1The Middle Chalcolithic site of Tel Tsaf (5200-4800 BC) yielded a rich assemblage of ceramics, chipped and ground stone assemblages, numerous figurines, faunal and floral remains, the earliest metal object found in the southern Levant, and seal impressions on clay (Dag and Garfinkel 2007; Garfinkel et al. 2007, 2014; Ben-Shlomo et al. 2009; Graham 2014; Streit and Garfinkel 2015b; Freikman and Garfinkel 2017).
2Among these finds are several unusual ceramic objects bearing depictions of various architectural details and rich painted or plastic decorations. These objects have been identified as fragments of architectural models. One of them, partly excavated by Garfinkel’s expedition and partly by that of Rosenberg, has already been published (Rosenberg et al. 2017). Eight additional objects of this kind are presented in this article for the first time.
3Small-scale representations of buildings or architectural details, usually made of clay and rarely of stone or other materials, are well attested and widely discussed in the archaeology of the Near East and Europe. They are discussed under different names, such as naoi/naïskoi, tabernacles, édicules, cubicular shrines or cubiculae, jar models, pot models, maquettes, and snake houses (Bretschnider 1991; Epstein 1998; Muller 2001, 2002, 2016; Zevit 2001; Kletter et al. 2015: 28; Katz 2016; Garfinkel and Mumcuoglu 2018a, 2018b). Technically, they all represent architectural constructions. As a general rule only features visible from the outside, or only the façade, are represented on them, but in some cases they also show installations and even people located inside. These models may be decorated with plastic or painted depictions of architectural elements, or zoomorphic or anthropomorphic creatures (Amiran 1978: 52-53, pls. 66, 115; Miroschedji 2011a: 89*-91*, 2011b; Kletter et al. 2015; Katz 2016). The interpretation of what these constructions represent is the subject of debate: while some of them are clearly dwelling houses, in different periods others may represent temples or granaries, or are even considered smaller versions of the ossuaries that in turn are models of full-scale architectural constructions. These architectural models, however, are mostly found in Middle Bronze Age or later strata, and only a few are attested in the Early Bronze Age or earlier periods. The assemblage presented here therefore helps to fill a lacuna in our understanding of this phenomenon in the late prehistoric southern Levant.
4Tel Tsaf is a large village site located in the Jordan Valley, Israel, and dated to the Middle Chalcolithic period (fig. 1). It was discovered and surveyed for the first time by Tzori in the 1950s (Tzori 1958: 49). Small-scale excavations at the site were conducted by Gophna on behalf of Tel Aviv University in 1979 (Gophna and Sadeh 1988-1989). Large-scale excavations were conducted at Tel Tsaf in 2004-2007 under the direction of Y. Garfinkel on behalf of the Hebrew University of Jerusalem. The site is currently under excavation by D. Rosenberg and F. Klimscha in a joint expedition of the University of Haifa and the German Archaeology Institute (Rosenberg et al. 2014). Radiometric measurements have indicated that the site dates from 5200-4800 BC (Streit and Garfinkel 2015a). This phase postdates the Wadi Rabah culture of the Early Chalcolithic and predates the Ghassulian culture of the Late Chalcolithic; it is hence termed “Middle Chalcolithic” (Garfinkel 1999, 2009, 2014).
Fig. 1. The location of Tel Tsaf in the central Jordan Valley.
5The excavations revealed large and impressive courtyard houses, numerous granaries with a total capacity of dozens of tons of grain, burials, a well, and various installations (fig. 2; Garfinkel et al. 2007, 2009, 2020; Garfinkel 2010; Ben-Shlomo et al. 2012). The Chalcolithic period is represented at Tel Tsaf by two strata, Stratum 3 (phases a-b) and Stratum 4 (phases a-c). During the excavation four large building complexes were partially excavated, two (I and II) belonging to Stratum 3 and three (I, III and IV) to Stratum 4, only Building Complex I continuing through both strata. The building complexes comprise rectangular or circular buildings used for living, storage, cooking, and other domestic activities, as well as circular structures that served as silos. These silos could contain a large amount of grain and were an important part of the economy of the site.
Fig. 2. The main excavation area of Tel Tsaf.
6Fragments of at least nine different items interpreted as the remains of architectural models were identified at Tel Tsaf. Hereinafter we present a detailed description of their morphology and spatial context, as well as the results of petrographic analysis.
7M1. Silo model (fig. 3). This previously published model (Rosenberg et al. 2017), made by the coiling technique, is in the form of a closed vessel reminiscent of the bag-shaped jar typical of this period (Garfinkel 1999: 173). Instead of a regular rim, however, it has a dome-shaped roof with a rectangular opening in the side. The vessel is decorated with some 40 clay balls applied to the dome. It was found in Building C70 of Architectural Complex I. This model is interpreted as a representation of a silo with clay balls depicting the cereal grains.
Fig. 3. M1: a nearly complete model of a silo.
8M2. Upper part of a building (fig. 4). Six fragments (Locus 584/Basket 1265, Stratum 4b) could be restored as the upper part of a building model, of which two sides are mostly preserved and the other two are fragmentary. The dimensions of the model can be reconstructed as 13 × 11 × ~12 cm. It was found in Locus 584 within the large rectangular Building C70. The model was manufactured from a delicate light gray/beige clay with few inclusions. On at least one side one can see that the coiling technique was used to produce this model (fig. 5), after which the outer surface, but not the inner one, was largely smoothed by hand. It was decorated with black and red paint. First, black stripes 7-8 mm wide were painted along three of the surviving intersections between the sides of the model, creating a hatched pattern. In addition, a tiny fragment of the black pigment remains in the lower part of the entrance on side 1. In the second stage, red decoration was painted on top of the black stripes. On one side the decoration is in the form of a solid block of red color, and on two of the remaining sides there is a net-like motif of crisscrossed lines. Although much of the model is missing, the shape of the structure can be cautiously reconstructed.
Fig. 4. M2: the upper part of a building, with painted decoration imitating the wooden beams that supported the roof.
Fig. 5. M2: coils seen on the inner side of the fragment.
9Side 1 (fig. 4 A, E, I). One of the edges of side 1 of the object, probably a square or rectangular panel, is relatively well preserved. It is decorated with a black stripe running along two of the three preserved edges. In addition, the inner part of side 1 is decorated by red lines about 1 cm wide in a net pattern, crossing at right angles.
10Side 2 (fig. 4 C, G, K). Although side 2 is severely damaged, the existing part preserves one black stripe running along its edge and remains of red lines, indicating that it was similar to side 1.
11Side 3 (fig. 4 B, F, J). This side is enclosed between sides 1 and 2, creating an angle of ca. 70˚ between them. The meeting point between these two sides protrudes slightly above side 3, thus creating an architectural detail reminiscent of a fronton and a pommel, the tip of which is broken off. This pommel was probably decorated by a geometric or zoomorphic detail, which has not survived. Side 3 is slightly convex. Its lower part, as well as the lower part of the fronton, is decorated with red paint.
12Side 4 (fig. 4 D, H). Only a small portion of side 4 is preserved. One can see, however, that part of it is not broken and forms the upper part of an opening located near the corner between sides 1 and 3. The preserved part of side 4 indicates that at least its upper part was decorated with red paint.
13This architectural model may represent a one-roomed structure of roughly square or rectangular plan with a gabled roof. The unbroken edge on side 3 may be interpreted in two ways, as the upper part of a window or the upper part of an entrance. In the former case this would be a building of longhouse type (one with the entrance located in the front, narrow wall), while in the latter case this would be a broadhouse building (one with the entrance located in one of the long walls), with the entrance in the corner of the building rather than its center. One should stress that the architectural model is dissimilar in shape to the elongated rectangular Building C70 where it was found, and hence it seems to represent a different building.
14M3. Fragment (fig. 6). A small fragment measuring 5 × 3.5 × 3.5 cm (Locus 383/Basket 795, Stratum 4). It was manufactured from a fine light gray clay with few small inclusions. It is fully covered by thick dark red paint. One may interpret it as a lintel or part of a fronton of a structure similar to model M2, with a small fragment of the door or window frame. This interpretation is not conclusive, since no clear connection to the roof is visible.
Fig. 6. M3: a small fragment of the upper part of a building, decorated with red slip.
15M4. Fragment (fig. 7). A small convex fragment measuring 6 × 4 × 0.4 cm (Locus 26/Basket 54, Stratum 4a). The fragment is made of a fine light gray to white clay with very few small inclusions. It was formed by the coiling technique; while the wall was thoroughly smoothed on the outer surface, the coils are clearly visible on the inside, a phenomenon that is unattested in the case of regular pottery. A single black stripe some 4 mm wide runs across the outer side of the fragment, with a pattern of parallel red lines on either side of the black stripe. The decoration, however, is almost totally faded and is barely visible today. The rounded wall can be interpreted as a representation of roofing or the rounded wall of a building similar to M1. However, due to the small size of the fragment this interpretation must remain inconclusive.
Fig 7. M4: a small fragment decorated with red paint.
16M5. Fragment (fig. 8). A small portion of a wall measuring 5.5 × 4.6 cm (Locus 662/Basket 1407, Stratum 4b). This fragment was made from a different type of raw material: instead of the delicate pale-colored clay often used to make architectural models as well as fine pottery vessels, it was made of the brown clay with numerous gray grits that was generally used for manufacturing storage vessels. The model was shaped by the coiling technique, the barely smoothed remains of which are clearly visible on the inner side of the fragment. The wall is slightly concave. The fragment is decorated by a pattern of parallel red stripes about 5 mm wide. An additional red line is painted at an angle of 30˚, creating a geometric design of some kind. This model was also decorated in black paint, remains of which are visible on one side of the preserved surface. This combination of unsmoothed coils and polychrome decoration on the inside is unattested on regular pottery, and hence this fragment may cautiously be interpreted as part of an architectural model. The fragment seems to represent part of the wall of a structure that may possibly have been circular, like the silo model.
Fig. 8. M5: a small fragment decorated with red paint.
17M6. Fragment (fig. 9). A small fragment of a model measuring 4 × 2.5 × 1.3 cm (Locus 291/Basket 596, Stratum 3a). It is made of a light brown clay with large inclusions that is more typical of the storage vessels found at the site. However, no such or similar profile was attested at the site. This fragment consists of a connection between a flat surface painted red and a convex surface covered with a design of parallel black stripes each 3-4 mm wide, joining a perpendicular line running along the connection with the second side. We may cautiously interpret it as the edge of the roof of a circular building, possibly a granary. However, the small size of the fragment precludes further speculation on this matter.
Fig. 9. M6: a small fragment of a corner, decorated with black paint.
18M7. Fragment (fig. 10). A fragment of a convex wall measuring 4.2 × 2.5 cm (Locus C377/Basket 785, Stratum 3a). This fragment is made of a fine beige clay with few small grits. It is painted with a pattern of thin dark red parallel lines, each 2-3 mm wide. Its clay differs from that of any pottery vessel found at the site, and the irregularity of its profile and its decoration are incompatible with pottery examples. Hence, despite its small size, we may cautiously interpret this fragment as belonging to an additional model.
Fig. 10. M7: a small fragment decorated with parallel red lines.
19M8. Fragment (fig. 11). A fragment measuring 7.6 × 4.9 × 3.6 cm (Locus 662/ Basket 1407, Stratum 4b). It is made of the brown clay with numerous gray grits that is characteristic of storage vessels. The fragment’s two sides (1 and 2, fig. 11C-E), meeting at an angle of roughly 90˚, were joined together. The joint was imperfectly smoothed by hand and is still easily detected. Only a small portion of the third side (3, fig. 11F, on the right) is preserved, but one can clearly see that it represents a lintel protruding above an opening. The most striking feature of this model is a number of plastic decorations applied to the joint between sides 1 and 2 (fig. 11A, B, D). Although this plastic decoration is severely damaged, one can detect a circular object with a diameter of some 22 mm, remains of three parallel stripes descending to side B, and an additional stripe at an angle of 90˚ to the former ones. All of these details are decorated by parallel short lines cut into the clay. In addition, remains of the red paint that originally covered this plastic decoration, and possibly the whole vessel, are still visible. All in all, it seems that this was a zoomorphic figure, possibly a feline, depicted as it climbs on the roof of the model toward the opening.
Fig. 11. M8: a small fragment decorated with incised plastic application.
20M9. Fragment (fig. 12). A fragment consisting of a corner between two walls, measuring 3.1 × 2.5 × 2.3 cm (Locus 221/Basket 416, unstratified). It is made of a fine whitish clay with a few small to medium-sized grits. This fragment is undecorated.
Fig. 12. M9: a small fragment of a corner, undecorated.
21As a whole, the fragments of this assemblage display several characteristics.
- The models were manufactured by the coiling technique, like the regular pottery vessels of the period, or by attaching separately prepared panels to one another. In most cases the artisan used a fine clay that is not typical of regular pottery vessels. Moreover, while the coils or joints between different parts are not visible from the outside, no serious attempts were made to smooth or obliterate them, and the coils are clearly visible on the inner walls of four of the nine models.
- In most cases sharp angles are seen at the joints between the sides of the fragments. Consequently, these fragments represent cuboid or in some cases spheroid models.
- Some fragments include plastic details that can be interpreted as representations of architectural elements, such as roofs, lintels, or openings.
- In at least four cases the models were carefully painted with parallel lines or complex net designs, possibly imitating wooden beams incorporated in the walls and roofs of the real structures.
- One of the fragments also bears remains of plastic figurative decoration.
22All things considered, we can propose an interpretation of the nine fragments presented above, together with the complete model published previously, as representations of actual architectural structures (Rosenberg et al. 2017). Although fairly limited, this evidence adds important information on the superstructure of prehistoric structures in our region, which for obvious reasons usually eludes archaeologists.
23As the assemblage of architectural models from Tel Tsaf published here is small, consisting of only nine fragments, the potential for spatial analysis is rather limited (figs. 13-14). However, a number of observations can be made.
- The models were found in Strata 3 and 4 of Tel Tsaf.
- Most of the fragments were found on floors or destruction layers inside rooms of buildings, namely Buildings C70, C612, C173, and C377. In addition, two fragments were found in the courtyard of Building Complex II.
- M2 was found in Building C70. This building was probably of special importance and was characterized by numerous finds, including a large quantity of exquisite pottery with “Tel Tsaf” decoration. This observation is even more striking with regard to the silo model (M1) found in Building C70 (Rosenberg et al. 2017). It reinforces the proposition that this building, which yielded the only two restorable models found at the site or any other Middle Chalcolithic site, had a special function differing from (or in addition to) such mundane functions as accommodation, cooking, or storage.
Fig. 13. Spatial distribution of the model fragments in Stratum 3.
Fig. 14. Spatial distribution of the model fragments in Stratum 4.
24Four of the fragments were analyzed by thin thin-section petrography at Ariel University (models M2, M4, M5, and M6; samples Tsaf 115-118). All of the fragments were manufactured from local clay used for regular pottery vessels (Petrographic Group 1, fig. 15, see below), most probably quarried in the vicinity of the site.
Fig. 15. Photographs of thin sections of architectural models M4 and M6 (taken with plain polarized light). Abbreviations: CC = calcareous concentration; CP = clay pellet; FR = microfossil; GR = grog; LS = limestone.
25Group 1 represents a generally highly calcareous clay, probably related to alluvial and fluvial soils, with various calcareous inclusions (fig. 15, “LS”, “CC”, limestone, chalk, micro-fossils calcareous concentrations) and clay pellets (fig. 15, no. 1, “CP”), grog or shale (fig. 15, no. 2, “GR”). Other components are very low in quantity. The porosity is very variable and is possibly related to both firing temperature and atmosphere, and quantities of organic material mixed into the clay. A sample of 129 pottery sherds from Tel Tsaf was studied by petrographic analysis. Clearly 107 samples belong to this group, and six additional ones possibly belong to it. This is by far the most common local clay used for pottery production, confirming that the building models were made locally. This fabric mostly has a silty calcareous matrix that is dark (sometimes opaque) under crossed polarized light and brown (sometimes more reddish) under regular polarized light; particles are mostly single-spaced, and voids vary from 10-20% of the slide area. The main inclusions are calcareous, calcareous concentrations, chalk, and limestone; these usually add up to 20-25% of slide area, and are in fine to medium size sand (up to 1 mm), grains are sub-angular to rounded in most cases. Microfossils appear in variable quantities. Another component appearing in most examples is clay concentrations that in some cases can be identified as grog due to their angular shape. In other cases these could be clay pellets or shale showing a more rounded shape and sometimes a gap around them due to firing. Quartz appears in small quantities, 1-5% in angular silt size. Other inclusions appearing in small quantities are opaque minerals and chert. This clay represents a fluvial or alluvial silty soil possibly representing mixtures of alluvial Holocene soils and marl Lisan Formation soils, both common on and around the site of Tel Tsaf. Certain mixtures between the alluvial and Lisan components may also exist. Somewhat similar clays have been reported for ancient pottery in the Beth Shean region and at Pottery Neolithic Sha‘ar Hagolan (Cohen-Weinberger 2002, 2007: 550, Group B; Cohen-Weinberger and Goren 2011: 216, Group TR, fig. 10, no. 1a), but in our fabric the mixture with alluvial soils and the lack of basalt and travertine represents a difference, and so this is likely a more local variant.
26Examples of architectural models found in non-mortuary contexts are extremely rare in the record of the prehistoric southern Levant. A fragment of an architectural model was reported at Jericho (Garstang and Garstang 1948: 66, pl. VIIb). It seems that this large clay model also depicted furniture or other details inside the building, and hence is more reminiscent of the models from Anatolia and Europe or “soul-houses” from Egypt (see below) than later models from the southern Levant. In the Late Chalcolithic period one may mention a fragment that possibly represents the façade of an architectural model with a male human figure applied to its left side found at the Sde-Eliyahu Water Reservoir (Davidovich and Garfinkel 2007: 443, figs. 4, 5). In the Early Bronze Age, a broadhouse model was found at Tel Arad (Amiran 1978: 52-53); notably, this model is painted with a geometric design somewhat similar to the large restorable model M2 from Tel Tsaf. A fragment of a façade decorated with an anthropomorphic figure was found at Early Bronze Age Tel Yarmuth (Miroschedji 1982).
27Mallowan and Cruikshank Rose reported a building model with gabled roof from the site of Al Ubaid (1935: 30-31, figs. 15, 17). According to the authors, similar roofs are still widespread in the area of the site of Arpachiyah in Iraq (Mallowan and Cruikshank Rose 1935: 30).
28In Anatolia, architectural models were found at Pre-Pottery Neolithic Çayönü and are thus among the earliest examples of clay architectural models in the world (Bıçakçı 1995; Erim-Özdoğan 2011a: 229, fig. 47). A Dark Faced Burnished “rectangular vessel” found in Stratum 2 of Sumaki Höyük in the Tigris Basin seems to be another example of an architectural model (Erim-Özdoğan 2011b: 29, fig. 33c).
29Models of buildings dated to the 4th millennium BC (the so-called “Temple period”) were found in Malta (Trump 2002: 192-195). One of them is unique, since it is made of stone.
30Numerous architectural models, many of them with gabled or domed roofs and openings described as “tectomorphs,” are attested in Southeastern Europe in both the Neolithic (Naumov 2013: 69-72; Bugaj et al. 2014) and the Chalcolithic, for example at Ovcharovo (Bailey 1990: 39), Podgoritsa (Todorova 1998: 389, fig. 10B), Popudnya (Palaguta and Starkova 2017), Volodymyrivka (Burdo et al. 2013), and other sites (Perrot and Ladiray 1980: 116; Palaguta 2012).
31One should also mention clay and wooden models of houses, sometimes called “soul houses”, which were placed in Egyptian tombs as early as the Early Dynastic period and which were related to the world of the dead (Petrie 1907; Piccione 1999; Barker 2020; Eschenbrenner-Diemer 2021).
32The similarity of the models from Tel Tsaf to the ossuaries found in large numbers during the Late Chalcolithic period in Israel is especially interesting (fig. 16). Although found in mortuary contexts, many of them have similar features, often depicting longhouses. Remarkably, ossuaries, which may in at least some cases represent houses, may have their own even smaller copies (Perrot and Ladiray 1980: 28). The roofs of many of these ossuaries are gabled, some of them bear a small knob on the fronton, and they are decorated with net patterns of painted stripes, making them very similar to model fragment M2 from Tel Tsaf. Such gabled roofs, net-pattern designs, and sometimes a pommel on top of the fronton are attested on many Late Chalcolithic ossuaries, uncovered at Hadera (Sukenik 1937), Benei Beraq (Kaplan 1963), Palmahim (Gophna and Lifshitz 1980), Azor and Ben Shemen (Perrot and Ladiray 1980), Kissufim (Goren 2002: 32, fig. 4, no. 8.1), Shoham (Brink 2005), Tayiba (Yannai and Porath 2006), Peqi‘in (Shalem et al. 2013), and elsewhere. In a way, model fragment M2 may be regarded, at least iconographically, as a small-scale predecessor of Late Chalcolithic ossuary roofs, and the architectural models of Tel Tsaf, and in particular the gabled roof of M2, may be the first attested evidence of the “diachronic progression from type to type” of the Chalcolithic funerary receptacles (Ilan and Rowan 2019: 265).
Fig. 16. A Late Chalcolithic ossuary with gabled roof from Azur.
33Our presentation of this unique assemblage of architectural models from Tel Tsaf raises three important questions: what is the origin of this phenomenon, what did these models symbolize in the eyes of the inhabitants of Tel Tsaf, and how were they used in the everyday life or rituals of the Tel Tsaf culture.
34The parallels presented above show that the fragments of architectural models of various types from Tel Tsaf comprise unique evidence among the archaeological assemblages from the prehistoric southern Levant published so far. The phenomenon of models of gabled structures has not so far been recognized in the domestic archaeological record of the southern Levant during the Neolithic or Chalcolithic period. Scholars studying this issue usually assume that structures in the prehistoric southern Levant had exclusively flat roofs made of wood and clay, mostly due to the relatively dry climate (Aurenche 1981: 153-154; Wright 2009: 128-129). Perrot and Ladiray, for instance, excluded the possibility of local development of the Late Chalcolithic ossuaries and proposed to seek their origins elsewhere (Perrot and Ladiray 1980: 115). Certainly, models of gabled structures are better represented in prehistoric Anatolia and Mesopotamia, and are especially common in Neolithic and Chalcolithic southeastern Europe, where clay gabled roofs are still widespread (Aurenche 1981: 155-156). In the Near East, in contrast, roofs are usually flat, made of beams and branches sealed with clay, sometimes supported by posts (North 1960: 21-23: Porath 1992: 44-45; Palaguta 2012: 53). However, although flat roofs are indeed predominant in the Near East, the previous existence of gabled roofs was already proposed by Sukenik (1937: 29). In addition, the ethnographic evidence shows that roofs of this type are attested in the traditional architecture of the southern Levant, and therefore may not have been alien to the Chalcolithic culture as well (Epstein 1998: 14-16, fig. 20).
35In the absence of additional evidence from the southern Levant, however, we may assume that the practice of making clay models of structures in general and gabled roofs in particular was foreign to or rare in our region in the Late Neolithic and Early Chalcolithic periods. We suggest that the practice, originating in Europe, was probably imported to Tel Tsaf via the Ubaid sites of Anatolia and Mesopotamia in the Middle Chalcolithic period, and was further incorporated into funerary ossuaries of the Late Chalcolithic period all over the southern Levant. The appearance of the models in these regions on the one hand, and new evidence for cultural and trade contacts between the Halaf and Ubaid cultures on the other, seem to reinforce this possibility (Rosenberg et al. 2010; Freikman et al. 2021).
36What type of architecture did these models represent? They certainly represented buildings, in which painted stripes imitated beams incorporated into the walls and roofs. They may represent dwelling houses (Sukenik 1937: 25; Albright 1940; Anati 1963; Callaway 1963; Mastin 1965; Perrot and Ladiray 1980: 111-113) or granaries (Bar-Yosef and Ayalon 2001). The second possibility is especially interesting in light of the recent interpretation of a rounded model from Tel Tsaf as a silo (Rosenberg et al. 2017). Unlike the rounded artifact, model M2 represents a rectangular building. This division, however, is not clear cut, as some of the rectangular buildings found at Tel Tsaf (for instance in Architectural Complex III in Area C) may be interpreted as storage facilities. The massive walls more than 1 m wide, with narrow spaces between them, may actually be foundations raised above the surface, while the actual storehouse was built on top of them. This kind of setting may be compared to buildings of similar function found in Mesopotamia (Streit 2012: 79-82; Freikman et al. 2021: 5). It is noteworthy that Bar-Yosef and Ayalon, based on ethnographic evidence, have interpreted the “legs” supporting the Late Chalcolithic ossuaries as poles intended to protect the grain stored in the real structures from humidity (Bar-Yosef and Ayalon 2001: 38-40). The possibility that the ossuaries symbolized granaries based on ethnographic evidence from India and Egypt was also raised by Porath (Singh 1977: 238-239; Porath 1992: 47). Therefore, we may propose that architectural model M2, and perhaps the models represented by the other fragments, were meant to represent another granary variant. It is unlikely that the models from Tel Tsaf were actual ossuaries like the Late Chalcolithic parallels: they are too small to contain bones, and none was found in a burial context or with human bones. However, it is possible that they symbolized the burials excavated in or near the nearby granaries, and that in the Late Chalcolithic period these two phenomena were physically merged, and bones were deposited inside the models that had become ossuaries.
37A much more difficult question is how were these models actually used by the inhabitants of Tel Tsaf. In the absence of additional data, no conclusive answer can be given. Moreover, it is unclear what was the purpose of investing in the tiny architectural details decorating their surfaces. Architectural models dated to the Middle Bronze Age and later have been found at different sites since the early 20th century (Bretschneider 1991; Muller 2002, 2016; Katz 2016). Although they were initially explained in various ways, for example as snake houses, today they are usually interpreted as temple or shrine models, especially since attributes of Canaanite deities have been found associated with two of them, at Ashkelon and Hazor. This explanation of the models of Tel Tsaf cannot be completely excluded. However, when dealing with much earlier examples such an explanation is hard to accept in the absence of more concrete evidence, such as anthropomorphic or zoomorphic figurines found in the same context, and other possibilities must be considered. For instance, these models could be explained as foundation deposits (Gallis 1985) or children’s toys (Bıçakçı 1995: 110). However, the quality of the workmanship and the effort invested in these models suggest that they were more significant than toys, and parallels found in ritual contexts in Mesopotamia as well as the similarities of the painted designs to the later ossuaries may imply that they were embedded in the pattern of religious beliefs in Chalcolithic society (Nativ 2008). In other words, we may speak of the models as an element in some kind of ritual of sympathetic magic. Such practices were proposed in the early 20th century by Frazer (1947), who claimed, on the basis of numerous anthropological observations, that all or most traditional societies share a belief in the ability to establish a supernatural connection with other entities through the imitation of the object. Various manifestations of this phenomenon are known in the ancient Near East from the Neolithic period onward (Freikman and Garfinkel 2009). In accordance with this model, we may propose that the miniature models represented real granaries and that their very existence, or some kind of manipulation of them, was believed to ensure the filling of the real granaries with grain. However, as they were found not in spaces that can be defined as “ritual” but in regular household contexts, it seems that rituals related to the models were performed by inhabitants of Tel Tsaf who belonged to the same extended family in the course of their everyday activities, rather than by the whole community during major festivals. It is possible that in the Late Chalcolithic period the memory of this ritual resulted in the transfer of these models of behavior into the mortuary context, thus connecting the world of the dead with the fertility rites of the living. In this regard the possible interpretation of the architectural models from Çayönü as funerary gifts is especially interesting (Bıçakçı 1995: 110).
38The finds published in this paper together present a phenomenon that is barely known in the southern Levant and seemingly alien to the local culture of the Late Neolithic to Middle Chalcolithic periods. In the absence of other evidence, the gabled and domed roofs depicted seem to point to southeastern Europe or Anatolia and Northern Mesopotamia as the source of a cultural influence that was imported in the context of trade contacts. Although models of structures do not appear in significant numbers in the archaeological assemblages of the southern Levant in the Chalcolithic period or Early Bronze Age, certain details, particularly the gabled roof, that are otherwise alien to the local cultural horizon do appear in numerous ossuaries in the Late Chalcolithic period. In general, we may propose an explanation for the function of the architectural models that is based on the concept of sympathetic magic. The models were part of a ritual practice intended to establish and ensure a magical link between the world of the ancestors and the fertility of the fields, eventually helping to fill the numerous granaries of Tel Tsaf with agricultural production. Some of the elements of this symbolism were inherited by Late Chalcolithic mortuary practices and found expression in the ossuaries of this period.