BaBesch 74 (1999)

The Late Roman Unguentaria of

R. Degeest, R. Ottenburgs, H. Kucha, W. Viaene, P. Degryse, M. Waelkens

1. INTRODUCTION them, as was for instance the case with the publica- tion of the Ernst von Sieglin collection8. The precise The city of Sagalassos, the ruins of which are situ- origin of this pottery type could never be determined, ated on the western slopes of the Taurus mountain although Hayes considered a possible production range in southern Minor, was already a town centre located somewhere in the Palestine-Lebanon in its own right when subdued area. The vessels bear a marked resemblance to ear- it by force in 333 B.C. After this event the city lier Nabataean ceramics, at least as far as ware fab- remained in existence for nearly a millennium until ric and colouring are concerned9. The vessel shapes it was finally abandoned somewhere around the are different however, and there is a hiatus of sev- middle of the 7th century. The site has been under- eral centuries between the wares. Also, publications going excavation by an international team directed of archaeological material from this region fail to by M. Waelkens since the beginning of the nineties1. produce significant numbers of these vessels. Among the many finds a series of slender fusiform Logically it could be assumed that greater numbers vessels, called “Late Roman Unguentaria” by J.W. of finds would be present near the production area. Hayes2, have always attracted attention in In fact based on the quantitative evidence so far Sagalassos, because they are so distinctive when available for , , Perge, and compared to the usual Sagalassos wares. The local Sagalassos, a production centre in or near Asia Minor products, both common and fine, are of a different now seems more likely10. The time frame of this spe- nature, as far as ware fabric composition, colour and cific type is as a whole not completely fixed, but other descriptive elements are concerned. In 1993, there is little doubt that the main concentrations are a first preliminary report on the presence of these to be found in the sixth and seventh centuries11. unguentaria was presented in the Sagalassos series The reason why this particular type of vessel is so of excavation reports3. In the present report we wish important at Sagalassos, notwithstanding the rela- to comment on the many examples that have come tively limited number of finds, can be attributed to to light since then and on the additional research that several factors. has been done. The first factor concerns the physical aspect: next Other types of unguentaria have been excavated at to the already remarked upon ease of identification, Sagalassos, but these were not deemed relevant to only one type, albeit with two variants, is found on the present study, as they fit into the much earlier this site. This type is called 8I110 in the Sagalassos tradition of Hellenistic to early Roman unguentaria4. classification scheme12. Both variants are only dis- The latter are mostly associated with funeral con- cernible in their lower part, i.e. the base, which com- texts and differ in shape and chronology to a con- monly is a massive spike13, while the rarer variant siderable extent from the late unguentaria. What initially attracted attention to the late unguen- 1 see Fig. 1. taria was the easy and virtually error-proof identifi- 2 Hayes 1992. cation of these vessels5. Even extremely small shards 3 Degeest 1993. could be recognized without trouble by their distinc- 4 For an overview of the earlier unguentaria see for instance Anderson-Stojanovic 1987. tive colour and finish. This vessel type was originally 5 Fig. 2. identified by J. Hayes as a separate and widely dis- 6 Hayes 1971. tributed vessel type from Late Imperial to Byzantine 7 Hayes 1968; Hayes 1992. times in the Eastern Mediterranean and beyond6. The 8 Pagenstecher 1913. 9 Hayes 1971. most complete publication concerning these vessels 10 Hayes 1992; Pülz 1995, 69-70; Atik 1995, 180-181. A was also from his hand, namely that of the excava- gazetteer of find locations can be found in the original article by tions at Saraçhane, the former church of Saint Hayes (Hayes 1971). Since then more finds have been published, Polyeuktos in Constantinople7. Although vessels of but we considered it outside the scope of this work to include them here. this type were rare in publications before their sys- 11 Hayes 1971, 245; Riley 1975, 37; Atik 1995, 181. tematic description and typing, earlier publications 12 Degeest, forthcoming. sometimes contained references and even images of 13 Fig. 3c-e.

247 has a flat, thick-walled base14. There is no known base of the triangle located on the upper part of the chronological difference in their spread, as both are vessels and the point near the spike. From the way found in the same contexts. Another type of unguen- in which the slip trickled down onto the lower part tarium has been identified by Riley as Type A, but of the body, it is clear that these are the impressions no traces of this type could be found at Sagalassos. of a triangular stand on which the leather hard ves- Type A unguentaria show a soft brown micaceous sel was placed to dry after immersion of the top half fabric that is very different from the very hard-fired in a slip solution. The three contact points were Type B15. clearly not at the same level, with two points pre- Secondly, the local stratigraphy at Sagalassos indi- sumably at the same level and higher than the third cates that the time frame for this ware is rather lim- point. In some instances the potter has used a small ited, with a very late appearance, i.e. presumably sponge or similar instrument to wipe down the ves- not before the end of the 5th century, and a disap- sel, since traces of such markings with a width of pearance at the latest around the middle of the 7th approximately 2 cm have been found. On a minor- century. ity of the vessels a stamp is present on the lower half Another contributing factor is the presence of a of the body. Often the clay was so wet at the time number of stamps, which may eventually lead to of the application of the die, that the stamp was par- further information on the history of Sagalassos and tially smudged. its trade related connections. To describe ceramic fabrics macroscopically at Sagalassos the descriptive algorithm developed by D. Peacock was used. This has proved to be a very 2. THE TYPOLOGY AND MINERALOGY OF THE LATE popular one over the years since its first publication UNGUENTARIA in 1977, because of its straightforwardness17. The only notable exception was the initial use of Before we start with the distribution of the unguen- Munsell Soil Color Charts for colour descriptions, taria, a description of the physical characteristics of instead of common name descriptions. In the end these vessels is appropriate. Whereas at first the former proved to be too time-consuming for the research was perforce limited to a purely macro- description of individual shards, and the practice is scopical description, the integrated approach of the now largely discontinued, except under special cir- Sagalassos project allowed a close collaboration cumstances, i.e. when new fabric ware groups are between the archaeological and geological branches, encountered. which resulted in more detailed information. Macroscopically the ware fabric group that com- The general shape is a slender fusiform one with prises these vessels is our group 818. This fabric is mostly a massive foot16. A rarer variant shows a very hard fired, many fragments show distinct sin- slightly wider string cut base with a thinner wall in tering. Rarely some fine whitish inclusions occur. this region. The vessel was clearly never intended The latter do not react to the standard hydrochloric to be stood upright by itself, as even the broader acid test19. based variant is unable to stand independently The colour is very irregular, with a range extending because of the irregular shape of the base. In from gray to maroon and purplish red, usually with Sagalassos we encountered mostly the same size of a colour difference between the core and margins. vessel, with a maximum height of 21-22 cm, a body The combination of a gray fired core and reddish width of 5-5.8 cm, a rim diameter of ca. 3 cm, and buff surface is most common. The colour gradation a base width of 1-3 cm, depending on the type of is facilitated by the thickness of the lower parts of base. The rim is thin and slightly flaring with a dis- these vessels, which can be more than 1 cm. The tinctive horizontal ridge at its base. The vessel was Munsell charts give often 10 R 5/6 or 7.5 YR 6/6 wheel thrown, apparently made in a single opera- reddish yellow as margin colour and 2.5 YR 5/0 gray tion, with a wet smoothed outer surface and dis- as core colour. The fracture is always conchoidal, tinctive interior spiralling as the shape of the vessel did not allow a proper smoothing of the interior, except at the very top. Not always immediately 14 Fig. 2. 15 Riley 1975: p. 36-37. It should be noted that this Type A is noticeable is the presence of a slip layer on the top more prevalent in Palestinian contexts than Type B, but is totally half. This is usually invisible on the sintered exam- absent in Sagalassos. ples. The latter are quite common and comprise as 16 Fig. 2, Pl. 1-2. much as 20% of the total shard number. The com- 17 Peacock 1977. 18 Degeest forthcoming. plete or near complete examples show another fea- 19 Hydrochloric acid is used to determine the presence of lime ture of their manufacture, namely three oval impres- in a fabric. As lime breaks down at temperatures over 750°C, sions set lengthwise in a triangular fashion with the its presence would indicate a low fired fabric.

248 much more so in fact than the Sagalassos wares20. has been excavated at Sagalassos, a distribution cov- The mineralogical analysis consisted of both chem- ering the whole of the site equally is impossible. ical bulk analysis and optical mineralogical compo- What is clear from the excavation stratigraphical sition determination21. To our knowledge no simi- series is that the finds are limited to the last phases lar analyses have been performed elsewhere on this of the occupation of the town, broadly from the late material. Chemically the fabric is characterized by 5th to the first half of the 7th century. 22 moderate amounts of SiO2 . The Al2O3 content of The main concentrations are found in areas that 18.5 % is the highest mean value recorded, with the were heavily occupied at that time, namely the so- exception of a single shard with no less than 24 %. called UAN site, a large complex of workshops and The iron content is also rather high in the fabric storage at the northern end of the upper agora24. This series of Sagalassos with 7.9 %. The CaO contents complex was built on the remains of an earlier mar- amount to 5.7 %. Low are also the amounts of Na2O ket building. The second most extensive concentra- (0.5 %) and the value for Loss-On-Ignition (L.O.I. tion was recovered from the layers to the north of 1.6 %). The variation coefficients are narrow for the late fortification wall that abutted the Doric tem- 25 SiO2 (0.09), Al2O3 (0.05), iron oxide (0.1), and TiO2 ple to the west . The areas in between or adjacent (0.12). A wider spread is shown by MnO (0.3), to these areas were also found to contain unguen- 26 MgO (0.29), Na2O (0.29), K2O (0.24), and P2O5 tarium shards, namely the upper itself , the (0.36). A wide spread was recorded for CaO (0.67) bouleuterion area between the upper agora and the and for L.O.I. (0.76). Doric temple27, and the northwestern heroon site to Mineralogically, no grog, chert, or pillow lava was the north or this28. found, elements that are typical for most Sagalassos Lesser concentrations were found in the excavations wares. Pyriboles accounted for 0.3 volume %. around the commercial center of the town, namely Biotite was present (1.6 %), as was iron oxide (0.2 the lower agora zone29, with to the east of the agora %), feldspar (2.2 %), quartz (3 %) and calcite (0.3 proper the main baths30. Other, similar concentra- %). Porosity was 0.2 %, i.e. the lowest value ever tions were found on nearly every site within the found for ceramic fabrics at Sagalassos. This can be original perimeter of the town, e.g. in a housing connected to the extensive sintering that is often zone to the northeast of the main bath building31. found in the ware. From smaller sites like the fill of the Late The relatively low feldspar and pyribole, i.e. amphi- Hellenistic fountain32, which was located outside the boles and pyroxenes, presence and the high quartz area of the late city walls, and from site W in the amount normally indicate fluviatile transport of the domestic quarters to the west of the public centre clay. Quartz may have been added as a filler, but the came also some examples of this ware33. narrow spread as indicated by the variation coeffi- cient for this mineral23 argues otherwise, unless one accepts an uncommonly high level of standardisa- 4. THE STAMPS FOUND ON THE UNGUENTARIA tion of the clay mixture. The latter seems unlikely in view of the often sloppy finish of the vessels in If we compare the number of shards with those from this ware. the largest published collection, namely that of Considering the differences in chemical/mineralog- ical composition, for example the total absence of chert and pillow lava, which to some extent can be considered to be trade marks of the local fabrics, 20 A conchoidal break, from the Latin concha or shell is typi- and the archaeological data, namely the relative cal for very hard homogeneous non-crystalline or partially crys- talline materials like glass. scarcity of the finds at Sagalassos, this was likely 21 All mineralogical/chemical analyses were done at the labo- an imported product for this area. ratory of Physico-chemical geology of the Katholieke Universiteit Leuven by W. Viaene, R. Ottenburgs, H. Kucha and P. Degryse. 22 Mean weight % was 56.4: see table 1. 3. THE DISTRIBUTION OF LATE UNGUENTARIA IN 23 See Table 1. SAGALASSOS 24 Fig. 1, no 5. 25 Fig. 1, no 2. 26 Fig. 1, no 5. To date 501 shards belonging to this type of vessel 27 Fig. 1, no 4. have been recovered from the excavations of the 28 Fig. 1, no 3. city of Sagalassos. Compared to the total number of 29 Fig. 1, no 9. 30 Fig. 1, no 8. excavated shards, which reaches many tens of thou- 31 Fig. 1, no 7. sands, this is a small number. Yet it is not insignif- 32 Fig. 1, no 6. icant. Due to the as yet relatively limited area which 33 Fig. 1, no 1.

249 Saraçhane in , where 504+ shards were 4.1. Monograms based on a rectangular frame- recovered, among which were some 50+ stamps34, work: Sagalassos shows a total of 501 shards, with only SA-91-DT-330-13241: previously published42, clear 21 stamps35. The stamps are more than fleetingly circular stamp of 6 or 7 interconnected letters, from interesting because of the fact that they are not lim- the area to the west of the Doric temple and north ited to this type of pottery and not even to ceramics of the late fortification wall (layer 3). in general. All stamps found so far at Sagalassos are SA-95-UA-148-143: the same monogram as the pre- of the so-called monogram type. vious one, probably even the same die was used. Monograms are found on rings, , seals, archi- The stamp is less deeply imprinted in the clay how- tectural elements, and on a variety of other objects36. ever. This was found on the upper agora. This type The main subject of the monograms are primarily of monogram was predominant until the late sixth names or invocations37. The large number of lead century44. The legible letters could be Q or F, U, T, seals with monograms are the primary source for E or P, O, M or S, and I. The same stamp was our knowledge of monograms of this kind, for the found on an unguentary from Ephesos45, dated from simple reason that, just like coins the reverse the sixth to the mid seventh century. contains often additional information on the owner of the seal and his official or private functions. In 4.2. Cross-shaped monograms many cases this allows a precise dating of the seals. In this group we can discern two variants. On the Monograms are to be found in many walks of life one hand we have square or rectangular stamps with in the Late Roman – Early Byzantine period, which fine or thick incised lines, while on the other hand is the one that concerns us. The most comprehen- the stamps are round with thick lines. sive and reliable source is formed by the lead seals38. The unguentaria monograms pose a diffi- culty, because there is no reverse side containing additional information. However, the study of the 34 Hayes 1992: 8. 35 See Table 2. lead seals allowed researchers to establish certain 36 Examples can be found in considerable numbers, as there datable elements based on the structure, iconogra- are coins like the pentanummium of Justin II (565-578) (Hendy phy and epigraphy of the stamps. This was based on 1986, 298, 300). Monograms can also be found on architecture, the assumption that fashions changed over time and for instance an extended series was part of the architectural decoration of the church of St. Polyeuktos in Istanbul erected that individuals would tend to respect the latter. in 524-527 (Harrison 1986, 406). Here we find them on cor- However, even if true for the majority of the cases, nices, piers, and screens (Harrison 1986, 162, fig. L). In fact allowances should always be made for personal from the remaining fragments 14 different monograms could preferences and conservatism39. A single stamp is be recovered, all attributable to the original church building. These monograms are by no means the only ones to be found therefore not by itself a reliable indicator for a given on architectural monuments. Others are present e.g. on the period. When studied as a group, the results become church of St. Mary at Ephesos and the martyrion of Philippos progressively more reliable as general trends will at Pamukkale, the ancient . Other places where emerge. One of these trends is the observation that monograms are found are for instance inscribed on utilitarian monogram seals are generally to be dated from the pottery like a jar found on the Yass Ada wreck from the early th th seventh century (Bass et al. 1982, 169, fig. 8). Monograms are 6 to the 8 century after Christ, with so-called not unexpectedly found on finger rings, e.g. some of the sixth block monograms prevailing in the 6th and the cross- century bronze rings excavated at Corinth (Davidson 1952, 237 shaped/cruciform ones appearing after 54040. nos. 1860-1862, Fig. 44, Pl. 103) or at Sardis (Waldbaum 1983, Pl. 48 nos. 826, 829). Virtually all our stamps can be put in one or the 37 Invocations can be of a religious nature like the often found other of these categories. Monogram seals reappear QEOTOKE BOJQJ, but can also be slogans or acclamations sometimes in later periods, but their style is clearly like a series found in which permit a reading like different from the early ones. NIKAI J TUXJ TWN PRASINWN i.e. The fortune of the Greens triumphs! (Roueché 1989, 226-228, pl. xl) The lecture of the monograms remains wrought with 38 Lead seals are the foremost source of monograms for several difficulties. Based on monograms on seals and cap- reasons. They are relatively abundant and have two faces. While itals it is known that they most often express a one side carries a monogram, the reverse side often shows a name, with more complicated ones giving a name more legible version of the former, though even there, and espe- cially for the early periods, many questions remain to be and title. However the lack of a reverse side for the answered. stamps on ceramics increases the difficulty of pro- 39 Oikonomides 1986, 151. ducing an acceptable reading for a name. 40 Zacos 1972, 367; Oikonomides 1986, 152-153; Seibt 1993, 21. According to their shape it is possible to classify 41 Pl. 4. 42 Degeest 1993, 184. our stamps into several main groups, none of which 43 th Fig. 3e, Pl. 3. has a border, an element that appears in the 6 cen- 44 Seibt 1993, 21. tury. 45 Gassner 1997, 171, Pl. 56 no. 710.

250 4.2.1. square or rectangular stamps 4.3.1. circular stamps SA-92-UA-19046: previously published47, the same SA-96-B-253-160: Diameter 1.6 cm. Found in the monogram was found at Saraçhane48, where it was bouleuterion area layer 3. Similar to a series at excavated from a context dated c. A.D. 650-67049. Saraçhane, but different monogram found in the Even taken into account the deformations due to dif- seventh century deposit 3061. The legible letters are ferent impressions the die of Saraçhane was not the F or Q, U, A, and possibly D and/or M. same as the Sagalassos one. SA-95-UA-322-362: Incomplete stamp. Diameter SA-94-UA-26650: identical stamp as the previous 1.6 cm. From the upper agora. The only certain let- example. Found in layer 3 on the upper agora site. ter is O. The letters seem to be E, L or U, A(?), and OU. SA-98-DA1-100-151: rectangular stamp with cut 4.3.2. square stamps corners. Lettering seems to consist of A, C, P, OU First a series of four identical stamps, made with the in ligature probably, and either N or H. This same die. The stamps measure 1 by 1.1 cm. In this stamped example was excavated from layer 4 of site case the square is closed. The stamp has a parallel DA1, a domestic quarter in the valley to the north at Saraçhane and in this case the similarity is more of the main baths of the city. This particular context than superficial, in all likelihood the same die was contained a fairly large quantity of unguentarium used63. The find was made in the seventh century fragments, among them four complete rims. deposit 3064. Reading the monogram is very uncer- SA-98-DA1-100-252: the same die as for the previ- tain, due to the extreme simplification of the letters. ous entry was used for this one. Both are from the Probably E, D, O, S, and X are present. same context, indicating that the original vessels SA-92-DT-27365: From the Doric temple area. belonged to the same assemblage. More complete SA-97-B2-10666: From layer 2 of the bouleuterion sets containing a dozen or more unguentaria and one excavation. example of a much larger size have been excavated SA-93-LA-13067: From layer 4 of the shops of the at Perge53. This is an indication for the way in which lower agora. these vessels were traded and/or used. SA-96-UA-131-568: Found in the tunnel at the north end of the upper agora. 4.2.2. circular stamps SA-98-H-52-169: Same types as the preceding SA-97-B1-2654: Diameter approximately 1.6 cm. series, but different lettering. From layer 5 of the From the bouleuterion topsoil layer. The cross has heroon site northeast of the Doric temple site. a small cross turned over 45° in the middle. The main arms have a letter connected to their extremi- 4.4. Monograms with a central “M” ties. The letters are P, E, X, A or perhaps a liga- Two identical circular stamps with a large letter M ture of TR, and OU. Possibly to be read as as the main element were also found. The dies used PETROU, were it not for the presence of the letter were the same The same stamp was found at X. The same monogram, but with a different treat- ment of the letters can be found at Saraçhane55. The 56 46 Pl. 5. latter is considered residual . 47 57 Degeest 1993, 184. SA-96-UAN-203 : Diameter 1.6 cm. From the 48 Hayes 1992, Pl. 17, no. 41. UAN site layer 1S. Although somewhat fainter, 49 Hayes 1992, 203. there is no doubt that this is the same stamp as the 50 Pl. 6. 51 Pl. 21. previous one, with the same die used. 52 58 59 Pl. 22. SA-91-N-666 : previously published . Diameter 53 N. Fırat, personal communication. 1.5 cm. From the fill of the Late Hellenistic foun- 54 Pl. 7. tain courtyard (layer 4) This smudged stamp has 55 Hayes 1992, Pl. 17, no. 45-46. thicker lines and larger letters than the others found, 56 Hayes 1992, 9. 57 Pl. 8. while at the same time the rays from the cross are 58 Pl. 9. missing. The letters are B and either J or H on one 59 Degeest 1993, 183-184. arm, while the top reads in ligature OU. 60 Fig. 3d, Pl. 10. 61 Harrison 1986, 38; Hayes 1992, 9, Pl. 71, no. 28-30, 91. 62 Pl. 11. 4.3. Monograms based on a square with diagonals 63 Hayes 1992, 9, Pl. 17, no. 33. and one open side 64 Hayes 1992, 9, 91. In this group we find at Sagalassos two variants, one 65 Pl. 12. 66 Pl. 13. a circular type with fine lines and clear lettering, the 67 Pl. 14. other a simplified square one with thick lines and 68 Fig. 3c, Pl. 15. large, mostly vestigial letters. 69 Pl. 20.

251 Saraçhane in the already mentioned deposit 30, dat- the vessels during the manufacturing process and are ing from the seventh century70. not later additions like painted or incised inscrip- SA-95-UA-139-571: Excavated in the upper agora tions. The question of their meaning remains unan- site layer 5. swered. Since the nature of the contents is also SA-97-B2-10172: From the bouleuterion excavation unknown, although a number of samples for organic layer 2. residue analysis have been prepared, no conclusions can be drawn that would be valid at this point. 4.5. Monograms based on “N” shape The observation that most of the stamps are found in Of this group only one was found. chronologically closely related contexts, with no SA-96-H-90-173: From the heroon excavation layer noticeable typological evolution and sometimes the 3. Parallels are present in Saraçhane, possibly even use of the same stamps, probably indicates a closely a full match74. All these parallels were retrieved spaced distribution in time and maybe a geographi- from deposit 30 (seventh century after Christ). cally related origin, which is certainly not Sagalassian, as evidenced by the chemical and mineralogical com- 4.6. Other forms position of the ware fabric. The next step in this SA-97-B1-26-175: An oval stamp, very faint, with a respect must consist of comparing the chemical/min- star consisting of 6 rays. Found in the topsoil of the eralogical data with those from other sites when they bouleuterion. The size is 0.8 by 1.4 cm. Possibly this become available, to test the validity of the above is not a stamp at all, but the impression made by the mentioned conclusion on the one hand and to try to lower part of the stand on which the vessel was put locate the area of production of the vessels. to dry after being partially immersed in a slip solu- When considering the stylistic treatment of the tion. monograms, the picture is equivocal. Regarding for SA-96-RB-194-1076: A square stamp of 0.9 by 0.95 instance the monograms based on a square with cm, without apparent motif. Found in the Roman diagonals, one could put forward the hypothesis that baths. Were it not for its positioning on the base it there is an evolution from a more legible, thin-lined would probably not be considered a stamp at all. original with discernible lettering to a simplified, thicker-lined successor stamp. This in turn could be a marker for chronological succession, which seems 5. CONCLUSIONS more likely than a geographical differentiation as the vessels are so uniform in their aspect and finish. Several conclusions can be drawn regarding the A second instance of evolution could be seen in the finds of Late Roman – Early Byzantine Unguentaria cross-shaped monograms, where also a variant with in Sagalassos. thin lines and widely separated letters is present, From the excavated material it is clear that there is with a thicker-lined one with closer spacing. only a limited number of stamps, not only in However, here the evolution seems to run from absolute numbers of finds compared to the total square to circle. If this observation holds true, then number of shards of unguentaria, but also in the there is no relation between the form of the stamp, number of different stamps, as several instances of circular or square, and chronology as in the first case the same stamps are found. the evolution goes from round to square and in the One conclusion is that they stamped only a small second case from square to circle. Whatever the case part of the total number of vessels of this type. This may be, from the archaeological context no conclu- is corroborated by the finds at other sites and has sive evidence can be drawn either way. probably a bearing on the purpose and manner of What seems to be certain is that the whole is closely distribution/trade of the vessels. So far the evidence connected in time and likely not earlier than the at Sagalassos does not allow us to get into more sixth century and not later than the middle of the detail concerning this aspect, but comparison with seventh century AD, based not only on the available other sites may eventually lead to firm clues in this stratigraphic evidence but also on the comparative matter. material, both from unguentaria and from lead seals. Related to the previous point is that of the meaning of the stamps. They are notoriously difficult to inter- pret. Most consist simply of names, but it is far from 70 Hayes 1992, 9, 91, Pl. 16 no. 12. obvious what the correct lecture of the names should 71 Pl. 16. be. In this respect one should note that Hayes pub- 72 Pl. 17. 73 Fig. 3b, Pl. 18. lished a stamp with an inscription that reads as 74 Hayes 1992, 9, Pl. 16, no. 13-14. bishop Proklos. Here we probably have a reference 75 Pl. 19. to a producer, not a potter. The stamps are made on 76 Fig. 3a.

252 Unless more narrowly dated contexts are found it Saraçhane in Istanbul: Fifth Preliminary Report, remains too ambitious to get into more detail as far Dumbarton Oaks Papers 22, 203-216. as chronology is concerned. Hayes, J.W. 1971, A New Type of Early Christian For the moment we must accept that additional Ampulla, Annual of the British School at 66, 243-248, Pl. 36-37. information on the origin of this ware is not to be Hayes, J.W. 1992, Excavations at Saraçhane in Istanbul found at Sagalassos. Further research will have to 2. The pottery, Princeton. be done further afield, as the potential of these Catalogue of Byzantine Seals at Dumbarton Oaks and attractive vessels is certainly not yet exhausted. in the Fogg Museum of Art 1. Italy, North of the Balkans, North of the Black Sea, J. Nesbitt and N. Oikonomides, eds. Washington, D.C. 1991. REFERENCES Oikonomides, N. 1986, A Collection of Dated Byzantine Lead Seals, Washington, D.C. Anderson-Stojanovic, V. 1987, The chronology and func- Pagenstecher, R. 1913, Die Gefäße in Stein und Ton. tion of ceramic unguentaria, AJA 91, 105-122. Knochenschnitzereien (Die Griechisch-Ägyptische Bass, G. 1982, Yassı Ada 1. A seventh century Byzantine Sammlung Ernst von Sieglin 3), Leipzig. shipwreck, College Station. Peacock, D. 1977, Ceramics in Roman and Medieval Davidson, G. 1952, The minor objects (Corinth. Results Archaeology, in Pottery and Early Commerce. of excavations conducted by the American School of Characterization and Trade in Roman and Later Classical Studies at Athens 12), Princeton. Ceramics, D. Peacock, ed., London – New York – San Degeest, R. 1993, Some Late Roman Unguentaria in Francisco, 21-33. Sagalassos, in Sagalassos 2 (Acta Archaeologica Poblome, J. forthcoming, Sagalassos Red Slip Ware. Lovaniensia Monographiae 6), M. Waelkens, ed., Typology and Chronology (Studies in Eastern Leuven, 183-189. Mediterranean Archaeology 2), Turnhout. Degeest, R. forthcoming, The common Wares of Roman Riley, J.A. 1975, The pottery from the first session of Sagalassos. From the First to the Seventh Century excavation in the Caesarea hippodrome, Bulletin of (Studies in Eastern Mediterranean Archaeology 3). the American Schools of Oriental Research 218, Gassner, V. 1997, Das Südtor der Tetragonos-Agora. 25-63. Keramik und Kleinfunde (Forschungen in Ephesos Roueché, C. 1989, The late Roman and Byzantine inscrip- 13,1,1), Vienna. tions including texts from the excavations at Harrison, R. 1986, Excavations at Saraçhane in Istanbul Aphrodisias conducted by Kenan T. Erim (Journal of 1. The excavations, structures, architectural decora- Roman Studies Monographs 5), Ann Arbor. tion, small finds, coins, bones, and molluscs, Princeton Seibt, W. 1993, Zur Problematik byzantinischer Mono- – Washington, D.C. – Guildford. grammsiegel. Mit Berücksichtigung der späten Hendy, M. 1986, The coins, in Excavations at Saraçhane Entwicklung, in Studies in Byzantine Sigillography 3, in Istanbul 1. The excavations, structures, architec- N. Oikonomides, ed., Washington, D.C., 19-28. tural decoration, small finds, coins, bones, and mol- Waldbaum, J. 1983, Metalwork from Sardis: the finds luscs, R. Harrison, Princeton – Washington, D.C. – through 1974 (Archaeological Exploration of Sardis Guildford, 278-373. Monographs 8), Cambridge, Mass. -London. Hayes, J.W. 1968, A Seventh-century Pottery Group, in Zacos, G. and A. Veglery 1972, Byzantine lead seals, R.M. Harrison and N. Fıratlı, Excavations at Basel.

253 SiO2 Al2O3 TOT Fe MnO MgO CaO Na2OK2O TiO2 P2O5 CO2 L.O.I. Fabric 1 (Sagalassos ware) n= 128 mean 52.57 16.13 8.28 0.10 6.53 7.79 0.97 2.75 0.88 0.26 0.86 3.14 median 52.67 16.19 8.30 0.10 6.59 7.71 0.99 2.76 0.88 0.24 0.62 3.02 mode 53.01 16.30 8.59 0.10 6.84 7.71 0.90 2.87 0.91 0.19 0.00 2.16 range 15.11 8.19 3.64 0.08 9.46 12.96 0.81 1.57 0.31 0.62 2.35 11.58 s 2.11 1.24 0.60 0.01 1.23 1.99 0.13 0.25 0.07 0.10 0.86 1.99 V 0.04 0.08 0.07 0.14 0.19 0.25 0.13 0.09 0.07 0.36 1.00 0.63 Fabric 8 (Unguentary ware) n=9 mean 56.49 18.51 7.97 0.10 3.85 5.71 0.59 3.44 0.96 0.16 0.00 1.67 median 58.48 18.02 7.51 0.09 3.60 5.44 0.57 3.80 0.96 0.18 0.00 1.11 mode NA NA NA 0.08 NA NA 0.57 NA 0.96 0.19 0.00 NA range 16.73 2.80 2.08 0.09 4.06 13.91 0.60 2.48 0.45 0.19 0.00 4.40 s 4.88 1.01 0.77 0.03 1.12 3.84 0.17 0.82 0.12 0.06 0.00 1.26 V 0.09 0.05 0.10 0.30 0.29 0.67 0.29 0.24 0.12 0.36 0.00 0.76

Table 1. Summarized data of the chemical bulk analysis of the main Sagalassos ware and of unguentarium ware (n= number of analysed samples, s= standard deviation, V = coefficient of variation, NA = not applicable)

254 stamp identification site layer date SA-91-DT-330-132 Doric temple layer 3 6th century SA-91-N-666 Hellenistic fountain layer 4 first half of 6th century courtyard SA-92-UA-190 Upper agora topsoil 5th to 7th century+ SA-92-DT-273 Doric temple topsoil 7th century or later+ SA-93-LA-130 Lower agora layer 4 end 5th to mid 7th century SA-94-UA-266 Upper agora layer 3 5th to 7th century SA-95-UA-139-5 Upper agora layer 5 5th to 7th century SA-95-UA-148-1 Upper agora layer 4 5th to 7th century SA-95-UA-322-3 Upper agora layer 2 5th to 7th century SA-96-UAN-203 Upper agora north layer 1S 6th to mid 7th century SA-96-H-90-1 Heroon layer 3 6th century SA-96-B-253-1 Bouleuterion layer 3 6th – 7th century SA-96-RB-194-10 Roman baths layer 4 6th – 7th century SA-96-UA-131-5 Upper agora tunnel surface SA-97-B1-26-1 Bouleuterion topsoil 6th – 7th century+ SA-97-B2-101 Bouleuterion layer 2 6th – 7th century SA-97-B2-106 Bouleuterion layer 2 6th – 7th century SA-97-B1-26-2 Bouleuterion topsoil 6th – 7th century+ SA-98-DA1-100-1 Domestic area 1 layer 4 6th – 7th century SA-98-DA1-100-2 Domestic area 1 layer 4 6th – 7th century SA-98-H-52-1 Heroon layer 5 end 6th – 7th century Total: 21 stamps

Table 2. List of all the unguentarium stamps found at Sagalassos with their stratigraphical date.

255 Fig. 1. Simplified plan of Sagalassos: 1 – Site W, 2 – Doric temple, 3 – Heroon, 4 – Bouleuterion, 5 – Upper Agora, 6 – Late Hellenistic fountain, 7 – Domestic area 1, 8 – Baths, 9 – Lower agora.

256 Fig. 2. Complete unguentaria found at Sagalassos: a: SA-96-UAN-120, b: SA-95-UAN-247-1, c: SA-96-UAN-51.

257 Fig. 3. Stamp drawings: a: SA-96-RB-194-10, b: SA-96-H-90-1, c: SA-96-UA-131-5, d: SA-96-B-253-1, e: SA-95- UA-148-1.

258 Plate 1. SA-94-UAN-198. Plate 2. SA-94-UA-349.

Plate 3. SA-95-UA-148. Plate 4. SA-91-DT-330-132.

Plate 5. SA-92-UA-190. Plate 6. SA-94-UA-266.

259 Plate 7. SA-97-B1-26. Plate 8. SA-96-UAN-203.

Plate 9. SA-91-N-666. Plate 10. SA-96-B-253-1.

Plate 11. SA-95-UA-322-3. Plate 12. SA-92-DT-273.

260 Plate 13. SA-97-B2-106. Plate 14. SA-93-LA-130.

Plate 15. SA-96-UA-131-5. Plate 16. SA-95-UA-139-5.

Plate 17. SA-97-B2-101. Plate 18. SA-96-H-90-1.

261 Plate 19. SA-97-BA-26-1. Plate 20. SA-98-H-52-1.

Plate 21. SA-98-DA1-100-1. Plate 22. SA-98-DA1-100-2.

262