundant numbers, is studied here statistically in an attempt to explain this phenomenon of transmission. As the cups are made locally, either they were manufactured by a sizeable population of Minoans living on Thera or they were produced by Akrotirians. Univariate, bivariate and multivariate statistical analyses are used on various measurements and form elements of the cups from Akrotiri, Knossos, Arkhanes, Mallia, Palaikastro, Kato Zakros, Myrtos Pyrgos, Kommos and Nerokourou. The results indicate that the conical cup on Thera is an eclectic item - in shape most like Knossos, in surface treatment most similar to Palaikastro and in function closest to the sites on the south coast of Crete and Zakros. Akrotiri had wide and extensive contacts with her Cretan neighbours to the south.

INTRODUCTION

'Contact' and 'transmission' are common words in archaeological contexts, used to cover a variety of situations and circumstances in which item X from area A appears at some later time in area B as well: one explanation might be that people from area B travel to area A, see something new, and take the thing home; a variation on this theme is that area B people see it in area A, like it and take the idea home; another possibility is that people from area A move to area B where they continue living as they did at home; on a more advanced level, people from area A trade or 'sell' something new or in demand to area B, either directly or through an intermediary from area C.

The aim of this article is to examine the conical cups found on Akrotiri, Thera, and investigate their nature and possible origin to see if any light can be shed on the types of connections which existed between Akrotiri and its neighbours to the south, that is, Crete, during the Late Bronze I period. The first issue to discuss is the origin of the conical cups. As has been stated elsewhere (Gillis 1988a) conical cups appear first on Crete as early as EM II - III and continue throughout the entire Bronze Age. As they appear on every single Cretan Bronze Age site, in every Minoan colony, and in contemporary civilizations, I feel it not unwarranted to consider them a Minoan phenomenon.

We are now faced with the situation given above, that of item X (conical cups), indigenous to area A (Crete), being found in area B (area B includes the Cyclades, the Dodecanese and the nearest coast of Asia Minor, the mainland and Kythera). For the purposes of this study, area B will be limited to include only Akrotiri. The question of their being imports is quickly resolved: my studies have shown that the vast majority of Minoan conical cups found in the Cyclades and on Kythera are made of local clay and are tempered with the same materials as the rest of the local ceramic production. Thus there is no question of the conical cups' being traded or brought from Crete, and we can rule out contact possibilities one and four, above, when seeking an explanation for their existence on Thera.

It is also necessary to examine the element of time (did item X in fact appear later in area B than in area A?) in order to rule out simultaneous occurrence in both places. In the Cyclades, the situation is a bit difficult. Akrotiri, being basically a one-period site, cannot provide much information about the time of arrival of these cups. Apart from LC I, one conical cup was found in an MC III pylon pit, and four very low, shallow cups from Karagiorgis cemetery date to 'early MC'. (Professor Doumas and Ms M. Marthari kindly showed these to me.) A large, handmade handleless cup from Christiana, from EC, bears little resemblance to conical cups. (Ms P. Sotirakopoulou has kindly informed me that in her studies of EC pottery, she has never come across conical cups.) To ensure that they did not arrive on Thera via one of the other two Cycladic islands possessing large Minoan influence, a quick survey of these two is necessary. Some 10 handleless cups from EC III / MC I were found from the early excavations at Phylakopi in the beginning of the century (the rest of the material was inaccessible when I was on Milos); these, however, were very different from conical cups, being large and practically cylindrical. The first conical cups I found belong to the modern excavation and date to MC III / LC I. In Ayia Irini, Kea, the earliest conical cup dates to MC II (and it is a possible later intrusion). The MC III / LC I period has some representatives, but vast numbers of conical cups appear first in early LC I. The civilizations to the east, the mainland and Kythera are out of the geographic sphere considered in this article and will not be taken up here. As there does not seem to be any time difference regarding the first occurrence of conical cups on the other relevant Cycladic islands, I believe it is fair to say that, with the exception of a few scattered examples, the Minoan conical cups burst upon the entire Cycladic world at the beginning of the LM I period, long after they had established themselves as one of the most common types of ceramic artefacts on Crete; further, we can most probably rule out the possibility that they came to Akrotiri from another Cycladic island.

We are now left with two possible explanations for the occurrence of Minoan conical cups on Akrotiri: that the idea of the cup was brought back to Akrotiri by travelling Akrotirians to be produced locally, or that Cretans settled in large numbers in Akrotiri in the transition from MC III to LC I and continued in their new home to make the cups they were used to back on Crete. The purpose of this paper is to examine the conical cups from Akrotiri and from a selected number of sites on Crete to see which of these two possibilities seems the more probable.

THE MATERIAL

Conical cups all share several characteristics which separate them from several other shapes: they are smaller than bowls, with a maximum height of 10 cm and a maximum rim diameter of 13.5 cm. Further, they never have feet or stems, and are handleless. Any vessel under 3 cm in height is considered a saucer if it has a wide rim diameter, or a miniature vessel if it does not, and consequently is not included. The vast majority of conical cups are plain, but some may be monochromed or have 'painted' decoration. Although the term 'conical cup', once applied, seems to have stuck, many conical cups have curving or other shaped walls. As we will see below, the great majority of conical cups included in this study are not conical at all.

This study includes 1534 conical cups: 550 from Akrotiri and 984 from Crete, divided among Knossos (425), Arkhanes (15), Mallia (48), Palaikastro (126), Kato Zakros (124), Myrtos Pyrgos (109), Kommos (74) and Nerokourou (63). The cups from Knossos come from the palace, the Knossos Village Excavation, Hogarth's Houses, the Royal Road, the Unexplored Mansion, the Royal Tomb and the Acropolis settlement. (For the break-down and for the division of cups by period, see Table 1.) The Cretan sites chosen were a compromise, as registering all conical cups on Crete is naturally an impossibility. I picked sites from all four coasts and from the inland; I took palace complexes, 'villas' and towns; I tried to keep to fairly modern excavations, as conical cups were usually thrown away unregistered in the old excavations, or if kept had no provenance or date.

Table 1: A breakdown of the conical cups from LB I into site and period

In almost all cases only whole cups or those complete enough to provide a full profile and allow a measurement of the rim diameter were registered. No small sherds are included. At the sites where there was a great number of cups, not all of them were always recorded. Furthermore it must be remembered that the percentage of the total amount of excavated conical cups which is catalogued and kept varies greatly from site to site, and that records are rarely made of the complete number found, nor of all their locations. Thus, while one might hope that the existing conical cups are representative of the whole amount, this can never be a certainty, and all results must be considered in this light.

The time period studied is MB III / LB I through LB I. Although Akrotiri is dated to LC I, equal to LM IA, even those Cretan cups from the transition period MM III / LM I were included to cover any possible time lags. The later LM I cups were included to guard against possible differences in dating from one site on Crete to another, even though, of course, the Akrotiri production ceased before the start of LM IB. When the excavator distinguished between LM IA and LM IB, the cups were recorded as such -otherwise, they were grouped under the category LM I. (See Table 1.)

METHODOLOGY

The question of whether Cretans moved to Akrotiri in numbers large enough to explain the vast quantities of conical cups found there, or if the Akrotirians started to produce them themselves, whether in an attempt to copy all things Minoan, a type of 'Minoanizing' that even extended to plain pottery, or as a result of a change in custom or practice necessitating new equipment, is a difficult one. The approach I have followed here, is to compare the physical properties of the cups on Akrotiri with those from selected places on Crete to see if there were any obvious similarities in shape, decoration and function. This was done by registering the material and then by analysing it statistically.

Registration:

As stated above, all, or a statistically significant sample of the whole or reasonably whole cups on Akrotiri and eight sites on Crete were recorded for the transition MB III / LB I and for LB I. The information registered included

1. measurements: height, base diameter, rim diameter, base thickness, weight empty, weight filled, the volume in cc of the vessel;
2. form elements represented by code: type of base, type of wall, type of rim, thickness or thinness of wall;
3. other information: type of manufacturing technique, placement and type of any traces of fire, surface treatment, existence of a hole in the base or wall, dating, wheel and string marks, all coded.

Everything was entered in the field on data sheets. The volumes and the ratios H:B diameter; H:R diameter, B diameter:R diameter were computed mechanically afterwards. All the information was entered or converted on a VAX computer.  (see Table 2 for the total information recorded for each cup).

Table 2: The information registered for each cup. The italicized items and the numbers after them in parentheses are the variables used in the multivariate analyses

Also recorded, but not included in the statistical analyses, was information concerning colour and hardness of fabric, details of surface treatment such as colour and design, inclusions and deliberate tempering, and any other special features or anomalies.

Analyses:

All the cups were grouped by site. No divisions were made within each site, as the purpose of the analyses is to compare the gross production on a site with that on another site. Bar graphs were made for the distribution of all the cups for single variables, e.g., height, rim diameter, base codes. Plots and/or cross-tables were made combining certain variables, e.g. a plot for wall type and volume, or height and weight, a cross-table of base codes to rim codes, to ultimately describe the most frequently occurring, characteristic shapes, decors and functions for the conical cups from each of the nine sites. While one can never check an individual cup in this way, one can see general patterns and broad trends. These simple statistical analyses were made using the Minitab program (MINITAB 1985; cf. Madsen 1985), a well-established program for such analyses.

While the above-mentioned method allows univariate or bivariate comparisons, a multivariate analysis was also desirable to see which site(s) were most similar to Akrotiri regarding form, decor and function and in connection with this, if the Cretan sites most like Akrotiri grouped with other sites in any logical pattern. A type of analysis such as Correspondence Analysis (Bølviken et al. 1982; Gillis in press) or Principle Components Analysis was not suitable here as the similarity of a single unit to the others is based on the relation of all the units to one another: that is, Akrotiri's similarity to the Cretan sites would be calculated not only by how it relates to the eight other places, but is also influenced by how these eight sites relate to each other.

A better method in this case is to calculate the value of similarity between units for each unit independently, arriving at a similarity matrix, where each unit maybe compared with all other units. Thus each variable was recorded as the percentage of the total number of recorded cups at a site possessing a certain attribute (e.g., 89 cups from Akrotiri have wall code 8, which is 16.2% of the total 550). The percentages for each variable were then standardized through division by the mean of each variable. Similarities between units were calculated by the City Block measurement of distances (Madsen 1985; for literature about similarity, see e.g. Gower 1971). Two identical units will have a similarity of 0 (zero), while the larger the number, the smaller the similarity. The maximum dissimilarity is a relative value dependent on the number of variables involved in the calculation of the dissimilarity measurement, among other things. The similarity matrices were then used for cluster analysis according to the UPGMA algorithm (Sokal et al. 1973).

THE CUPS - PHYSICAL PROPERTIES

Both measurements and form elements represented in code are included in the study of the physical properties of the cups. As practically all form elements appear on all sizes, even though any one particular site may not have examples of all these possibilities, the two types of measurement are not particularly dependent on each other and can be regarded separately.

Measurements:

The measurements of interest are height, weight empty, base diameter, rim diameter and volume (computed by subtracting the empty weight in grams of a cup from its weight filled with sand and then by using the weight of 10 cc of sand as the standard at each site, converting the grams to cc). These were studied individually in bar graphs and in pairs, in plots and cross-tables.

For obvious reasons it is impossible to reproduce every bar graph and plot for all the nine sites. The detailed information for Akrotiri can be found in Gillis (in press); that for the other sites, in a forthcoming work.

Regarding the height, as can be seen from Table 4, the majority of all conical cups have an average height of 3.8 - 4.2 cm. (Note that the figures in Tables 4 - 8 for the peaks represent midpoints. The increments vary. Here, with an increment of 5, for example, the figure 40 is the midpoint between 38 and 42 mm.) Akrotiri and Pyrgos have two distinct peaks, indicating two separate groups (Zakros has a possible larger group as well). Knossos indicates standardization, but has more cups included in the curve. By 'peak' is meant the bar which is longest in the graph. If there are two 'high-points', then the graph is double-peaked. By a 'wide spread' is meant that several more or less equal bars share the position of 'longest'. If no bar is particulary longer than the others, or shares its longer length with other bars, there is no clear peak (although of course all curves have a peak in some way). See Figure 9 - 12 for bar graphs illustrating one-peaked, double-peaked, wide-spread, and 'no-clear-peak' distributions. The lower group from Akrotiri seems to be smaller on the average than all others (50% of all cups are between 3.3 - 4.2 cm), while the cups from Kommos, Zakros, Pyrgos and Palaikastro are slightly larger. Akrotiri, (Arkhanes), Mallia, Nerokourou and Palaikastro indicate good standardization, while Kommos, Zakros and Pyrgos have a wider spread.

The weight (in grams) can be a good indication of the degree of skill of the potter. For those sites with good standardization of height, one would expect to see a comparable degree of standardization in weight. In Table 5, we see that the weight varies from site to site between 55 and 95 grams, with the most frequent between 75 - 85 grams. Akrotiri has only one peak, representing only 17% of the total. This peak is long and very skewed, finally tapering off at 280 grams. Fifty percent of the total at Akrotiri is between 55 - 95: it is fair to say that the cups must be quite standardized as to weight. This relation between height and weight for Akrotiri is borne out by a very high correlation coefficient of 0.874 indicating that the potter was quite skilled. Knossos, Mallia, Zakros, Pyrgos and Palaikastro show good standardization of weight, but Knossos and Pyrgos show no clear peaks. Pyrgos and Arkhanes are lighter than the others. Kommos has a wide spread, with small peaks, and Nerokourou has no real peaks at all.

Table 4: The most frequently occuring heigth measurement in millimetres and its percentage of the total number of cups based on bar graphs representing all the measured heigths for the sites. The numbers given for the peaks are midpoints, with increments of 5 - thus, the number 40 represents all cups between 38 and 42 mm high.

Table 5: The most frequently occuring weight in grams and its percentage of the total number of cups based on bar graphs representing all the measured weigths for the sites. The numbers given for the peaks are midpoints, with increments of 10 - thus, the number 80 represents all cups weighing between 76 and 85 grams.

As expected most of those sites with height standardization also have weight standardization: Akrotiri, Palaikastro, Mallia, Arkhanes and Knossos.

Regarding the base diameter, Table 6, the measurements here too vary somewhat from place to place, with the average at 3.6 - 3.8 cm. Akrotiri has a special position - the base diameters are generally bigger than any of the Cretan sites - the (equal) peaks are at 40 and 43, each being 25% of the total. In general the total range of diameter size at Akrotiri is larger than the comparable ranges on Crete, where the vast majority of bases have 3.3 - 4.1 cm diameter.

Those sites with good standardization regarding base diameters are Mallia, Nerokourou, (Arkhanes), Zakros, Pyrgos and Palaikastro. Akrotiri and Knossos show standardization also, though not as good. Once again, Kommos indicates a wide spread, with no clear peaks.

Table 6: The most frequently occuring base diameter measurement in millimetres and its percentage of the total number of cups based on bar graphs representing all the measured base diameters for the sites. The numbers given for the peaks are midpoints, with increments of 3 - thus, the number 37 represents all cups with base diameters between 35.5 and 38.4 mm.

The rim diameters, Table 7, are much more uniform, with a majority of the sites having rims of 7.5 - 8.5 cm in diameter. Akrotiri shows strong affinities with Zakros regarding the percentage of cups with rim diameters ≥9.7 cm; the larger range that the majority of the cups at these two sites fall into; and the larger peak. (In the Akrotiri study (Gillis in press), I discovered that the special group of cups with large volumes that clustered together all had rims of 10 cm or more in diameter. Therefore I used that figure to delimit 'extra-large' rim diameters. For technical reasons (how the increments fell), I have used 9.8 cm as the cut-off rather than 10 cm in this paper. Both numbers are of course arbitrary rather than absolute.)

The difference between Akrotiri and Zakros is that the peak at the former is clearly defined, while there is little difference in numbers between the peak at the latter and the other three groups included, in the 75% of cups which fall between 7.8 - 9.7 cm. The large figure for Pyrgos is explained by the large quantity - 28% - of cups ≥9.7 cm in diameter. The cups at Palaikastro also tend to have larger diameters than the average. We see that the average rim diameter is 80. Neither Mallia nor Arkhanes have cups larger than 9.7 cm in rim diameter.

The sites exhibiting good standardization are Akrotiri, (Arkhanes), Nerokourou, Zakros, and Palaikastro; those with standardization are Knossos and Mallia; once again Kommos shows a large spread with little indication of standardization. Pyrgos also has a large spread, and a very skewed peak, but this might be influenced by the large number of large-diameter cups.

Table 7: The most frequently occuring rim diameter measurement in millimetres and its percentage of the total number of cups based on bar graphs representing all the measured rim diameters for the sites. The numbers given for the peaks are midpoints, with increments of 5 - thus, the number 80 represents all cups with rim diameters between 78 and 82 mm.

Akrotiri has two peaks for volume as does Zakros and Pyrgos (see Table 8). As these sites also have two peaks regarding height, this is not unexpected. The average volume is 8 cc, with larger volumes for Zakros, Pyrgos and Palaikastro. As the heights and rim diameters of these three sites are also bigger than the average, this fact does not come as a surprise either. More unusual is the fact that Palaikastro has no cup with ayolume greater than 14 cc - this is explained by the 62% of the total amount of cups which fall between 9 and 11 cc. Regarding the size in general, with the exception of Knossos the sites seem to be clearly divided into those with only small conical cups and those with small and quite large ones. The reason that 15cc was chosen as the division between large and small was that it corresponds to a Greek wine glass and a Swedish coffee cup, in my opinion the smallest size one can drink from (Turkish coffee cups and whisky glasses notwithstanding). As indicated in a previous study of the conical cups from Akrotiri (Gillis in press), the function of the larger conical cups seems to have been different from that of the smaller ones.

Table 8: The most frequently occuring volume measurement in cubic centilitres and its percentage of the total number of cups based on bar graphs representing all the measured volumes for the sites. The numbers given for the peaks are midpoints, with increments of 2 - thus, the number 8 represents all cups with volumes between 7 and 9 cc.

Form elements:

All the form elements for base, wall, and rim were included in the initial statistical analyses (see Table 2, variables C8 - C11, and Fig. 1, 2 and 3). These were studied first as univariates in bar graphs, to ascertain the prevalence of each type within the variable (Table 3) and to see which type of base, wall and rim was most common for each site. Then each variable was combined with the others in cross - table form for each site - base:wall, base:rim and wall:rim (See Fig. 13 for an example of a typical cross-table).

The bar graphs yielded the following: the base which is the most common for all the sites in this period, and accounts for more than 50% of the cups from every site, is type 2, the raised base. The wall which appears most often is the curving wall, code 3, although in some sites, other wall shapes take precedence. The straight rim is also the most common for all the sites, being furthermore in the majority for all but Kommos. This is also reflected in Table 3. 

A compilation of the results of the bar graphs and cross-tables allowed me to make composites of the most typical cups for each site, along with other important or distinguishing characteristics (see Fig. 7). This compilation must be considered as just that, a statistical reconstruction of 'reality', made by combining the most frequent base:wall combination with the most frequent wall:rim one and checking with the most frequent base:rim grouping to ensure that there were no inconsistencies. Where two or more combinations are identical or very close, both variations are given in the figure. If there is a clear 'biggest combination' followed by one which is lower but still greater than the rest, the second is also given as an alternative. As is seen here, the most common cup in the LB I period has a raised base, curving wall and a straight rim (given as 2-3-1 in the figure). The only exceptions to this are Akrotiri and Knossos, which both have raised bases, slightly curved walls and straight rims (2-4-1) as the most common shape.

Decoration:

As can be seen from Table 3, the percentages of decorated conical cups and the variety of their distribution varies greatly from site to site. Some sites like Mallia and Akrotiri are notable for their almost complete absence of decorated cups, while others like Zakros are equally remarkable for having one third of all conical cups decorated, most with painted decoration. Some of this may be due to differences in excavation ideology and storage possibilities, but in the end one must rely on the material one has at hand as the starting point.

A composite picture showing which types of cups were decorated ('paint', monochrome or 'real' slip, as distinguished from 'self-slip') revealed that these cups did not have the same shapes as the common, plain ones (see Fig. 8). 'Real' slip in my definition (Gillis 1988c) is a surface layer usually applied to a vessel by dipping a dried cup before firing into a well levigated solution. When fired, this surface often has a different colour from that of the vessel. 'Self-slip' is created by the potter as the last step in throwing the pot, by smoothing the surface with water as the pot revolves on the wheel.) While the purposes of painting, slipping or monochroming might differ, as might those of slipping and 'self-slipping', this is not an issue which can profitably be taken up here. However, one can note that it is rare that the same shape is used for paint, slip and monochrome; usually, just the opposite is the case. The cups with paint on the inside only (code 5), are found at only two sites: Knossos with one (1-7-2) and Pyrgos with two (1-3-1 and 1-9-4). These are shown in Fig. 8. The examples given for Nerokourou-paint, Mallia-slip, Palaikastro and Zakros-monochrome, involve one cup only and are therefore not very representative.

While the composite picture of the form elements indicated a large degree of similarity in the shapes of the cups, there is very little uniformity from site to site regarding which forms are decorated. This seems to be a very individual matter. Regarding paint, Zakros and Pyrgos seem similar, while the rest are unique. They are also somewhat alike regarding slip, and in their lack of monochromed cups. There are no two sites that are similar to each other regarding slip or monochrome. Regarding the types of cups chosen in general to decorate, Akrotiri is more or less dissimilar to all the others regarding paint, while there is a good resemblance to Knossos regarding slip and a lesser one for monochroming.

Function:

Regarding traces of fire on the cups, while these marks provide clear evidence of something, it is not always easy to know of what. Those cups which are completely blackened inside and out would seem to indicate some form of function for contexts where there was not a great fire.

Those cups with blackened rims would reasonably have been used as lamps. All the other traces - partial inside, all outside, etc. - could be indications of function, of secondary burning or even of conditions in the earth. Therefore I included only codes 1 (burnt all over) and 2+8 together (rim). As can be seen in Table 3, most sites did not use conical cups extensively as lamps in LB I - only in Mallia and possibly Knossos did they seem to have been used as lamps to any great extent, while in Pyrgos, there are none with rim fire at all.

Most of the lamps surprisingly enough were thin-walled (less than 7 mm at the base and less than 6 mm in the middle of the wall). The most common rim by far was the straight one, although others were also used. Curving and slightly curving walls are predominant, although straight walls are also evident, and flaring walls at Zakros.

Furthermore, the amount of fire marks on vessels at a site is not helpful. Only 2% of all the cups at Pyrgos had traces of fire, and 7% at Arkhanes and Kommos, while 75% of those at Nerokourou revealed evidence of burn marks of one sort or another. While this seems to be significant, it is difficult to interpret in what way. Are we dealing with catastrophe layers, ceramic/non-ceramic ahernatives or functional differences?

PATTERNS OF SIMILARITY

Certain tendencies seem to emerge, but before any conclusions can be drawn, other types of analyses must also be used. As stated before, it was decided to compare the sites in pairs by converting the amounts present in each item into percentages, standardizing them, making a similarity matrix and then using the matrix as the basis for a cluster analysis.

As not all the variables or codes within some of the variables were informative, only those italicized in Table 2 were used. The figures in parentheses after them indicate the number of the variable. In total 25 variables were tested for 9 units, or sites, the 25th being all cups with a volume ≥ 14 cc, volume being considered a measure of function, as discussed above. The total data set is seen in Table 9.

Then the similarities were calculated. The variables were divided into three areas: shape (variables 1 - 18), decoration (variables 19 - 22) and function (variables 23 - 25). Table 10 in the column labelled Ak indicates the similarity of the Cretan sites in relation to Akrotiri concerning shape and decoration together. The numbers show how similar or dissimilar each site is to Akrotiri - the higher the number, the higher the degree of dissimilarity. Even though shape and decor are usually considered to go together, and were probably carried out by the same hand, we considered them together and separately, just to see if there was a difference: Table 11, column Ak, shows the similarity of the other sites to Akrotiri regarding shape, Table 12, column Ak, decoration. Table 13, column Ak, illustrates distances for function.

The distances (degree of dissimilarity) were then converted into matrices (the left parts of Tables 10 - 13). The matrices in combination with the Ak column show which sites Akrotiri is most like, as well as internal, Minoan similarities. The higher the number, the greater the dissimilarity.

Table 10, matrix for shape and decoration, indicates that there is a fair degree of similarity between Akrotiri and Knossos, with Palaikastro in second place. Pyrgos is the most dissimilar to Akrotiri in shape and decor together. Taken as a whole, there is no great similarity between Akrotiri and any of the other sites.

A look at Tables 11 and 12, where shape and decoration are split, presents another story: in regard to shape alone, Knossos of all the Cretan sites is still most similar to Akrotiri. Palaikastro is still in second place, but is now tied with Nerokourou. The matrix for decoration tells quite another story, however. Palaikastro, with a dissimilarity of 4, is almost identical to Akrotiri. It is followed by Mallia, Kommos and Arkhanes, with Knossos running a poor filth. There are no great differences, however, with only Pyrgos differing to any great extent. The considerably greater distance between Akrotiri and Pyrgos compared to the rest of the units is perhaps the result of one of the less suitable properties of the use of standardized data. It is a result of the effect of the presence of 2% of cups exhibiting decoration code 5 (painted inside only) at Pyrgos and a total absence of this variable at all the other sites with the exception of Knossos, where there was one cup with decoration code 5.

The pattern of decoration provided by the similarity matrix differs quite considerably from the compilation of decorated cups (Fig. 8) based on form, which was reported above. This apparent discrepancy can be explained by how one views decorated cups - as a shape upon which something may or may not be added, or as the pattern of various types of decoration in relation to each other, regardless of the medium upon which they are applied. The former approach corresponds to the composite drawing, Fig. 8, while the latter represents the patterns seen in the percentages, Table 3.

The function of the Akrotiri cups, as expressed through the use of fire and volumes greater than 14 cc is seen in Table 13, where it is most similar to that of the cups from Kommos, Pyrgos and Zakros.

Thus the analysis of the similarity matrices indicates that Akrotiri is most similar in shape to sites on the north coast: Knossos, followed by Palaikastro, Arkhanes and Nerokourou. As the information from Arkhanes is based on only 15 conical cups, all results concerning it must be taken with reservation. In terms of decoration another pattern emerges, with Palaikastro closest, followed by Mallia and Kommos, while Akrotiri equals the sites along the south coast and the east as regards function (Kommos, Pyrgos and Zakros). Thus it would appear that the conical cups at Akrotiri were thrown like the ones at Knossos, decorated like those at Palaikastro and used like those from the south coast.

The next step is to group the Cretan units by similarity to each other and to study the likeness of the Akrotiri assemblage to these groups. As mentioned above, a cluster analysis was used, based on the similarity matrices seen in Tables 10 - 13. The cluster analyses are represented here in the form of dendrograms (Fig. 14 - 17). In them we can see what we already have discovered, i.e., which Cretan sites group together and which groups Akrotiri most resembles concerning shape, decoration and function.

Concerning shape and decoration taken together (Table 9 variables 1 - 22, and Fig. 14), Knossos and Palaikastro form one group, Nerokourou, Arkhanes and Kommos another. In this second group, Nerokourou and Arkhanes are quite similar, more so than Knossos and Palaikastro to each other. Akrotiri is most similar to the first group, less similar to the second one, and least similar to the rest of the units, which are all three more or less distinct. The relative highness of the range of the numbers (136 - 334) representing the degree of dissimilarity between Akrotiri and the Cretan sites indicates that no Minoan site is very like Akrotiri, with Zakros and Pyrgos most different from Akrotiri, each other and the rest of the sites.

The dendrogram illustrating the similarities for shape alone (Fig. 15) is more or less the same as the one for shape and decoration, although Pyrgos is now added to the group that is most similar to Akrotiri. This similarity to Fig. 14 is explained by the large number of shape variables (18) compared with decoration (4). Here Mallia and Zakros are very different from Akrotiri and from each other as well. Zakros is most dissimilar to all the other sites (cf. Table 11).

In Fig. 16, the dendrogram for decoration only, the grouping is considerably different. Here there is a group consisting of Palaikastro, Kommos and Mallia which is similar to Akrotiri. Arkhanes and Nerokourou form another group. Knossos shares some similarities with the other sites, but no clear grouping is seen, while Pyrgos is quite different from the rest. As mentioned above in the discussion of Table 11, the explanation for the position of Pyrgos is due in part to statistical method.

The last dendrogram, Fig. 17, shows the similarities of function. Here three distinct groups are formed: the sites on the north coast and Palaikastro, the sites on the south coast and Zakros, and the single site in the west. Akrotiri is closest to the southern group.

RESULTS AND IMPLICATIONS

The phenomenon of trade and exchange of objects and ideas in the Bronze Age is a very complex one. While the existence of a great number of Minoan conical cups in the Cycladic town of Akrotiri is a fact, their arrival there is a cause for speculation and educated guesswork. In an attempt to explain their appearance more factually, four hypothetical models of 'transmission' were suggested. Two of these must be discarded at once, as they concern trade of Cretan conical cups to Thera. The evidence of the physical cups - made locally on Thera of Theran clay with Theran temper and inclusions - leaves no room for discussion: the physical cups were not imported from Crete, or anywhere else, for that matter. This leaves two possibilities to examine: that the Minoans moved to Akrotiri and settled en masse in the city, continuing to make their very handy little cups for private use as always (or have them made), or that the idea of the cups was transmitted to Thera, where it was adopted, adapted and manufactured by the Therans (Akrotirians).

If one assumes that a large contingent of Minoans settled in Akrotiri, large enough to warrant having their own potters, one would assume that the pottery now made there would be quite similar, if not identical, to that of the mother city or site. A pan-Minoan settlement is of course theoretically possible, but not very likely. If the idea was transmitted, rather than the actual objects, then one can assume that the Theran cups would resemble Minoan conical cups in general and be recognizable as such, but would very likely have a local flavour, being made by local craftsmen, and would not be an imitation of the cups from one Cretan site alone. What does the study indicate?

First of all, there seems to be remarkable similarity in the dimensions of cups between sites and between Crete and Akrotiri. While certain sites had their own peculiarities in one way or another (larger volumes at Pyrgos, for example), the majority of cups at the majority of sites seemed to fall within well-defined, narrow limits. Conical cups could not be mistaken anywhere for something else, and would be familiar everywhere.

Regarding their manufacture, we see that several sites exhibit a high degree of standardization, that is, that the majority of cups fall within narrow ranges regarding the dimensions for height, weight, base diameter and rim diameter (and even volume) and that a large percentage of the total was usually included in the peak measurement (see e.g. Fig. 9). These sites are Akrotiri, Mallia, Nerokourou, Zakros and Palaikastro. Knossos and Pyrgos showed some degree of standardization but not as high as at the first-named sites. Arkhanes exhibited good standardization, but as only 15 cups were included in the analysis, no conclusions may be drawn. Kommos was characterized by a lack of standardization, often with wide ranges and no clear measurement peaks.

This element of standardization can be somewhat misleading, however: in the case of Akrotiri and Nerokourou, it is quite possible that only one potter was active. While the figures indicate that he possessed skill in his work, they do not speak of any sort of uniformity among conical cup makers. On the other hand, Mallia, Zakros and Palaikastro, being palaces or large complexes, might illustrate just this phenomenon - a type of conical cup koine. Knossos shows standardization but to a lesser extent than the other sites mentioned above; however, the sources of my Knossian material are varied and widespread, both geographically and socially, and the cups are undoubtedly the work of several, perhaps even a great many, potters. Pyrgos presented an interesting case - the cups seemed to be easily divided into two clear groups regarding height, rim diameter and volume, which coincided with a division in their find context - the smaller ones came partly from the settlement, partly from the tomb, while the large ones were exclusively from the tomb. (This is to be taken up in greater detail in a forthcoming work.) Thus, the seeming lack of good standardization at Knossos and Pyrgos can be explained by other factors. The case of Kommos presents a problem: either there were a great number of very individualistic potters active at the same time, or the potter(s) who were working were not concerned with uniformity, due to lack of skill or interest. In general, the cups at Kommos were well turned and well fired, which seems to rule out the question of ceramic incompetence. Perhaps conical cups were of lesser importance in Kommos and thus were not made in such great quantities: the potter never perfected this shape. The fact of the relatively small total number of conical cups points in this direction.

The question of a sort of conical cup koine can be approached in another way, however: the combinations of elements into a whole. As was seen in the compilation of form (Fig. 7), the conical cups from almost all the sites had one shape which appeared more often than any other for this period: raised base, curving wall and straight rim. The only exceptions were Akrotiri and Knossos, which both had raised base, slightly curved wall and straight rim as the most common shape.

It could be expected that this similarity in form would be reflected by a similar resemblance in the manner of decorating (or not decorating). This is not the case, however, as can be seen both from the compilation of decorated forms (Fig. 8) and from the percentages of decoration given in Table 3.

Here there is no uniformity at all: no site is similar to any other regarding what shape is to be decorated with what. Even the patterns of surface treatment, measured in the similarity matrix and represented in the dendrogram, while indicating an extremely close relation between Akrotiri and Palaikastro, and no great (absolute) differences between sites, show enormous variation, from 30%, of all registered cups at Zakros having some painted decoration (Table 9 variable 19) to 92% of all cups at Mallia having no decoration at all, only self-slip.

Regarding function, this is difficult to determine without the benefit of the related finds in their architectural contexts. I have chosen to use only those variables which would be indicative of some specific function - rim fire indicating lamps, and large volume indicating drinking or storage, as mentioned above. I have also chosen to include the cups which are completely burnt inside and out.

While I cannot at this time give any definite function which they might have had, I feel that they are special enough to be indicative of something, and have therefore included them. Holes in the bottom, while rare in conical cups in this period, are found. However, as their function as funnels could vary so greatly - in storage contexts or as part of a religious ceremony, as seemingly with rhyta also - I preferred not to use them without the benefit of context. Similarities in function on Akrotiri are found at Kommos, Pyrgos and Zakros.

Thus a pattern emerges: conical cups from Akrotiri are close in shape to Knossos, followed by Palaikastro, while in terms of surface treatment they seem very similar to Palaikastro as far as the proportions of each type of decoration used, but more or less unlike all the other sites with regard to the shapes which are to be decorated, and with what. Functionally, based on the few diagnostic variables employed here, the cups from Akrotiri seem to be used much more as they were on the south coast of Crete and Zakros than anywhere else.

Let us now return to the initial problem: whether Minoan potters living on Thera produced the conical cups or whether local potters made them -either for Minoans or for Akrotirians. First of all, one must ask why a Cretan potter might be found in Akrotiri. Two obvious possibilities come to mind:

1. a large, all-encompassing population of Minoans lived in Akrotiri with its own 'service elements' such as craftsmen;
2. an itinerant Minoan potter had set up temporary or permanent residence in Akrotiri. In either case, however, one would expect that the conical cups the Minoan potter produced for his 'buyers' would resemble the cups he had made at home, which, as seen from the evidence presented above, is not the case here. Thus, one could say that the conical cups from Akrotiri do not seem to have been made by Minoan potters.

Assuming, therefore, that local potters were producing the conical cups themselves, one naturally asks for whom, and why: whether it was in response to 'orders' from resident Minoans on Thera, a general 'Minoanization' process or even perhaps a local response to newly introduced customs or needs. This is of course difficult to determine, and cannot be answered on the basis of the study of one kind of ceramic production. If we test the hypothesis that the cups were made by local potters for Minoans, using the find contexts of this originally Minoan object as a criterion one must assume that Minoans lived in all the houses thus far excavated. This indicates a very large or completely Minoan population. If this were the case, as argued above, one would expect to find Minoan craftsmen also, and thus the above mentioned similarities to the home production. Therefore one must perhaps reject the hypothesis that any sort of larger Minoan population lived in Akrotiri.

The other possibility, that local potters produced the cups for fellow Akrotirians and not for resident Minoans, is to me the logical one to explain the occurrence of these objects in Akrotiri.

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

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