British School at Athens is collaborating with JSTOR to digitize, preserve and extend access to The Annual of the British School at Athens.

Transcription

1 Byzantine and Allied Pottery: A Contribution by Chemical Analysis to Problems of Origin and Distribution Author(s): A. H. S. Megaw and R. E. Jones Reviewed work(s): Source: The Annual of the British School at Athens, Vol. 78 (1983), pp Published by: British School at Athens Stable URL: Accessed: 15/01/ :01 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. British School at Athens is collaborating with JSTOR to digitize, preserve and extend access to The Annual of the British School at Athens.

2 BYZANTINE AND ALLIED POTTERY: A CONTRIBUTION BY CHEMICAL ANALYSIS TO PROBLEMS OF ORIGIN AND DISTRIBUTION (PLATES 24-30) INTRODUCTORY REPORT FORTY years ago, Charles Morgan1 complained: 'That many of the finest wares recovered on various European sites, at Pergamum, and elsewhere, are products of single centres of manufacture is apparent from their uniformity of fabric, but thus far no one of these has been identified.' Since the position has not much improved today, it occurred to us that the origins of, and the patterns of trade in, Byzantine and allied ceramics might be clarified by the same technique of spectrographic analysis which has made it possible to assign to the Peloponnese, Crete, and Cyprus their separate productions of pottery decorated in similar Mycenaean styles. As a consequence, the analyses of which the results are reported here were made in the Marc and Ismene Fitch Laboratory.2 The identification by analysis of the clay used at production centres known from the presence of potters' kilns, or otherwise, should lead to the localization of distinctive wares if their clay compositions were established. But the pottery employing a particular technique and style of decoration was evidently produced at more than one place, and a complete picture can only emerge after the composition characteristics of the clay of all the possible centres and of all the unlocalized wares have been established. The investigation reported here was limited to analyses of fourteen batches of specimens, seven of which derive from well-attested production Abbreviations Bakirtzis BP Catling CMGP Ch. Bakirtzis, 'Didymoteichon: un centre de ceramique postbyzantine', Balkan Studies 2 (Ig980) Ch. Morgan II, The Byzantine Pottery: Corinth XI (Princeton 1942). H. W. Catling, 'An Early Byzantine Pottery Factory at Dhiorios in Cyprus', Levant 4 (1972) J. du Plat Taylor and A. H. S. Megaw, 'Cypriot Medieval Glazed Pottery-Notes for a Preliminary Classification', RDAC Dikigoropoulos A. I. Dikigoropoulos and A. H. S. Megaw, and Megaw 'Early Glazed Pottery from Polis', RDAC Frantz M. Alison Frantz, 'Middle Byzantine Pottery in Athens', Hesperia 7 (1938) Great Palace I Robert B. K. Stevenson, 'The Pottery, ' in The Great Palace of the Byzantine Emperors... First Report... (Oxford-London 1947). Hippodrome I D. Talbot Rice, 'The Byzantine Pottery' in Preliminary Report upon the Excavations in the Hippodrome of Constantinople 1928 (London 1928). Lane Arthur Lane, 'Medieval finds at Al Mina in North Syria', Archaeologia 137 (1937) Megaw 1968 A. H. S. Megaw, 'Zeuxippus Ware', BSA 63 (1968) Megaw 1975 A. H. S. Megaw, 'An Early Thirteeth Century Aegean Glazed Ware', Studies in Memory of David Talbot Rice (Edinburgh 1975) Peschlow Urs Peschlow, 'Byzantinische Keramik aus Istanbul', Istanbuler Mitteilungen 27/28 (1978) Rudolf W. Rudolf, 'Excavations at Porto Cheli and Vicinity: The Early Byzantine Remains', Hesperia 48 (1979) Taylor J. du P. Taylor, 'Medieval Graves in Cyprus', Ars Islamica 5 (1938) 55 if 1 BP 2. 2 For a preliminary report see A. H. S. Megaw and R. E. Jones, 'Spectrographic Analyses of Byzantine and Allied Pottery', Actes XVIe CIEB (Vienna 1981) (forthcoming).

3 236 A. H. S. MEGAW AND R. E. JONES sites (Batches A-C, E-G, H(2)) and seven wares awaiting localization. The illustrated catalogue of the specimens is followed by the technical presentation consisting of an account of the methods used, the analytical data, and the interpretation of the results. This work is no more than a beginning, but it is hoped that the results obtained will encourage others who have access to relevant material and the necessary technical facilities to extend our limited coverage by similar analyses. The selection of the batches we have examined was limited to what was reasonably accessible. We are grateful both to those who have kindly made material available and to those responsible in the Greek Archaeological Service, the Benaki Museum, and the Cyprus Department of Antiquities who authorized the taking of samples for analysis. Their co-operation is individually acknowledged below. We also wish to thank Mrs. Ph. Lembessi and Miss E. Louka for considerable technical assistance in the Fitch Laboratory, Dr. A. M. Pollard of the Research Laboratory for Archaeology at Oxford for help with the multivariate analysis of the data, and Dr. Judith Herrin, for whose assistance, particularly regarding Byzantine White Ware, we are most grateful. Pre-eminent among relevant kiln sites in Greece3 are those at Corinth excavated and published by the American School of Classical Studies.4 A selection of the associated wasters, representing various early twelfth-century types of decoration, forms our Batch A. Coarse and fine wares are equally useful in establishing the characteristics of the clays used in particular areas, and our earliest batch comprises amphora fragments from kilns that supplied Halieis during its brief reoccupation at the time of the Slav migrations (Batch H(2)). Analysis of similar fragments (Batch H(I)) found with seventh-century coins on a nearby 'isle of refuge' indicates a wider distribution. An inland pottery factory near Dhiorios in Cyprus, which survived the mid-seventh-century Arab raids on coastal sites, is the source of a range of kitchen-ware (Batch G). Its products were widely distributed in the Island and it is the suspected source of an alien cooking-pot found at Carthage.5 For the most part, naturally, our investigation concentrated on glazed table-wares, chief among which must be counted the white-bodied vessels, often with polychrome underglaze decoration, commonly associated with Constantinople, though their place of manufacture has yet to be determined. Our samples come from the American excavations at Corinth (Batch D), and were selected from the contents of a sealed deposit dated by coins to the tenth century. Their identity with the White Ware that was used in the Capital was confirmed when control samples picked up in Istanbul were found to have the same clay composition (Batch J). The factory, or factories, that produced the ware can hardly have been far from the city since both the vessels and the related white-bodied wall-tiles occur there in such great quantities," not to mention the excavation in Istanbul of a waster (as yet unpublished) at Kalenderhan&. In addition there is the supporting evidence from a white clay used today by potters at Anadolu Hisari whose composition is similar to that of the White Ware; nevertheless, it remains to analyse the clay from other sources in the region of the city, such as those that were used for the Turkish Golden Horn ware,' and from any that exist in the vicinity of Izmit.8 And the story would be incomplete without analysis of the production of the Byzantine 3 The short list of recorded Byzantine kilns given by R. M. Cook (BSA 56 (1961) 64 ff.) can now be expanded. Notable are those at Lefkadia (PAE (i959) 87) and Trikala (A. Delt 20 (1965) Chron ), in addition to the three in which our Batches A, F, and H(2) originated. 4 BP 7ff" 5 J. W. Hayes in Colloques Internationaux du CNRS No. 578: Salamine de Chypre. Histoire et Archbologie (1978) fig e.g. for tiles, E. S. Ettinghausen, 'Byzantine Tiles from the Basilica in the Topkapu Sarayi', Cahiers Archiologiques 7 (1954) D. T. Rice, Byzantine Glazed Pottery (Oxford 1930) Cf. D. T. Rice, 'Byzantine Polychrome Pottery', Cahiers Archiologiques 7 (1954) 75 n. I.

4 BYZANTINE AND ALLIED POTTERY: A CHEMICAL ANALYSIS 237 master craftsmen who, with local assistants, made the tiles for the monuments of Preslav,9 probably using a clay from a neighbouring source.10 We were fortunate to be able to take samples in the Benaki Museum of the Byzantine plain, red-bodied, sgraffito slip-ware from sherds collected in Istanbul by Theodore Makridy (Batch M). Not surprisingly, their clay composition proved to be quite distinct from that of the comparable and contemporary vessels made at Corinth (Batch A). While the place of manufacture of the parent ware remains unknown, it is significant that for the mid-twelfthcentury consignment of closely related pottery shipwrecked off Pelagonnisos (Batch N) an apparently similar clay was used. The same is true, as seen in FIG. 13, of rather later imported Byzantine glazed pottery from a sealed pithos group excavated in the Athenian Agora (Batch O), in which the coarse ware had a quite different, 'Attic', composition (Batch P). It would not be surprising if such export varieties of the twelfth-century Byzantine pottery, as well as the best quality specimens excavated in Istanbul, all originated near the city, for Makridy's collection included a probable waster (M7), and until recently a red-bodied pottery was made on the Bosphorus, in the Tophan6 area. There is, unfortunately, a complication arising from the similarity of their compositions with the products of Thessaloniki; here a pottery industry of somewhat later date is well attested by tripod stilts and wasters, and the results of analysis of these and the characteristic thirteenth-fourteenth century bird bowls (Batch E) suggest that Thessaloniki should be considered as an alternative source at least of the Athens glazed ware (Batch O). Thebes, on the other hand, although its products were not part of the present investigation, can probably be discounted as the source of this glazed ware since the compositions of prehistoric pottery from Thebes do not in any way resemble those of Batch O. With the clay compositions of several characteristic Byzantine wares in our hands, it seemed useful to examine examples of the decorative thirteenth-century Protomaiolica ware, which was claimed for Byzantium when recorded for the first time at Corinth" nor has that claim been abandoned altogether, to judge by a recent presentation of the examples set in walls of the Merbaka church.12 Discoveries of the ware on Crusader sites13 and in south Italy'4 have suggested other possible sources, particularly as the technique used--overpainting on an opaque coating of tin glaze-is alien to securely identified Byzantine pottery. The specimens examined (Batch L) were selected from the Corinth finds. The results support the view that this is a south Italian ware, inspired by imports of Islamic pottery from north Africa decorated in the same technique. For its buff-coloured clay is quite unlike that of the Byzantine White Ware, as the diagram in FIG. 14 indicates, and although it was found to have kinship with Corinthian clay (both, for example, are highly calcareous), its composition coincided more closely with that of Hellenistic Black (and semi-)glaze pottery from Taranto and other Apulian sites. We also examined specimens of the very active production of lead-glaze pottery in Cyprus under the Lusignan dynasty, which reached sites beyond the Levant as far as Sicily. They come from two factories well attested by wasters: the first near Paphos, which was active in 9 K. Miatev, Die Keramik von Preslav (Sofia 1936). 10 I. Akrabova-Zhandova, 'Un atelier de c6ramique peinte au sud de l'6glise ronde a Preslav', Bulletin of the Archaeological Institute (Sofia) 20 (I955) 487-5IO; T. Totev, 'Le Monastire de Tuzlalaka-un centre de c6ramique peinte a Preslav au Ixe-xe s.', Archeologija Bilgarska (1976) A. Philadelphius, Deltion Christianikis Archaeologikis Eterias i (1924), 22 ff. 12 G. Nikolakopoulos, Ta KEpalELKa 7"r7r Havayiag rou ME'pp.rraKa r?7 NawrAolaa (Athens '979)- 13 D. Pringle, 'Some More Proto-Maiolica from 'Athlit (Pilgrim's Castle) and a Discussion of its Distribution in the Levant', BSAJ 14 (1980) D. Whitehouse, 'Proto-Maiolica', Faenza 66 (1980)

5 238 A. H. S. MEGAW AND R. E. JONES the early phase in the thirteenth-century (Batch C) and the second covering the later stages of production in the fifteenth century at Lapithos (Batch B), where a factory remained active in the present century. The clays used at the two centres were found to be quite distinct and neither of them matches that used in the early factory near Dhiorios (Batch G). The last and the latest material covered by our investigation was the output of an important pottery centre of the Ottoman period at Didymoteichon in Thrace. The material sampled comes from plain green-glazed vessels found in the neighbourhood of kilns which had been in use in the early nineteenth century (Batch F). This centre is likely to have been continuously active since the Palaeologue period. In living memory caiques used to sail up the Evros to load its products and convey them to distant ports.15 Finally, in order to test the usefulness of a project of this kind to an excavator who encounters medieval wares, we analysed a selection of single sherds of different types found at Paphos (Batch K), and the resulting clay compositions were compared with those obtained for the fourteen main batches included in the project. The results are discussed in the Appendix. CATALOGUE In order to facilitate identifications of wares examined, photographs are included of all the unpublished specimens analysed (PLATES 24-30). Where they were large enough to indicate the form of the vessels to which they belonged, either profiles are reproduced at the scale I : 3 (FIGs. I-IO) or reference is made to similar published profiles. Dimensions are given in centimetres. The batches are listed in the order in which these were analysed. Batch A. Corinth With few exceptions, the specimens analysed are wasters found in the Agora, in the area of the factories uncovered in the American School excavations and known by the designations Ag.S.C.1936, Ag.S.Stoa 1936, and Ag.S.C.1938.'" For their selection and for the photographs reproduced in PLATES 25,4 and 27,1, we are indebted to Charles Williams and his staff at Corinth. They are representative of the slip-ware vessels made there in the late eleventh and the first half of the twelfth century. They had been decorated either with slip-painting or in the sgraffito technique on slipped biscuits, but had been rejected before glazing, doubtless on account of accidents in the first firing. Also included are two more (AI3 and I4) which found their way into a bothros more recently excavated. In addition, four fragments from the factory sites which, exceptionally, did receive their glazing in a second firing (AI6, 17, 19, and 20) and an unslipped and virtually undecorated jug (AI8) were also included; it is not certain that the latter was intended to be glazed, but it was found with indisputable wasters. The batch displays a considerable range of body colour in the break, as does the ouput of the Corinthian factories as a whole, from red (e.g. A9) to the more usual buff, depending probably on the firing temperature; but the compositions were shown to vary but little with the variations in body colour, nor do the results for specimens that had received the glaze-firing differ materially from the remainder. AI. BP 720 fig. 18a (inv. C , area ofkiln Ag.S.C. 1938). Rim fragment of bowl. Form as BP fig. 7Ic. Slip-painted. Lacks glaze. Max. dimension 9'4- A2. PLATE 25,4 (inv. C.38.II9, area of kiln Ag.S.C.1938). Small rim fragment of bowl, form as BP fig. 95e. Red slip painting. Lacks glaze. Max. dimension Fragments have been seen (by Megaw) among the unpublished pottery, destined for the Haaretz Museum in A3. PLATE 25,4; BP 1034 (inv. C , area of kiln Ag.S.C. 1938). Fragment from centre of bowl, form as BP fig. 95c. Overall sgraffito. Lacks glaze. Max. dimension 10"2. A4. PLATE 25,4; BP 1033 (inv. C , E. of kiln Ag.S.C. 1938). Wall fragment of bowl, form as BPfig. 95e. Tel-Aviv, from the ruined manor Kh. Mazra'a, near the coast north of Someleria. 16 BP Ioff. fig. 8.

6 BYZANTINE AND ALLIED POTTERY: A CHEMICAL ANALYSIS 239 Overall sgraffito (bird?). Lacks glaze. Max. dimension 1-'7. A5. BP 6o8 figs. 65j, 157b (inv. C , deposit near kiln Ag.S.C. I936). Section of bowl. Red slip painting, bird. Lacks glaze. Est. diam A6. BP 1028 figs. 95e, I57a (inv. C , deposit near kiln Ag.S.C. 1936). Bowl. Overall sgraffito, bird. Lacks glaze. Diam. 270o. A7. PLATE 25,4 (inv. C.37.I856A, between kilns, Ag.S.Stoa 1936 and Ag.S.C.1938). Two joined fragments of bowl. Overall sgraffito (bird?). Lacks glaze. Max. dimension I4'2. A8. BP 723 fig. 157c (inv. C , deposit near kiln Ag.S.C.1936). Fragment from the centre of bowl. Slippainted 'oyster dots' as AI4. Lacks glaze. Diam. of foot 6-6. Ag. BP 234 fig. 157d (inv. C , deposit near kiln Ag.S.Stoa 1936). Cup with handle. Red biscuit. Undecorated. Lacks glaze. Diam Axo. BP 666 fig. 157e (inv. C , deposit near kiln Ag.S.C.1936). Bowl. Sgraffito, wavy lines, 'oyster pots' in red slip. Trial piece? Lacks glaze. Diam. 20o-I. AI. PLATE 25,4 (inv. C.37.I322A, stray near kiln Ag.S.Stoa 1936). Rim fragment of bowl, form as BP fig. 95e. Sgraffito scroll border. Lacks glaze. Max. dimension 8-3. AI2. PLATE 25,4 (inv. C.37.I322B, stray near kiln Ag.S.Stoa 1936). Base fragment of plate, form as BP fig. 95b. Trace of sgraffito. Lacks glaze. Max. dimension 8-o. Ai3. PLATE 25,4 (inv. C.75.7, West Shops S. end grid 7oD, bothros fill below pebble floor). Small fragment, near base of bowl, form as BP fig. 95d. Sgraffito medallions. Lacks glaze. Max. dimension 5'5- Ai4. PLATE 25,4 (inv. C.75.6, find-spot as AI3). Small fragment from wall of bowl. Reddish biscuit. Slip-painted 'oyster dots' as A8. Lacks glaze. Max. dimension 4'9. AI5. PLATE 25,4 (inv. C.38.II7, area of kiln Ag.S.C.1938). Rim fragment of bowl, form as BP fig. 95e. Slip-painted. Lacks glaze. Max. dimension 13.'4 AI6. (Inv. C , deposit near kiln Ag.S.C.1936.) Plain glazed bowl, form as BP fig. 65a. Pale green glaze. Not a waster. AI7. PLATE 25,4 (inv. C , area of kiln Ag.S.C.1938). Fragment from base of small bowl. Body light red to chocolate, blackened. Slip-painted; beast in medallion. Pool of discoloured glaze on the floor. Max. dimension 9-o. Waster. Ai8. PLATE 27,I (inv. C , deposit near kiln Ag.S.Stoa I936). Jug, in form akin to a type of White Ware jug (Hippodrome Ifig. 22) and to a glazed jug of Stage V (Great Palace Ipl. 27, i). No slip. Traces of painted zig-zag band. No glaze. Preserved ht AIg. PLATE 25,4 (uncatalogued, box 1285, Ag.S.C.1936). Fragment from wall of bowl, form probably as BP fig. 68i or 68j. Red slip dots within sgraffito outlines ('Measles Ware'). Yellowish glaze showing yellow-brown on unslipped exterior. Not a waster. Max. dimension 6-3. A20. PLATE 25,4 (uncatalogued, box 1285, Ag.S.C. 1936). Base fragment of cup, similar in form to BP fig. 95g or 95h. Sgraffito scrollwork medallion. Yellow glaze in interior only. Not a waster. Max. dimension Batch B. Lapithos, Cyprus The material from which this selection was made was exposed in 1958 by road-widening in the Ayios Loukas quarter of Lapithos and is now in the Department of Antiquities' Cyprus Survey store in Nicosia (inv. CS 91o). Further road-work in 1971 cut deeper into the deposit and exposed part of a kiln wall, within which pots had been abandoned in position as fired, one above the other but separated by tripod stilts (inv. CS I8o8).17 The fragments analysed belong to a class of bowls or goblets, slipped and decorated both inside and out, with rather sparse sgraffito decoration, which in the finished product was enhanced with touches of green and yellow-brown (the purple-brown of BI6 is exceptional) in the colourless overall glaze, a class assigned to the fifteenth century.18s Except where otherwise stated, the exterior rim decoration is illustrated. The slip was found by analysis to have a high lead content. Most of our specimens had been rejected before glazing. Six others, which had also received their glaze and sparse colouring, are from the same find and probably represent breakages during or after the second firing (BI4-i6, 18, 20, and 21); two more were evidently rejected on account of glaze discoloration through overfiring (BI7 and 19). One of the glazed pieces (Bi5) belongs to a class of small bowl on a low foot with minimal sgraffito decoration touched up with green only, in a clear overall glaze; it seems to have remained popular throughout the Island in the sixteenth century.19 Another fragment was a casualty in the second 17 We are indebted to Dr. V. Karageorghis for permission to analyse this selection, and to Mr. M. Loulloupis for relevant information. 18 CMGP 9 Group VII pl. viii CMGP Io Group VIIIB.

7 240 A. H. S. MEGAW AND R. E. JONES firing (B3), in which the rim of the bowl next above it in the kiln fused on to its glazed exterior. Bx. PLATE 28,3. Tripod firing stilt. Max. dimension 6-8. B2. PLATE 28,3. Tripod firing stilt. Max. dimension 7-1. B3. PLATE 28,3. Glazed fragment from side and vertical rim of goblet as CMGP pl. ii 33. Overfired waster with rim fragment of another (est. diam. 1-5) adhering. Est. diam. 13-o, max. dimension B4. PLATE 28,3. Fragment from vertical rim of similar goblet. Lacks external glaze. Est. diam. of rim 14-o, max. dimension 9'5. B5. PLATE 28, 3. Another. Form as CMGP pl. ii 32. Lacks glaze. Est. diam. of rim 14-o, max. dimension 8-2. B6. PLATE 28,3 and FIG. I. Fragment of bowl with wide, everted rim. Lacks glaze. Est. diam. 20-0, max. dimension io-6. B7. PLATE 28,3 (interior) and FIG. i. Fragment of bowl with narrow, everted rim. Lacks glaze. Est. diam. I8-o, max. dimension 5'3- B8. PLATE 28,3 (interior) and FIG. I. Fragment of simple rim of small bowl; lacks glaze. Est. diam. 15, max. dimension 3-8. B9. PLATE 28,3 (interior). Fragment from vertical rim of goblet as B4. Lacks glaze. Est. diam. i5-0, max. dimension 6.5. Bxo. PLATE 28,4 and FIG. I. Fragment from a slightly inset rim of goblet. Lacks glaze. Est. diam. of rim 140o, max. dimension 70o. Bi i. PLATE 28,4 and FIG. I. Fragment from inset rim of goblet. Lacks glaze. Est. diam. of rim I I-o, max. dimension 45-. B6 B7 B8 810 B11 FIG. I. Batch B, Lapithos (1:3) BI2. PLATE 28,4. Another similar. Lacks interior glaze; exterior glaze overlaps break at the top. Est. diam. 15-0, max. dimension 4'7- Bx3. PLATE 28,4. Foot and lower part of side of goblet similar to CMGP pl. ii 28. Interior, sgraffito gridiron medallion. Exterior slip to foot. Lacks glaze. Diam. of foot 5-o, ht BI4. PLATE 28,4. Foot fragment of goblet with exterior ridge as CMGP pl. ii 63. Interior, thin slip and scant sgraffito. Colourless exterior glaze on slip down to ridge on foot; spot of yellow in interior glaze. Est. diam. of foot 9.0. Ht. of foot 5-0. Bx5. PLATE 28,4. Centre fragment of small bowl, form as CMGP pl. ii 46. At centre small sgraffito medallion.20 Colourless glaze overall, touches of green in interior. Diam. of foot 7-5, max. dimension 6.9. Bx6. PLATE 28,4 (interior). Fragment from thick wall (max. 1.8) near foot of large goblet. Sgraffito medallion with touches of green and purple-brown in colourless overall glaze. Tripod scar. Exterior undecorated. Max. dimension 5-6. BI7. PLATE 28,4 (interior). Fragment of vertical rim of goblet as B4. Overall greenish glaze, discoloured by overfiring. Est. diam. 13, max. dimension 6-5- Bi8. PLATE 28,4 (interior). Section of bowl of goblet as CMGP pl. ii 44. Touches of yellow and green inside and out in overall glaze. Tripod scar. Est. diam. of rim I i-o, max. dimension Big. PLATE 28,4 (interior). Wall and rim fragment of goblet similar to B 8 but with more flaring rim. Exterior undecorated, but yellow spots at the rim. Yellow-brown touches on interior sgraffito in glaze discoloured by overfiring. Est. diam. of rim I9.o, max dimension 8-5. B2o. PLATE 28,4. Fragment from slightly flaring rim of goblet. Green touches in yellowish overall glaze. Est. diam. 12o0, max. dimension 5o. B2I. PLATE 28,4. Fragment from rim of goblet as CMGP pl. ii 27. Strong green and brown touches inside and out in colourless glaze. Est. diam. 15-0, max. dimension 4-8. Batch C. Lemba, Cyprus The twenty-three specimens examined come from the head of a deep gulley between the village of Lemba, 4 km north of Paphos, and the medieval site Ayia Marina.21 The kiln waste collected there by Hadjisavvas, which attests a factory at no great distance, is in the Cyprus Survey store of the Cyprus Department of Antiquities in Nicosia (inv. CS 2289). The distorted 20 Similar to Taylor fig. I I. 21 S. Hadjisavvas, 'An Archaeological Survey of Paphos', RDAC Subsequently, wasters have been found at Ayia Marina itself (information from Franz Maier).

8 BYZANTINE AND ALLIED POTTERY: A CHEMICAL ANALYSIS 241 bowl centre C6, on to which the tripod stilt C5 is fused, is of particular interest as attaching to that factory the form of ring-foot with out-turned lip (FIG. 2), a detail it shares with C7 and C8. This is a feature of Cypriot bowls of thirteenth-century date and of equivalents from Crusader sites on the mainland22 and, in particular, of a group in the Cyprus Museum from the Paphos District.23 Among our fragments are several with other characteristics of that group: the externally concave rim of CI I, an unusually thin slip (C9 and CI8), frequent use of the gouge with or without the fine point (C6, 7, 9, 19-22), and a tendency to overfiring which gives the body a blackened core (C8, I o, and 13). Indeed, setting aside the recognizable stranger CI6 and another alien piece C23, which was not analysed, the results warrant the connection of all the Lemba batch with the local factory, and this although few of them display the obvious defects of wasters. C7 C6 C9 FIG. 2. Batch C, Lemba (1:3) C2o and C2I were included with some reluctance because, unlike the characteristic pieces, which are dark purple when not black in the core, these two are a dull light red; but both were found to have the same composition as the remainder, as does the tripod C3, which is even lighter in colour. Whether all products of the 'Paphos' class were made at the factory that dumped its waste at Lemba, and whether all our Lemba batch dates from the thirteenth century, remain open questions. Ci. PLATE 28,1. Fragmentary tripod stilt. Brick-red body. Max. dimension 7-0. C2. PLATE 28, I. Another. Light red body, darkened to purple on top, to dull olive green on the underside. Max. dimension 5'4. C3. PLATE 28, I. Another. Buff body. Max. dimension 5"1. C4. PLATE 28,1. Another. Body blackened thoughout and partially vitrified. Max. dimension 6-2. C5. PLATE 28, I. Another. Dull reddish brown. Max. dimension Fused onto C C6. PLATE 28, I and FIG. 2. Centre of bowl, ring-foot with outturned lip. Blackened and distorted by overfiring. Thin slip. Gouged hatching in central medallion (obscured by stilt fused in place). Max. dimension 8.2. C7. PLATE 28,2 and FIG. 2. Part of bowl centre, ring-foot with out-turned lip. Thin slip. Outline of central medallion and lines within incised with the narrow gouge. Colourless glaze, no added colour. Max. dimension 6-8. C8. PLATE 28,2. Small fragment of ring-foot as C6 (est. diam. Io) from centre of bowl. Black core in break. Scrap of orange glaze on slip preserved in interior. Max. dimension 5-2. C9. PLATE 28,2 and FIG. 2. Small fragment from floor of bowl within ring-foot, nipple on the unglazed underside. Grey body with red core. Thin slip. Gouged concentric circles. Orange glaze. Max. dimension 2.9. CIO. PLATE 28,2. Fragment from centre of bowl as C7, ring-foot missing. Body purple-red, black core. Thin slip. Radial sgraffito lines from centre, scribbles in alternate segments. Dabs of green in dirty yellow glaze showing chocolate on unslipped exterior. Max. dimension 5'5- CII. PLATE 28,2 (interior). Externally concave rim fragment of carinated bowl, as CMGP fig. 2e. Blackened body. Band of slip on exterior below rim (also interior). Yellowish glaze overall. Max. dimension 4-6. CI2. PLATE 28,2. Small fragment from steep wall of small bowl, as CMGP fig. 2a, to junction with missing horizontal rim. Dull purple-brown body. Colourless glaze shows yellow-brown on unslipped exterior. Max. dimension 2.7. CI3. PLATE 28,2. Small fragment from horizontal rim of similar bowl. Purple-red body, red in core. Touch of green on sparse sgraffito. Exterior not glazed. Max. dimension 3.0. CI4. PLATE 28,2. Small rim fragment of simple profile (as Dikigoropoulos and Megaw fig. 8g) from small bowl. Red body. Touch of green over band of gouged lines in yellowish glaze showing light chocolate on exterior. Max. dimension CMGP 5 fig Dikigoropoulos and Megaw

9 242 A. H. S. MEGAW AND R. E. JONES C15. PLATE 28,2. Undecorated fragment from wall ofjug. No white slip on blackened dark brown body. Erratic glaze on exterior only, showing dark grey-green. Max. dimension 33-. CI6. PLATE 28,2. Fragment from wall of bowl. Grey-brown body. Slip-painted concentric coils. Colourless glaze on interior only, showing light chocolate on body. Max. dimension 6-8. The composition is atypical of the batch; see p. 261 and FIG. 12. CI7. PLATE 28,2. Fragment from wall of bowl. Dark brown body, reddish exterior. Part of slip-painted spiral, probably as Taylor fig. 39. Colourless glaze showing chocolate on body. No slip or glaze on exterior. Max. dimension 4"1. CI8. PLATE 28,2. Small fragment from wall of bowl. Dark brown body. Slip-painted ovals. yellowish glaze, showing dark olive green over body. No slip or glaze on exterior. Max. dimension 4-4- CI9. PLATE 28,2. Small fragment from wall of bowl. Blackened body. Radial lines gouged through interior slip, which shows dull cream colour under matt glaze (bubbled in incisions, dark olive green on unslipped exterior). Waster. Max. dimension 49-. C20. PLATE 28,I. Fragment from wall of bowl. Light red body. Radial sgraffito decoration heavily stained with green and brown in pale yellow-green glaze, which extends to the upper part of the unslipped exterior. Max. dimension 5"5- C21. PLATE 28,I. Small fragment from wall of goblet. Light red body. Sgraffito lines through slip under bright green glaze inside and out. Max. dimension 3-5- C22. PLATE 28, I. Fragment from centre of small bowl. Thin flaring ring-foot (damaged). Light purple-red body. Gouged concentric circles or spiral, under pale yellow glaze. No slip or glaze on exterior. Evidently imitates Zeuxippus Ware class IA (Megaw ). Max. dimension 5-3- C23. PLATE 28, I. Fragment from wall of bowl of greenpainted ware datable to late 15th and 16th cents. (Taylor 84). Not analysed. Batch D. Byzantine White Ware from Corinth With the exception of D2o, which was a surface find, the entire batch, kindly selected and made available by Charles Williams, comes from an unpublished pit-group designated , which is datable to the tenth-century by coins. This ware, which is readily distinguished by its fabric, was much used in Corinth before the local factories became active, though it continued to be imported on a smaller scale in the twelfth century.24 In order to check that the ware once current in Byzantium and that imported at Corinth are the same, three specimens in the British School collection which were picked up in Istanbul were also analysed (Batch J below); they were found to have the same clay composition as Batch D. In form also, the vessels represented by our specimens match those found in Istanbul, where the corpus of published examples has recently been enlarged by the series found in the excavations to the south of St. Eirene.25 In our selection, plain dark green glazes predominate, though petal ware (D8), impressed ware (DII and DI8), incised (D4 and DI3), and polychrome (D2) are represented. But whatever the technique and whatever the form of vessel (cup, plate, chafing dish, or jug), the clay composition was consistent. DI. PLATE 25,2. Fragment from centre of bowl. Green glaze overall, except within flaring ring-foot, as BP fig. 29e. Est. diam. of foot 6-5. Max. dimension 4-2. D2. PLATE 25,1. Rim fragment from plate, as BP fig. 47a. Polychrome decoration in green and yellow within black outlines. Canary glaze on exterior. Est. diam. 22-o. Max. dimension 5-3- D3- PLATE 25,I. Rim fragment of plate with blob of red clay, as Peschlow no. 33. On underside, zigzag formed by excising triangles of the body through a brown-red band. Pale green glaze overall. Est. diam. 25o0. Max. dimension 40o. D4. PLATE 25,1. Fragment from pinched mouth of juglet. Vertical incisions through a wash of orange-red on exterior. Yellow glaze overall. Ht. of fragment 4-2. D5. PLATE 25,2 and FIG. 3. Two joined fragments from flatbase jug. Interior, yellow glaze with specks of brown. 24 BP Peschlow 380 ff. Exterior, splashes of the same and on underside. Est. diam. of base 8-o. Max. dimension i r o. D6. PLATE 25,2 and FIG. 3. Base of chafing dish, as Peschlow no. 85. Flaring foot. Green glaze on exterior only, extending erratically to underside of foot. Diam. of foot 9-7. Max. dimension 9'7- D7. PLATE 25, 2 and FIG. 3. Two joined fragments from base and wall of flat-base cup(?). Mottled green glaze inside and out. Est. diam. of base 5-2. Max. dimension 8-1. D8. PLATE 25,I and FIG. 3. Fragment from lower wall of globular 'petal ware' jug(?). Alternate rows of white and red 'petals'. Pale yellow glaze overall. Est. diam. at top of fragment Max. dimension 4"1. Dg. PLATE 25, 1. Handle and rim fragment of a medium cup similar to Peschlow no. 51. Yellow glaze overall. Est. diam. 16-o. Max. dimension 5.2.

10 BYZANTINE AND ALLIED POTTERY: A CHEMICAL ANALYSIS 243 D8 DIo. PLATE 25,1. Two joined rim fragments of a large plate as Peschlow no. 39. Mottled green glaze overall. Est. diam Max. dimension 7-4. DII. PLATE 25,2. Centre of cup as Peschlow nos. 49 and 51 with stubs of two handles extending to ring-foot, flaring at the bottom. Lightly impressed rosette at the centre. Yellow glaze overall except within ring-foot (est. diam. 6-5). Max. dimension 74-. DI2. PLATE 25,2. Centre of small bowl or cup with splayed ring-foot as Peschlow no. 55. Yellow glaze on interior only, except for splash within ring-foot (est. diam. 5-o). Max. dimension 5-o. Dx3. PLATE 25,1. Rim fragment, as Peschlow no. 69, probably of fruit-stand. Wavy line incised through brown-red band on rim. On the underside, another brown-red band, undecorated. Yellow glaze overall. Est. diam Max. dimension 3'4. Dx4. PLATE 25,1. Rim fragment of chafing dish, as Peschlow no. 81, but larger, with stump of handle. Pale greenishyellow glaze overall. Est. diam. 20oo. Max. dimension 3-6. D5 D6 D7 FIG. 3. Batch D, Byzantine White Ware (1: 3) DI5. PLATE 25, I1. Rim fragment of cup, profile as BP fig. 4od. Mottled green and yellow glaze overall. Est. diam. 15-o. Max. dimension 5.2. DI6. PLATE 25,1. Handle and rim fragment of cup, as Peschlow no. 51. Mottled green glaze overall. Est. diam. 16-o. Max. dimension 6-7. Dx7. PLATE 25,1. Rim fragment of chafing dish, as Peschlow no. 81, but larger, with stump of handle. Green glaze overall. Est. diam. I8-o. Max. dimension 3-7. Dx8. PLATE 25,2. Fragment from centre and flaring ring-foot of cup, probably as Peschlow no. 52. Trace of impressed design. Dark green glaze overall, except within foot. Max. dimension 4-2. Dig. PLATE 25,1. Fragment from wall of carinated cup, probably as Peschlow no. 49. Green glaze overall with specks of red. Est. diam. at carination 14-o. Max. dimension 5'4. D2o. PLATE 25,1. Rim fragment of bowl with narrow ledge for lid, as Peschlow no. 40, but deeper. Dark yellow-green glaze overall with specks of red. Est. diam Max. dimension 5.6. Batch E. Thessaloniki The early assumption that Thessaloniki was a centre of production of Byzantine slip-ware26 has been fully justified by recent discoveries.27 Apart from isolated finds of tripod stilts and wasters, kiln waste was found in some quantity in 1973 on the Pharangi building site near the Rotunda. That material not being accessible, our Batch E was selected from the old pottery collection of the Byzantine Ephorate at the Rotunda, through the kindness of Chrysoula Tsioumis and with the assistance of Dimitra Bakirtzis. Nine of our specimens are wasters or firing stilts. Though not from kiln sites, they have provenances as follows: Serapeion (inside west wall), EI 3; Agora (near St. Demetrius), E4, E9-12; excavation for sewers, E7, E8; Vardari (outside west gate), EI4. Two more wasters (E5, E6) and five tripods (EI5-19) from the city, but without provenance, were also included. In addition, three centres from finished bird bowls of hemispherical form (EI-3) were examined because they match the waster E4; also two finished bowl centres with orange glaze (E20-I) because, like the bird bowls, they are characteristic of pottery finds in Thessaloniki. That they were indeed made there is evident from the consistent clay composition throughout the batch. Ex. PLATE 29,1 (inv. BK I 187). Centre of bird bowl. Incised technique with body of bird (excluding wing) darkened by removal of slip. Touches of green in pale greenish 26 Rice, op. cit. (n. 7) 7 and A. Xyngopoulos, 'Byzantine pottery from Olynthus', Olynthus 5 (1933) D. Pananikola-Bakirtzis, 'Epyaar774pa EoOaAwlp. glaze, which does not extend to unslipped exterior. Not a waster. Diam. of low ring-foot 59.- Max. dimension 9 1. Kepap.ELKir 9c7T OeEcuaAov'K7: HpO7rES TrTqpop'ES' in Milanges Stylianos Pillkanides (forthcoming) and Ch. Bakirtzis and D. Bakirtzis, Byzantino-Bulgarica 7 (1981) 434.

11 244 A. H. S. MEGAW AND R. E. JONES E2. PLATE 29,I (inv. BK 862/129, from Vardari). Another. Orange glaze, except on unslipped exterior. Not a waster. Diam. of splayed ring-foot 4.9. Max. dimension 5-6. E3. PLATE 29,1 (inv. BK 273). Another. Interior, orange glaze. Vertical gouged incisions through exterior slip under pale yellow glaze with touches of green. Not a waster. Max. dimension 6-8. E4. PLATE 29,1 (inv. BK 909/71 from Agora). Another. Tall and narrow ring-foot bent outwards to form wide flange. Exterior incision as E3. Lacks glaze. Diam. of foot 5-0. Max. dimension 9-o. E5. PLATE 29, I (inv. BK 2152). Bowl centre on low ring-foot. Undecorated. Slip inside and out. Lacks glaze. Diam. of foot 5'9. Max. dimension 7-o. E6. PLATE 29, I (inv. BK 2232). Overfired rim fragment of hemispherical bowl. Slip on interior only. Band on lip now shiny brown. Elsewhere overfiring has left a colourless matt surface. Max. dimension 7-5. E7. PLATE 29,I (inv. BK 871/82 from sewer excavations). Tripod stilt, fused on to E8. Max. dimension 7-0. E8. PLATE 29,I (inv. as E7). Centre of plate. Low ring-foot. Slip (if any) indistinguishable as a result of overfiring. Glaze now greenish brown and cracked. Diam. of foot Max. dimension E9. PLATE 29,I (from Agora). Tripod stilt of coarse fabric with blunt points, blackened. Max. dimension 6-2. Eio. PLATE 29,I (from Agora). Another, blackened. Max. dimension 5"4- Eii. PLATE 29,2 (from Agora). Another. Dark buff. Max. dimension 5-4- E12. PLATE 29,2 (from Agora). Another. Grey-brown. Max. dimension 6-o. EI3. PLATE 29,2 (from Serapeion). Tripod stilt of finer fabric with sharp points. Light red. Max. dimension 6-3. Ex4. PLATE 29,2 (from Vardari). Fragment of another. Light red. Max. dimension 6-I. Ei5. PLATE 29,2. Tripod of coarse fabric, blackened. Max. dimension 6-6. EI6. PLATE 29,2. Fragment of another. Dark buff. Max. dimension 7-2. EI7. PLATE 29,2. Tripod stilt of finer fabric. Light red. Max. dimension 6-6. EI8. PLATE 29,2. Another. Max. dimension 6-6. Ei9. PLATE 29,2. Another, fragmentary. Max. dimension 7-0. E2o. PLATE 29, 2. Centre fragment of small bowl. Low splayed ring-foot. Concentric circles incised with gouge through slip. Orange glaze not extending to unslipped exterior. Diam. of foot 6-o. Max. dimension 7-0. E21. PLATE 29,2. Another. Similar decoration and glaze. Diam. of foot 4-6. Max. dimension 8-0. Batch F. Didymoteichon In this once important centre of pottery production (200 factories are mentioned by the seventeenth-century traveller Evliya 1elebi) two kilns and much of their waste datable to the early nineteenth century were recently exposed in a building site just outside the walls, by the main gate.28 We were able to analyse a representative selection thanks to the kindness of Charalambos Bakirtzis. The body of the vessels is normally buff to pink in colour and they are sometimes decorated with impressed circles on the floor of bowls or small buttons of clay with stamped patterns applied to jugs. They are normally covered with a white slip and finished with green glaze, whether bowls, jugs, or lamps. The tripod stilts are mould-made with simple devices on the underside. Very similar pottery was found in the excavation of abandoned cave-basements within the walls. FI. PLATE 29,3. Fragmentary mould-made tripod stilt with device on the underside, as Bakirtzis figs and pl. v 5. Max. dimension 9-6. F2. PLATE 29,3. Another. Max. dimension 7-8. F3. PLATE 29,3. Another. Max. dimension 5-4. F4. PLATE 29,3. Rim fragment of bowl, as Bakirtzis pl. vi 2. Dark green glaze on slip (interior only), overfired brown on the rim. Est. diam Max. dimension 9-2. F5. PLATE 29,3 (interior). Fragment from wall of slightly carinated bowl, as Bakirtzis pl. v i. Slip on interior only. Lacks glaze. Max. dimension 4"9. F6. PLATE 29,3 (exterior). Rim fragment from simple bowl, as Bakirtzis pl. iii 3. Interior slip extends to upper part of exterior. Lacks glaze. Est. diam. 18-o. Max. dimension 5-9- F7. PLATE 29,3 (exterior). Fragment from wall of similar bowl. Slip extends lower on exterior. Lacks glaze. Max. dimension 6-9. F8. PLATE 29,3 (exterior). Rim fragment from another. Slip overall. Lacks glaze. Est. diam Max. dimension 6.6. F9. PLATE 29,3. Fragment from lower part of lamp, as Bakirtzis pl. iv 9 and 12. Green glaze direct on body over part only of the interior. Splash of slip on exterior. Diam. of flat base 6-o. Max. dimension 8-2. Fio. PLATE 29,3 (interior). Rim fragment from similar lamp. Erratic pink slip overall. Dull green glaze on interior and lip only. Est. diam. 9-o. Max. dimension 7'4. FIi. PLATE 29,3. Tubular spout from side-spouted jug, as Bakirtzis pl. vi 3. Dark green glaze evidently direct on body. Max. dimension Bakirtzis 148 ff.

12 BYZANTINE AND ALLIED POTTERY: A CHEMICAL ANALYSIS 245 F12. PLATE 29,4 (interior). Notched rim fragment of bowl as F4. Slip and apple-green glaze extending erratically over lip. Est. diam. I7.5. Max. dimension 6-9. Fi3. PLATE 29,4. Rim fragment of bowl with everted rim, as Bakirtzis pl. iv 6. Thin slip on interior and lip. Pale yellow glaze with spots of green. Est. diam Max. dimension I. Io. Fx4. PLATE 29,4. Narrow neck ofjug with pinched mouth, as Bakirtzis pl. vi 4. Green glaze, apparently without slip, extending to interior of mouth. Ht. from junction with shoulder 6-6. Max. dimension 9-0. Fx5. PLATE 29,4 (interior). Fragment from wall of bowl near base, as Bakirtzis pl. vi Io. Green glaze on interior, no slip apparent. Max. dimension 5'4. Fx6. PLATE 29,4. Fragment from upper wall of bowl. Applegreen glaze on interior only, no slip apparent. Max. dimension 5'2. Fi7. PLATE 29, 4 (exterior). Fragment from simple rim of hemispherical bowl. Dark green glaze on interior and lip, on slip extending lower on exterior. Est. diam. 2 1o. Max. dimension 6.6. FI8. PLATE 29,4. Fragment from centre of plate or shallow bowl with relief circle (diameter 5-o) impressed in the floor, as Bakirtzis fig. 26 and pl. iv 6. Tripod scar. Green glaze on interior slip. Diam. of low ring-foot 6-4. Max. dimension 7-0. Fxg. PLATE 29,4. Another. Impressed circle diameter 2-9. Tripod scar. Mottled green glaze not completely covering interior slip. Diam. of low ring-foot 6-7. Max. dimension 8-5. F2o. PLATE 29,4. Another. Impressed circle diameter 2-9. Dull green glaze on slipped interior. Diam. of low ring-foot 6-5- Max. dimension 6-o. F21. PLATE 29,4. Another without impressed circle. Dull green glaze on slipped interior. Diam. oflow ring-foot (of profile similar to Bakirtzis pl. vi 12) 6-5. Max. dimension 6-o. Batch G. Dhiorios, Cyprus This batch comes from a pottery factory at the site Mersineri, which was very active in the seventh century in the production of thinly-potted casseroles and cooking-pots. The abundant material recovered from Hector Catling's excavation of the factory is now in the Cyprus Survey store of the Department of Antiquities in Nicosia. We are indebted to Vassos Karageorghis, Director of the Department, for permission to take a selection on loan for analysis. Our pieces come from the latest of the kilns, the only one of rectangular form,29 the use of which may have extended into the eighth-century. The ware is red, but, as is to be expected among throw-outs, many of our specimens were blackened by overfiring, where not actually distorted (e.g. GII). GI. PLATE 24,4. Fragment of frying-pan with horizontal rim, as Catling fig. 23 P47 1. Thin blackened body. No wheelridging. Est. diam. 190o. Max. dimension io-. G2. PLATE 24,4. Another. Thin red body, external wheelridging. Est. diam. 22-o. Max. dimension 6-5. G3. PLATE 24,4. Fragment from rim and wall of carinated frying-pan. Horizontal rim similar to Catling fig. 23 P124. Thin red body. Wheel-ridging inside and out at the top only. Est. diam. 22-o. Max. dimension 95-. G4. PLATE 24,4. Fragment, probably from lid with part of knob handle as Catling fig. 27 PI34. Thick red body, plain. Max. dimension 7-5. G5. PLATE 24,4. Thick rim fragment from cooking-pot with vertical handle, as Catling fig. 33 PI I I. External wheelridging. Red body. Est. diam. 19-o. Max. dimension 95.- G6. PLATE 24,4. Rim fragment of casserole similar to Catling fig. 27 P195, with finger-pressed lug handle. Red body. Est. diam. 240o. Max. dimension 6.9. G7. PLATE 24,4. Another, rim profile similar to Catling fig. 36 PI99. Exterior wheel-ridging. Est. diam Max. dimension 8.2. G8. PLATE 24,4. Vertical handle of cooking-pot, as Catling fig. 32 P127. Max. dimension Catling 52 and fig. 31. G9. PLATE 24,4. Another. Blackened and distorted. Max. dimension 6-7. Gxo. PLATE 24,4. Casserole fragment with horizontal handle and section of rim, as Catling fig. 32 Pi28. Exterior wheel-ridging. Est. diam. 2I1o. Max. dimension i1.9. Gii. PLATE 24,5. Cooking-pot rim fragment, profile similar to Catling fig. 27 P93. Exterior wheel-ridging. Blackened and distorted. Max. dimension 6-8. G12. PLATE 24,5. Another. Lipped profile as Catling fig. 34 P495. Exterior wheel-ridging on thin, blackened body. Est. diam. of rim 23-o. Max. dimension o103. G13. PLATE 24,5. Another. Profile as Catling fig. 30 PI53. Exterior wheel-ridging. Thin, reddish grey body. Est. diam. of rim 15o0. Max. dimension 9"7. Gx4. PLATE 24,5. Fragment of casserole rim with lip-ledge as Catling fig. 32 P205. Close external wheel-ridging. Blackened. Est. diam. 21o. Max. dimension 6.5- GI5. PLATE 24,5. Another, as Catling fig. 32 P129. Wide exterior wheel-ridging. Red/black body. Est. diam Max. dimension 5-4- GI6. PLATE 24,5. Fragment of cooking-pot rim as Catling fig. 34 P419. Plain exterior. Blackened. Est. diam. 18-o. Max. dimension 5-6.

13 246 A. H. S. MEGAW AND R. E. JONES Gx7. PLATE 24,5. Fragment of casserole rim with ledge for lid as Catling fig. 35 PI36. Exterior wheel-ridging. Blackened. Est. diam. I8-o. Max. dimension 5-8. GI8. PLATE 24,5 (interior). Another as GI7, but with wider lid-ledge. Exterior wheel-ridging. Red body, blackened exterior. Est. diam. 24. Max. dimension 7-5- GIg. PLATE 24,5. Fragment of cooking-pot rim with lip as Catling fig. 20 PI73. Exterior wheel-ridging, very thin red body, blackened exterior. Est. diam. 24-o. Max. dimension 7-7- G2o. PLATE 24,5. Another, beaded form with stub of handle as Catling fig. 7 P97. Exterior wheel-ridging. Thin red body. Est. diam. I9-o. Max. dimension 7-7. G21. PLATE 24,5. Another, beaded form as Catling fig. 7 P89. Exterior wheel-ridging. Thick body, blackened. Est. diam Max. dimension 9"2. Batch H(I). Chinitsa, Argolid (Offshore Island) Chinitsa is opposite Porto Cheli (ancient Halieis), about a kilometre offshore.30 The three unglazed coarse ware sherds in this batch were analysed thanks to the initiative of Timothy Gregory. They are representative of the pottery found with early seventh-century coins at the centre of the island, on the landward side, during a survey based on the American School excavation at Halieis,31 a site ranking as one of the 'Isles of refuge' studied by Sinclair Hood.32 The clay composition of this batch indicates that the vessels used on Chinitsa had most likely originated in the kilns on the mainland from which our Batch H(2) was obtained. The three Chinitsa specimens come from the same type of globular amphora that is common to the site of Halieis and the material from the kilns.33 H(I)I. PLATE 24,I. Fragment from lower neck and shoulder of small amphora, as Rudolf fig. 3. Buff-pink body. Wide wheel-ridging on neck only. Est. diam. of neck (at top of fragment) 6-o. Max. dimension 13.o. H(I)2. PLATE 24, I1. Neck end of handle of similar amphora. Pink body. Max. dimension 7-0. H(I)3. PLATE 24, I1. Another. Pink body, buff exterior. Max. dimension 8-6. Batch H(2). Kounoupi Kilns, Porto Cheli The locality Kounoupi is on the coast of the Argolid to the east of Porto Cheli (ancient Halieis). Remains of two kilns were discovered there during the Argolid Exploration Project (site B-19) based on the American School excavations on the ancient site.34 Our batch of specimens from the massive deposit of obvious wasters and other unglazed coarse ware sherds round about the kilns was included thanks to the co-operation of Timothy Gregory. Their date is fixed by the finding of identical pottery at Halieis, where a brief recovery is attested in the late sixth and early seventh centuries,35 and by the finding of similar fragments with early seventh-century coins on Chinitsa Island (Batch H(I)). With the exception of H(2)27-9, all our specimens come from globular amphorae, as Rudolf fig. 3. The three exceptions are not matched by the published pottery from the settlement. H(2)4. PLATE 24,2. Distorted and partly vitrified fragment from shoulder of amphora with stub of handle. Buff body. Max. dimension i i.3. H(2)6. PLATE 24,2. Fragment from base of neck of amphora, partly vitrified after fragmentation. Max. dimension 6-8. H(2)8. PLATE 24,2. Fragment from lower neck and shoulder of amphora. Buff to pink body. Max. dimension 9-o. H(2)9. PLATE 24,2. Fragment from shoulder of amphora. Deep grooving above, plain below. Pink clay. Max. dimension 9-9. H(2)IO. PLATE 24,2. Another. Deep, close grooving throughout. Max. dimension 7-3. H(2)II. PLATE 24,2. Another. Deep grooving throughout. Pink body, thick Max. dimension 5-3. (I.I). H(2)I2. PLATE 24,3. Another. Alternate deep and shallow grooving. Pink body. Max. dimension 5'3- H(2)13. PLATE 24,3. Another. Grooving as the preceding. Pink body. Max. dimension 4'7- H(2)14. PLATE 24,3. Another. Between plain bands, deep grooving smoothed over. Dark buff body. Max. dimension 7-0. H(2)I5. PLATE 24,3. Fragment from belly of amphora. Above, deep grooving fading out downwards and smoothed over; below, at maximum diameter, plain. Body pink to buff (exterior). Max. dimension I Rudolf 304 n Rudolf 304 n M. S. F. Hood, BSA 65 (1970) Rudolf fig Rudolf 304 n Rudolf 303ff.

14 BYZANTINE AND ALLIED POTTERY: A CHEMICAL ANALYSIS 247 H(2)17. PLATE 24,3. Another. Deep grooving above max. diameter, plain below. Coarse pink body, pitted. Max. dimension o104. H(2)I8. PLATE 24,3. Fragment from shoulder of large amphora. Deep and rounded grooving. Pink body. Max. dimension H(2)21. PLATE 24,3. Fragment from splayed rim of amphora, as Rudolf fig. 3. Est. diam. of rim 9-0. Max. dimension 6.3. H(2)23. PLATE 24,3. Fragment from shoulder of amphora. Between plain bands, wavy and straight grooving, as Rudolf fig. 5,I I. Max. dimension 5-0. H(2)27. PLATE 24,3. Fragment from bottom of basin(?) with small low foot, as FIG. 4, 29, but thinner. Coarse buff body with grits. Overfired. Est. diam. of slightly concave foot 5'7. Max. dimension 5-6. H(2)28. PLATE 24,3 and FIG. 4. Bottom of similar vessel. Yellowish buff body. Diam. of low flat foot 5.6. Max. dimension io-8. H(2)29. PLATE 24,3 and FIG. 4. Bottom of another, but thicker. Pink buff body. Diam. of slightly concave foot 5'4- Max. dimension 8-3. H(2)28 H(2)29 FIG. 4. Batch H(2), Kounoupi (1: 3) J3 J1 FIG. 5. Batch J, Istanbul, White Ware (i: 3) Batch J. Byzantine White Ware from Istanbul Three fragments from the British School's Sherd Collection, which were picked up in Istanbul, were analysed as a control on Batch D, the Byzantine White Ware from Corinth. The results were sufficiently close to establish that what Corinth imported was the ware used in the capital. J2 JI. PLATE 30, I and FIG. 5. Fragment from splayed ring-foot of plate, as Peschlow no. 3. Trace of impressed design in floor. Yellowish glaze direct on white body (floor only). Est. diam. of foot Max. dimension 6.2. J2. PLATE 30, I and FIG. 5. Fragment from wall of cup at junction with ring-foot. Yellowish glaze direct on white body (exterior, dull yellow). Est. diam. at junction with foot 50o. Max. dimension 2-5- J3. PLATE 30, I and FIG. 5. Fragment from thin, almost horizontal rim of polychrome plate, similar to Peschlow no. o01. Running spiral border in thin brown line between green bands direct on white body, under yellow glaze. Edge and underside unglazed. Est. diam Max. dimension 3o.0 Batch K. Paphos, Cyprus As a test exercise, these single specimens of pottery current in medieval Paphos have been analysed in order to check whether, by comparison with the clay characteristics so far established for known centres of production, it is possible to determine the place of manufacture of single examples of particular wares wherever found. The results are discussed in the Appendix. Saranda Kolones is the site of a castle destroyed by earthquake in 1222, the excavation of which is approaching completion.36 Ayia Solomoni is an underground chapel 36 A. H. S. Megaw, 'Supplementary Excavations on a Castle Site at Paphos, Cyprus, ', DOP 26 (1972) and id., 'Saranda Kolones, 1981', RDAC

15 248 A. H. S. MEGAW AND R. E. JONES complex, which was partially filled in the sixteenth century evidently with material from a medieval midden. The pottery from both sites is in the Paphos Archaeological Museum. We are indebted to the Director of the Department of Antiquities for permission to take this batch of fragments on loan for analysis. K8 K10 K9 K11 K12 FIG. 6. Batch K, Paphos, Miscellaneous (1 :3) KI. PLATE 30,2 (Saranda Kolones, inv. FC 633). Fragment from wall of plate of Byzantine Plain Sgraffito Ware. Light red body. Slip on interior only. Concentric cufesque sgraffito border. Pale yellow glaze on interior only. Max. dimension 3-4. K2. PLATE 30,2 (Saranda Kolones, inv. FC 2393). Fragment from wall of plate of Byzantine Brown and Green Painted ware. Light red body. Slip on interior only. Brown dot within green loop, ring of brown dots outside it. Colourless glaze on interior only. Max. dimension 40o. K3. PLATE 30,2 (Saranda Kolnes, inv. FC 873). Fragment from centre of Zeuxippus Ware bowl of Class IA with splayed ring-foot, as Megaw 1968 fig. I no Body light red, grey on surface. Concentric circles gouged at centre through slip, under pale yellow glaze. Exterior, drips of slip only. Est. diam. of foot 6-o. Max. dimension 5"5- K4. PLATE 30,2 (Saranda Kolones, inv. FC 188). Thin wall fragment from Zeuxippus Ware small bowl of Class II (Megaw ). Body red, blackened internally. Small isolated medallion engraved through slip under yellow glaze, touch of dark yellow on medallion. Exterior, slip and yellow glaze on upper part. Max. dimension 4"2. K5. PLATE 30,2 (Saranda Kolones, inv. FC 309). Fragment with horizontal rim from plate of type with low ring-foot (Megaw 1975 passim). Form as Megaw 1975 fig. 2b. Red body. Pale yellow glaze on slip (interior only) with patch of green. Max. dimension 4-7. K6. PLATE 30,2 (Ayia Solomoni, inv. AS 1694). Rim fragment of Syrian Multi-glaze Sgraffito bowl with downturned lip, as Lane fig. 8j, here notched externally. Red body. Sgraffito border engraved through slip under greenish glaze with spots of strong green and brown. Exterior, slip and yellow glaze, green touches on lip. Max. dimension 7.0. K7. PLATE 30,3 (Saranda Kolones, inv. FC 141 I). Fragment of thick-walled, tall undecorated storage jar, as RDAC 197 I123 fig. 2,4. Light red body. No slip. Greenish interior glaze showing brown on body. Drip of glaze on exterior showing green. Max. dimension 6-0. K8. PLATE 30,3 and FIG. 6 (Saranda Kolones, inv. FC87I). Fragment from centre of thin-walled plate with low ringfoot. Same form and ware as RDAC 197 I 123 fig. 2,2 and pl. xxxiii I; found also at Khirbel al Mafjar (QDAP x 70). Dark purple-red body, blackened internally. Pair of lines gouged through thin internal slip show dark red under yellow glaze with grey patches. Exterior plain with patch of glaze within foot. Est. diam of foot 9-8. Max. dimension 6-9. K9. PLATE 30,3 and FIG. 6 (Ayia Solomoni, inv. AS 1694). Fragment from centre of bowl on rather high ring-foot with out-turned lip similar to C6-8 of the Lemba batch. Red body with blackened core. Hatching incised through slip with gouge under pale yellow glaze with touches of green. Unslipped exterior, but glazed except within foot. Est. diam. of ring-foot 70o, ht. to floor at centre 3-o.0 Max. dimension 70o. Kxo. PLATE 30,3 and FIG. 6 (Ayia Solomoni, inv. AS 1694). Another. Light red body. Slip-painted ring of ovals under colourless internal glaze. Exterior plain. Ht. to floor at centre 2-5. Est. diam. of foot 6-5. Max. dimension 5-8. KIi. PLATE 30,3 and FIG. 6 (Ayia Solomoni, inv. AS 1694). Fragment from centre of bowl on low ring-foot. Light red body, slipped inside and out (except within foot). Central medallion with alternate straight and wavy sgraffito lines, the latter coloured green (the centre one) and yellow in colourless glaze. Tripod scar. Diam. of foot 5'4- Max. dimension 8-9. K12. PLATE 30,3 and FIG. 6 (Ayia Solomoni, inv. AS 1694). Fragment from centre of bowl, as Taylor fig. 18 and form 8, but splayed ring-foot. Light red body. Interior slip. Central sgraffito medallion (hatched and plain bands) and parts of pendant semicircles. Brush-strokes of green in colourless glaze. Exterior plain. Ht. to floor at centre 2*1, diam. of foot 5-4. Max. dimension 7'4- KI3. PLATE 30,3 (Saranda Kolones, inv. FC 1370/3). Small fragment from bottom of cooking-pot of type A (Megaw, op. cit. (n. 36) (1972) 334 fig. D). Thin, dark red body, blackened except on exterior surface. Interior glaze showing dark chocolate. Max. dimension 3-2.

16 BYZANTINE AND ALLIED POTTERY: A CHEMICAL ANALYSIS 249 Kx4. PLATE 30,3 (Saranda Kolones, inv. FC 1052). Rim Atlit (QDAP iii i44 and pl. lvii 3). Dull purple-red body, fragment of cooking-pot with attachment of strap blackened in the core. Max. dimension handle, evidently of globular form, a type found also at 8.o. Batch L. Protomaiolica from Corinth This batch was selected, with the help of Charles Williams and his staff, from specimens of Morgan's Group I included in his catalogue, with one exception (L6). Ten of them are here illustrated for the first time. The fabric has been described in detail.37 On the characteristic opaque tin glaze, the decoration is overpainted in blue, yellow, and yellow-brown with manganese outlines. The ware continues to turn up at Corinth in thirteenth-century contexts.38 The principal forms are illustrated in BP fig. 83. Lx. PLATE 27,2. BP 790 pl. xxxva (inv. C ). Fragment from floor of plate. Low ring-foot pierced for suspension. Right shoulder and arm of horseman(?). Est. diam. of foot Io-o. Max. dimension 9-o. L2. PLATE 27,2. BP 791 pl. xxxvf (inv. C ). Fragment from simple rim of plate. Hands holding a firebrand(?). Est. diam. 24o0. Max. dimension 9-o. L3. PLATE 27,2. BP 792 pl. xxxvd (inv. C.33.9oo). Fragment from side of plate with everted rim (missing). Within zigzag and scroll border, right leg of warrior. Max. dimension 9.I. L4. PLATE 27,2. BP 794 pl. xxxvc (inv. C ). Fragment from side of large plate. Lion's head. Max. dimension 9-9. L5. PLATE 27,2. BP 796 pl. xxxvi (inv. C.37.I889). Fragment from side of plate. Top of animal's head. Max. dimension 6-7. L6. PLATE 27,2 (inv. C.38.90). Fourjoined fragments of bowl with everted rim. Form as BP fig. 83b. Parts of four radial leaves, blue outlined in manganese brown, blue circles between them. Est. diam. I5-0, diam. of low foot 6-o. Max. dimension L7. PLATE 27,2. BP 8oi pl. xxxve (inv. C and 56i). Two fragments, from floor and near rim of plate. Low ring-foot pierced for suspension. Blue cross, between the narrow arms of which converge the heads of fish; yellow fish-tail on the smaller fragment. Est. diam. 25-o. L8. PLATE 27,2. BP 806 (inv. C ). Fragment from side and everted rim of plate. Dull yellow leaves between blue leaves as L6, blue scroll border on rim. Est. diam Max. dimension 9-o. L9. PLATE 27,3. BP 821 (inv. C ). Small bowl with low ring-foot and thickened rim as BP fig. 83b. Central yellow gridiron medallion, blue chevron border. Diam. I4-8, ht. 6-I, diam. of foot 6-3. Lxo. PLATE 27,2. BP 813 (inv. C ). Section of plate as BP fig. 83a, with everted rim and low ring-foot (diam. 7.'0). Manganese dots in blue gridiron. Est. diam Max. dimension 130o. Lxi. PLATE 27,2. BP 817 (inv. C.36.9I6). Fragment from centre of plate. Form similar to BP fig. 83a. Blue band radiating from one round central gridiron medallion. Est. diam. of foot Io-o. Max. dimension 70o. L12. PLATE 27,3. BP 822 pl. xxxvib (inv. C ). Greater part of small bowl as L9. Diam. of foot 5'7. Est. diam. 15-o. Li3. PLATE 27,4. BP 843 pl. xxxvii (inv. C.38.72). Bowl centre. Form similar to BP fig. 83e. Solomon's knot. Black hatching alternates with blue. Diam. of foot 7-2. Lx4. PLATE 27,4. BP 844 pl. xxxvie (inv. C ). Bowl centre on low foot. Chequer-board, the filled squares blue or yellow. Diam. of foot 6-9. LI5. PLATE 27,4. BP 849 pl. xxxvid (inv. C ). Bowl centre. Form similar to L14. Four radial yellow blades, blue circles between. Diam. of foot 6-7. Lx6. PLATE 27,4. BP 819 fig. 83b (inv. C ). Small bowl as L9. Diam Lx7. PLATE 27,4. BP 853 (inv. C ). Small bowl centre fragment. Cross with circles between the arms in blue medallion, encircled by yellow-brown dots. Est. diam. of foot 4-5. Max. dimension 5-7. Li8. PLATE 27,4. BP 858 (inv. C ). Rim fragment of large cup, as BP fig. 83f, with stumps of handle. Radial daggers, alternately blue and yellow brown. Est. diam Max. dimension 9-3- Lxg. PLATE 27,4 (upside-down). BP 86I (inv. C.37.I355). Fragment from vertical neck ofjug, as BP fig. 83g. Very thin glaze. Vertical blades formed by intersecting arcs, yellow-brown dots. Est. diam. at bottom of neck 70o. Max. dimension 9-5- L2o. PLATE 27,4. BP 862 (inv. C.33.56o). Another. Vertical blades (all blue) formed by intersecting arcs, yellowbrown dots. Est. diam. at bottom of neck 8-o. Max. dimension 7-4. L2r. PLATE 27,4. BP 863 (inv. C ). Another. Scale decoration in blue, manganese scribbling on shoulder. Est. diam. at bottom of neck 6-5. Max. dimension 6-7. Batch M. Byzantine Uncoloured Sgraffito Ware from Istanbul To secure specimens of this most characteristic class of Byzantine twelfth-century tableware from Byzantium itself, we had recourse to a series of representative fragments collected by 37 BP 105-3s T. Stillwell-Mackay, 'More Byzantine and Frankish Pottery from Corinth', Hesperia 36 (1967)

17 250 A. H. S. MEGAW AND R. E. JONES Theodore Makridy and now in the Benaki Museum. For permission to analyse twenty specimens and to illustrate them we are indebted to Angelos Delivorrias, for help in selecting them to Mrs. Phopho Mavrikios. Some of the small rim fragments that were included preserve only minimal parts of the central sgraffito decoration (MI4, 17, and 20) and others none at all. But all qualify for inclusion by similarities of form (FIG. 7) or fabric, the latter usually characterized by the smooth, almost soapy, texture of the body where exposed. The two jug fragments (M3 and 6) are rarities, but their decoration is unmistakable. MI. PLATE 25,3 (inv ). Fragment from side of large plate. Compact red body. Thin slip inside and out. Zigzag sgraffito band (outer diam. c. i8-o) under yellowish glaze. Max. dimension 4-6. M2. PLATE 25,3 (inv. i9908). Fragment from side of plate. Light red body with some cavities, smooth 'soapy' finish. Slip and pale yellow glaze on interior only. Sgraffito plumage. Max. dimension 5-5. M3. PLATE 25,3 and FIG. 7 (inv ). Shoulder fragment of jug. Compact light red body. Slip and pale yellowish glaze on exterior only. Wavy sgraffito band. Est. greatest diam. I6-o. Max. dimension 4-1. M4. PLATE 25,3 (inv ). Fragment from side of plate. Dull red body, 'soapy' finish. Slip and glaze on interior only. Scribbled sgraffito band (est. inner diam. I6-o). Max. dimension 4'5. M10 Mil M14 M17 M3 M20 M12 M15 M18 M7 M 13 M16 M19 M6 M8 M9 FIG. 7. Batch M, Byzantine Plain Sgraffito (I: 3) M5. PLATE 25,3 (inv ). Another. Dull red body blackened at interior face. Sgraffito band (est. inner diam. I6-o), under yellowish glaze. Exterior 'soapy' finish with very thin dressing of slip and no glaze. Max. dimension 3o.0 M6. PLATE 25,3 and FIG. 7 (inv ). Fragment from side of thick-walled globular jug. Dull red body with exploded lime inclusion. Slip and pale yellow-green glaze on exterior only. Deeply engraved cufesque band. Est. greatest diam Max. dimension 5'5. M7. PLATE 25,3 and FIG. 7 (inv ). Fragment from floor of plate from near ring-foot. Compact dull red body. Thin slip interior only. Unidentified sgraffito decora- tion. Lacks glaze (waster?). Max. dimension 4'5. M8. PLATE 25,3 and FIG. 7 (inv ). Fragment from floor of plate. Compact light red body. Low ring-foot. Slip only on interior. Patch of sgraffito plumage under pale yellow glaze. Est. diam. offoot i oo. Max. dimension 4'5. Mg. PLATE 25,3 and FIG. 7 (inv a). Fragment from floor of plate. Compact light red body with exploded lime inclusions. Low ring-foot. Interior, outline of sgraffito medallion. Exterior, no slip on smooth 'soapy' finish. Est. diam. of foot Max. dimension 100o. MIo. FIG. 7 (inv ). Undecorated rim fragment of plate. Red body. Slip and greenish-yellow glaze extend to outside of rim and outside of lip respectively. Est. diam Max. dimension 5'7- MII. FIG. 7 (inv a). Undecorated rim fragment of a smaller plate. Red body with white inclusions. Slip extends to entire external curvature of rim, yellowish glaze just over the lip. Est. diam. 9.o0. Max. dimension 4-8. MI2. FIG. 7 (inv a). Another. Dull red body with a few cavities. Slip extends just over lip, colourless glaze has dripped below it showing orange-brown on body. Est. diam Max. dimension 5"1-

18 BYZANTINE AND ALLIED POTTERY: A CHEMICAL ANALYSIS 251 MI3. FIG. 7 (inv ). Another. Dull red body. Thin slip extends just over lip, with drips of colourless glaze extending lower. Est. diam. 240o. Max. dimension 4-6. MI4. FIG. 7 (inv. I9929c). Another. Similar body. Slip and colourless glaze extend just over lip. Sgraffito line at bottom of rim. Est. diam. 26-o0. Max. dimension 2-4. MI5. FIG. 7 (inv ). Another. A few red grit inclusions in similar body. Thin spotty slip and pale yellowish glaze extend over lip, the slip lower and wiped lower still. Est. diam Max. dimension 3-o. M16. FIG. 7 (inv ). Another. Profile as Great Palace I pl. 10o,5. Compact red body. Slip and yellowish glaze inside and out. Est. diam Max. dimension 4-7. MI7. FIG. 7 (inv ). Undecorated rim fragment of bowl. Compact dull red body. Only drip of slip on exterior, where colourless glaze extending over lip, shows orangebrown on body. Est. diam Max. dimension 3-7. MI8. FIG. 7 (inv ). Another. Slip on exterior of rim and wiped lower still. Yellowish glaze extends over lip. Est. diam Max. dimension 5-8. MI9. FIG. 7 (inv b). Fragment from side of large bowl. Thin slip and pale yellowish glaze on interior. Est. diam M2o. FIG. 7 (inv. I99g5b). Fragment from side of bowl at junction with foot. Dull red body, smooth exterior despite some cavities. Slip and pale yellow glaze on interior only. Strong incision encircles floor (est. diam. I I'o). Est. diameter of ring-foot Io-0. Batch N. Byzantine Sgraffito Ware from Pelagonnisos Wreck In 1971 the first large-scale underwater operation of the Greek Archaeological Service recovered 8oo vessels and many fragments from a cargo of mid-twelfth-century Byzantine pottery wrecked off Pelagonnisos in the Northern Sporades.39 We are indebted to Pavlos Lazarides, Director of the Byzantine Museum, for making arrangements for us to take samples from a selection of twenty fragments from this remarkable find, and for authorizing their publication in the present context. Some of them come from the familiar type of plain sgraffito plate with a central medallion and concentric bands of ornament, but a number of the rim fragments are without ornament. Some of the latter probably come from vessels decorated with a central device (commonly a bird or a beast) in the free field manner, the remainder from undecorated pieces. A few are smoke-stained, suggesting that together with pottery of high quality the cargo included a proportion of 'seconds'. All have suffered, but to a varying degree, from the accretion of marine growths now carefully removed. The normal body colour is orange. A characteristic form of rim is initially rather thick, sometimes even thicker than the wall below, but brought to a point at the lip (N5, 6, io, and 17). This profile recurs in our batch from Istanbul (FIG. 7, MI3) and is also found in the twelfth-century glazed pottery from the Athenian Agora,40 though there is no example in our small Batch O. Our pieces with touches of green (NII and 20) or brown (N4) are a reminder that examples of the painted-sgraffito manner were also recovered.41 NI. PLATE 26, I and FIG. 8 (inv. IO72). Rim fragment of plate. Orange-red body. Slip extends over lip only to external flange. Sgraffito elongated animal, probably one of several forming a border. Glaze interior only, discoloured grey. Est. diam Max. dimension N2. PLATE 26,1 and FIG. 8 (inv. 964). Fragment from rim and side of bowl. Orange-red body with some cavities. Slip extends over lip. Edge only preserved of free field sgraffito motif. Glaze blackened in patches and disfigured by white spots. Est. diam Max. dimension N3. PLATE 26,2 and FIG. 8 (inv. 998). Another. Similar body. Slip overall. Sgraffito central medallion and rinceau border at rim. Glaze inside and out, blackened and decayed. Est. diam Max. dimension o C. B. Kritsas, 'T7 B oavoclvdv vavdylov IHAayovvr4aov- AAovvi'aov', Athens Annals of Archaeology 4 (I97i) 176 ff. N4. PLATE 26,2 and FIG. 8 (inv. 992). Another. Gritty orangered body, blackened at core, a few exploded white inclusions. Slip on interior only. Sgraffito scrollwork border and band at rim. 'Cream' glaze, now matt, on interior only. Est. diam Max. dimension o107. N5. PLATE 26,2 and FIG. 8 (inv. 855). Rim fragment of plate. Body as N4. Slip overlapping lip. Sgraffito border; twisted ribbon on imbricated ground. Effaced yellowish glaze on interior only, blackened at one point. Est. diam Max. dimension N6. PLATE 26, I and FIG. 8 (inv. 864). Another. Orange-red body with many cavities, darkened core, exterior rough, exploded white inclusions. Slip overlaps lip. Sgraffito outline of central medallion. Yellowish glaze, now matt, interior only. Est. diam Max. dimension Frantz fig. 32 A65, A79; fig. 33 CI, C3, and C5. 41 Kritsas, op. cit. (n. 39) I8o.

19 252 A. H. S. MEGAW AND R. E. JONES NI N1 N5 N7 N10 N13 N2 N18 N17 N3N N6 N il N16 N19 N4 N15 N12 N20 N7. PLATE 26, I1 and FIG. 8. Another. Similar body, less rough. Slip overlaps lip. Edge of sgraffito free field motif. Crackled yellow glaze overlaps lip. Est. diam Max. dimension o107. N8. (Inv. 971.) Undecorated rim fragment of bowl as N1 5. Body as N6. Slip overlaps lip. Effaced glaze on interior only. Est. diam. 2o-o. Max. dimension 1o-o. N9. Another. Same form and body; slip overlaps lip. Glaze interior only, worn. Est. diam. 19-o. Max. dimension 8.3. Nio. FIG. 8 (inv. 934). Rim fragment of bowl. Red body with small cavities, white inclusions, and darkened core. Crackled yellow glaze on interior only. Max. dimension 14'5- Nxi. PLATE 26,1 and FIG. 8 (inv. Io73). Another. Body as N6. Much of the yellowish interior glaze lost, traces on exterior. On interior only, spots of green. Est. diam. 21 -o. Max. dimension NI2. FIG. 8 (inv. 965). Another. Profile as NI5, but deeper. Body as NIo. Thin interior slip, drips only on exterior. Yellowish glaze overlaps lip, darker where slip thinner. Est. diam Max diam. 15-i. NI3. FIG. 8 (inv. 659). Another. Light red body, smooth finish. Slip and yellow-brown glaze overlap lip. Est. diam. I6-o. Max. dimension NI4. Small fragment from wall of bowl. Slip and effaced glaze FIG. 8. Batch N, Pelagonnisos Wreck (1: 3) inside and out. Patch of brown in interior. Max. dimension 6-8. NI5. PLATE 26,2 and FIG. 8. Section of bowl on low ring-foot. Thin slip on interior only. Sgraffito central medallion with rinceau border. Yellowish glaze interior only, blackened in parts. Ht Est. diam. 27-0, est. diam. of foot I Max. dimension I.o NI6. PLATE 26,2 and FIG. 8. Rim fragment of plate. Slip, missing in places, overlaps lip. Sgraffito border at rim, zigzag filled out with scrollwork. Glaze on interior only, partly effaced. est. diam. 30-0o. Max. dimension 8-5. NI7. FIG. 8. Rim fragment of bowl. Compact orange-red body, smooth exterior. Slip overlaps lip. Touches of brown underglaze colour on lip. Grey-stained glaze on interior, drips on exterior. Est. diam Max. dimension NI8. FIG. 8 (inv. 955). Rim fragment of plate. Normal orangered body, with cavities. Much of slip and glaze (interior only) now lost. Est. diam Max. dimension 6.o. NI9. FIG. 8. Another. Body yellowish on exterior, normal orange-red core, a few cavities. Thin slip interior only. Patches of dark green colour under yellowish glaze on interior only, in one area discoloured grey. Est. diam Max. dimension I5-o. N2o. FIG. 8. Another. Body colour ranges from red core to orange surface. Thin slip overlaps lip, which has a stroke of green under the yellowish glaze (interior only). Batch O. Byzantine Glazed Ware found in Athens In the absence of material from a kiln site, a selection was made on the initiative of Timothy Gregory from a well-documented late twelfth-century pottery group found in the 1937 season of the American excavations in the Agora. The context was the upper part of the fill of a pithos found immediately west of the Stoa of Attalos and designated Sigma 434, Deposit PI I. i. The five principal items from this group, with which coins of Manuel I were associated,

20 BYZANTINE AND ALLIED POTTERY: A CHEMICAL ANALYSIS 253 comprise Alison Frantz's Group E.42 The ten glazed specimens which we analysed proved to have clay compositions so different from the typically Attic constituents of the clearly local coarse ware with which they were found (Batch P) that this glazed batch must be regarded as imports. And this is doubtless true of the fine warrior plate in the incised technique from the same context FIG. 9. Batch O, Athens, Imported Glazed Ware (I: 3) 01. PLATE 26,3 and FIG. 9 (inv. AS Undecorated rim 543)- fragment from plate. Red body with white inclusions. Slip inside and out, where thin. Thick yellow glaze on interior, 'cream' glaze on the underside. Est. diam. 30-o. Max. dimension PLATE 26,3 and FIG. 9 (inv. AS 544). Another. Red body, evidently wiped externally with slip for smooth finish. Interior slip overlaps lip, as does interior glaze showing cream colour. Est. diam. 24-o. Max. dimension 6-o. 03. PLATE 26,3 and FIG. 9 (inv. AS 545). Fragment from floor of plate adjoining foot. Red body. Broad circles incised at centre (est. diam. 8-0) and concentric sgraffito border incised through interior slip under yellow glaze. Exterior, thin slip under greenish glaze leaving rough surface. Max. dimension PLATE 26,3 and FIG. 9 (inv. AS 546). Fragment from rim of plate. Red body. Edge of sgraffito border through interior slip, overlapping lip. Brown-yellow glaze inside and out. Est. diam. I9.o. Max. dimension PLATE 26,3 and FIG. 9 (inv. AS 547). Undecorated fragment from wall (adjoining foot) of small cup, similar in form to BP fig. Io2(?). Red body with wheel-ridging. Slip inside and out. Interior glaze pale yellow, exterior 'cream'. Est. diam. of foot 3-o. Max. dimension PLATE 26,3 and FIG. 9 (inv. AS 548). Undecorated fragment from floor of plate on low ring-foot. Red body. Slip overall. Interior glaze yellowish, exterior colourless. Est. diam. of foot 12-o. Max. dimension PLATE 26,3 and FIG. 9 (inv. AS 549). Fragment from floor of plate adjoining foot. Red body, thin wiping of slip(?) gives smooth finish. Close concentric circles incised through slip, which with orange glaze extends on exterior to outer corner offragment. Max. dimension PLATE 26,3 and FIG. 9 (inv. AS 550). Undecorated shoulder fragment from jug. Red body. No slip. Glaze showing brown on exterior only. Est. diam. at bottom of fragment 14-0o. Max. dimension 5'7. Same vessel as Oio(?). Possibly an intrusion from the lower part of the pithos fill. 09. PLATE 26,3 and FIG. 9 (inv. AS 551). Fragment from wall of large bowl. Pink to light red body, 'soapy' surface. Incised decoration through slip (interior only) under colourless glaze showing 'cream'. Est. diam. at top of fragment Max. dimension 6 I. OIo. PLATE 26,3 and FIG. 9 (inv. AS 552). Undecorated fragment from flat bottom of jug. Red body. No slip. Glaze showing brown, on exterior only, including underside. Est. diam. of base Max. dimension 5-0. Same vessel as 08(?). Batch P. Coarse Ware found in Athens Selected coarse fragments (designated Sigma 433) from the same context as Batch O (Frantz Group E). PI. PLATE 26,4 and FIG. Io (inv. AS 553). Fragment from diam. of top of fragment (not the widest part of the vessel) bottom of globular amphora. Very pitted, micaceous red Max. dimension I 1.9. body. Interior wheel-ridging round centre point. Est. 42 Frantz Frantz 465 fig. 30 E2.

21 254 A. H. S. MEGAW AND R. E. JONES P2. PLATE 26,4 and FIG. Io (inv. AS 554). Rim fragment from cooking-pot. Red body, cleaner than PI. Shallow wheelridging (cf. Frantz B6). Est. diam. of rim 1i6o. Max. dimension i1-3. P3. PLATE 26,4 and FIG. Io (inv. AS 555). Fragment from globular neck of amphora (cf. that of the smaller amphora Frantz D4). Body as P2. Est. diam. of neck 100. est. diam. of rim 7-5. Ht P4- PLATE 26,4 and FIG. io (inv. AS 556). Fragment from bottom of basin with stub of handle. Very coarse red body with white and other inclusions. Est. diam. of base Max. dimension 5-o. P5. PLATE 26,4 and FIG. IO (inv. AS 557). Handle and part of rim (top surface sloping in) of small cooking-pot with shallow wheel-ridging (cf. Frantz B6, without wheelridging). Est. diam. of rim Max. dimension 9-5. P6. PLATE 26,4 and FIG. I0 (inv. AS 558). Rim fragment from cooking-pot as P8. Red body, blackened core. No wheelridging (cf. Frantz E5, from same context, with no ridging on shoulder). Est. diam. of rim. 26-o. Max. dimension 8'4. P7- PLATE 26,4 (inv. AS 559). Another, as P8. Similar red body, black towards interior. No wheel-ridging. Max. dimension 5-1. P8. PLATE 26,4 and FIG. IO (inv. AS 560). Another, vertical rim. Red body, blackened core. No wheel-ridging. Est. diam. of rim 250o. Max. dimension Pg. PLATE 26,4 (inv. AS 561). Another, as P8. Est. diam. of rim 26-o. Max. dimension 6-7. PIo. PLATE 26,4 and FIG. Io (inv. AS 562). Fragment from bottom of globular amphora. Body as PI. Interior wheelridging round centre-point. Max. dimension 8-2. P8 P2 P3 P4 P5 P1 P 10 FIG. Io. Batch P, Athens, Local Coarse Ware (I "3) TECHNICAL PRESENTATION The application of chemical analysis to the investigation of pottery origin is based on a number of requirements, of which the ability to characterize (or 'fingerprint') the pottery of the production centres of interest and the need to differentiate their respective compositions are paramount. Satisfying the former requirement is accomplished ideally by analysis of ceramic material that is clearly associated with a kiln; in this way, the derived chemical compositions can assume direct, undisputed locational significance. The factors that are relevant in differentiating the characteristic compositions of each production centre are essentially geochemical and instrumental. Within this framework, examination of the approach adopted in the present analytical programme suggests that conditions were, at the outset, favourable to a satisfactory outcome. The principal aim was to define stylistically and chronologically recognizable classes of pottery, most of them of fine fabric, with respect to the chemical composition of the clay, using the information obtained either to classify groups directly according to origin or to establish relationships between them. Not only were seven batches of specimens derived from kiln sites, but the geographical coverage of the production centres was sufficiently wide to expect,

22 BYZANTINE AND ALLIED POTTERY: A CHEMICAL ANALYSIS 255 a priori, significant inter-site composition differences on geochemical grounds. Moreover, the analytical method employed, optical emission spectroscopy, has been shown on previous occasions to be suited to this type of problem.44 Naturally, more sensitive techniques are required in order to differentiate compositions on an intra-site and intra-region basis. Finally, for comparative purposes, the Fitch Laboratory has been able to draw upon its large data bank relating to pottery of earlier date from centres throughout the Aegean and, to a lesser degree, from Cyprus. No attempt was made in this study to extend the characterization process either with respect to other properties of the clay45 or in a technological direction, for example, by analysing the pigments and glazes. The spectrographic technique followed the procedures outlined by Schweizer46 and described in detail by Jones.47 The preparation of the samples presented some problems: many specimens, being associated with kilns either as wasters or tripod stilts, had been overfired and were thus very hard. For these and, in fact, the majority of samples, small flakes were broken off the sherds with pliers and, the slip and/or glaze having been removed, they were ground slowly by hand in an agate mortar. In the case of Batches J and M the sample was obtained by drilling with a tungsten carbide drill-head. Of the mg48 of powder collected and subsequently heated at i io OC for two hours, 15 mg were withdrawn for analysis. The TABLE I Al Ca Mg Fe Ti Na Mn Cr Ni Coefficient of variation (%) o o0o07 oo0 I OIo I'7 7'5 o065 i' '031 0'012 chemical compositions were determined with respect to the nine elements (expressed as oxides) that are, and continue to be, routinely measured by the Fitch Laboratory and the Research Laboratory for Archaeology at Oxford in their provenance studies. Estimates of the precision of the analytical technique have been made periodically in the Laboratory by analysis of twenty samples of a standard sherd from Mycenae; its range of values, expressed as coefficient (standard deviation/mean x Ioo), is given in Table I.49 Also appearing in Table i, and relevant in this connection, are the compositions of two samples, 08 and O10, that appear to come from the same vessel. A combination of errors resulting mainly from the analytical procedure, but including the sampling process, are responsible for the small discrepancies between the respective contents of most elements; on the other hand, the discrepancies in Cr, which is admittedly the element with the poorest reproducibility of measurement, and in Al are such as to suggest that the two sherds may not be from the same vessel.50 The decision to have roughly twenty specimens in each batch was based on experience as well as on practical and statistical considerations. The principal factors that justified this 44 See, for example, Catling et al., BSA 75 (1980) 49-I13 and R. E. Jones and C. B. Mee, JFA 5 (1978) See, for example, L. Lazzarini et al., 'Chemical, Mineralogical and Moessbauer Studies of Venetian and Paduan Renaissance Sgraffito Ceramics', Archaeometry 22 (I980) A. J. N. W. Prag, F. Schweizer, and J. Ll. Williams, Archaeometry 16 (1974) R. E. Jones, Greek and Cypriot Pottery: a Review of Scientific Studies, forthcoming. 48 At least 150 mg in the case of the coarse wares, especially the Dhiorios batch. 49 A full account of both the precision and accuracy of the analytical procedure using optical emission spectroscopy is given by Jones, op. cit. (n. 47). 50 This matter is discussed in detail by Jones, op. cit. (n. 47).

23 256 A. H. S. MEGAW AND R. E. JONES decision were the control exercised over the uniformity of the members of each batch with respect to date and fabric type and the fact that for the Greek mainland, at least, none of the batches in this study, except Didymoteichon, constituted new 'points on the pottery map'. In the case of a previously unexplored region, such as Thrace, the Laboratory has found that analysis of not less than twenty specimens of uniform fabric and date usually gives an adequate indication of the range of compositions to be expected from a production centre; the example of Didymoteichon is no exception. As a first stage in the treatment of the data, a visual check was made on the homogeneity of compositions within the main batches of specimens. Although the dispersion of element contents within each batch displayed, not unexpectedly, some variability, it was soon established that in general one composition group accounted for all member samples. In the case of those batches that included wasters and other overfired material, this result was a useful demonstration that variable firing temperature had not affected the contents of the nine elements measured in this study. It is largely because the compositions within all the main batches are well defined that the individual compositions are not presented in this article; instead, their mean composition and standard deviation have been computed and they are set out in Table 2.51 It should be noted that one sample has been omitted from the calculations of the Lemba group characteristics (see FIG. 12). The compositions of Batch J, which are given individually in Table 3, correspond closely to those of Batch D. Mention has already been made that for a given element the dispersion of its contents among the groups is uneven; for several batches the values of dispersion are comparable with, or lower than, the corresponding values for composition groups determined by the Laboratory of decorated pottery of earlier (Late Bronze Age to Hellenistic) date. The compositions of one batch, N, deserve scrutiny in this respect for evidence of possible contamination from concretions resulting from sea burial. That this pottery was thoroughly washed in fresh water was verified by examination of the sodium contents, which were comparable with those in the other batches. The Mg contents, on the other hand, are likely to have been affected, and one indication is that their dispersion is larger than would be expected generally from fine pottery.52 Secondly, the possible common origin of this batch and the Sgraffito from Istanbul (M) has a bearing on this issue. The two groups are similar in their compositions for all elements except Mg, whose mean content is higher in the Pelagonnisos group. There is, thirdly, evidence from an independent source to support the claim of Mg enrichment; comparing the Mg contents of some Bucchero and Campanian pottery from land and shipwreck contexts, Picon53 observed that they were higher among the latter. We can now discuss some of the results in the light of relevant data from the Laboratory's archive. At Corinth the clays used in Byzantine times were evidently more calcareous than, but otherwise similar to, those encountered in some fourth-century B.C. plain wares54 from the city and in early Mycenaean pottery from Korakou nearby55 (Table 4). It is encouraging to find a correspondence between the compositions of the White Ware (Batch D) and 51 The composition characteristics of the groups presented by the authors, op. cit. (n. 2), were based on calculations of the preliminary data. Small discrepancies are to be found between the two data sets. 52 The dispersions of the other major and minor, but not the trace, element contents are also relatively high, but less so than that of Mg. 53 M. Picon, 'Remarques preliminaires sur deux types d'alteration de la composition chimique des ceramiques au cours du temps', FIGLINA I (1976) (Documents du Laboratoire de ceramologie de Lyon). 54 These were analysed in connection with a study of Corinthian transport amphorae in collaboration with Drs. C. G. Koehler, A. Kostikas, and A. Simopoulos. 55 See Mountjoy et al., BSA 73 (1978) 159 ff.

24 BYZANTINE AND ALLIED POTTERY: A CHEMICAL ANALYSIS 257 TABLE 2. Summary statistics of the main batches percentage oxide Find-Spot Batch N Al Ca Mg Fe Ti Na Mn Cr Ni Corinth A 20o 2II 2-5 7' I o-o026 s.d o 0o7 1-2 o0o8 0o o0oo6 5"5 Lapithos B o085 I8" o025 s.d I.I Io 0' oo005 0-oo005 Lemba C o0 6-8 I oo o-o66 o-o02 o-oo7 s.d Corinth (White Ware) D 20o > 25 <1I0 o o oo005 s.d II Thessaloniki E '3 2A4 8-o '15 0o104 oo s.d. 17"3 2" I1I o0io Didymoteichon F i o o031 s.d. 3' o o Dhiorios G o.62 0' s.d o 0. II o43 oo63 0-oo004 Kounoupi H '4 220-o o.o65 0-o40 s.d. 3-I 5I " o010 Corinth (Protomaiolica) L ' o II s.d. 3T3 5-o 0-7 II 0o.o8 0o77 ooi6 0-oo005 0-oo003 Istanbul (Sgraffito) M I o o oI5 s.d "3 1" o0o Pelagonnisos N '4 40o o oo18 s.d ,7 2'7 2'3 2' o21I 0oo6 o0oo6 Athens (Fine ware) O Io '5 7'5 o o019 0o013 s.d oo6 0-oo002 Athens (Coarse ware) P 1o 9 3'7 3'3 o I8" o102 0o s.d * I o029 o-o Ii (9 mean; s.d. standard deviation; < less than; > greater than) TABLE 3. The sample compositions of Batches J and K percentage oxide Sample Al Ca Mg Fe Ti Na Mn Cr Ni Batch J I o5 4-1 I 1-9 o-i o-oi8 0oo I I o-oi6 0' I- I o77 o-60o Batch K I I'4 7T3 0' o "084 I " o55 I" oo014 0o ' o oo " "4 15"7 30o o o39 1i o o014 oo010o " o o01o o-o o i o.oio o0oo6 8 22" i o ,7 13"2 I'3 5'3 o06o o010o o ' ' o-o64 o0012 0o007 II I919 9 II6 5'4 87 o oo o-oi o o "4 0o o'1o3 o0o25 o01o ' o o-oo o-00oo6 4"3 4"2

25 258 A. H. S. MEGAW AND R. E. JONES TABLE 4. Summary statistics of the comparative groups percentage oxide Group N Al Ca Mg Fe Ti Na Mn Cr Ni Corinth 22 X 14'7 3'3 9'5 0 I8"5 o'74 0o89 o0o (4th-cent. a.c. plain wares) s.d o o0o9 o0-48 o-o o-o1r Korakou 3" '7 3T3 8-5 o o0-26 (LH I-II fine wares) s.d o-8 I1] o-io o-63 o-oi8 o-o o o00o8 Bosphorus, white clay I <0' o90 < < 00oo 23"7 0"5 Bosphorus, clay mix I <o05 0o6 8.5 I.15 <0.2 o 017 0o Langadha brick I I8.2 2'3 2' o028 o0012 Thessaloniki brick I io o130 o-ox6 4" Franchthi cave *7 7 0'67 o 118 o-o092 o o034 (Neolithic wares) s.d I6 o o o Athens 23 X I7' '2 9-6 o O (SOS amphorae) s.d '4 0-7 I14 0' ' o01o Athens 20o "7 8 o085 o092 o-o96 o0o (Hellenistic BG) s.d o7 4o0 o0-i 0o15 0-oi6 o-o0 I 0o012 Thebes 20 t 18'7 14'3 5'4 9-0 o o (LH IIIB fine wares) s.d. 3'9 I "7 Taranto *3 17* o 0o77 o-96 o-o o-oi6 (Hellenistic semi-glaze) s.d '3 1-6 o-o8 o- 8 0-o23 0-oo003 o-oo6 4"5 Megara Hyblaea '4 II o o-o079 0o (Types A and B fabrics) s.d o o oo (; mean; s.d. standard deviation; < less than) a modern white clay from Anavatk6y on the Bosphorus56 that forms one component of a clay mixture prepared by potters working at Anadolu Hisan on the Bosphorus.57 A specimen from a modern brick-works at Langadha some 20 km north of Thessaloniki matches well with Batch E, the correspondence being less good with a modern brick made to the south-west of the city. The site closest to Kounoupi geographically, if not chronologically, for which analytical data are available is the Franchthi cave.58 Here chemical analysis of the many and varied classes of Neolithic pottery has identified one dominating composition type which is likely to be representative of the clays from the vicinity of the cave. The Kounoupi and Franchthi compositions differ in two elements, Ca and, especially, Mn being higher at Kounoupi than at Franchthi, but in other respects the compositions are similar (Table 4). The feature that characterizes Attic clays, relatively high values of the two elements Cr and Ni, is shared by Batch P. That these coarse wares are local products is supported by the -, We are grateful to Mlle D. Kassab for supplying samples of these clays. Analysis at the Research Laboratory for Archaeology at Oxford of a white clay used by potters today at iznik revealed that, unlike the Bosphorus clay, it was highly calcareous (H. Hatcher, personal communication). See n We may note some analyses of the White Ware reported by Talbot Rice in Studies in Islamic Art and Architecture in Honour of K. A. C. Creswell (Cairo 1965) 197 n. 4. The mineralogical composition of the best polychrome White Ware (his type 2) was given as: much quartz, a little calcite, a little y-al,o0, a little calcium aluminate, and a little feldspar or anorthite. That of his type i (pink or buff body, very hard) was: much quartz, some mica, possibly a little y-alo,. He also reported that iznik i6th-cent, pottery was clearly different from the White Ware since it contained much quartz, a little crystobalite, and a little diopside (198 n. 4). 58 Chemical and petrological analyses of pottery from the cave have been carried out in collaboration with Dr. K. D. Vitelli, Mrs. Ph. Pomoni, and Dr. V. Perdikatsis.

26 BYZANTINE AND ALLIED POTTERY: A CHEMICAL ANALYSIS IN H/ E A -2 F -4-6 P O CV1 FIG. I I. Canonical variable plot of the discriminant analysis of the Corinth (A), Thessaloniki (E), Didymoteichon (F), Kounoupi (H), and Athens (P) batches. The first two canonical variables, CVI and CV2, account for 50 and 23% of the total dispersion respectively. Ca, Ni, and Mn are the principal discriminating elements overall similarity of composition with Attic pottery of earlier date-sos amphorae of the Archaic period59 and Hellenistic Black Glaze pottery60--despite their disparity in date and fabric.61 It remains finally to apply to the data a technique of multivariate statistics, discriminant analysis, that admirably meets the requirements of differentiating the compositions classified according to groups. In all applications of this analysis the full complement of nine element contents was used, the Mn, Cr, and Ni contents being log transformed.62 The results are presented visually in the form of a plot of the first two canonical variables in which circles are drawn enclosing c. 90 per cent of the member samples of each group. In FIGS. I I-14 the centroid of each circle is marked by the batch letter. The first discriminant analysis attempted to differentiate, for heuristic purposes, the five mainland groups; as is shown in FIG. I I, this was achieved but for some overlap between the Corinth and Kounoupi groups. Ca, Ni, and Mn appeared as the principal discriminating elements. The three Cypriot groups are also capable of discrimination (FIG. 12), principally along the first canonical variable that accounts 69 A. W. Johnston and R. E. Jones, 'The SOS Amphora', BSA 73 (1978) 10o Prag et al., op. cit. (n. 46) Table The Fitch Laboratory's corpus of Attic pottery ranging in date from the Mycenaean period to the present material of the I2th cent. A.D. reveals small but discernible variations in composition, reflecting the use of different clays and clay mixes in the Athens region as functions of time and type of vessel. The impressive results obtained in an analytical programme carried out at the Brookhaven National Laboratory in collaboration with Mrs. D. Thompson on material of Proto- geometric to Hellenistic date are highly pertinent to this issue; an account of this work was given by D. Fillieres at the Archaeometry Symposium in Brookhaven in i An exposition of discriminant analysis in its application to provenance studies of pottery in the Aegean is given by Cherry in Mountjoy et al., op. cit. (n. 55) and in Catling et al., op. cit. (n. 44). A more general account is provided by Pollard in Chapter 2 ofjones, op. cit. (n. 47). The BMDP7M stepwise discriminant analysis programme was used on the Oxford University Computing Service's ICL 2980 computer.

27 260 A. H. S. MEGAW AND R. E. JONES 61 O 4 2 G B 0 C C FIG. 12. Canonical variable plot of the discriminant analysis of the Lapithos (B), Lemba (c), and Dhiorios (G) batches. CI6 is the anomalous Lemba sample. The first two canonical variables, CVI and CV2, account for 69 and 31 % of the total dispersion respectively. Mg and Cr are the principal discriminating elements t mo A -4-6 P C CV1 FIG. 13. Canonical variable plot of the discriminant analysis of the Corinth (A), Istanbul (M), Pelagonnisos (N), Athens fine ware (o), and Athens coarse ware (P) batches. The first two canonical variables, CVI and CV2, account for 59 and 36% of the total dispersion respectively. Ca, Ni, and Fe are the principal discriminating elements

28 BYZANTINE AND ALLIED POTTERY: A CHEMICAL ANALYSIS 261 cm N 2 A\ 0 D ---r -2-1c CV 1 FIG. 14. Canonical variable plot of the discriminant analysis of the Corinth (A), White Ware (D), Protomaiolica (L), and Taranto (T) batches. The first two canonical variables, CV and CV2, account for 87 and 8% of the total dispersion respectively. Ca is the principal discriminating element for 69 per cent of the total dispersion. CI6 is markedly atypical among the Lemba compositions because of its high trace-element contents (Cr o-o51 per cent and Ni 0o028 per cent). Moving to the groups of uncertain origin, visual assessment of the compositions of the imported glazed wares from Athens, the Istanbul Sgraffito, and the Pelagonnisos wreck material (Batches O, M, and N) reveals general similarity, a situation that is reflected in the discriminant analysis of these and the Corinth and Athens coarse ware groups (FIG. 13). This might be sufficient to justify a common origin for Batches O, M, and N were it not for the fact that their compositions overlap with the Thessaloniki material (Batch E). The application of discriminant analysis to the four groups, that is, Batches E, M, N, and 0, does not unequivocally support the preferential alignment of any one group with another, for example, the Athens and Thessaloniki groups, on the one hand, and the Pelagonnisos and Istanbul groups, on the other. It is expected that this problem may be resolved as the Laboratory moves over to atomic absorption spectrometry, an analytical technique that is capable of greater precision than the present spectrographic method, and when material becomes available for analysis from other major centres, both on the mainland and on the Asiatic coast, particularly if kiln sites are found there. In the meantime, it is of interest that one such centre on the mainland, Thebes, can be discounted as the source of the Athens group since this group's compositions are well differentiated from those of (Mycenaean) Theban pottery in Ca, Mg, Cr, and Ni (Table 4) See Catling et al., op. cit. (n. 44) Table 4.

29 262 A. H. S. MEGAW AND R. E. JONES Finally, we come to the Protomaiolica ware. Comparing its composition visually with that of the Byzantine White Ware (D), it is clear that there is no connection between the two wares. As regards the Corinth batch (A), on the other hand, the differences are less discernible, both wares being made of a highly calcareous clay. A closer correspondence is evident with south Italian pottery; the Laboratory's reference data for Italy and Sicily are restricted to a number of sites in Apulia, probable production centres of Hellenistic glazed wares,64 and Megara Hyblaea in Sicily,65 and it is with the Apulian compositions that the matching is found (Table 4). Taking a group of Hellenistic semi-glaze pottery from Taranto as a control for southern Apulia, the results of discriminant analysis (FIG. 14) tend to support this view, the Protomaiolica lying closer to the Taranto than the Corinth group. The discrimination between these three groups occurs on the second canonical variable and as such it is small compared with that separating them from the Byzantine White Ware. A more thorough assessment of the origin of the present group of Protomaiolica will undoubtedly have to await the analysis of finds of this ware in south Italy and Sicily. APPENDIX Batch K. Sherds excavated at Paphos These fourteen fragments, each representative of a particular type of pottery found on two sites at Paphos, were analysed to see how far their origin could be clarified or their classification confirmed by reference to the compositions established for the production centres and wares covered by the sample batches in the main enquiry. Some caution was necessary, however, in interpreting the composition data, owing to the difficulties that frequently arise in making assignments of provenance to individual pieces. To appreciate this point, one should bear in mind, firstly, the relative lack of precision of the spectrographic technique and, secondly, the fact that for a variety of reasons the composition of the single specimen may not be typical of its parent population. The compositions of the fourteen pieces, which are set out in Table 3, were both classified and then compared with the reference data by univariate means. It should be stressed that the Fitch Laboratory's data bank scarcely extends outside the Aegean and Cyprus. Three samples had been identified provisionally as imports from the mainland Crusader states, an area outside the limits of our investigation: K6, 8, and 13. It was no surprise to find that their compositions do not match any of those established for the main batches. Although their calcium contents are uniformly low, the compositions are otherwise not comparable; the iron and titanium contents of K8 are exceptionally high. On stylistic grounds, a possible source of the cooking-pot fragment, K I3, is Acre, where there was a street of cooking-pot makers in the thirteenth century.66 If that is indeed their source, it is clear that the industry there was established before 1222, when the Paphos castle was destroyed. The plain sgraffito fragment of Byzantine type, likewise from the castle, KI, has a composition that lies within the concentration ranges of the glazed pieces from the pithos group in the Athenian Agora (Batch 0) for all elements, the matching being less good with the twelfth-century batches from Istanbul (M), Thessaloniki (E), and the Pelagonnisos wreck (N). But the source (or sources) of the Athens glazed sherds has not been established, and it is to be remembered that this group is dated by coins later in the century than the decoration on the other groups and the Paphos sherd would be expected. 64 From Prag et al., op. cit. (n. 46) Table H. Treziny and R. E. Jones, MEFRA 91 (1979) D. Pringle, 'The Medieval Pottery of Palestine and Transjordan (AD ): an Introduction, Gazetteer and Bibliography', Medieval Ceramics (Bulletin of the Medieval Pottery Research Group) (198i) 45-

30 BYZANTINE AND ALLIED POTTERY: A CHEMICAL ANALYSIS 263 The composition of the Brown and Green Painted plate fragment K2 is similar and, as in the case of KI, its decoration is appropriate for a date hardly later than the mid twelfth century. By their composition, both sherds are likely to be imports from the Aegean, Corinth as a source being excluded. A different composition, not matched in any of our main batches, was recorded for two fragments of the high-quality Zeuxippus Ware, K3 and 4, datable around 1200 but before the destruction in 1222 of the castle where they were found. If it was correct to localize this ware at no great distance from Constantinople,67 it can hardly have been at the same centre as the 'classic' twelfth-century Byzantine pottery represented by KI and 2. Certainly, the origin of K3 and 4 is not local, although it is interesting to note that Zeuxippus Ware appears to have been imitated at Lemba (C22). The composition of the rim fragment K5 of another early thirteenth-century ware is difficult to match; it shares some features in common with that of the Zeuxippus Ware sherds, though the fabric is much coarser in quality. This association, coupled with the earlier Aegean examples of the decoration of the class to which K5 belongs (07 and 09 in our Athens batch are such Aegean examples)68 gives some support to the localization of Zeuxippus Ware also in the Aegean area. The allocation of the type of later thirteenth-century bowl represented by the base fragments K9 and Io (FIG. 6) to the west of Cyprus seems now confirmed by their similarity in clay composition to our Batch C from Lemba, near Paphos, which furthermore includes a waster of identical form (FIG. 2, C6). Both come from the Ayia Solomoni deposit and their date is indicated by the lack of this type of bowl in the 1222 destruction contexts on the castle site. On account of its similar composition, we may also assign provisionally to the same local source a series of rather earlier and very different vessels represented by K7. These are tall storage jars with thick rounded bases and glazed interiors from the Paphos castle, where numerous examples were sealed by the 1222 destruction. Whatever their function, at that date they would be the earliest identified products of the Paphos industry. The two bowl fragments from the unstratified deposit at Ayia Solomoni, K II and 12, resemble in their composition the samples in our Batch B from a Lapithos factory active at the close of the fifteenth-century. By that time the Paphos industry was in decline, if not extinct, to judge by the appearance of the alien fragment C23 among the much earlier kiln waste from Lemba, and by the appearance in the neighbourhood at Kouklia69 of the latest Lapithos style represented by our fragment BI5. Consequently, the appearance of pottery with Lapithos connections among the later material from Ayia Solomoni is not surprising. Little can be said of the cooking-pot rim KI4 other than that in composition it does not resemble KI 3; of the three Cypriot groups examined, Lapithos comes closest, but the matching is poor. It is likely that the results of a programme of petrological analysis on cooking wares of this period would be more revealing. However, taking into account the differentiation of local products, those made elsewhere in Cyprus, and those imported, which analysis has made possible, it is permissible to conclude that even a small contribution like the present one to an archive of chemical compositions can materially assist the interpretation of Byzantine and allied wares encountered in the course of excavation in the eastern Mediterranean area. A. H. S. MEGAW R. E. JONES 67 Megaw Megaw ' F. G. Maier, RDAC I979 pl. xvii 2.

31 3 2 w PLATE 24 B.S.A. 78 B E U m a1 N% I. Batch H(I), Chinitsa H m yrl U/ b i7"i lwn 27 mw m U mo 2. Batch H(2), Kounoupi 3. Batch H(2), Kounoupi - 2! s ::::::: 7i--~-~niiii 18 1 i!'iia!i Batch G, Dhiorios Batch G, Dhiorios BYZANTINE AND ALLIED POTTERY

32 l B.S.A. 78 PLATE 25 2 : :~:-:: IMM: 3 14~ ::~~: i. Batch D, Byzantine White Ware 2. Batch D, Byzantine White Ware E H H I U Batch M, Istanbul 4. Batch A, Corinth BYZANTINE AND ALLIED POTTERY

33 11 PLATE 26 B.S.A """':"":":I 16 3ll 2 7 i4iiiiiii 15 i. Batch N, Pelagonnisos Wreck 2. Batch N, Pelagonnisos Wreck 1 2 r a l Batch O, Athens, Imported Glazed Ware 4. Batch P, Athens, Local Coarse Ware BYZANTINE AND ALLIED POTTERY

34 4 B.S.A. 78 PLATE I. Batch A, Corinth 2 3 8~i~, o!oiiio Batch L, Protomaiolica Batch L, Protomaiolica 4. Batch L, Protomaiolica BYZANTINE AND ALLIED POTTERY 19

35 PLATE 28 B.S.A :::r::::: I::1:: 3i::: ~i _i 4L i1iii iiiii~iiiii 7R.~ Irg i. Batch C, Lemba 2. Batch C, Lemba 10 11* 12 3 ~a~a~ 4 4i!~iiiiiiiiiiii w y 2o1' ~~8i:DI~ Batch B, Lapithos BYZANTINE AND ALLIED POTTERY 4. Batch B, Lapithos

36 to B.S.A. 78 PLATE N 4 S~~liii :::'::--::: 17iiii iiiiiii:itsi.iiii oi iii 14 i i:i::i41i :ii I. Batch E, Thessaloniki 2. Batch E, Thessaloniki 12b I'EIEI iiiiiiiiiiiiiii!oi!io!!!iiiio!ilili!;!i Ma i~i_-:r-:::::::- 7 '71 15I 1 6 iiii! y i 8 41-iioiiiiil~ i~~ W" Batch F, Didymoteichon BYZANTINE AND ALLIED POTTERY 4. Batch F, Didymoteichon

37 Miscellaneous Paphos, K, Batch 3. POTTERY ALLIED AND BYZANTINE Ware White Miscellaneous Istanbul, J, Paphos, K, Batch i. Batch 2.