1. The Development of a Cypriot Late Antique Ceramic Chronology: Analysis and Critique

Transcription

1 1. The Development of a Cypriot Late Antique Ceramic Chronology: Analysis and Critique 1.1 Introduction; methodological approach and background The methodological approach and arrangement of this thesis is not the typical one used for Late Roman/Early Byzantine material from an Eastern Mediterranean site. This was a conscious decision made by the author after careful thought and analysis of past and current work. By introduction it is important to place this thesis in perspective for the benefit of those who read and use the material presented in future. In any new analysis of archaeological material an archaeologist necessarily approaches the task with background experiences influencing decisions made. Often in pottery studies these influences are not made explicit, leaving it to the reader to try to interpret. Accordingly, I intend to briefly detail my background as an important influence upon the methodological approach taken in this project. Prior to beginning this thesis I have had several years experience in excavation and artifact analysis. My background in artifact analysis includes an Honours and a Masters thesis focusing on Bronze Age material from Cyprus. Before embarking upon this project I also had some experience in excavation and analysis of Late Roman and Byzantine material from Pella in Jordan and sites in Paphos. In theory excavation and ceramic analysis techniques are basically similar for any period. However, the intensive focus used for analysing Early/Middle Bronze Age material and the detailed and careful excavation techniques employed were important influences. The focus for the Bronze Age pottery was to obtain extensive information from a large collection of well stratified material in order to construct a reliable typology and chronology. This took the form of examining thousands of small sherds starting from a base of hardly any reliable past work to build upon. This thesis will present new material from a recent excavation in Cyprus, along with fresh ideas about typology and chronology in the Late Roman period. The excavations are in a sector of the Nea Paphos Theatre site in Southwest Cyprus. This is part of a larger project concentrating upon uncovering the remains of the Hellenistic Theatre by a team from the University of Sydney. At first encounter the site the material was not as well-known, however past experience gave me a basis from which to learn about a new group of artifacts. 1

2 What was a surprise was the technique I observed used by an eminent ceramicist inspecting a group of excavated Roman material from the Paphos site. The certainty with which the pottery was closely dated by century, merely from browsing through it, was interesting. For the expert no details of context or stratigraphy were needed, looking at the Fine wares and Amphora sherds suggested a date range almost immediately. There was no need to use identifiable coins to refine the dating because the Fine ware chronology was a secure and reliable tool for those who had spent years analysing Late Roman pottery. While admiring the breadth of experience that enabled the ceramicist to approach the material so confidently and quickly, I could not help wanting to investigate how the chronology and typology was constructed. The excavation of three specific trenches on the theatre site and the well stratified Roman material recovered was an opportunity to indulge this interest. From the beginning the material was not in the hands of an experienced Late Roman ceramicist. Some might consider the lack of experience a disadvantage but I have attempted to turn it into an opportunity. The on-site analysis of the excavated material began free of any preconceived ideas about typology and dating. I do not consider this a deficiency but rather an important opportunity to explore new ideas and perhaps challenge established ones. The plan was to present a full analysis of the whole corpus of material, including coins, from the three trenches. This is a key approach rarely attempted for Cypriot material and an ideal vehicle for a PhD thesis rather than a final site report. In fact the thesis is probably the best place to allow the time and words to present a very full analysis. Part of the methodology and presentation is deliberately different from the standard Late Roman site report. This is not meant as an attack but is part of a process of re-evaluatiing nearly a century of past work. This progression is very necessary for all archaeological material and is particularly needed for the study of Late Roman Cyprus. Scholars sometimes need to approach material from varied directions in order to test each others methodology and conclusions. This is a process I am very familiar with as regards Bronze Age Cyprus where there is a vigorous approach taken to assessing each others work in print. This is not the case for Late Roman Cyprus in my experience. Rather there is an atmosphere of general consensus in print. In using this thesis to question if the consensus is justifiable the intent is as a valid methodological approach to new material and not to be purposely hypercritical. Indeed the aim of this thesis is to re-evaluate past and current work and present new data to back up any conclusions. 2

3 The full details of the excavation of Area Three on the Paphos Theatre site, which is the focus of this thesis, are contained in Chapter Three. Before presenting the recent material, it is important first to define the context in which it is set. The basis for any analysis of new material is usually the current ceramic typologies of the Late Antique period in Cyprus. Today there is a growing body of comparative published material covering most types of ceramics and other artefacts. However, for most excavated assemblages the Fine wares are the first indicators used to date deposits in the absence (or presence) of any coin evidence. A focus on the work of John Hayes is essential to any analysis of the development of Late Antique ceramic chronologies in the Mediterranean. His seminal publication Late Roman Pottery (LRP), published in 1972, describes the basic ceramic typologies for the Red Slip Fine wares prevalent around the Mediterranean in the Late Roman and Early Byzantine periods. (Hayes, 1972) He has continued to modify and add to this work over the last thirty-five years. (Hayes, 1980b) At the time of its publication, this work provided a much-needed framework for Fine ware ceramics upon which future work could be based. The typologies described in LRP are still the benchmark, and in many cases the first and only reference, used to date deposits from sites all over the Mediterranean. (Fentress, 1998, p.5) While work has continued since 1972 on improving and modifying the African Red Slip ware typology, the basic model for the Fine wares produced in Cyprus remains essentially the same. In nearly thirty years, there has been little challenge to the validity of the Cypriot Red Slip Ware (CRSW) typology and chronology first developed by Hayes. The CRSW was never the main focus of LRP and the typology was based upon the evidence currently available at the time. If LRP was created as a useful working model to be improved and expanded upon by others when new material became available, then this has not happened. The question is whether this lack of challenge reflects the essential correctness of the original typology or merely that not enough newly excavated material has been analysed to dispute it. The history of the creation and interpretation of the CRSW typology and chronology is of significance for this thesis. The aim of this chapter is to investigate how and why Hayes began developing his ideas about CRSW and to look at other scholars work which may have influenced the formation of the typology. One line of investigation will be a site in rthern Cyprus, discovered in the 1950s, the Kornos Cave. In the publication of the site, it is clear that Hayes discussed with the author the identification of the pottery found in the cave. (Catling & Dikigoropoulos, 1970, Footnote 1) Indeed Hayes uses the 3

4 Kornos material as vital evidence in his own subsequent publication of the CRSW typology. (Hayes, 1972, p.381) The two CRSW bowls found in the Kornos Cave, plus another from Salamis, form the basis for a description of a common bowl shape in the typology. Hayes named these medium sized deep bowls Form 9 and the date range attributed to the form comes from the work done at the Kornos Cave. As will be expanded upon later, this dating is one of the foundations upon which the chronology of the whole CRSW typology was constructed. The Kornos Cave pottery is often cited by subsequent publications as reliable comparative evidence for dating CRSW Form 9 bowls and other types of pottery. (Flourentzos, 1996, p.25, Giudice, et al., 1999, Fig.4a: 3, p.292-3, Manning, S. W, S.J. Monks, D.A. Sewell and S. Demesticha, 2000, p.249, Rautman, 2000, p.321) As the Kornos Cave excavation was an important basis for the CRSW chronology and typology, it deserves a vigorous reanalysis. Such a process will establish if, over thirty years after publication, the material deserves its reputation as high quality comparative evidence. 1.2 The Kornos Cave The Kornos cave site was discovered in 1955 by a small unit of Royal Marines engaged in a security operation in the western end of the Kyrenia Range. (Catling & Dikigoropoulos, 1970, p.37) (Fig.1) It was searched in a routine operation by the marines who noticed a number of whole pottery vessels lying on the ground inside the cave; a soldier removed one. A local policeman then took the vessel to the Cyprus Museum. (Catling & Dikigoropoulos, 1970, p.39) Ten days later the head of the Cyprus Survey team, Hector Catling, visited the site. Catling and the policeman made a preliminary study of the cave and its contents but left without touching any of the artefacts. (Catling & Dikigoropoulos, 1970, p.39) The cave was then left for seven weeks before a small archaeological team from the Department of Antiquities and some caving enthusiasts from the British Military did a one-day search operation. This included completing a sketch plan, photographing the objects in situ and removing most to the Cyprus Museum. (Catling & Dikigoropoulos, 1970, p.39) (Fig.2.1) Catling was not part of this exercise and did not go back to the cave until nearly three years after the original discovery. By that time, three of the four objects left in situ had disappeared and Catling recorded his suspicion that souvenir hunters had been through the cave. (Catling & Dikigoropoulos, 1970, p.39) Finally another two men went back five years after the original discovery and found seventeen bronze coins and the remains of two keys, which had not been noticed before. (Catling & Dikigoropoulos, 1970, p.39) This 4

5 exploration history clearly illustrates that the excavation of the cave was neither systematic nor uncontaminated. Cave robbers could have removed more than the three pieces definitely known to have disappeared and rock falls prevented some areas of the cave from being fully investigated. (Catling & Dikigoropoulos, 1970, p.42-43) The cave takes its name from the high peak nearby called Kornos in the Kyrenia Range. The surrounding mountain slopes are steep and heavily forested. This whole area has been densely occupied continuously since the Early Bronze Age. The cave itself is a complex series of interconnecting chambers and small galleries with pottery, metal artefacts and animal bones found scattered throughout. There are many such natural caves in the area and they are assumed to have been used for habitation rather than as tombs. The majority of the objects found were located in the main chamber. These include a single amphora, fine ware bowls, plain ware jugs, cooking pots, iron artefacts and a lamp. (Catling & Dikigoropoulos, 1970, p.42) (Fig.2.2) In a bowl-shaped rock cleft in the main chamber the final group of searchers found the seventeen bronze coins and close by on the chamber floor two keys. (Catling & Dikigoropoulos, 1970, p.42) The coin group, which may originally have been gathered together in a box or purse, is dated to between A.D.610 and 644. (Catling & Dikigoropoulos, 1970, p.52) One small bronze key is illustrated and the other iron key is not. Both come from the main chamber floor where a range of other metal objects including knives, tweezers? and finger rings were found scattered. (Catling & Dikigoropoulos, 1970, Fig.5:28, p.50-51) Catling considered the cave to have remained undisturbed from when the last artefacts were left, perhaps the end of the seventh century, until rediscovery in The facts that the delicate stalactites in the cave were undamaged and the artefacts themselves intact, were considered good evidence for assuming the cave to be a sealed context. (Catling & Dikigoropoulos, 1970, p.43) Even if the terminal date for the deposition of the artefacts is reasonable, how can the length of the period of occupation of the cave be estimated? The cave was always considered an occupation site rather than a tomb context. Catling assumed that a solitary person used all the material found in the cave during one short period of occupation. (Catling & Dikigoropoulos, 1970, p.44) He was unconvinced that any group of people could live in the cave for an extended period. (Catling & Dikigoropoulos, 1970, p.43-44) However, there is no conclusive proof that people did not use the cave sporadically or permanently over many decades. It is clear that the occupant(s) must have eventually left and taken most things. The cave could also have been looted shortly after it was abandoned. (Catling & Dikigoropoulos, 1970, p.44) 5

6 Therefore, the existing evidence cannot support an assumption that the material from the cave can be characterised as a closed group. Catling assumed an absolute terminal date for all of the cave s contents as provided by the coins, and proceeded to confirm this by dating the other artefacts. The Fine ware ceramics were dated with reference to comparative examples and by consulting with Hayes. At the time, he was preparing a thesis on the Red Slip Fine wares of Late Antiquity, including the local Cypriot version. (Catling & Dikigoropoulos, 1970, p.44-45) The CRSW that had been found on the island before the 1960s included examples from Kormakiti on the north coast. (Fig.1) Catling used this comparative evidence but provided no details about the excavation on this site. He did note that some CRSW pottery was found in a fill deposit along with several coins. (Catling & Dikigoropoulos, 1970, p.46) The earliest was a coin of Heraclius A.D.613/14 and the latest was a coin of A.D.742/3. Based on this evidence Catling suggested a range of A.D for the main production phase of CRSW, while acknowledging that the excavated evidence was imperfect. (Catling & Dikigoropoulos, 1970, p.46) However, these are the exact dates later published in the CRSW section of LRP, relying heavily upon the Cypriot evidence assembled by Catling. Catling had a latest coin date of the mid-seventh century and a pottery date range of A.D for the Kornos Cave. However, as already mentioned he had developed the idea of a single short occupation and so deduced that all the pottery could be classed as a homogeneous group. If a local hermit or monk used the cave at the very end of the seventh century, as Catling believed, then the two hundredyear date range for the pottery was too long. He needed to confirm a date for the pottery more specifically to a shorter period around the mid-seventh century. (Catling & Dikigoropoulos, 1970, p.59) 1.3 Dating the Pottery- The Salamis Deposit Catling s methodology was to compare what he believed to be a reliable closed group of pottery from Salamis to the Kornos group. (Catling & Dikigoropoulos, 1970, p.53) If they were similar, this would provide the independent evidence to confirm his hypothesis that the Kornos Cave group represents a short period of occupation. As already mentioned there are some problems with the integrity of the Kornos Cave excavations that may negate its status as an uncontaminated closed context. Catling s methodological approach, based upon the principles of crossdating, is only sound if the status of the closed group from Salamis can be confirmed. 6

7 The Salamis material came from excavations conducted in the Public Baths by the Department of Antiquities in (Catling & Dikigoropoulos, 1970, p.54) The complex was first built as a Gymnasium but was adapted in the Byzantine period to be the main Public Baths. During the excavation, a built bench, perhaps constructed in the last phase of repairs to the building, was found against the west portico wall of the bath. (Catling & Dikigoropoulos, 1970, p.54) A section drawing of part of the Salamis excavation is provided in the Kornos Cave article along with a discussion of the results. (Catling & Dikigoropoulos, 1970, Fig. 6, p.53) (Fig.3) The bench is shown as being composed of two distinct layers, 9a is loose earth and rubble and 9b a very hard packed earth. Originally, the bench was faced with sandstone and the top sealed with plaster. The excavator was certain that other later material on top did not contaminate the fill from the bench. Both layers of the bench are thought to have been deposited at the same time, as coins of Heraclius A.D were found in both. (Catling & Dikigoropoulos, 1970, p.54) The section drawing presents some problems in interpretation. (Fig.3) It demonstrates that the wall and attached bench were exposed to the elements and heavily eroded before being covered by thick layers of sand. The top level of bench fill 9a is apparently composed of loose rubble but it is shown with very straight edges. Level 6 is also described as loose earth and debris but the section shows a definite break between it and the edge of the 9a loose bench fill. Other layers shown in the section drawing also seem problematic. For example, level 4a is a hard packed earth lens sticking out into the section and surrounded by the major loose sand fill of level 2. By no means can this section of the excavation be seen as a sealed context. However, the two layers of the bench itself are characterized as a closed context and so considered important for dating purposes. The pottery assemblage from the two bench levels are presented together despite the fact that it comes from different layers. (Fig.4.1) Only thirteen ceramic items are illustrated from this bench group, including six fragments from CRSW bowls, two cooking pot fragments and five plain ware bases. (Catling & Dikigoropoulos, 1970, Fig.7, p.54) It is not clear if this is everything that was found in both layers. A single lamp was also found in the bench deposit, in which layer it is not stated, dated to the first half of the eighth century. (Catling & Dikigoropoulos, 1970, p.55) (Fig.4.2) The lamp was dated by an independent authority at the time. This type would be classified today as a Cypriot Type 19 series 5, its date range difficult to pinpoint. (Oziol, 1977, Pl.44:799, 800 & 804, p.266-7) Examples from all over the Eastern Mediterranean can be dated from the fourth to the eighth centuries but the 7

8 main period of production seems to have been the sixth to seventh centuries. (Oziol, 1977, p.256) Evidence from other areas of the Salamis public baths suggests building alterations were carried out up to around A.D.700 following the Arab raids of A.D.649. (Catling & Dikigoropoulos, 1970, p.55) If the bench was constructed during these fifty years then the lamp, as originally dated, is too late to be in the bench fill. 1 Catling & Dikigoropoulos suggest the bench fill itself was composed of debris from the Arab raids. (Catling & Dikigoropoulos, 1970, p.55) The earliest coin from the bench is of Justinian A.D , and the latest is a Constans II A.D However, a date range of A.D is suggested for the material from the bench. (Catling & Dikigoropoulos, 1970, p.55) The A.D.650 date comes from assuming the fill is the remains of the Arab raids and the A.D.725 date is a result of trying to fit in the later date of the lamp. By suggesting that the material from the bench is a closed homogeneous group most scholars tend to inevitably date all the material to the latest available absolute date. The bench was made up of two different kinds of fill that could have come from anywhere nearby. The material used to form the bench may have been collected after A.D.650 but the pottery in the fill does not all have to date to this period as well. Both types of bench fill could have included a mix of earlier and contemporary pottery and therefore neither should be described as a sealed deposit. Certainly, one might be able to give a terminal date for this material but the earliest date must be open to question. By using this methodology, there is a tendency to consistently late-date the ceramics and consequently one cannot investigate the idea that certain types could have developed earlier. There are close similarities between some of the Salamis bench pottery and the group from the Kornos Cave, but how important is this in terms of relative chronology? These are two groups of material, which are not from reliably sealed deposits but are assumed homogeneous groups of a similar date. In both cases, the coin evidence was utilized to give a latest date and then applied as a general date for all the ceramic types. Catling wanted the Salamis group to be the dependable closed group to confirm his hypothesis about the Kornos group. However, the Salamis example is not wholly reliable itself and therefore cannot be used to support any conclusions about the Kornos Group. It is possible that both groups are mixed 1 In fact, the wide date range and lack of detailed typology for this type of lamp means it is not useful for a process of defining a relative date for the pottery assemblage from the bench deposits. 8

9 contexts containing material from the early sixth to seventh centuries. Just like the Salamis bench fills, the material found in the Kornos Cave may be the remains of accumulation over several hundred years. Catling may be correct in his assumptions about both groups but a reanalysis shows that the evidence is not indisputable. One of the aims of this thesis is to present data that confirms the suggestion made in this chapter; that neither group are closed homogeneous assemblages with a narrow date range. The final question is whether the Kornos and Salamis assemblages present good enough evidence with which to construct a typology and chronology for CRSW. The consequences of the dates assigned to the ceramics from the Kornos Cave and the Salamis bench have been far-reaching, including the fact that these groups formed an essential part of the evidence used by Hayes to construct his CRSW typology. A good example of these ramifications is found at the important Cypriot site of Dhiorios. In his publication of that site, Catling used the Kornos Cave evidence to date the ceramics. 1.4 Dhiorios: Dating the Pottery Catling conducted a small rescue excavation at Dhiorios, a site close to the Kornos Cave in rth Cyprus, in (Fig.5) Initial investigations revealed a probable kiln site with a lot of dumped pottery that needed to be dated. (Catling, 1972, p.3) A three-week rescue excavation opened a variety of areas including part of a kiln, storerooms and a possible olive press. (Catling, 1972, p.3-4) The trenches covered only a small part of what surface survey suggested was a large site. (Fig.6) As might be expected from a kiln site, a huge quantity of pottery was recovered, much of it apparently kiln waste. Of this assemblage, only four hundred items were sorted, examined and published. (Catling, 1972, p.4) Several detailed sections and photos of the trenches are provided in the site report. They show surviving walls lying just below the topsoil and the associated deposits must have been much disturbed by ploughing. Some areas had a deeper stratigraphy that the sections indicate to be very complex, with many overlapping small lenses of soil. (Catling, 1972, Fig.18 p.28, & Fig.19, p.30) (Fig.7) The output of the kilns was specifically Cooking ware in a limited range of shapes. The excavation of a nearby storeroom recovered plain wares, amphorae and fine wares in smaller numbers, it is difficult to know if any were made on the site. 9

10 Catling valiantly attempted to divide all the ceramics into phases and described four periods of occupation on the site. There were no coins found during the excavation and only a few lamp sherds. Therefore, the dating of the kiln pottery relied on the small number of Fine ware sherds found during the excavation. (Catling, 1972, p.79) Catling detected a Period I that dates from the Late Hellenistic to the fourth century A.D. This was not the period in which the kilns were used. (Catling, 1972, p.79) The pottery factory production is divided into Period II and then Periods III and IV as a group. Small amounts of CRSW were found in Period II deposits as well as in Periods III and IV. Using the evidence from the Kornos Cave, Catling restates his view that the general date range of CRSW was A.D The Kornos pottery is dated around A.D.650 but Catling felt the CRSW at Dhiorios was of inferior quality and so must be at least 50 years later, placing the end of Period IV at A.D.750. (Catling, 1972, p.80) If the latest date is A.D.750 then the earliest date, as represented by Period II, could be as early as A.D.550 according to Catling s date range. However, Catling felt that the deposits on the site could only cover about one hundred years of use. (Catling, 1972, p.80) Therefore, the kiln production begins around A.D.650 and ends about one hundred years later. This discussion illustrates how the high dating of the Kornos material has heavily influenced the analysis of the Dhiorios site. The result is that Dhiorios is assumed to have been inhabited from about 50 B.C. to A.D.300; the fourth, fifth and sixth centuries are missing, with the site back in production in the seventh and eighth centuries. Catling made it clear that his chronology was not fixed and may have to be lowered. He also acknowledged that much more work needed to be done on the chronology of Late Roman Pottery in Cyprus; no one has so far taken up the challenge. (Catling, 1972, p.80) Catling presents a huge range of well-published material from the Dhiorios site; of importance is the fact that nearly all is locally made Cooking ware. This type of ware had never received such detailed treatment before nor has it since, so this site is often used as comparative evidence by other ceramicists. Yet, a reanalysis shows that however impressive the exhaustive detail of the publication is, if the assumptions underlying the dating of the site are not valid, then the usefulness of the pottery as comparative dating evidence is limited. The material presented in this thesis will show that the dating of Periods II, III and IV at Dhiorios is at least 10

11 two hundred years too high. A more detailed analysis of the ceramics from Dhiorios, particularly the Cooking ware, is included in Chapter Seven. 2 In summary, the analysis of this older published material from Cyprus has shown that long-held assumptions about the integrity of certain ceramic material can no longer be taken at face value. The evidence does not come from carefully controlled excavations and quality stratigraphic deposits. An analysis of the methodologies used to arrive at the dates for the Cypriot ceramics shows that they are closely dated mostly by assertion only. As a result, the CRSW types were grouped around the mid-sixth to seventh centuries. This leaves the fourth and fifth centuries with virtually no local Fine ware production on the island. This important anomaly has existed for the last thirty years at least and needs to be explained. 1.5 Typology and Chronology: Deconstructing the Hayes Model The basic methodology used by Hayes for constructing the Late Roman Fine ware typologies was to collect examples from sites all over the Mediterranean. The hope was that the pieces were from reliably dated contexts so the relative dates would generally agree. Thus, a clear type-series with neat associated date ranges could be created to be user-friendly for all. (Fentress, 1998, p.5) This process of chronological seriation to establish a sequence of ceramic types has a long tradition. (Rice, 1987, p.436) There are several points about this methodology that Hayes did not specifically address in LRP such as: How many examples are necessary for a legitimate sized sample for each form? Can a single piece from a site or deposit be a representative example for a specific form? Should each example be accompanied by essential stratigraphic information so it can be assessed independently for its integrity? The Cypriot typology that Hayes presented in LRP is less detailed than the African Red Slip ware (ARS) one that was the main subject of his research. (Hayes, 1972, p ) At the time, there was a larger amount of excavated ARS pottery to include in the analysis. For the other Fine ware groups such as Egyptian Red Slip ware (ERS), Late Roman C ( Phocaean Red Slip Ware), and Late Roman D (CRSW) there was a much more limited corpus of examples. (Hayes, 1998, p.9) 2 The author did receive permission from the Cypriot Department of Antiquities to study the pottery from Dhiorios as well as that from the Kornos Cave. Unfortunately, the material was found not to be in the Cyprus Museum as hoped but located in the occupied rth of the island and therefore inaccessible. 11

12 The typology frameworks for these minor Red Slip types imitate those developed for the ARS. Therefore, the CRSW typology is divided into Form numbers from 1 to 12 based upon shape differences. Because this was the first attempt at constructing a typology for the ware, the Form numbers also have inherent chronological assumptions with 1 being the earliest type and 9 to 12 the latest. For the CRSW typology, Hayes included examples from Cyprus itself but many more from sites that he was familiar with around the Eastern and Western Mediterranean. The main question to resolve about this methodology is whether one or two pieces from distant sites like the Athenian Agora or Antioch is good evidence to construct the typology of a type originating in Cyprus. Forty-seven examples are utilised as evidence to support the framework of the CRSW typology and chronology; only nine are from Cypriot sites. examples from Cyprus are provided for Forms 1 through to 8. This is particularly disquieting for Form 1, which is supposed to be the initial form of the whole CRSW range. Hayes suggested a late fourth century (or earlier) date for this shape as found at Abou Mina, in Egypt. (Hayes, 1972, p.373) How much earlier might Form 1 have first appeared in the Cypriot workshops? In the absence of published new evidence, scholars have continued to use the Hayes date range for this earliest form. The three Cypriot pieces used to date CRSW Form 9 are from the Kornos Cave and the Salamis bench deposit. As discussed above, the evidence for the date of these can be questioned. Another four Cypriot examples included in the typology are from the Agios Georgios excavations at Peyia and are used to date Forms 10, 11 and 12. Hayes used Megaw s dating to affirm that the pieces come from a destruction deposit dated to the middle of the seventh century. (Hayes, 1972, p.383, Footnote 1) There is no further detail given of the context or reliability of this excavation. The Peyia pieces may be from an excellent sealed context but this cannot be accepted without any supporting published evidence. The final two Cypriot examples also have no contextual data to support their inclusion as datable examples. For Form 11 an example from the Cyprus Survey is quoted with no other details and for Form 10 an unpublished piece from Ayios Philon. (Hayes, 1972, p.382-3) The small number of Cypriot examples used by Hayes was probably all he could find to work with at the time. Nevertheless, as this discussion has shown there is a possibility that some are not dependable examples. The many foreign examples utilized often include only basic information about their context. Obviously many are pieces about which Hayes acquired a personal knowledge even if they are from unpublished sites. Should this material, that is obviously a Cypriot product imported to the sites, be more or less important in constructing the typology? Hayes does not make a 12

13 distinction between the two sources of evidence. For the development of a basic framework typology for CRSW, Hayes had to make use of the available evidence at the time. Thirty years later, it is from Cypriot sites that reliable deposits with a large number of examples of each CRSW type can be found. From this larger body of material comes the opportunity to investigate the internal variability of each type and refine the dating. The methodology of naming the Cypriot types in a numerical sequence has had specific consequences. The numbering of types from 1 to 2 thence to 9, as the basic sequence, has made it difficult to explore the possibility of greater complexity. Hayes admitted that the developmental sequence of Forms such as 3, 4 and 5 are not obvious based on then current evidence. (Hayes, 1972, p ) He was definite about a CRSW sequence beginning with Form 1, starting in the late fourth century (or earlier) and ending around A.D.475. (Hayes, 1972, p.373) Form 2 was thought to have developed in the late fifth century, after Form 1 ceased production, and goes through to the mid-sixth century. Hayes was certain that this form is a direct copy of ARS Form 84. (Hayes, 1972, p.375) However, he dated the main production phase of ARS Form 84 considerably earlier, between A.D (Hayes, 1972, p.130, Atlante, 1981a, p.69) What is the theory behind a seventy-five year gap between the development of the form in rth Africa and it being copied, produced and exported in Cyprus? If Form 84 were a popular shape, would the Cypriot market really wait until it was superseded in African workshops before producing a good copy to export? CRSW Form 2 is a highly developed shape with the addition of rouletting and stamped decoration. It does seem a big leap between producing the simple earlier Form 1 shapes to mass production of this new complex type directly copying an import. However, perhaps there were other shape developments after Form 1 and before Form 2 and not the abrupt change Hayes describes. This might explain the development of the form more clearly than suggesting it was simply a direct copy of an African type, which was already old-fashioned before CRSW Form 2 was first produced. The principal shape of the Cypriot typology is the medium to large bowl named Form 9. It was characterised as a direct derivative from the Form 2 bowl shape. (Hayes, 1972, p.376) This means that Form 9 begins in the mid-sixth century, when Form 2 is no longer in production, and ends around A.D.700. (Hayes, 1972, p.382) Hayes did note that the earliest examples of Form 9 are of a better quality in shape and finish than the latest Form 2, but could not give a reason why. (Hayes, 1972, p.381) The only real reason that Form 9 was characterized as a direct successor to Form 2 is the late date assigned for the production of the form. inherent shape 13

14 characteristics suggest a sequence from Form 2 directly to Form 9. In fact, some of the small-sized, early Form 9 bowls are closer to Form 1 in shape. Hayes was very confident that the late dating for Form 9 was secure. As mentioned above, the main piece of Cypriot evidence utilised is the material from the Kornos Cave and Salamis, described as being from an excellent closed group of the 640 s. (Hayes, 1972, p.382) As argued above, the start date for Form 9 production is still open to question. If it is assumed that the form is a derivative of Form 2, a date after A.D.550 must follow for the first production of the form. This late dating affects the phasing of other associated shapes identified by Hayes. Forms 10 and 11 are both positioned late in the CRSW sequence because the very few examples cited were found in association with Form 9. (Hayes, 1972, p.382-3) Evidence will be presented in this thesis of Form 9, 10 and 11 examples in a sealed context dated by coin evidence to around the fifth century. This data will support the suggestion that the CRSW typology should be revised with Form 1 followed chronologically by Forms 8, 9, 10, 11 and Form 2 simultaneously. Popular shapes like Form 9 start earlier than Hayes envisioned and continue through to the seventh century. The Cypriot typology may not be the singular progression as suggested by Hayes but a more complex production of several forms simultaneously for both the home market and export. Popular shapes were produced for a long period with new possible shapes introduced at intervals throughout. There may also be an added element of regionalism around the island, with different factories introducing new shapes. 3 Hayes worked on the basic premise that the major shape developments that took place in CRSW were directly influenced by the African Fine ware production. He stated, African Red Slip Ware is the leading fine ware in most parts of the Mediterranean world and seems to have dictated pottery fashions throughout the area for much of this time. (Hayes, 1972, p.13) In LRP, he detailed a complex African Fine ware ceramic production industry. However, perhaps because of a lack of published evidence, the Cypriot typology illustrates a more simplified industry with little internal impetus. Cannot CRSW be part of a single unbroken tradition, whose origins may be traced back to Hellenistic times, the same as Hayes envisioned for ARS (Hayes, 1972, p.13)? Even today, Hayes still argues that the CRSW vessel forms derive from ARS models, particularly the early types. (Hayes, 3 Regional variation in pottery tradition is a very strong trend in Cyprus stretching back to the Bronze Age. Much more quality evidence is needed from sites all across the island before further study of this process can be undertaken for the CRSW. However, it is important to note that there is every chance that regional differences were significant in the Late Roman period. 14

15 1998, p.10) The dating range that Hayes applied to CRSW leaves at least a two hundred-year gap between the ceramic production of Cypriot Sigillata (CS) and the later Fine ware industry. During this gap, Hayes inferred that the knowledge and skills base of the ceramic industry was lost. Only the impetus from a prosperous ARS production, with new types reaching Cyprus, influenced a resumption of Cypriot production. If good quality evidence could be presented to show that this gap in Cypriot production did not exist, this would challenge the accepted supposition. Such evidence might confirm a suggestion that the late dating of CRSW types has artificially created the gap in production. If the date of the start of CRSW production could be lowered and evidence showed a more complex development of different shapes over a long period, then the major influence for CRSW shapes and production would be CS types and not the African industry. In this thesis, it will be argued that the Cypriot Fine ware industry was a continuous production with no large gap between the later Sigillata types and a development into early CRSW forms. This CRSW production then continued for a long period into the late seventh century. Hayes described the CRSW pottery in general as characterised by heavy rather crude potting and rough irregular rouletted decoration and a production centre in some unknown location on the island. (Hayes, 1972, p.371) However, the many more examples found since the publication of LRP from different locations on the island suggests that several production centres may have been located in various regions. In this thesis, evidence will be presented to suggest that the Paphos region was a vigorous production centre, as would be expected from a former capitol. Perhaps competing with another region in the northeast of the island focusing around Salamis. While Form 9 was a principal CRSW shape produced early in Paphos, Form 2 was possibly a product from the Salamis region specifically copied for the export markets on the Syro-Palestinian coast. 4 Today we know that the quality of the overall Cypriot product is more variable than that described early in the 1970 s by Hayes. There is no doubt that some examples from the sixth and seventh centuries can be rather roughly finished but the earlier product of the fourth and fifth centuries is generally of high quality in both potting and decorative finish. The product was obviously a sought-after trade item as revealed by the many locations around the Mediterranean in which it was found. If 4 This point will be further expanded upon in Chapter Five, Section

16 it were generally a poor-quality product then why would it have been so popular across Lower Egypt and the Syro-Palestinian coastal towns where there were already local Fine ware pieces available? While Hayes may have meant the CRSW typology to be an initial working model that could be expanded upon, this has not happened in reality, despite some thirty years of excavation since it was published. (Hayes, 1998, p.9-10) There is a demonstrable tendency for some excavators to use this typology and chronology as the only tool to date deposits, without investigating the possibility that their evidence may be part of what is needed to help modify and expand upon it. (Fentress, 1998, p.5) Deconstructing the CRSW typology Hayes created reveals that it is not absolutely inaccurate. It is a simplified scheme, which includes some assumptions based upon insecure evidence. Methodologically it is essential that a ceramic typology with the accompanying chronology be based upon a large number of securely datable examples from wellstratified deposits. Ideally, the contexts from which the examples come must be demonstrably uncontaminated deposits. Detail about the way they were excavated and the artefacts associated should be available for scrutiny. If the foundations upon which the CRSW typology was constructed are possibly insecure, the next question must be: Has any evidence from published excavations in the last thirty years contributed anything in support or dispute of the Hayes model? This issue will be explored in Chapter Two. 16

17 2. Thirty Years of Site Reports: Review and Evaluation 2.1 Introduction The analysis of the new material could not have been achieved without reference to past and current work on ceramic research in the Eastern Mediterranean. In Chapter One some of the inherent problems with the Hayes typology and chronology for Cypriot Fine ware ceramics were discussed. The purpose of this chapter is to appraise the published results from other sites in the region with the view to assessing which might be most relevant for a process of refining the Cypriot typology. This evaluation includes comment upon the methodological approach to ceramic analysis used for each site and any contribution towards a reappraisal of current typologies. Of necessity, several sites in Cyprus, particularly in the Paphos region, are the focus of most interest for comparative purposes. Amongst the large number of site reports from around the Eastern and Western Mediterranean available for review, the second part of the chapter deliberately concentrates upon those with most significance for this project. 2.2 The Cypriot Sites5 In Chapter One, the reputation of the early excavation evidence from the Kornos Cave, Dhiorios and Salamis was questioned. Further evidence from systematic long-term excavation of settlement sites did not really exist in Cyprus before The Swedish Cyprus Expedition had done the main work conducting a series of fastpaced excavations at sites all around the island. The results were published in 1956 covering material from the Hellenistic and Roman periods. (Vessberg & Westholm, 1956) There was a significant amount of Late Roman and Byzantine material available for study, but most consisted of finds from rescue excavations of large communal tombs. Often these had been looted and/or flooded and the contents rearranged over centuries of use by successive generations. The material is interesting because of the large number of whole vessels preserved, as opposed to the more fragmentary assemblages from settlement sites. However, in terms of 5 The site names used in this thesis for Cypriot locations follow those used in the publication of the sites. In Cyprus the double barrelled names usually mean the first is the general place name and in italics (or not) after this is the specific locality name. 17

18 efficacy for providing evidence to construct a detailed ceramic typology this material was inadequate. (Hayes, 1977a, p.96) The tombs at Kambi Vasa, on the southern slopes of the Troodos massif, are an example of the frustrating situation of good quality material to which only very general dates can be assigned. (Du Plat Taylor, 1958) (Fig.8) The tomb groups included a wide range of pottery and glass as well as many coins. Detailed attempts were made to try to identify separate burials and the accompanying artefacts to go with each skeleton. (Du Plat Taylor, 1958, p.45) However, there are special problems involved in constructing a tomb chronology and often settlement material is needed for comparative evidence. There are questions to be asked with reference to the artefacts themselves such as; How old they were when deposited, and are there specific types of artefact more prevalent in a tomb assemblage? They could be new items or possessions already a generation old when deposited in the tomb. In the Cypriot context, tomb material is not often used as reliable comparative evidence for the Late Roman and Early Byzantine periods. The main use seems to be providing examples of whole vessels to compare with stratified fragments found in settlement excavations. The publication of new tomb material from Paphos may provide some important fresh evidence in the near future. The published evidence from Cyprus discussed below comes from settlement excavations around the island conducted over the last thirty years. A review of the current state of knowledge is necessary to establish where the new material from Paphos fits into the presently accepted representation of the Late Antique in Cyprus Kalavasos-Kopetra The site of Kalavasos-Kopetra is located in the Vasilikos valley along the south coast of Cyprus. (Fig.8) The project began in 1987 and involved several seasons of intensive survey and excavation of selected areas. (McClellan & Rautman, 1994, p.289) The aim of the project was to explore a smaller provincial Late Roman settlement to provide a contrast to previous excavations mostly based around larger coastal towns. (Rautman & McClellan., 1992, p.265) After several seasons, the authors emphasized that their main goal, before final publication, was to refine the site s stratigraphy. (McClellan & Rautman, 1994, p.289) At present there has been no attempt to use the results from this site to question or refine any current Late Roman/Early Byzantine ceramic typology. 18

19 In fact, it is explicitly stated that the conventional dating of the fine wares would be used to establish the chronology for Kalavasos-Kopetra. (Rautman & McClellan, 1992, p.265) The Fine ware pottery was catalogued according to Hayes standard types and the general date of A.D assigned to most of the CRSW follows the traditional chronology. (Rautman & McClellan, 1992, p.265) There is no discussion in the preliminary reports on the subject of why the excavators view the Hayes typology as securely dated. This site is an example of how LRP is considered an invaluable tool for Cypriot sites with few coins and artefacts found in specific deposits. In the preliminary report of Area II at Kopetra, the site of a Basilica, a short catalogue of selected artefacts is included. (McClellan & Rautman, 1994, p.293) It must be assumed that these specific pieces help date the phasing of the building. The Fine ware examples illustrated are compared either directly to LRP examples or to pieces from the Late Roman site of Anemurium on the south coast of Turkey. (McClellan & Rautman, 1994, p. 293,.1-8) As will be discussed later in this chapter, the dating of the CRSW from Anemurium is directly based upon the Hayes standard types. evidence except the Hayes typology is used to substantiate the late sixth and early seventh century date assigned to the final phase of the Basilica in Area II.. The pottery evidence from the deposits in Area VI at Kopetra establishes two phases ending in the late seventh century. (McClellan & Rautman, 1994, p.303) The short catalogue of artefacts draws its datable comparative evidence from restricted sources; LRP and Anemurium for the Fine wares; the Salamis bench deposit for a Plain ware jug base; the Dhiorios site for a Cooking pot. (McClellan & Rautman, 1994, p.306) As discussed in Chapter One, these examples may not actually be securely dated comparative evidence. If this were true the dating of some levels on this site may be too late. Certainly, as will be illustrated in later chapters, some material from this excavation is comparable to the new material from Paphos suggesting that some revision is necessary. The most recent publication concerning the Kopetra site attempts to place this site within the context of the whole Late Roman island. (Rautman, 2000) The conclusions reached about the dating of this site being single period and all the assumptions made about the pottery types found is based upon the standard Hayes CRSW chronology. (Rautman, 2000, p.320) This leads to a continuous pushing forward of the chronology and influences all the subsequent statements made in this article about Late Roman Cyprus. Rautman continues to support a theory that it was only in the mid sixth century that CRSW became the prominent Fine ware on the island. He acknowledges that there is new evidence for the earlier production of the ware but affirms that the Kopetra pottery, not especially well stratified, supports the Hayes dating scheme. (Rautman, 2000, 19