Version classiqueVersion mobile

Multidisciplinary approaches to food and foodways in the medieval Eastern Mediterranean

Sylvie Yona Waksman

Cyprus and the Levant

Some thoughts on sugar production and sugar pots in the Fatimid, Crusader/Ayyubid and Early Mamluk periods in Jordan

Richard Jones et Anthony Grey


Excavations during the last decade at the Tawahin es-Sukkar near Safi, to the south of the Dead Sea in Jordan, revealed structures connected with sugar cane milling and the refining of sugar. Study of the sugar pots from the 2002 excavations at the Tawahin and the adjacent settlement of Khirbet Shaykh ‘Isa (which can be equated with ancient Zoara) indicates that sugar production was taking place on a significant scale from the 11th century to the late 14th century.
This paper presents the morphological and material characteristics of these sugar pots, comparing them with their counterparts from elsewhere in Jordan and Palestine. This is considered within the broad model of production and consumption, reviewing the evidence, for example, for centres where storage and secondary refining occurred separately. The issue of the extent to which sugar pots travelled with their contents is discussed, as is the potential of integrating typology with fabric analysis to begin elucidating some of the steps within the chain that led from primary production to secondary refining to selling, first within southern Jordan (encompassing the southern Ghor and the region administered by Kerak), and second, more widely within the Levant.

Texte intégral


1Our understanding of the medieval sugar industry in the Eastern Mediterranean has been growing over the last few decades as the remains of the industry’s infrastructure have been recovered by excavation and as finds of the industry’s ceramic proxy – the sugar pot – have been made in the course of both excavation and survey. Most progress has been made in Cyprus where three sugar refineries – at Kouklia-Stavros (von Wartburg 2001), Episkopi-Seraya and Kolossi (Solomidou-Ieronymidou 2015) – have been excavated, encompassing in date the late 13th to the end of the 15th century. These sites provide us with the fullest view anywhere in the Mediterranean of the layout and function of sugar installations at that time. The equivalents in Palestine and Jordan are generally of earlier date (but extending to the late Mamluk period at the end of the 15th century) and less well preserved, their structures having often long been dismantled or put to new or other uses. Nevertheless, the examples at Lower Horbat Manot in western Galilee (Stern 2001), Tell Abu Sarbut in the Jordan Valley (Steiner 2008) and Tawahin es-Sukkar at Jericho (Taha 2009) (fig. 1a) are all valuable in their different, complementary ways.

2The aim of this paper is to briefly introduce a relatively new site – Tawahin es-Sukkar – near Safi south of the Dead Sea in Jordan (fig. 1b), whose partial excavation has given an opportunity to look anew at the whole process involved from crushing to refining; it operated between the 12th and the early 15th century. Any view that the Dead Sea lies too far from the Eastern Mediterranean to be relevant to this volume’s focus is mitigated by the fact that the sugar industry was a regional phenomenon; the manifestation of this industry in the Dead Sea area was probably little different from that along the Jordan Valley or western Galilee close to the Mediterranean coast. Alongside the Tawahin is the settlement of Khirbet Shaykh ‘Isa (hereafter KSI) which was both a centre of economic importance during Ayyubid to Mamluk times and the support town for the adjacent sugar industry.

Fig. 1 – Satellite image (from Google Earth) of (a) the Eastern Mediterranean showing some of the main sites mentioned in the text, and (b) Tawahin es-Sukkar, Khirbet Shaykh ‘Isa, Safi, Feifeh, Mazra’a and Kerak in the south Dead Sea area.

Fig. 1 – Satellite image (from Google Earth) of (a) the Eastern Mediterranean showing some of the main sites mentioned in the text, and (b) Tawahin es-Sukkar, Khirbet Shaykh ‘Isa, Safi, Feifeh, Mazra’a and Kerak in the south Dead Sea area.

3As the springboard for this enquiry we use the evidence from the 2002 excavation at the Tawahin es-Sukkar and KSI. That excavation and the finds therefrom are published by Jones (2017), and fuller data will be forthcoming from the further excavations at both sites by Konstantinos Politis of the Hellenic Society for Near Eastern Studies and his collaborators. Having outlined the sugar vessels from that site, drawing on comparanda at other sites in Jordan and Palestine, this ceramic evidence is then used in an exploratory manner to investigate the routes by which sugar – a new commodity at this time, reached its final destination, the consumer. Where was the refining of the crude sugar carried out, and where was the storage? What distances were involved and how did the sugar travel? These are questions to which documentary records give us a tantalizing but still partial picture; the challenge is for archaeology to provide complementary information which, we suggest, may come from the sugar vessels themselves. As a case study, we use the distribution of sugar during the Mamluk period in the Kerak province in which Tawahin es-Sukkar and KSI are situated. In the final section sugar is treated as a commodity that formed a component, albeit of unknown magnitude, in the diet at KSI and at contemporary settlements in Jordan. Diet is thereby considered within the broad spectrum of the faunal and botanical evidence as well as the food preparation vessels.

The archaeological sites of Tawahin es-Sukkar and Khirbat Shaykh ‘Isa (fig. 1b, 2a-c)

4The attractive site of Tawahin es-Sukkar (TeS), its name meaning sugar mill, has long been known as a medieval sugar installation owing to its relatively good state of preservation. Lying today in a rich alluvial fan close to the modern town of Safi, it is situated in a dramatic landscape, not only at the lowest point on Earth but with the Dead Sea to its north, the high steep ridge up to the Karak plateau to the east and the similarly high scarp to the west, looking up to Masada. Its location on a knoll optimised on the one hand access to the water supply coming from the Wadi al‑Hasa (fig. 1b) and on the other visibility and thus control of the sugar cane plantations that would presumably have lain to the north and west. The most prominent features of the site are the water chutes and mill rooms (fig. 2a: MW). The eastern mill room was shown on excavation to be 4.8 m across and its large ashlar block walls survived to 3.2 m in height (James & Photos-Jones 2017, fig. 5.2d). A substantial millstone base (3.25 m diameter, 2.5 m internal diameter), made of a single piece of limestone, was found in-situ within this room, sitting on top of a flagstone floor; there was a small circular drain hole in its rim. The upper millstone, 1.4 m in diameter, 0.4 m thick and with a central square hole 0.5 m wide, lay across the edge of the millstone base. Two halves of a collar stone were also found above the millstone. Below was a room with a vaulted roof (Politis et al. 2005, pp. 321‑323) that must have housed the horizontal mill wheel which was powered by water descending the chute at the back of the room. The broadly similar mill room lying to the west has since been excavated by the Department of Antiquities of Jordan (see, consulted 10/12/2019) and by K.D. Politis (Corbett et al. 2016, fig. 15).

Fig. 2 – (a) Site plan of Tawahin es-Sukkar, Safi, showing the chutes (C), buttresses (B), mill wall (MW), AW adobe wall (from a a probable late Ottoman fort) and dump, and (b) the trench locations at Khirbet Shaykh ‘Isa; (c) Satellite image (from Google Earth) of the Tawahin es-Sukkar, Khirbet Shaykh ‘Isa and Wadi al‑Hasa.

Fig. 2 – (a) Site plan of Tawahin es-Sukkar, Safi, showing the chutes (C), buttresses (B), mill wall (MW), AW adobe wall (from a a probable late Ottoman fort) and dump, and (b) the trench locations at Khirbet Shaykh ‘Isa; (c) Satellite image (from Google Earth) of the Tawahin es-Sukkar, Khirbet Shaykh ‘Isa and Wadi al‑Hasa.

5The other locus of attention in 2002 was the main dump (fig. 2a), representing the waste from the industrial processes taking place at the refinery as well as debris from construction/demolition phases of the nearby building (James & Photos-Jones 2017). The chief interest of this dump was the presence of industrial waste whose subsequent analysis was able to shed light on the refining processes (Photos-Jones 2017). Other groups of deposits within the dump were mixed clay-rich layers with pottery and mud brick, charcoal-rich layers, compact ash and dumps of exclusively sugar-pot waste. The pottery recovered from the Tawahin indicated that the mill operated between the 13th and 15th centuries.

6Immediately adjacent to the Tawahin is the multi-period settlement of Khirbat Shaykh ‘Isa (KSI) (fig. 2b, c) which is usually regarded as the site of biblical Zoara as observed, for instance, in the Madaba mosaic. Early visitors to the site included Frank (1934) who reported its considerable size – 300 m (E‑W) and 160 m (N‑S) – and the presence of a flat-top mound (now gone), but its importance was not fully appreciated until the early 1980s when irrigation channels were laid out exposing probable Byzantine structures (Politis 1998). The 2002 excavations focused on five small trenches selected on the basis of the results of magnetic survey over most of the currently extant site, ca 2.7 ha (Jones et al. 2000, p. 531, fig. 6; Photos-Jones et al. 2002, pp. 595‑599, fig. 5a, b). These trenches revealed structural remains, all of them consistent with a date in the 10th to 15th centuries, and in the deep stratigraphy of one of them (Trench III) there was also an earlier, 6th-9th century phase (James 2017). Two trenches (V and VII) towards the eastern end of the site revealed industrial remains in the form of kiln lining and wasters (see below). Sugar vessels were not only manufactured in this sector but their common occurrence in most of the other trenches strongly suggests that KSI, a market centre of some importance during the 1200-1500 AD period, was the support town for the sugar installation at the Tawahin. Since 2002, excavations at KSI situated mainly in the western sector of the site have uncovered substantial building remains revealing sequences of the 8th to ca 15th century AD (Politis et al. 2009, pp. 303‑308).

Sugar wares

7The main corpus of vessels from KSI dates to the 14th century Mamluk period with earlier vessels from the Crusader-Ayyubid 12th-13th century period. The antecedents are uncertain, and body sherds that appear to be from sugar vessels (in ware type and wall thickness) from KSI Trench III in particular could be from either early versions or basins/bowls. The main function of these basins/bowls was the decanting of the sugar syrup into the sugar pots and/or for some settling process, but they could also have been used for washing moulds ready for reuse, as was the case at Kouklia (von Wartburg 1983, p. 309). As at other sites in Jordan – Kerak, but apparently not Tell Abu Sarbut – basins are found together with, and in the same fabric as, sugar pots. Of the 17,779 sherds weighing 534 kg recovered from the excavations, 77% of them were sugar pots at TeS and 55% of them at KSI. Fig. 3 below also shows how the moulds are considerably more plentiful than the molasses jars.

Fig. 3 – Relative percentage proportions of the ware types at Khirbet Shaykh ‘Isa.

Fig. 3 – Relative percentage proportions of the ware types at Khirbet Shaykh ‘Isa.

Shapes (fig. 4)

Fig. 4 – Moulds and molasses jar at Tawahin es-Sukkar and Khirbet Shaykh ‘Isa (Grey 2017, fig. 3.21c).

Fig. 4 – Moulds and molasses jar at Tawahin es-Sukkar and Khirbet Shaykh ‘Isa (Grey 2017, fig. 3.21c).


8Shapes vary from conical with straight sides to bell-shaped with a pinched-in waist. Overall sizes vary with concomitant weight variation, the trend being towards an increase in size in the Mamluk period. Grey’s rim form typology (Grey 2017, tabl. 3.4) is here simplified (tabl. 1).


Tabl. 1 – Moulds and molasses jars illustrated in fig. 4.

109. Out-turned rim in pink-buff fabric. Rim diameter 36 cm. The same form from Kerak (Milwright 2008, fig. 10: 6, 7).
110. Out-turned rim. Rim diameter 30 cm.
111. A nearly plain rim which has slight external thickening in reddish ware. Rim diameter 36 cm. Parallel: Zir’in (Grey 1994, p. 57, fig. 8: 7).
112. Outwardly folded rim in buff ware. Rim diameter 35 cm.
113. Base of sugar pot with ridged body and basal hole without lip. Parallel: Horbat Manot (Stern 2001, fig. 7: 10, 12th-13th centuries).
114. Base of sugar pot with ridged body in yellowish sandy ware with basal hole with lip.
Molasses jars
115. Very well-finished, smooth body in orange fabric. Rim diameter 16 cm. Parallel: Horbat Manot (Stern 2001, fig. 7: 11, Crusader period, with smooth body).
116. Hooked rim in fairly fine, thin-walled, smooth-bodied ware. Rim diameter 14 cm. Possibly an earlier form. Same form from Kerak (Milwright 2008, fig. 12: 3).
117. KSI 509 <193> MJ7 with flattened, incurved and externally thickened rim and smooth body in pale fabric. Rim diameter 12 cm.
118. Externally bevelled rim and ridged body. Rim diameter 12 cm. Parallel: Horbat Manot (Stern 2001, fig. 7: 10, Crusader period).
119. Externally bevelled rim and smooth body. Rim diameter 14 cm. Parallel: Horbat Manot (Stern 2001, fig. 12: 4, Mamluk period).
120. Externally thickened and rounded rim which is distorted from misfiring and cream in colour. Rim diameter 10 cm.
121. Heavy externally thickened and bevelled/hooked rim. This is a heavy-duty vessel, thick-walled with ridged upper body. Rim diameter 16 cm. Parallel: Horbat Manot (Stern 2001, fig. 12: 4).
122. Externally bevelled rim and ridged body. Rim diameter 14 cm. Parallel: Horbat Manot (Stern 2001, fig. 7: 12, Crusader period); Kerak (Milwright 2008, fig. 12: 2).

Molasses jars

9Two main variants: (a) a thick-walled, heavy, ridged vessel with heavy rim, probably of later date than (b) which is much lighter, smooth-bodied vessel with externally beveled rim.

10At TeS there were 607 and 96 rims of sugar moulds and molasses jar respectively, and the corresponding figures at KSI were not very different: 644 and 130. These figures can be compared with those at Horvat Manot: Crusader 2,276 and 46; Mamluk 411 and 151; Ottoman 630 and 114 (Stern 2001, tabl. 3).

11Rim diameters for the sugar pots from KSI range from 33 to 46 cm or more and molasses jars from 12 to 16 cm, reflecting considerable variation in vessel size which is further corroborated by the corresponding weight measurements. Attention is drawn to an example of two holes pierced through a mould from the exterior before firing (Jones 2017, fig. 3.20b); its significance is as yet unclear but similar instances have been recorded at Tell Abu Sarbut (LaGro 2009, fig. 15) and at early sites in West Galilee (Stern et al. 2015).

12The molasses jars showed two major variants in size: a thick-walled, heavy, ridged vessel with a heavy rim, and a smaller, much lighter, smooth-bodied vessel with externally bevelled rim. There is a clear chronological dimension to this variation, also observed at Horbat Manot, the former type being 14th century Mamluk and the latter Crusader-Ayyubid 12th-13th century in date. But the factor of size of both moulds and jars also surely relates to the different grades of sugar being produced, with the evidence from Cyprus in particular indicating that the finest sugar grade was manufactured in the smallest size(s) (von Wartburg 2014, pp. 239‑245). Abundant though the sugar vessels are at TeS and KSI, this material from the 2002 excavation is insufficiently securely stratified to propose a definitive typo-chronology; this will have to await the cumulative results from subsequent seasons of excavation. The wide variation observed among the moulds and molasses jars in terms of hardness should not be unexpected (Grey 2017, fig. 6.12) and was confirmed petrographically (see below) given that both types were no doubt fired in the same kiln. However, at Kerak (see below) Milwright (2008, p. 159) observed that examples of molasses jars were apparently harder than moulds, suggesting that this may have been deliberate to allow the jars to travel more securely with their contents; this interesting point is taken up below. A final comment on the issue of size is that the high ratio of moulds to molasses jars may relate in the case of jars of small size having concomitant smaller rim diameters experiencing fewer breaks per jar than the moulds.

13A feature of the sugar pot sherds in Trench II at TeS was the frequency of off-white accretions, some of them ashy, adhering to their interior. The results of their analysis showed they were composed of calcite which was likely employed for the purification of the crude sugar juice following boiling. No organic phase was identified by Fourier-transform Infrared spectroscopy (Jones 2017, p. 69, tabl. 3.7). Chemical analysis by XRF revealed that their calcium to strontium ratios were unusually high; that such values were found to occur in sediments taken from the nearby Lisan peninsula (fig. 1b) is an indication that this may have been the source of the very fine-grained textured calcite (Photos-Jones 2017, p. 185, tabl. 8.5; Jones 2017, tabl. 3.7). Less common were instances of red staining on the interior which were iron rich, but again there was no organic content (Jones 2017, tabl. 3.7).

14The sugar pot fabric has been characterised by Louise Joyner (in Grey 2017) as Quartz Sand and Silt; calcareous composition. More precisely, the fabric comprises a continuum between a moderately packed silty matrix with common sub-rounded to rounded medium to fine sand-sized quartz grains, to a fabric with sparse well-rounded medium sand-sized monocrystalline quartz grains set in a loosely packed groundmass containing sparse coarse silt-sized monocrystalline quartz and calcimudstone fragments. The firing temperatures were found to be variable. Clay collected from the Wadi al‑Hasa close to KSI (fig. 2c) was broadly similar to, but coarser than the sugar pots in fabric composition. Chemical analysis of eleven elements by pXRF shows (fig. 5) the sugar pots forming a tight cluster characterised by a calcareous composition and similar to the modern clays. These pots are separate from (a) the cooking pots and other local wares which have a low calcareous composition and (b) the likely imports.

Fig. 5 – Classification by principal components analysis of the chemical composition data (obtained by pXRF) for sugar pots (S, yellow), clays (CL, blue) and cooking pots (CP, purple) from Tawahin es-Sukkar and Khirbet Shaykh ‘Isa. Data from Grey 2017, fig. 6.14b and Appendix II.

Fig. 5 – Classification by principal components analysis of the chemical composition data (obtained by pXRF) for sugar pots (S, yellow), clays (CL, blue) and cooking pots (CP, purple) from Tawahin es-Sukkar and Khirbet Shaykh ‘Isa. Data from Grey 2017, fig. 6.14b and Appendix II.

15Survey and excavations at both KSI and TeS revealed strong evidence of ceramic production in the form of wasters and kiln lining. The magnetic survey of ca 2.7 ha at KSI indicated several hotspots, at least some of which were consistent with the presence of kilns or related structures (Photos-Jones et al. 2002, fig. 5a-b). These magnetic anomalies were pronounced in two discrete areas at the east end of the site (fig. 2b), but subsequent excavation revealed that the finds of wasters and kiln lining were in fact more common towards the west and centre in Trenches I, II and VII than in Trenches IV and V. At KSI there were fifteen wasters (7 sugar pots, 2 jugs, 1 bowl, 5 either kiln wall or sugar pot fragments, and in the waste dump at TeS 6 distorted sugar pots, 3 overfired sugar pots, 1 exploded sugar pot and 1 distorted jug) (fig. 6). It is premature to identify the location of an industrial quarter but in view of the presence of wasters in the dump at TeS there may have been kilns for sugar pot production close to the mill complex as well as at one or more locations at KSI. Indeed, a few small magnetic hotspots were encountered in the field immediately to the north of TeS where surface finds of sugar pots and furnace debris were prominent (Jones et al. 2000, fig. 6, area B: anomalies 2, 4, 7 and 8).

Fig. 6 – (a) Sugar pot waster (Khirbet Shaykh ‘Isa surface find) and (b) jug waster (Khirbet Shaykh ‘Isa Trench I).

Fig. 6 – (a) Sugar pot waster (Khirbet Shaykh ‘Isa surface find) and (b) jug waster (Khirbet Shaykh ‘Isa Trench I).

Transport and distribution

16As a counterpoint to the efforts towards understanding the mechanics of manufacturing sugar in the Levant is the need to investigate the industry’s organisation following primary production, that is, the sugar’s destinations from that original location and by what means the sugar reached those destinations. Interesting and important as these issues are for the Mamluk period, they are not fully or even well illuminated by the documentary sources of the time (Walker 2011, pp. 86, 220) and as a result the initiative falls more to the archaeological record.

17To take organisation first and restricting attention mainly to the Mamluk period in Jordan, the archaeological record has identified primary production centres, of which TeS is a prime example. Some or even many of these centres also carried out a certain level of refining, but it is likely that the partially treated sugar received further refining elsewhere. In the apparent absence of archaeological data about the location of such refining centres, reference can be made to the 15th century writer Qalqashandi who mentions the sugar from primary centres in the Jordan Valley being transported to Damascus, a large urban area where the matbakh, the boiling place, was situated (Milwright 2008, p. 164).

18Archaeological evidence may also indicate the involvement of the storage/distribution centre. A likely example is the citadel at Hisban (fig. 1a) which on current evidence had no role in primary refining as no moulds have been recovered (B. Walker pers. comm.). Having reached Hisban probably in the typical molasses jars, sugar/molasses was presumably then transferred, perhaps following further refining at an as yet unknown location, to the large jars of the type found in the citadel’s store room (Walker 2003, fig. 3). The shape (and size) of these jars would have made them suitable for attachment, via a rope round their middle, to a pack animal (Walker 2010, p. 148), perhaps a camel. The sugar carried in the jars at Hisban has been estimated at ca 571 kg (Walker 2007, p. 191), destined for redistribution to local markets and probably further afield.

19Moving west into Palestine, another example of a storage centre comes from Crusader Acre where Stern (2012) has reported the large-scale storage of 13th century sugar vessels, in particular in the halls of the Hospitaller compound. The final component was the consumer who could have obtained the sugar product at any of the points along the network, from the places of primary production to markets large and small and the urban centres. Those such as Acre and Beirut on the Mediterranean coast were also surely exporting sugar products overseas. The markets served local concentrations of population as well as mobile populations travelling along pilgrimage routes, such as the Hajj.

20Having identified the likely main components of the network, the next step would ideally be to create a graphical scheme of the components’ inter-relationships within parts of the Levant. But that remains premature until such time as a level of knowledge drawn from, on the one hand, documentary accounts about the structure of society in Mamluk Jordan, as for example outlined by Walker (2011, pp. 86, 220), and on the other a fuller archaeological picture has been achieved. In pursuit of the latter we propose that the sugar vessels should play a larger role and thus more could be done to extract their information potential. To reiterate a few basic characteristics about the vessels, a feature of most (primary) sugar production centres is the high frequency of mould fragments consistently outnumbering the molasses jars (at Tawahin and KSI the ratio is ca 6:1). This is usually ascribed to the frequent need to break the mould to extract the sugar loaf despite efforts to design or treat the interior of the mould such that the loaf would ease its way out of the intact mould. In the early phases this extraction process was probably inefficient, hence the breakage, whereas later on it improved. From several lines of enquiry, it is generally recognised that the vessels’ sizes varied according to the quality of sugar being processed (Milwright 2008, p. 160). Less straightforward, however, is the issue of how sugar travelled from the primary production centre to its various destinations. The reality is that this probably occurred in several ways including as the sugar loaf itself or as granules either wrapped in a sack or placed in a crate ready for export overland or by boat to an overseas destination. Alternatively, the sugar could have been transported in the vessels themselves, both moulds and molasses jars. Despite the not insignificant weight of the moulds, typically 6 kg, this mode of transport would have only been suitable over short distances, for instance from primary production centres to local markets or alternatively direct to consumers at locations where vessels have been found but lacking direct evidence of primary production such as milling, pressing or boiling; Ibrahim et al. (1976; 1988) identified numerous instances of such locations in their East Jordan Valley surveys. The case cited above at Hisban is another relevant example, and also implicitly supporting this argument is the situation at the Tawahin-KSI complex itself where it can be reasonably supposed that since both sugar and sugar pots were produced there the vessels’ function was primarily industrial but to some measure also for transport.

21To reflect the complexity of the dynamics of sugar distribution the discussion now retains the focus on the vessels and explores the potential of using their fabric as an indicator of source, using data from Kerak (fig. 1b) and Tawahin-KSI as the example. Arising from both archaeological investigations in and around the castle at Kerak and historical considerations drawn from Mamluk texts (Milwright 2006; 2008; Walker 2011, pp. 86‑90, 220‑222), much is known about its economic and political importance as a provincial capital in the Mamluk period, situated on the high plateau above the Dead Sea. Excavations in the castle recovered both moulds and molasses jars, which in the absence of evidence of sugar refining there were all products in Milwright’s view (2006, pp. 159‑161; 2008, p. 248) of regional, probably short-distance distribution. This view may be interpreted in two ways: the vessels reached Kerak either containing their products ready for consumption or empty. The latter view is prompted by the results of typological and petrographic study in Galilee reported by Stern et al. (2015; in this volume) and Shapiro (2012; 2019; in this volume) pointing to the control that a large regional centre held over production centres lying within its orbit in western Galilee and the way that control changed over time. More specifically, Crusader Akko (Acre) stockpiled moulds that were manufactured in its vicinity (Shapiro 2012) and then distributed for use to local production centres, whereas the Mamluk period saw control over production of the vessels used at those same centres moving inland (Shapiro 2019).

22But to return to Kerak, Mason and Milwright’s (1998) petrographic study of a variety of wares included a few sugar vessels which they found contained some distinctive composition features:

  • sample 6 (mould): higher basalt content with high carbonate and argillite, sharing similarities with a kiln tripod (marked with green glaze) from Shawbak (ca 30 km south of Tawahin);
  • sample 17 (molasses jar): distinctive quartz resembles one example of a molasses jar they had from Tawahin;
  • sample 18 (mould): basalt-bearing fabric, but unique, having relatively high carbonate and brown fissile argillite;
  • sample 19 (mould): unique sandstone (and related aplastics) fabric, different from all other Karak fabrics.

23Not only are these examples at Kerak, admittedly few in number, diverse in composition, but the two sugar vessels found by Mason and Milwright at Tawahin-KSI, which as shown above is itself securely classified as a ceramic production centre, differed from each other; one (Mason sample 10002) resembled sample 17, and the other (Mason sample 10001) was similar to Joyner’s Quartz Sand and Silt; calcareous composition. This is an unexpected result in the light of the uniformity of compositions for all the moulds and molasses jars which we have analysed at TeS and KSI. Whichever model is appropriate – vessels reaching Kerak either with or without content, or indeed a combination of both scenarios – the results from Kerak paint an intriguing picture of intra-regional connections, involving several supplying centres, one of which, reasonably enough as argued above, was surely Tawahin. Shawbak, lying ca 70 km south of Kerak, although not a strong contender as a sugar production centre (Jones 2017, tabl. 3.1), is possibly implicated by the composition of sample 6. Here we suggest two additional candidate locations: Mazra’a and Feifeh situated to the north and south respectively of Tawahin (here fig. 1b; Jones 2017, tabl. 3.1, fig. 3.8) where surface remains include millstone bases and scatters of sugar vessel sherds.

24The frequency of basaltic inclusions among several of the wares, not only sugar vessels, analysed by Mason is consistent with the presence of basalt outcrops occurring to the north and south of Kerak (Watts et al. 2004, fig. 1). In particular, basaltic fragments may be expected to be transported down the Wadi Kerak reaching the clay-like sediments near Mazra’a; although there is no evidence (because there has been no excavation) of ceramic production at Mazra’a itself, this location could be a candidate source for one or other of the basalt-rich sugar vessels analysed at Kerak. Less encouraging would be the expectation of differentiating petrographically between products of Tawahin and Feifeh owing to their spatial proximity. But the more general point is the potential for fabric analysis of sugar vessels to shed light on sources. A petrographic and chemical study drawing on a large number of typologically well-defined sugar vessels found at a range of sites – production and consumption – in the Levant could shed valuable light on both intra and inter-regional distribution.

Food and foodways

25For the inhabitants of KSI during the time span of the Tawahin’s operations, sugar was presumably the standard sweetener, while previously honey would have largely fulfilled that function. But to complement the emphasis in this paper thus far on one commodity, sugar, mention can be made of the wider diet of the population at KSI. Sheep, goat and cattle bones feature strongly, more so in the early period levels at KSI (6th-9th century) which probably had a more subsistent and pastoral economy than the more urbanised and agricultural situation in the 10th-15th centuries; analysis of the sheep bone points to butchering but not slaughter on site, thereby indicating a picture of a moderately urbanised society (Henton 2017). To this diet was added camel (evidence of its butchering is evident), fish (parrot fish from the Red Sea and catfish) and domestic fowl. At least some sections of society also exploited pig products, both in the early and late phases of settlement at KSI. Some wild food was consumed. Analysis of the very limited botanical finds revealed small amounts of barley (hulled, six-row), date palm and Assyrian plum (Miller 2017). Comparable assemblages are limited, but note may be made of the following: in the Mamluk levels at Khirbat Faris on the Kerak plateau wheat remains from different stages of crop processing and their weeds predominate (Charles & Hoppé 1997-1998) and from the 13th century onwards a few exotic crops such as cotton, sorghum, watermelon, pistachio, and citrus fruits start to appear (McQuitty 2007, p. 159). At Mamluk Hisban (Walker 2011, p. 177) wheat, barley, lentils, bitter vetch, broad beans, olives and grapes are reported.

26Reviewing the published faunal data for Mamluk assemblages at Hisban, Dhiban and Kerak and bearing in mind varying levels and non-uniform contexts of recovery, Brown (2016, tabl. 4.3) found the ranges of relative % representations of sheep/goat-cattle-pig were 77‑89, 11‑18 and 0‑5.9 respectively with the corresponding figures for Shawbak standing apart: 98, 2 and 0. The sheep to goat ratio at the four sites lay within the ranges 1:1.1 (Hisban) and 1:2.1 (Dhiban), contrasting with the value for KSI of 1:0.75. Since a ratio of between 1:1 and 1:1.7 is the preferred range in order to preserve herd viability and security (Redding 1984), the ratio for KSI suggests either some pressure on herders from an urban community or elite to raise more sheep, or that there is a selective uptake of sheep from rural herds by urban dwellers. An alternative interpretation that the environment was moister and more suited to sheep is at least consistent with the available environmental data on the Dead Sea levels (Klein & Flohn 1987, fig. 3, p. 156). Of interest is that the figures for sheep-goat and cattle are not greatly different for the preceding Crusader and/or Ayyubid periods at Wu’ayra, Wadi Farasa (both in the Petra area), Shawbak and Hisban (Brown 2016, tabl. 4.3); pig is better represented in the latter periods, although the figure may be skewed by the anomalously high presence in certain contexts at Crusader Wu’ayra. Sheep-goat representation rises to over 90% in the Ottoman period.

27The physical evidence for cooking at KSI was limited to three ovens (tabuns) for roasting and (bread) baking, a simple hearth and ashy layers containing domestic debris. Cooking pots, which are the most common ceramic class after sugar moulds (fig. 3) (Grey 2017, pp. 130‑132), comprise:

  • Wheelmade cooking vessels (5.47%) which are abundant in the early periods at KSI up to the Fatimid period before the 10th century; their lids have a steam hole in the lid knob (red wares) or next to the lid knob (grey wares), indicating these pots were for boiling.
  • Coarse handmade cooking vessels (2.2% in Tr II at KSI) either Plain (or with cordon) (late 10th to 11th-12th centuries) or rarely Painted (13th century). Their frequently globular shape makes them suitable for boiling/stewing, while open “frying pans” were for frying or preparing items, perhaps sauces or small quantities of vegetables.
  • Glazed cooking vessels (0.04%). First appearing in the 9th century, these vessels are present in 12th and 13th century deposits becoming less common in the 14th century in Palestine and Jordan. The open form, including examples of open “frying pans”, is much more frequent than the closed form.

28Proceeding briefly to other shapes, attention can be drawn to some jugs and small jars which may have held herbs for an infusion drink in the pre-coffee and tea era; the neck filter on these vessels, perhaps requiring the use of fine cloth like muslin, would probably also have had the more common function of screening out grit and sand from fresh drinking water. Painted jugs were for water and possibly milk at the meal; tableware would also have included large bowls for fruit, olives and the like.

Concluding remarks

29This paper has placed the southern Dead Sea area onto the map of medieval sugar production in Palestine and Jordan, alongside that in the Jordan Valley, in Galilee and elsewhere. The combined archaeological evidence points to a broad commonality in the process of manufacturing sugar during the Crusader/Ayyubid and Mamluk periods across the region, meaning that the steps involved were essentially similar. On the other hand, the nature of the space in which those steps took place varied, ranging from the adaptation of one or a series of pre-existing structures to the needs of sugar manufacture, as at Lower Horbat Manot, to the situation at the Tawahin es-Sukkar which seems to have been built from its initial construction for that function alone. Another important characteristic of the Tawahin, and one which is not apparently shared elsewhere, is the presence of the so-called support town, Khirbet Shaykh ‘Isa, which provided an industry meeting at least some of the demand for the manufacture of sugar pots, as well as perhaps part of the workforce. In economic terms, Khirbet Shaykh ‘Isa was well placed geographically to market sugar and no doubt other products, situated as it was in the early Mamluk period with routes westward into Palestine, eastwards via Kerak to both the King’s Highway, with its connections north to Damascus, and the Hajj route. Movement of goods to the south could take either the southern direction of the Kingsway Highway to Aqaba on the Red Sea or more directly the route along the Wadi Araba.

30One of the industry’s features that did change, and in a chronological fashion, was the ceramic equipment, the sugar vessels. From the evidence at several sites their shape changed from the Crusader/Ayyubid to the Mamluk period, although this has not yet been fully or securely documented at the Tawahin. However, whereas that modification in morphology seems to have been rather uniform across the region as a whole, the control over production of both sugar and its associated ceramics manifested itself differently. As mentioned above, the change that occurred in the production location of the sugar vessels used at sites in western Galilee from Crusader to Mamluk and the important political consequences of that change are not (yet) apparent at the Tawahin; here the indications are that local production continued throughout that time span. It is premature to consider the implications of that contrasting situation. Instead that issue of the source of the sugar vessels has been used here as the proxy for investigating the network of interconnections between sugar producer and consumer within the Mamluk province of Kerak. Owing to the relative geological diversity within this region, fabric composition has begun to demonstrate that this network was not straightforward; Kerak as the provincial capital had links, even controlling links, with more than one component of the sugar industry, whether it was a producer or not. The presence of the Tawahin in that network is all the more likely as, together with Khirbet Shaykh ‘Isa, it combined a sugar industry with a market town. Another candidate member of the network, but still very much to be tested, is Mazra’a lying ca 30 km to the north. For the moment, then, the hypothesis is that Kerak controlled a network of sugar producers in the region, marketing the product along the routes identified above; some of this sugar evidently reached the Mediterranean coast destined for export as Pegolotti records its reputation for high quality (Milwright 2008, p. 121), but local and regional markets may have met much of the remaining demand. It remains to reiterate that fabric, particularly petrographic, analysis of typologically well-defined sugar vessels found at producer and consumer sites has a valuable role to play. The results obtained to date, although limited in number, have been encouraging.

31If the brief review of the faunal and plant remains at Khirbet Shaykh ‘Isa and contemporary sites in the region has value, it is to place sugar in some kind of context, a reminder that sugar was not part of a subsistence economy but rather a specialized product of what was, if not continuously, a lucrative industry. Those environmental remains hint at a relatively varied diet enjoyed by at least some inhabitants of Khirbet Shaykh ‘Isa, similar to that observed at the larger centres elsewhere in Mamluk Jordan, but, to judge from the quantity of handmade pottery, there would have been a larger and poorer element of population at Khirbet Shaykh ‘Isa whose access to food was much more restricted.


32Richard Jones thanks Yona Waksman for the invitation to attend the meeting in Lyon. He is indebted to Edna Stern and Anastasia Shapiro for discussions during the meeting in Lyon, and is also grateful to the former for an advance view of her forthcoming publications.


Brown 2016: Brown R.M., “Faunal distributions from the southern Highlands of Transjordan: regional and historical perspectives on the representations and roles of animals in the Middle Islamic period”, in S. McPhillips, P.D. Wordsworth (ed.), Landscapes of the Islamic world: Archaeology, history, and ethnography, Philadelphia, 2016, pp. 71‑93.

Charles & Hoppé 1997‑1998: Charles M.P., Hoppé C.F., “The archaeobotanical remains”, in McQuitty et al. 1997-1998, pp. 194‑198, 223‑225.

Corbett et al. 2016: Corbett G.J., Keller D.R., Porter B.A., Shelton C.P., “Archaeology in Jordan, 2014 and 2015 seasons”, American journal of archaeology 120, 2016, pp. 631‑672.

Frank 1934: Frank F., “Aus der Araba I Reiseberichte”, Deutschen Palästina-Vereins 57, 1934, pp. 191‑280.

Grey 1994: Grey A., “The pottery of the later periods: An interim report”, in D. Ussishkin, J. Woodhead, “Excavations at Tel Jezreel (1992‑1993): Second preliminary report”, Levant 26, 1994, pp. 51‑62.

Grey 2017: Grey A.D., “The ceramics from Tawahin es-Sukkar and Khirbet Shaykh ‘Isa”, in Jones 2017, pp. 112‑155.

Henton 2017: Henton E., “Animal bone”, in Jones 2017, pp. 169‑177.

Ibrahim et al. 1976: Ibrahim M.M., Sauer J.A., Yassine Kh., “The East Jordan Valley Survey, 1975 (part one)”, Bulletin of the American schools of Oriental research 222, 1976, pp. 41‑66.

Ibrahim et al. 1988: Ibrahim M.M., Sauer J.A., Yassine Kh., “The East Jordan Valley Survey 1976 (part two)”, in Kh. Yassine (ed.), Archaeology of Jordan: Essays and reports, Amman, 1988, pp. 189‑207.

James & Photos-Jones 2017: James H., Photos-Jones E., “The excavations at Tawahin es-Sukkar (Safi) and Khirbet Shaykh ‘Isa”, in Jones 2017, pp. 95‑111.

Jones 2017: Jones R.E., Sweet waste. Medieval sugar production in the Mediterranean viewed from the 2002 excavation at Tawahin es-Sukkar, Safi, Jordan, Glasgow, 2017.

Jones et al. 2000: Jones R.E., Tompsett G., Politis K.D., Photos-Jones E., “The Tawahin as-Sukkar and Khirbat Ash-Shaykh ‘Isa project phase 1: The surveys”, Annual Department of Antiquities of Jordan 44, 2000, pp. 523‑534.

Klein & Flohn 1987: Klein C., Flohn H., “Contributions to the knowledge of the fluctuations of the Dead Sea level”, Theoretical and applied climatology 38, 1987, pp. 151‑156.

LaGro 2009: LaGro H.E., “Ayyubid-Mamluk sugar pottery from Tell Abu Sarbut, Jordan”, Leiden journal of pottery studies 25, 2009, pp. 63‑102.

McQuitty 2007: McQuitty A., “Khirbat Faris: Vernacular architecture on the Karak Plateau, Jordan”, Mamluk studies 11, 2007, pp. 157‑171.

McQuitty et al. 1997-1998: McQuitty A.M., Sarley Pontin M.A., Khoury M., Charles M.P., Hoppé C.F., “Mamluk Khirbat Faris”, ARAM 9‑10, 1997-1998, pp. 181‑226.

Mason & Milwright 1998: Mason R.B., Milwright M., “Petrography of Middle Islamic pottery from Karak”, Levant 30, 1998, pp. 175‑190.

Miller 2017: Miller J., “Botanical remains”, in Jones 2017, pp. 181‑182.

Milwright 2006: Milwright M., “Central and southern Jordan in the Ayyubid period: Historical and archaeological perspectives”, Journal of the Royal Asiatic Society 16, 2006, pp. 1‑27.

Milwright 2008: Milwright M., The fortress of the Raven: Karak in the Middle Islamic period (1100-1650), Leiden, 2008.

Photos-Jones 2017: Photos-Jones E., “The waste in Trench II, Tawahin es-Sukkar”, in Jones 2017, pp.182‑188.

Photos-Jones et al. 2002: Photos-Jones E., Politis K.D., James H.F., Hall A.J., Jones R.E., Hamer J., “The sugar industry in the southern Jordan Valley: An interim report on the pilot season of excavations, geophysical and geological surveys at Tawahin as-Sukkar and Khirbat Ash-Shaykh Isa in Ghawr As-Safi”, Annual Department of Antiquities of Jordan 46, 2002, pp. 591‑614.

Politis 1998: Politis K.D., “Survey and rescue collections in the Ghawr As-Safi”, Annual Department of Antiquities of Jordan 42, 1998, pp. 627‑634.

Politis et al. 2005: Politis K.D., Kelly A., Hull D., Foote R., “Survey and excavations in the Ghawr as-Safi 2004”, Annual Department of Antiquities of Jordan 49, 2005, pp. 313‑326.

Politis et al. 2009: Politis K.D., Sampson A., O’Hea M., “Ghawr as-Safi survey and excavations 2008-2009”, Annual Department of Antiquities of Jordan 53, 2009, pp. 297‑309.

Redding 1984: Redding R., “Theoretical determinants of a herder’s decisions: Modelling variation in the sheep/goat ratio”, in J. Clutton-Brock, C. Grigson (ed.), Animals and archaeology, vol. 3, Early herders and their flocks, Oxford, 1984, pp. 223‑241.

Shapiro 2012: Shapiro A., “Petrographic analysis of the Crusader-period pottery”, in E.J. Stern, ‘Akko I. The 1991-1998 excavations: The Crusader-period pottery, vol. 1, IAA reports 51, Jerusalem, 2012, pp. 103‑126.

Shapiro 2019: Shapiro A., “Petrographic examination of selected pottery vessels from the al‑Waṭṭa Quarter, Safed (Ẓefat)”, ‘Atiqot 97, 2019, pp. 225‑233.

Shapiro et al., in this volume: Shapiro A., Stern E.J., Getzov N., Waksman S.Y., “Ceramic evidence for sugar production in the ‘Akko plain: Typology and provenance studies”, in Waksman 2020, pp. 163‑190.

Solomidou-Ieronymidou 2015: Solomidou-Ieronymidou M., “Sugar mills and sugar production in medieval Cyprus”, in S. Rogge, M. Grünbart (ed.), Medieval Cyprus: A place of cultural encounter, Münster, 2015, pp. 147‑174.

Steiner 2008: Steiner M.L., “Tell abu Sarbut: The occupation of a rural site in the Ayyubid/Mamluk periods”, in M. Steiner, E.J. van der Steen (ed.), Sacred and sweet: Studies on the material culture of Tell Deir ‘Alla and Tell Abu Sarbut, Ancient Near Eastern studies supplement series 24, Leuven, 2008, pp. 157‑196.

Stern 2001: Stern E.J., “The excavation of Lower Horbat Manot: A medieval sugar production site”, ‘Atiqot 42, 2001, pp. 277‑308.

Stern 2012: Stern E.J., ‘Akko I. The 1991-1998 excavations: The Crusader-period pottery, 2 vol., IAA reports 51, Jerusalem, 2012.

Stern et al. 2015: Stern E.J., Getzov N., Shapiro A., Smithline H., “Sugar production in the ‘Akko plain from the Fatimid to the Early Ottoman period”, in K.D. Politis (ed.), The origins of the sugar industry and the transmission of ancient Greek and medieval Arab science and technology from the Near East to Europe, Athens, 2015.

Stern et al., in this volume: Stern E.J., Waksman S.Y., Shapiro A., “The impact of the Crusades on ceramic production and use in the southern Levant: Continuity or change?”, in Waksman 2020, pp. 113‑146.

Taha 2009: Taha H., “Aspects of sugar production in Jericho, Jordan Valley”, in E.K. Kaptijn, L.P. Petit (ed.), A timeless vale: Archaeological and related essays on the Jordan Valley in honour of Gerrit van der Kooij on the occasion of his sixty-fifth birthday, Archaeological studies Leiden University 19, Leiden, 2009, pp. 181‑191.

Von Wartburg 1983: von Wartburg M.‑L., “The medieval sugar industry in Cyprus: Results of recent excavations”, Antiquaries journal 63, 1983, pp298‑314.

Von Wartburg 2001: von Wartburg M.‑L., “The archaeology of cane sugar production: A survey of twenty years of research in Cyprus”, Antiquaries journal 81, 2001, pp. 305‑335.

Von Wartburg 2014: von Wartburg M.‑L., “Ubiquity and conformity: A comparative study of sugar pottery excavated in Cyprus”, in D. Papanikola-Bakirtzi, N. Coureas (ed.), Cypriot medieval ceramics: Reconsiderations and new perspectives, Nicosia, 2014, pp. 213‑245.

Walker 2003: Walker B.J., “Mamluk investment in southern Bilād Al‑Shām in the 8th/14th century: The case of Ḥisbān”, Journal of Near Eastern studies 62, 2003, pp. 241‑261.

Walker 2007: Walker B.J., “Sowing the seeds of rural decline?: Agriculture as an economic barometer for Late Mamluk Jordan”, Mamluk studies review 11, 2007, pp. 173‑199.

Walker 2010: Walker B.J., “From ceramics to social theory: Reflections on Mamluk archaeology today”, Mamluk studies review 14, 2010, pp. 109‑157.

Walker 2011: Walker B.J., Jordan in the late Middle Ages: Transformation of the Mamluk frontier, Chicago, 2011.

Watts et al. 2004: Watts B.G., Bennett M.E., Kopp O.C., Mattingly G.L., “Geochemistry and petrography of basalt grindstones from the Karak Plateau, central Jordan”, Geoarchaeology 19/1, 2004, pp. 47‑69.

Table des illustrations

Titre Fig. 1 – Satellite image (from Google Earth) of (a) the Eastern Mediterranean showing some of the main sites mentioned in the text, and (b) Tawahin es-Sukkar, Khirbet Shaykh ‘Isa, Safi, Feifeh, Mazra’a and Kerak in the south Dead Sea area.
Fichier image/jpeg, 699k
Fichier image/jpeg, 2,5M
Titre Fig. 2 – (a) Site plan of Tawahin es-Sukkar, Safi, showing the chutes (C), buttresses (B), mill wall (MW), AW adobe wall (from a a probable late Ottoman fort) and dump, and (b) the trench locations at Khirbet Shaykh ‘Isa; (c) Satellite image (from Google Earth) of the Tawahin es-Sukkar, Khirbet Shaykh ‘Isa and Wadi al‑Hasa.
Fichier image/jpeg, 1,6M
Fichier image/jpeg, 414k
Fichier image/jpeg, 1,1M
Titre Fig. 3 – Relative percentage proportions of the ware types at Khirbet Shaykh ‘Isa.
Fichier image/jpeg, 191k
Titre Fig. 4 – Moulds and molasses jar at Tawahin es-Sukkar and Khirbet Shaykh ‘Isa (Grey 2017, fig. 3.21c).
Fichier image/jpeg, 766k
Titre Fig. 5 – Classification by principal components analysis of the chemical composition data (obtained by pXRF) for sugar pots (S, yellow), clays (CL, blue) and cooking pots (CP, purple) from Tawahin es-Sukkar and Khirbet Shaykh ‘Isa. Data from Grey 2017, fig. 6.14b and Appendix II.
Fichier image/jpeg, 422k
Titre Fig. 6 – (a) Sugar pot waster (Khirbet Shaykh ‘Isa surface find) and (b) jug waster (Khirbet Shaykh ‘Isa Trench I).
Fichier image/jpeg, 987k
Fichier image/jpeg, 1022k


University of Glasgow, Archaeology Department, Gregory Building, Glasgow G12 8QQ, United Kingdom,

Museum of London Archaeology, United Kingdom (retired),


Volume papier
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search