CHAPTER IV: THE CERAMIC SEQUENCE. IV.1 The sample

Transcription

1 CHAPTER IV: THE CERAMIC SEQUENCE Pottery is an archaeologist s delight. D.P. Braun (1983: 108). This chapter will deal with the presentation, description and chronological attribution of the Late Bronze Age ceramics from Tell Sabi Abyad. I will provide a general overview of their physical appearance, as well as some indications of how these ceramics compare to those of other contemporary sites in the region. I will also discuss whether, and how, the ceramic corpus changes from level to level. I will not discuss the reasons for ceramic change. Pottery production is often said to be a very conservative craft, not very open to changes. Over time, however, change did occur. The reasons behind changes in materials, technology, shape, style or decoration are many, and may include a changing organization of the craft, a change in the tastes and demands of the market or consumer group, more or less involvement of authorities in production, a change of diets and food habits, coincidental changes over time, and so on (e.g. Rice 1987: ; McGovern 1989; Gosselain 1998; Eerkens and Lipo 2005). Chapters V and VI will discuss some of the possible mechanisms related to changes in the ceramic production at Sabi Abyad, mainly with regard to production organization and function or use. IV.1 The sample The ceramic sequence of Sabi Abyad is based on all sherds from undisturbed, unmixed contexts only. As has been described in Chapters II and III, the levels distinguished in the stratigraphy of Sabi Abyad represent coherent building phases. For the chronological picture of the ceramics only material securely ascribed to a single level and undisturbed contexts are used. Material from mixed levels and contexts is not included unless a special shape or feature is involved. It is then illustrated in the catalogue for reasons of comparison with other sites, but this material is kept out of the discussion of chronology in this chapter. At the moment of writing, the stratigraphical information available could not yet be used to discern between in situ contexts and other contexts, so this information is not included here. There are two other reasons not to base the chronological sequence on in situ floor contexts only, but to include secondary fills of streets, rooms, pits, etc. as well. First, the inclusion of these contexts increases the sample size, which is especially important for levels 7 and 3. Second, the inclusion of different contexts reduces a functional bias in the sample and therefore reduces the risk that differences between assemblages are more related to the use of space than to chronological position. It is expected that in each level a small portion of the material will be intrusive from other levels even if no disturbances were noted. Since, as we will see, the Middle Assyrian ceramics show a lot of similarities between levels, it was impossible to filter out these intrusive sherds. Prehistoric sherds included in the mud bricks made by the Assyrians were easily recognizable and already taken out of the sample before description (this material was partly published in Nieuwenhuyse 1997). Islamic and Hellenistic / Roman pottery derived from levels 2 and 1 was often recognized when single fragments were intrusive in earlier levels, although this proved difficult when shapes and/or wares resembled the Middle Assyrian pottery. These ceramics will be published elsewhere (Duistermaat and De Jong in prep.). A total number of 13,050 diagnostic sherds that are securely attributed to one of the five Late Bronze Age levels at Sabi Abyad was used for this chronological analysis. This is two-thirds of the total amount of sherds described. The distribution of these sherds over the different levels is as follows: 59

2 Level No. % of total % of secure prov Total secure prov Mixed levels No level assigned Total not included Total Table IV.1: Amount of sherds from secure contexts. It is clear that the differences in sample size between levels 6 to 4 on the one hand and levels 7 and 3 on the other are very big. The differences in sample size are caused by the limited exposure, small settlement size and/or limited preservation of the settlement in levels 7 and 3 compared to the extensive occupation in levels 6 to 4. In addition, as was described in Chapter II, the original focus of data collection did not aim at a chronological cross-section but rather at a complete assessment of one single level (5), resulting in an overproportional amount of described diagnostics for this level. The architectural and stratigraphical sequence has been discussed in Chapter III. Level 7 predates the presence of the Middle Assyrian administration at the site, and is most probably of Mitanni date. Levels 6 to 4 comprise the main occupation at the site, when the Middle Assyrian grand viziers ran a dunnu estate at the site. In level 3 some isolated houses are built on the site, but they still date from the Late Bronze Age. IV.2 Research methods The ceramics are first described per level, characterizing the different aspects of the pottery. These descriptions include information about clay and inclusions, shaping techniques, rim and base shapes and types, surface treatment and decoration, and firing. The illustrations in figures IV.1 IV.120 are chosen to form a representative collection of shapes and decorations for each level. Next, the continuity or changes between levels are discussed. The relative dating of the Sabi Abyad material is based on comparisons with published material from other sites. The elaborate and detailed discussion of the Middle Assyrian ceramics from Tell Sheikh Hamad / Dur Katlimmu (Pfälzner 1995) concentrated mainly on the ceramic sequence at the site and the comparisons with other sites. One of the most important results of Pfälzner s study was the definition of a ceramic chronology for the Mitanni and Middle Assyrian periods, the division of the Middle Assyrian period in three ceramic stages on the basis of shapes and wares ( mittel Assyrisch I, II, III, here called Middle Assyrian (MA) I, II, III) and the study of the geographical spread of these ceramic traditions in Northern Mesopotamia. The present study builds on these foundations. In presenting the relative chronology references are made to Pfälzner s Middle Assyrian ceramic phases. Comparisons with material from other sites are included only for important sites published after 1995, 15 or for special cases in which a detailed comparison is more informative than a general reference to Pfälzner s framework. Comparisons with the Sheikh Hamad material are so numerous that mentioning them all in the text with each shape would unnecessarily burden the text. They are usually listed with the illustrations in figs. IV Comparisons with other sites, especially non-middle Assyrian sites, are mentioned in the text and with the illustrations. Finally, several absolute dates obtained from cuneiform texts and 14 C samples are presented. IV.3 The Mitanni ceramics 15 Most important are the publications of ceramics from Tell Brak (Oates et al. 1997) and Tell Rimah (Postgate et al. 1997). New excavations at Tell Taban (Ohnuma et al. 1998, 2000, Ohnuma and Numoto 2001), Ziyaret Tepe (Matney et al. 2003, 2002, Matney and Rainville 2005) and Giricano Höyük (Schachner 2003, 2002) now contribute especially to our knowledge of the end of the Late Bronze Age. The ceramic assemblages from these sites still await full publication, as is the case with Tell Barri and Tell Chuera (Boesze in prep.). 60

3 Level 7 (figs.iv.1 IV.11). A total of 344 diagnostic sherds could be assigned to the first Late Bronze Age occupation at Tell Sabi Abyad. As we have seen in Chapter III, the settlement at that time consisted of a tower and some minor buildings surrounding it, most probably representing a dimtu (Wiggermann 2000: 184, cf. also Koliński 2001). Level 7 has only been reached within the limits of the Middle Assyrian tower walls, where excavation continued inside the Middle Assyrian rooms until the prehistoric tell was reached, and at isolated spots outside the tower. The settlement was clearly much smaller than the later Middle Assyrian occupation. No sub-levels were distinguished in the stratigraphy for level 7. Clay and inclusions In level 7 the majority of the ceramics has organic (chaff) inclusions (wares H, I, J, together 84.6%). Almost 15% has mineral inclusions only. Coarse wares D, E and F, with large mineral inclusions, are represented by approximately 6% of all cases. Wares D and E are cooking wares, and exclusively consist of closed pots (fig. IV.5). Fine wares A, B and C occur in 9% of all cases. Two sherds show unidentified inclusions not classified in one of the existing groups. One (fig. IV.6.h) has a rather coarse mixture of possibly grog and calcite inclusions. The other, a handle fragment (fig. IV.9.j), includes calcite and shiny deep-black particles with a molten appearance. 16 A base fragment (fig. IV.10.j) also has some black glimmering particles, apart from calcium and fine sand (B) inclusions. For two sherds in this group it was noted that the amount of inclusions is very large. About half (5) of the sherds in ware group H has relatively many calcite inclusions. These include two bases and 3 rims, 17 of which at least two are singular and untypical shapes at Sabi Abyad. In ware group I, the largest group, 5 sherds were noted to have relatively many calcite inclusions. 18 The results of the thin-section analyses (Appendix D) have shown that the occurrence of large amounts of calcite aggregates is typical for the level 7 pottery. Two sherds have many sand inclusions and two have very few inclusions. 19 These sherds come from rounded bowls, jars, and bases. In group J, one flat base was noted to have large orange and black particles next to the organic and sand inclusions, and one unidentified rim sherd was noted to have very few inclusions. So, in total, the inclusions could not be identified in 2 sherds, while in 19 sherds (5.5%) the inclusions are a bit different from the rest of their group, mainly concerning the amount of inclusions. A group of 18 sherds from level 7 was selected for thin-section analysis of the fabric. These sherds come from ware groups A, B, C, D, E, F and I. 20 The results of the thin-section analyses showed that most level 7 sherds analysed are made of a slightly different clay than the locally produced vessels from later levels (Appendix D). A rather large proportion of the pottery in level 7 seems to have been imported to the site from sites nearby or further away. 16 Both of these sherds were not further analysed. 17 Fig. IV.6.a; fig. IV.4.d; fig. IV.8.a. 18 Fig. IV.2.j; fig. IV.8.i (= thin-section sample 42, thin-section ware group A1a), fig. IV.3.d (= thin-section sample 39, thin-section ware group B2 and not local to the Balikh). 19 Fig. IV.3.a. 20 Sample nos. 02, 05, 32-44, in Appendix D and Table V.2. 61

4 No. % A Calcium B Calcium and sand C Fine sand X Subtotal 9.0 D Coarse sand E Coarse calcite F Shell Subtotal 5.8 H Organic inclusions and calcium I Organic inclusions and calcium and sand J Organic inclusions and fine sand Y Subtotal 84.6? Unidentified inclusions Total Table IV.2: Inclusions in level 7 Shapes When looking at the general shape classes, we see that more than one third of all fragments belongs to bowls, while about 10% comes from pots. If we look at rim sherds only, about two-thirds of all rims are from bowls. Especially the cooking pots are numerous in the group of pots. No. % % of rims Bowls Pots Jars Goblets Bottles Loose base fragments Diagnostic bodysherds and others Total Table IV.3: Shapes in level 7 A total number of 36 different rim types were distinguished in the assemblage (Table IV.9). The top ten of the most popular rim types is made up of the following types: 111 (14.9%), 122 (11.1%), 212 (9.1%), 123 (7.7%), 125 (6.3%), 322 (6.3%), 131 (5.3%), 132 (4.3%), 315 (3.4%), 411 (3.4%). These ten types together form 71.6% of the whole assemblage. Five rim types occur only in level 7 and no longer in later levels (117, 1213, 312 (this particular rim shape with incision, fig. IV.8.a) 1214 (this particular shape) and 319). They are shaded in table IV.9. For three types (129, 2211 and 326), all but one example occur in level 7. Types that do occur in later levels have then often slightly different. There are very few large storage vessels from level 7. No. % Carinated bowls % Rounded bowls % Straight-sided bowls % Deep bowls 8 6.3% Total % Table IV.4: Different bowl shapes in level 7 Bowls The group of rounded bowls is largest in level 7, including almost half of all bowls, while the group of deep bowls is comparatively small. The most popular shape among bowls is the carinated bowl type 111 (n=31, or 14.5% of all rims, 24.6% of all bowl rims). The smaller examples of this type (111a, n=13, six are completely preserved, cf. fig. IV.1.a, b) are identical to the Middle Assyrian small carinated bowls in later 62

5 levels at Sabi Abyad and at other sites, with flat string-cut bases and simple rounded rims. The larger carinated bowls (fig. IV.1.e-i) are a bit different from their later counterparts (and not all are locally produced; cf. Appendix D sample 41, fig. IV.1.e). Their rims are slightly thickened on the outside, and the carination on the outside is relatively shallow. The bowls are rather shallow and the inside is smoothly curved and not carinated. One bowl is exceptional (fig. IV.1.c, thin-section sample 48 in Appendix D, not from the Balikh Valley), showing a sharp carination and a thickened rim, a ring base and a carefully burnished surface. Carinated bowls with a straight, rather high part of the wall above the carination (type 113) occur in small numbers in level 7. One of them is decorated with some low thin ridges on the part above the carination. At other sites bowls like these have occurred since the beginning of the Late Bronze Age. However, in the earlier part of the LBA the rims of these bowls are more often squarish, hammer-headed, while in the later part of the Mitanni period rims are simply rounded as they are here (cf. Oates et al. 1997, Duistermaat in prep.). The level 7 type 113 bowls (fig. IV.1.j-l) are buff and reddish in colour and simply smoothed, not burnished, but their shapes are comparable to the grey or other-coloured (red, buff, brown) burnished bowls of the Mitanni period. We will see more examples of these bowls in later levels. Carinated bowls with their walls above the carination turning inward (type 117) are characteristic of the later Mitanni period, and occur only in level 7 at Sabi Abyad. The rims are folded over (fig. IV.1.m, n). A thin-section sample (no. 32 in Appendix D) of the bowl in fig. IV.1.n proved that it was made locally in the Balikh Valley. Rounded bowls are a characteristic of the level 7 assemblage. Eight different rim types of rounded bowls exist, but most rounded bowls belong to rim types 122, 123 and 125 (together n=52, 24.3% of all rims, 41.3% of all bowl rims). Only one of them is burnished (fig. IV.2.e), and nine (17.3% of types 122, 123 and 125 together, fig. IV.2.b, e, j-l, fig. IV.3.m, u) show the characteristic Mitanni dark-red or red-brown painted stripes on the rim, inside or on both in and outside. 21 At Tell Bderi about 21% of conical bowls has a red painted band along the rim (Pfälzner 1995: abb. 79a, Ware 8). At Tell Brak the number of red-rimmed bowls is steadily increasing from levels 5 to 2, but it is not clear what their share is among rounded bowls. It is, however, clear that at all sites unpainted bowls are more numerous. Also at Brak burnishing of red-rimmed bowls seems to be a characteristic of the earlier Mitanni levels, while in the later levels burnishing almost disappears (Oates et al. 1997: 73). These characteristic red-rimmed bowls appear no more in the later levels at Sabi Abyad. Rounded bowls of the same shape, however, are also part of the Middle Assyrian assemblages (see below). A thin-section sample (no. 44 in Appendix D) of one of the smaller type 122 bowls shows that it was most probably made in the Balikh Valley. Another connection with the Middle Assyrian assemblage at Sabi Abyad is formed by the occurrence of straight-sided bowls type 131 and 132 in level 7. Like the small carinated bowls type 111a, these straight-sided bowls are identical to their Middle Assyrian counterparts. Twenty bowls of these two types occur in level 7 (9.3% of all rims, 15.9 % of all bowl rims). No illustrations were available for type 131 bowls from this level, the most popular among straight-sided bowls in level 7. Type 132 is illustrated in figs. IV.4.e, f. A thin-section sample from a type 132 bowl (sample no. 02 in Appendix D) suggested that it was made locally in the Balikh Valley. Deep bowls (type 145) in level 7 have a rather rounded shape, and are decorated with a plain applied horizontal band (fig. IV.4.j, k). Pots Closed pots with types 211 and 212 in level 7 are mainly pots made of cooking ware, with coarse mineral inclusions. Of the twenty pots of these types, only three are made of wares H, I and J, and they are slightly different in shape as well (fig. IV.6.a, b), while the remaining 17 are made of wares D and E. Wares D and E are exclusively used for closed pots. 22 Thin-section analyses of four cooking pots (sample nos. 33, 35, 36, 46 in Appendix D, fig. IV.5.a, e, i, l) show that two could come from the Balikh region while two (nos. 33, 36, fig. IV.5.a, i) are imported from further away. All four are different from each other, pointing to several different workshops. All show slightly incurving rims that are thickened on the outside (fig. IV.5). The 21 One plain type 123 bowl (fig. IV.3.d) was not of local origin, cf. Appendix D sample Two other diagnostics had coarse mineral inclusions: a rough platter (fig. IV.7.j) and a body sherd, probably from a cooking pot (fig. IV.9.g). 63

6 number of cooking pots is relatively high (9.3% of all rims, 57.1 % of all pot rims), especially considering that more than half of them (n=12) were found in the tower. Only two fragments of large storage vessels (type 213, fig. IV.6.c, d) were found in room 1 of the tower. These are most probably handmade. One of them (fig. IV.6.c) shows numerous carelessly applied blackish lines on the shoulder (see below). Thirteen rims come from pots with an open shape and straight, vertical walls (6% of all rims, 37.1% of all pot rims). Five have squarish, almost hammer-shaped rims that will become more popular in the Middle Assyrian levels, although the rims in level 7 are generally a bit plumper (fig. IV.6.e). Five examples of the well-known small cylindrical pots called grain measures (type 225) have been found in level 7. These pots with thin walls and bent or folded thin rims occur as early as in Old Babylonian or late Middle Bronze Age times in Northern Mesopotamia, and we will see that at Sabi Abyad they continue into the Middle Assyrian levels. Often they are decorated. The decoration, more than the shape, is characteristic for the period as we shall see below. In level 7 one type 225 pot is decorated with Nuzi-style decoration (fig. IV.7.a), two with parallel horizontal red bands and stripes on the rim (fig. IV.6.b, d), and one with applied bands of clay with finger impressions (fig. IV.6.e). One pot is left undecorated (fig. IV.6.c). Three pots have slightly thickened and smooth rims with an oval section (type 2211). Only four fragments of this type were found at Sabi Abyad, and all except one were found in level 7 (the fourth example came from level 6). This shape therefore seems to be characteristic for level 7. The thin-section analysis made of one of these pots (fig. IV.7.h, sample 34 in Appendix D) proves that it is a cooking pot with crushed shell inclusions, possibly made in the region. Jars A total of 39 rim sherds belong to jars (18.2% of all rims). In level 7 a majority of jars has a clear neck (n=21, 53.8% of all jar rims), while a smaller number has a ribbon rim sitting directly on the shoulder of the vessel without a clear neck (n=17, 43.6% of all jars). The latter group becomes more popular in Middle Assyrian times, as we will see below. Jars with handles do not occur in level 7. Among jars with a neck there are seven different rim types. A characteristic shape is the small, thinwalled jar with Nuzi style decoration (type 314, fig. IV.8.b). Thin-section analyses (sample 40 in Appendix D) showed that it was most probably locally made. Most popular is the jar with a squarish or flattened rim, type 315 (fig. IV.8.d-i). Sometimes these jars show a raised ridge along the transition from shoulder to neck. Thin-section analyses of four type 315 jars (samples 05, 37 (fig. IV.8.g), 42 (fig. IV.8.i) and 47 (fig. IV.8.d) in Appendix D) indicate that one (no. 37) was not locally produced but possibly came from the Euphrates area. Jar type 319, with a strongly outward-bent and slightly pointed rim (fig. IV.8.k-m), occurs only in level 7 and may be characteristic of the period. In the group of jars with a ribbon rim the oval rims with a vertical outer side (type 322) are the most popular (n=13). Although they are very similar to the later Middle Assyrian typical jar rims, they generally tend to be a little more squarish in section (fig. IV.9.a, b). Rims with concave sides (type 321) or slanting rims (type 323) each occur only once. A special shape is represented by two rims of jars with strongly inward-sloping, pointed rims (type 326, fig. IV.9.c, d). Only one other example of this shape occurred at Sabi Abyad, in level 5. And so this type may be characteristic for level 7 as well. Goblets Ten rims and eight bases (fig. IV.10.a-f) belong to goblets. Both the V-shaped goblet as well as the S-shaped goblet are present in level 7, and the rims are very similar to their later Middle Assyrian counterparts. Shouldered goblets, with a sharp transition from a globular body to a straight neck, do not occur at Sabi Abyad. However, they are typical for the Mitanni period at, for example, Tell Brak (up to level 3, cf. Oates et al. 1997: 71). At Brak in level 2 they no longer occur (ibid.). At Sabi Abyad V-shaped goblets (type 411) seem to be more popular than S-shaped goblets (type 421). Straight-sided, U-shaped goblets are at Sabi Abyad only represented by two pedestal base fragments. Pedestal bases on goblets are characteristic of this period, and occur no more in Middle Assyrian times. One of these is painted in Nuzi-style and was most probably produced locally (fig. IV.10.e, thin-section sample no. 43 in Appendix D). As to goblet bases, the 64

7 almost complete absence of the typical Middle Assyrian nipple base is interesting. All but one goblet have smaller or larger knob-shaped bases. Base type No. %? Total Table IV.5: Base types in level 7, loose bases and complete shapes included. Bases The greater majority of bases in level 7 is of the ring-base kind (type 741, fig. IV.11). Small and larger bowls, pots and jars all show ring bases. In general, the ring is well-shaped and rather pronounced in comparison with later Middle Assyrian shapes, and often the ring is a bit squarish in section. Flat bases occur a lot as well, with all shapes and sizes (fig. IV.10.g-m). Many of them are simply string-cut, but a rather large proportion is carefully scraped and flattened after string-cutting. In three cases holes were made in the flat base, twice before firing (fig. IV.10.h, k) and once after firing (fig. IV.10.g). Body sherds Several diagnostic body sherds (fig. IV.9.e-m) show interesting decorations or shapes. One fragment of a bottle neck with a handle and a body fragment of a small jar have a dark-red slip and vertical burnish (fig. IV.9.l, m, see also below). One other body sherd is vertically burnished. One body fragment shows impressions of a rope made before firing, probably during the shaping process, while another shows an impression or potters mark (fig. IV.9.i, see also Chapter V). Among the painted body fragments, one fragment has Nuzi-style decoration (fig. IV.9.e), one has dark-brown painted horizontal bands (fig. IV.9.f), and one cooking pot fragment has brown and orange painted blobs (fig. IV.9.g). The irregular crayon lines described above occur as well. Other body sherds show horizontally applied ridges (fig. IV.9.k) or incised lines. Surface treatment and decoration In level 7, only two sherds show a dark-red slip (see above, and fig. IV.9.l, m). One of them, as is clear from the thin-section analysis, was not produced at Sabi Abyad but came from further away (fig. IV.9.m, Appendix D sample 38; moreover, raw materials for red slips are not available in the Jezira region). At Brak slipped sherds seem to date mainly from the later levels (level 2, see Oates et al. 1997: fig. 110 and fig. 204), but in small numbers (up to 15 sherds in level 2), and it is also suggested there that red-slipped pottery is not local but comes from Anatolian regions (ibid.: 74). All other pottery at Sabi Abyad is not slipped. However, the surfaces are generally carefully smoothed and some are very smooth indeed. A total of 12 diagnostics, or 3.5% of all diagnostics, is burnished. These include a carinated bowl (type 111 fig. IV.1.c), a rounded bowl (type 122, fig. IV.2.e), five cooking pots (type 212, fig. IV.5.j-n) and a grain measure (type 255, fig. IV.7.d), as well as a bottle fragment (fig. IV.9.l). Apart from the cooking pots, where the burnishing was perhaps done for functional reasons, burnishing seems to be rare in level 7. Traces of scraping are rare, since they were mostly obliterated by very careful smoothing during the shaping process. Most pottery in level 7 is undecorated. However, 31 diagnostics show different types of decoration (9% of all diagnostics). Most popular is the characteristic painted red band decoration along the rims of bowls or on the body of type 225 pots. On bowls this type of decoration seems to be typical for the Mitanni period, but on goblets and small jars the style continues well into the Middle Assyrian period and even into the Early Iron Age (see below, and also Duistermaat in prep, Pfälzner 1995: 46, 239, Postgate et al. 1997: 53, 54). The very easily recognizable Nuzi-style decoration occurs in level 7 as well. It consists of geometrical motifs (mainly mmm-shapes and spiralling bands, but perhaps also birds? Cf. fig. IV.7.a) in whitish paint on a background 65

8 of dark-red or dark-brown, painted bands. It occurs on a grain measure (type 225), a small jar and a goblet base as well as on a body sherd probably from a small jar. This type of decoration seems to be chronologically limited to the Mitanni period, until the 13 th century BC (cf. Postgate et al. 1997: 54, 55). Nuzi decoration catches the eye, but it must be remembered that at all sites it occurs in very small numbers only. At Brak, at the height of the occurrence of Nuzi style decoration in level 2, still only 38 sherds show this kind of decoration (Oates et al. 1997: figure 92). The presence of several sherds at the small settlement of Sabi Abyad, all of them made from local materials (samples 40, 43, 45 in Appendix D, figs. IV.8.b, IV.10.e, IV.98.c), is therefore interesting. No. % Painted horizontal red bands Painted Nuzi style decoration Painted crayon lines Painted blobs Incised horizontal line(s) Applied horizontal band(s) Total decorated Table IV.6: Decoration in level 7. A peculiar kind of decoration is formed by parallel or crossing, carelessly applied lines of a blackish material. It looks as if they were made by lightly rubbing a crayon over the surface, or by some kind of material rubbing the surface. Perhaps they were caused by the use of the vessel (e.g. by ropes or bands tied to the vessel?) and are not really a decoration. These traces were found on a large storage vessel, a jar base (fig. IV.6.c, IV.11.j) and several body sherds (fig. IV.9.h). Incised horizontal lines occur in small numbers, and horizontal applied ridges or bands are applied to a large bowl, a jar and a grain measure (fig. IV.4.j, IV.7.e, IV.8.d) as well as to several body sherds (fig. IV.9.k). Firing In level 7 the majority (84%) of the ceramics was well fired at medium temperatures. More than 89% of these are made of wares with organic inclusions (wares H, I, J), which is in line with the general predominance of these ware groups (see above). Only 23 sherds have light cream or greenish (n=21) surface and core colours and were fired at higher temperatures, but only one body sherd had been fired at very high temperatures, causing the clay body to sinter % of high-fired sherds are made of fine wares (wares A, B, C). A total of 32 diagnostics has dark-brown surface colours and brownish or grey core colours, and was fired at a lower temperature. Almost half of these (n=14) are pots made of wares D and E, with coarse mineral inclusions, probably cooking pots. The others are mainly base fragments (both types 731 and 741), and six rims of bowls. Fine-ware sherds (wares A, B, C) were never fired at low temperatures. No. % High Medium Low Total Table IV.7: Firing temperatures in level 7. Lime spalling occurred in 9 sherds (2.6%), all made of ware I except one (ware B). All were fired at medium temperatures, except P93-30 (fig. IV.1.m), which was fired at a higher temperature. A small majority (64.5%) of the ceramics was fired in completely oxidizing kiln circumstances. Fine-ware sherds (wares A, B, C) were fired only in completely oxidizing circumstances. Surface colours range from reddish to orange and buff. Characteristic for the period, however, are the relatively large amounts of sherds with deep orange to orange-red surface colours and grey cores, made of wares with organic inclusions (wares 23 No kiln wasters (or any other indication of local pottery production) were found in level 7. 66

9 H, I, J) (cf. Duistermaat in prep., Oates et al. 1997: 157, appearing from level 5 onwards at Brak). In level 7 approximately 75% of the ceramics fired in incompletely oxidizing circumstances shows this characteristic. 24 Shapes fired in a reducing atmosphere, with completely grey colours, include several bowl types (types 111, 123, 131, 145), pots (212, 2211), jars (315, 322) and loose bases (731). So-called grey-burnished bowls, which are rather common in earlier Mitanni levels at other sites and which also appear in Middle Assyrian Sabi Abyad, are absent from level 7 (as they are from Brak level 2; cf. Duistermaat in prep., and Oates et al. 1997). No. % Oxidizing Incompletely oxidizing (grey core) Reducing Total Table IV.8: Firing atmospheres in level The thin-section analyses (Appendix D) seem to suggest that the pottery in level 7 was generally fired at slightly lower temperatures than the vessels from later levels, especially level 4. The lower temperatures may therefore account for the darker reddish and orange surface colours. 67

10 Table IV.9: Level 7, proportions of ware per type, frequencies of types. Shaded types occur only in level 7? A B C D E F H I J Total % of rims % 6.5% 3.2% 90.3% 100.0% % % % % 4.3% 8.7% 56.5% 30.4% 100.0% % 12.5% 75.0% 12.5% 100.0% % 61.5% 38.5% 100.0% % % % % % % 77.8% 22.2% 100.0% % % 66.7% 33.3% 100.0% % All bowls % 2.4% 1.6% 3.2% 79.4% 13.5% 100.0% % % 31.6% 52.6% 5.3% 5.3% 5.3% 100.0% % % % 33.3% 33.3% 33.3% 100.0% % 225? % % 33.3% 66.7% 100.0% All pots % 2.8% 16.7% 30.6% 2.8% 2.8% 36.1% 8.3% 100.0% % % 312? % 68

11 ? A B C D E F H I J Total % of rims % % % 14.3% 85.7% 100.0% 315? % % % % % % % 50.0% 50.0% 100.0% All jars % 2.6% 2.6% 5.1% 89.7% 100.0% % 85.7% 14.3% 100.0% % 33.3% 66.7% 100.0% All rims % % of bases % % % 14.3% 14.3% 57.1% 14.3% 100.0% % 2.0% 6.0% 6.0% 4.0% 70.0% 12.0% 100.0% % 2.7% 4.1% 81.1% 12.2% 100.0% All bases % 1.5% 8.0% 2.9% 3.6% 72.3% 11.7% 100.0% Loose bases % 7.0% 0.9% 4.4% 72.8% 13.2% 100.0% Bottles Trays Miniatures % 50.0% 100.0% Other diagn % 6.7% 6.7% 66.7% 13.3% 100.0% Total diagn % 0.6% 6.1% 2.3% 2.0% 3.5% 0.3% 3.5% 70.3% 10.8% 100.0% 69

12 IV.4 The Middle Assyrian ceramics Level 6 (figs. IV.12 IV.35). In total 2295 diagnostics in the database are securely attributed to level 6. However, part of this collection was not described by myself. In Chapter II and in Appendix D it is argued that the differences between my own descriptions and those of others do not have a basis in the petrography of the sherds. However, especially with the identification of fine-sand inclusions, the data looks different in my database and that of others. To keep the information on the level 6 corpus presented here (and based on databases compiled by others) comparable to that of the other levels, the sections discussing the clay and inclusions and the firing circumstances will be based on database information from 842 diagnostics described by myself only. Because ware groups X and Y are petrographically sound groupings that can be made with both databases, the data on these larger groups has been included for all sherds from level 6 irrespective of who described them (see also Chapter II for more information). The level 6 occupation represents the first Middle Assyrian settlement at the site. After a hiatus following the level 7 occupation, the inhabitants rebuilt the level 7 tower, dug a square moat and built the first dunnu. The level 6 occupation at the site was the largest in the Late Bronze Age history of the site (cf. Chapter III). Clay and inclusions As in level 7 the majority of the ceramics has organic (chaff) inclusions (wares H, I, J, together 90.7%). The amount is a bit higher than in level 7, mainly due to the now only very rare occurences of wares with coarse mineral inclusions. 25 As in level 7 fine wares (A, B, C) are used in 9% of all cases. In level 6 the clays that were used seem to be a little more homogeneous in composition than in level 7. Now only 2.2% of all sherds were noted to have more (calcium or sand) inclusions than the average. This is especially true for wares H (around 10% contain many calcium inclusions) and C (approximately 10% has many sand inclusions). The only coarse-ware sherd contains organic inclusions as well as coarse calcite consisting of very coarse, large (6 mm) angular white and grey calcite/stone particles. It is not clear what the original vessel shape of this coarse body sherd was. Only one sherd from level 6 was analysed in thin section. 26 This is a sherd from a grey-burnished carinated bowl with straight walls above the carination (type 113), with impressed circle decoration (Fig. IV.17.b). This bowl was most probably made locally (see Appendix D). No. % A Calcium B Calcium and sand C Fine sand X subtotal (n=242; 10.5%) 9.1 H Organic inclusions and calcium I Organic inclusions and calcium and sand J Organic inclusions and fine sand Y subtotal (n=2040; 88.9%) 90.7 D Coarse sand (1) E Coarse calcite (2) K Organic inclusions and coarse sand (2) L Organic inclusions and coarse calcite 1 (2) 0.1 Coarse total (n=7; 0.3%) Total % Table IV.10: Inclusions in level In the collection described by myself only one coarse-ware sherd was found (ware L, having both coarse calcite and organic inclusions, 0.1%), while the collection described by others contains 3 examples of wares D and E (0.1 %) and 2 examples of ware K (0.1%). 26 M : P (sample no. 16 in Appendix D); a sample of 15 sherds was selected for thin-section and chemical analysis by Ewout Koek, coming mostly from level 6, described in 2005 and not included in this thesis. These sherds will be analysed by Ildiko Boesze in the framework of her thesis on the Middle Assyrian ceramics from Tell Chuera (Boesze in prep.). 70

13 Values for the corpus described by the author (values for all level 6 sherds are added between brackets). Shapes In level 6 two-thirds of all rims belong to bowls, while the group of jars is the second-largest shape group. The number of pots has decreased when compared to level 7, mainly due to the almost complete absence of cooking pots in level 6. The number of goblets and pot stands has increased. No. % % of rims Bowls Pots Jars Goblets Pot stands Strainers Bottles Trays Loose base fragments Diagnostic bodysherds and others Total % 100% Table IV.11: Shapes in level 6. The number of different rim types increased sharply to 52 different types. This is probably mainly due to the sharp increase of the sample size rather than to an increase in diversity, since many of the new types occur only once or twice (cf. Table V.20: the diversity actually decreased in level 6). There are 23 new types in level 6 that did not occur in level 7 (bold underlined in Table IV.17). Moreover, ten types did occur before in level 7, but either had a slightly different shape from their counterparts in level 6, or occurred in such small numbers compared to the later levels that they may be considered intrusive in level 7. This is the case for types 113, 121, 145, 211, 221, 222, 311, 312, 321 and 323. One rim type, 1416, occurs only in level 6 (shaded in Table IV.17). For rim types 118, 1217, 1410, 1412, 1413, 2110, 227 the majority comes from level 6, while isolated other sherds of the same types occur either in mixed contexts or in contexts to which no level has been assigned yet, or in other Middle Assyrian levels (5 or 4). In level 6 the top ten of the most popular rim types is made up of the following types: 111 (41.8%), 131 (7.5%), 322 (6.9%), 311 (4.4%), 411 (3.9%), 421 (3.2%), 221 (3.1%), 611 (2.8%), 132 (2.3%), 222 (2.1%). These ten types, all typical of the Middle Assyrian period, together form 78% of all rims in level 6. Notably, rounded bowls (types 122, 123, 125), so popular in level 7, still occur in level 6 but are not represented in the top ten anymore. Instead, carinated and straight-sided bowls (types 111, 131, 132), jars with simple rims and with rolled rims (311, 322), and goblets (411, 421) are now the most popular shapes. No. % Carinated bowls % Rounded bowls % Straight-sided bowls % Deep bowls % Total % Table IV.12: Different bowl shapes in level 6. Bowls In the Middle Assyrian levels, starting from level 6, carinated bowls become far more popular than rounded bowls. Now almost 70% of all bowls is of the carinated type. The most popular shape among bowls (n=681, or 41.5% of all rims, 63.1% of all bowl rims, 91.5% of all carinated bowls) is the carinated bowl type 111. This shape occurs in three distinct size groups: 111a (small bowls, n=178; 26.1% of type 111), b (middle-sized bowls, n=162; 23.8% of type 111) and 111c (larger bowls, n=322; 47.3% of type 111). 28 The 27 This subgroup is similar to the Knickwandnäpfe at Tell Sheikh Hamad, cf. Pfälzner These two subgroups are similar to the Standard Knickwandschalen at Tell Sheikh Hamad; cf. Pfälzner

14 small type 111a bowls are characterized in level 6 by rounded or often pointed or thin rims, and rather straight walls above the carination (figs. IV.12, 13). Concave walls above the carination do occur, but in small numbers only (fig. IV.12.ak-av). The carinations themselves are generally not very sharp, and are often placed in the lower half of the bowl. Squarish rims are rare in these bowls, as are rims thickened on the outside. The middle-sized and larger bowls 111b and 111c sometimes have rounded rims, but more often rims that are squarish or triangular in section (fig. IV.14.p-ad). A few rims of larger bowls have rather pronounced but rounded carination and a rim slightly thickened on the inside (fig. IV.16.g-h, m, comparable to a rim from Tell Hammam et-turkman VIIIB, Smit 1988: pl. 154 no. 8). Carinated bowls type 112 are basically a variant of the more common type 111 bowls. The rims of the type 112 bowls are strongly bent over outwards (fig. IV.16.o-q). They are new in level 6, just like two single examples of types 114 (an almost rounded bowl with the rim bent over outwards in a sharp carination) and 118 (a carinated bowl with relatively thick wall and pronounced but rounded carination, and thickened rounded rim) (fig. IV.17.k, IV.17.l-n). No comparisons were found for the latter two shapes. Carinated bowls with a straight, rather high part of the wall above the carination (type 113, fig. IV.17.a-j) occur in all levels at Sabi Abyad, but are most popular in the Middle Assyrian levels 6 to 3 (with the maximum in level 5). They are not only more numerous than in level 7 (13.2% of this rim type occurs in level 6, against 1.2% in level 7), but they also differ in shape details. In level 6 these bowls have simple rounded or slightly squarish rims, or sometimes rims slightly thickened on the outside. About 12% have burnished surfaces, and about a quarter of these bowls is grey/black or dark-brown in colour and decorated with impressed circles on the wall, sometimes combined with impressed triangles on the rim that were originally filled with a white gypsum/lime paste (fig. IV.17.b,d; see also below; a good comparison comes from Tell Hammam et-turkman VIIIA, Smit 1988: pl. 153 no. 88). Other bowls are buff or orange in colour. These bowls always have ring bases. A thin-section sample (no. 16 in Appendix D, fig. IV.17.b) showed that this bowl was probably made locally. The number of rounded bowls decreased sharply in level 6, to only 6.9% of all bowls. There are now six different rounded-bowl rim types, two of which are new in level 6 (types 127 and (fig. IV.18.e-f); furthermore, these two types occur in mixed contexts or level 5 only). Most rounded bowls still belong to types 122, 123 and 125, but these types together now only represent 5.6% of all bowls (compared to 41.3% in level 7). These bowls are now completely undecorated, and the painted red rims characteristic for level 7 no longer occur (fig. IV.18.a-d). Rounded bowls with a pinched rim (type 121) occur in all levels, but are most popular in levels 6 to 3 (with the maximum in level 5). Two examples of this type occurred already in level 7, but 9.8% of all 121 bowls occurs in level 6 and this number increases later in level 5. As in level 7, about 15% of all bowls is of the straight-sided type. In level 6 two new types appear: 134 and 135 (fig. IV.20.m-p, IV.20.q). The large majority of straight-sided bowls, however, is made up by bowl types 131 with outward-sloping rim (11.3% of all bowls, 73.5% of all straight-sided bowls, fig. IV.18.g-l, IV.19) and 132 with outward-sloping rim thickened on the inside (3.4% of all bowls, 22.3% of all straight-sided bowls). Type 131 rims are smooth, sloping outward, thickened on the outside and sometimes also a bit thickened on the inside, with a generally triangular section. Some rims are a bit rilled on top (fig. IV.18.j, IV.19.c, g). Type 132 rims are thickened on the inside, with mostly rounded or pointed shape, and triangular sections (fig. IV.20.a-h). Deep bowls begin to occur in larger numbers in level 6. In level 7 there were a few examples of deep bowls, but although the types are similar to several types in level 6, the vessels show minor differences in rim shape. In level 6 there are 13 different types of deep bowls, a sharp increase compared to level 7. Especially characteristic are deep bowls type 141 and 142, with square or rectangular hammer rims running horizontally or sloping inward respectively (30.8% and 13.0% of all deep bowls respectively, fig. IV.21, IV.22.a-c), new in level 6 and typical for the Middle Assyrian period. Other typically Middle Assyrian shapes are type 143 (15.2% of all deep bowls, fig. IV.22.d-e), with an outward-sloping rim strongly thickened on the inside and outside, and large bowls type 145 (14.1% of all deep bowls, fig. IV.23), with a thickened upper wall and rim. Type 145 bowl P is a lot less deep than its later counterparts, which are generally more closed in shape (fig. IV.23.a). The decorations typical for these bowls, incised wavy lines and applied rope bands, occur from level 6 onwards with this type. New type 148 (fig. IV.24.a-b) occurs in small 29 No comparisons could be found for the type 1217 bowl, of which another identical example was found in square L12 (P93-376, no level assigned yet). 72

15 numbers and is only shortlived, appearing mainly in levels 6 and 5, just like types 1410, 1411, new type 1414 with a rounded rim thickened on the outside and often incised decoration on the upper wall, and new type 1415 with outward-sloping, squarish hammer rim (fig. IV.24.c-d, IV.24.e-f, IV.25.a-c, IV.25.d). Type 1413 (fig. IV.24.h), with a thickened ridge just under the rim, and type 1416 (fig. IV.25.e), with a strongly bentover rim, are unique in level 6. All these types ( ) are as yet without comparisons from other sites. Pots In level 6 almost one quarter of all pots is of the closed type, while three-quarters has straight walls. This is a marked contrast with level 7, where most pots were of the closed type. Although there was one example of a closed pot with inward-sloping rim in level 7, it was made of cooking ware and was different in shape from the same type 211 pots in level 6. Type 211 pots in level 6 are now globular in shape, with rounded rims sloping inward (fig. IV.25.f-i). Most closed pots in level 6 are of type 212, with a rim thickened on the outside. Pot P96-487, with one ribbed handle (or originally perhaps two?) and incised decorations, attracts the attention, and is closely comparable to a similar pot in level 5 (fig. IV.26.a, cf. fig. IV.64.g). This pot, unknown at other Middle Assyrian sites, is reminiscent of shapes found at Late Bronze Age sites along the Euphrates, although close comparisons have not been found. 30 Type 2110, closed pots with a very low, upwards-bent pointed rim, occurs only once in level 6 (fig. IV.26.d) and once in level 5. In contrast with level 7, there are only few possible cooking pots among the level 6 closed pots (fig. IV.26.b, c). Most pots in level 6 have straight walls and an open shape (75.9% of all pot rims). The majority is of type 221 (40.6% of all pots, 53.1% of all straight-walled pots), with a rectangular, horizontal hammer-shaped rim (fig. IV.26.f-i). The large variant of this type (221b) belongs to a large storage vessel (fig. IV.27.a). This type (221b) is the straight-walled counterpart of deep bowls type 145. Next are the type 222 pots (28.1% of all pots, 36.7% of all straight-walled pots), with similar hammer-shaped rims that slope inwards (fig. IV.27.b-e, IV.28.a-d). Both these types (221 and 222) are comparable to the deep bowls type 141 and 142 respectively, and often the distinction between the two classes is difficult to make. The 221 and 222 pots are characteristic of the Middle Assyrian assemblage at Sabi Abyad, and although examples of these types occur in all levels, the majority occurs in levels 6 to 4. Type 226, a large thick storage-vessel rim thickened both inside and outside, is new in level 6 and from now on appears in all later levels, with a maximum in level 4. Type 227 is a rare shape at Sabi Abyad and until now without comparison. Two examples occur in level 6 (fig IV.27.g). Jars In level 6 the number of jars decreased slightly compared to level 7, and comprises now about 15% of all rim sherds. However, it is the second-largest shape group. A marked difference with the level 7 jars is that in level 6 most jars have no neck and a ribbon rim sitting directly on the shoulder of the vessel (around 55% of all jars, compared to 44% of jars with a clear neck). Among jars with a clear neck, types 311 and 312 with simple rounded rims are most popular (65.5% and 15% of all jars with neck, 29.2% and 6.7% of all jars, respectively) (fig. IV.28.e-j, fig. IV.29.a-b). Although isolated examples of these types occurred in level 7 they are characteristic of the Middle Assyrian assemblages, and are most popular in levels 6 to 4. As in later levels, the 311 jars can be divided into two size groups, small and large. They have either ring bases or disc-shaped bases. Other jars with necks belong to types 315 (13.3% of jars with necks, fig. IV.29.f-h) and 318 (3.5% of jars with neck, fig. IV.29.i-k), with squarish rim either horizontal or sloping inwards. These types already occurred in level 7 and continue into Middle Assyrian times. The type 314 jar in fig. IV.29.e has a possible comparison in a jar from Tell Shioukh Fawqani (Bachelot 1999: pl. 10 no. 2). In the group of jars without a neck, the oval rim with vertical or slightly rounded exterior (322) is still most popular (80.7% of all jars without neck), while smaller numbers of rims with concave sides (321) or with sloping sides (323) occur (approximately 10% each). In 321 rims the concave part of the ribbon rim starts rather low, while the rim in general is oriented vertically (fig. IV.29.l-p, especially IV.29.n). Type 322 rims are more oval and rounded than in level 7, and sometimes a bit triangular in section (fig. IV.30, IV.31.a- 30 Tell Fray (Pfälzner 1995: taf. 185a), but with mineral inclusions only; Emar Oberstadt phase II (Finkbeiner et al. 2001: abb. 11, i), but not decorated like the Sabi Abyad pots; Hadidi (Dornemann 1981: fig. 8 no. 5), but that one is much bigger. 73