(The "Aharoni Fortress" Near Quseima and the "Israelite Fortresses" in the Negev, Zeev Meshel, 1994 AD)

The first part of this article is the final report on the excavations at the Iron Age site on a high hill near Quseima, dominating the Dharb Ghazza road to Eilat and Sinai. The second part returns to an old archaeological controversy: who built the sites known as "Israelite fortresses," when, and why? The article argues that the model of self-initiated nomad sedentarization recently proposed by Finkelstein, Herzog, and Eitam has many weak points and does not answer the questions.

THE EXCAVATIONS

The site, named for the late Y. Aharoni, be-longs to the category of "Israelite fortresses," of which some 50 have been discovered in the Negev Highlands. From the earliest stages of research in this area, initiated mainly by Aharoni, to the most recent studies (see, in particular, Finkelstein 1984; Finkelstein 1988; Eitam 1988; Finkelstein and Perevolotsky 1990 and bibliography), no definitive answers have been given to such questions as a precise dating of these fortresses within the Iron Age, the identity of their builders, and the purpose of the sites. The fortress reported here may shed some new light, at least, on the third of these questions.

The Site, Its Location and Discovery

The site stands atop a high, flat, rocky hill (map reference 0871 2120, UTM 6288 3967), the highest point of a steep spur sealing off part of the Valley of Quseima; it is the northwestern approach to Qadesh Barnea (figure 1). The hill is the western end of the spur, which drops down steeply to the north and west, less steeply to the south, and very gently to the east (figures 2-4). Just below the summit, to the west, is a deep saddle, between the spur and the continuation of the ridge toward the north. Converging on the saddle from the north and south is a broad fan of paths, which comes together to form a well-defined road in the pass itself. This "road" is so clearly defined that, viewed from the air, it looks like a motorway. It is indeed so marked erroneously even on the latest maps,1 although no motor vehicle has ever negoti-ated it. It marks an ancient route in this area, perhaps the biblical "road to Shur," represented by parallel lines, some designated with Arabic names, which drain into the valley at the foot of the hill. The Quseima region was always an important junction (Meshel 1979a), and that is still the case today.

EDITOR'S NOTE: This article was originally published in Hebrew by Zeev Meshel and Avner Goren, under the title "`Aharoni Fortress' Another 'Israelite Fortress' in the Negev and the Problem of these 'Fortresses. Eretz-Israel 23 (1992) (Avraham Biran volume): 196-215. The paper appears here under the sole authorship of Zeev Meshel, with the permission of Avner Goren.

Because of the relative altitude, the steepness of the slopes, and the broad view, any observer at the site experiences a "fortress feeling." The defensive advantages of the site are obvious, in fact more clear-cut and convincing than at other similar sites.

The site was discovered in 1975 by officers from an Israel Defense Forces base at Quseima, themselves amateur archaeologists. Following their

Fig. 1. The map of the region and the roads.

report, we surveyed the fortress in January 1976, noting its size, natural fortifications, visual command of the roads at its foot, similarity to the fortresses of the Negev Highlands, and its position at the westernmost of all such fortresses (Meshel 1976: 45-46). Since then the site has been marked on the map of fortresses, where it fits in well with the pattern of distribution (Meshel 1977: 111; 1981: 361; cf. also tables and maps in Finkelstein 1984; 1985; Eitam 1988; however, it is not marked in the maps and lists of Cohen 1979; 1986: 130, map D).

To obtain more detailed information, particularly because the site was in danger of damage, we carried out exploratory excavations there in 1981.2 Two phases of occupation were discovered: the first, a main stage during which the building was erected was in the Iron Age, somewhere between the end of the 11th century B.C. and the 10th century B.C. The sparsity of finds precludes any more precise dating. During a secondary stage, several original walls were made higher and some new walls built; the masonry is irregular, consisting of piles of stones, showing no regular courses or straight lines. One gets the impression of temporary quarters, put up by nomads. A few Roman Byzantine sherds may belong to this stage.

The general plan of the fortress (figure 5) shows quite clearly that it was not completely enclosed by casemates but rather was partly enclosed by a

Fig. 3. An aerial view of the fortress. Looking north.

single wall. Attached were several casemate rooms, some with inner courtyards, and a uniquely shaped gate in an impressive state of preservation.

Except in the east, the building occupies the entire area of the summit, its general plan adhering closely to the topography. The wall runs along the edge of the flat, rocky summit, directly overlooking the steep slopes of the hill. In the east there is no natural boundary just a wall built across the flat rock. Only from that direction could the site be approached easily; in fact, it was there that the gate was discovered. The site is therefore yet another example of the Israelite fortresses in the Negev,3 whose size and shape were both essentially determined by the features of the terrain. In this case the shape was that of an irregular rectangle,4 of maximum dimensions 100 x 40 m. The total area of the site is more than 3500 m2, making it the largest of its kind, several times larger than most of the other fortresses (for a summary of the data on the fortresses, see Cohen 1986: 330; Finkelstein 1984: 191, with bibliography).

Most of the enclosed area is an open courtyard, with the western part separated from the main part by a thin wall; the floor is natural rock (figure 6). The rock hard Eocene limestone (Nizzana formation) forms smooth, stepped platforms, on which the walls were founded. The summit is not completely level

Fig. 4. An aerial view of the fortress. Looking east.

Fig. 5. General plan.

but slopes down gently toward the south, forming two low, discontinuous terraces, each a nonuniform height of approximately 0.5 m. The difference in height between the northern and southern edges of the summit is as much as 6 to 7 m.

The stones of which the walls are built were taken from the local rock, which splits naturally into blocks, many of them more or less rectangular in shape. Many such monoliths, some quite large, are built into the walls, usually at vulnerable positions such as doorposts and corners. They are sometimes stood on end as columns a characteristic of such sites. The width of the walls is not uniform, neither is the building technique.

Gate Complex (Loci 7, 8, 17). The indirect gate (figures 7-9) has two parts: an outer part, 6.40 m x 2.00 m (Locus 7), and an inner part, 5.00 x 3.00 m (Locus 8). Its walls are founded on bedrock, which slopes down from altitude ca. 2.70 m at the gate entrance to altitude 3.37 m in the south. (All altitudes are cited relative to a local 0.00 reference

point, at the maximum preserved height of the northern wall. All points measured are lower and are thus specified by a negative number.) Because of the sloping ground it was necessary to level off Locus 7 with earth, on which flat paving stones were laid. In the southern part, which is not part of the gateway passage, no paving stones were found, only a few patches of a beaten earth floor (-2.94 m), on which a small carnelian bead was found. Excavation down to bedrock (-3.37 m) showed that the thickness of the fill here reached 0.45 m. The fill contained a few sherds of jars and one sherd of Negev ware, as well as a few bone fragments. On the rock itself were ash stains of a fire; we ascribe these to the building stage of the site.

The entire area of the gate was full of debris, lying in such a way as to indicate destruction in an earthquake. The collapsing walls severely damaged the stone paving, which therefore was loose. In the gate entrance, the paving runs up to the rock, and its stones were covered here with a thin layer of ashes. Among the few sherds found in the debris, down to floor level, were a fragment of a burnished bowl, the neck of a juglet, and the rim of a jar. No sherds of Negev ware were found.

The walls of the gatehouse are the highest and best preserved in the site, an indication of their original strength and height. Wall 12, built of large- (up to 1 m) and medium-sized stones, fit well to one another without bonding material (as in all walls at the site), is preserved to a height of seven to ten courses (2.15 m). Judging from the debris, the original height must have been tencourses or approximately 2.75 m. The northern end of the wall is very solidly built. The stones are large, and some of them are set as headers and stretchers (figure 8). This wall is preserved to a height of seven courses. The opposite wall (Wall 10), which is extremely thick (2.10 m) in its northern part, is built of large stones on the outside and fill inside. The thickness may indicate that it was a platform for a tower, defending the gate square (Locus 17) and commanding a view to the east, to-ward the only convenient access to the site. A further indication that this was the case is the fact that the area west of the wall (Locus 16) was buried under deep piles of rubble, probably collapsed from the upper part of the wall. If our conjecture is correct, it follows that this wall or platform was not only the widest but also the highest in the site. The southern part of the wall is narrower (1.80 m), like Wall 1 to the south of the gate, and partly destroyed. A striking feature of this part of the wall is a clearly defined face in the upper two courses, well built halfway across the wall thickness, running north from the gate for 2.5 to 3 m till its disappearance. The height (1.20 m) and location of this face suggest that it might be the western side of a long, narrow space within the wall, to house the bolt locking the door of the inner gate. This may have created a vulnerable point, explaining why the wall was partly destroyed. Alongside the destroyed section, in line with the northern end of Wall 12, is a seemingly natural depression in the rock, measuring 0.35 x 0.25 x 0.15 m. Since a socket was discovered in the inner doorway of the

Fig. 7. The gate complex.

gate (see below), this depression may well have been the socket for a door hinge.

A stepped threshold, built mainly of two long, flat stones, leads into the inner part of the gate (Locus 8). The latter is formed by two parallel walls (Walls 8 and 9) that fit perpendicularly into the eastern wall. Each wall is 3.5 to 4 m long and 0.9 to 1.0 m thick; they are preserved in some places to a height of sixto eight courses. The lower part of Wall 8 also serves as a retaining wall for the earth fill used here, too, to level the natural rock in the passageway and for the beaten earth floor. The western continuation of Wall 8 (Walls 15 and 17) is a low, thin retaining wall that served the same purpose. The floor here is not always clearly defined; on or near it were found two limestone mallets, a few fragments of jars, and two

Fig. 9. The gate. Looking from inside east to Loci 8 and 7.

sherds of Negev ware. Wall 9 also serves as a retaining wall for the fill to its north, where the level was higher than in Locus 8.

The inner passageway of the gate was made nar-rower by the addition of a solid pilaster (Wall 13), 1.80 x 0.75 m. Between it and the threshold, in situ, the excavators found a stone socket for the door hinge (figure 10). It is 0.35 m in diameter and contains two well-polished holes, one inside the other, of diameters 0.12 m and 0.09 m, respectively. The socket, made of the local limestone, was inserted in the earth alongside a few stones that stood on their narrow ends. Two similarly positioned stones were standing at the other (northern) side of the threshold, but no further socket was discovered there. These stones may have supported a wooden door frame, if one existed.

To the best of our knowledge, this is the first indirect gate deliberately built on such a plan, discovered in a site of this category, with various indications that the gate could be locked: the socket for the hinge and the housing for a possible bolt. The closest architectural comparison is the gate of Horvat Teman (Kuntillet `Ajrud, Meshel 1992b:

Fig. 10. The threshold and the stone socket for the door hinge. Looking east.

Fig. 12. Locus 9. Looking west.

105), which is, however, some 150 to 200 years later than the present site.5 Wooden remains found at Locus 7 were defined as Pistacia atlantica and Populus (Appendix 1). They may be the remains of a wooden ceiling of the gate.

Locus 9 (figures 11, 12). A small room adjoined the southern wall, its inner measurements 6.60 x 2.20 m. The floor, part natural rock and part beaten earth, is incomplete and slopes down to the south. The northeastern corner was apparently slightly raised. Along the room, more or less in the middle, stand three stones, possibly bases for wooden columns to support the ceiling. Along the southern and western walls is a stone-built bench, ca. 0.30 m in width and 0.20 to 0.25 m high. The sparse finds include a small clay wheel with a hole in the middle, found on the floor, perhaps part of a spindle; a few pieces of an ostrich egg shell, four jar fragments; and a few ribbed fragments from the Roman-Byzantine period, all in the debris and earth above floor level.

The walls of the room are not uniformly built and are less regular than shown in the plan. Almost all of the southern wall (Wall 1) had fallen down the steep slope on whose rim it was built; it is 1.50 m wide and built of fairly large stones. The other walls show clear indications of two building stages. For example, the three lowest courses of the western wall, which is 0.40 to 0.50 m wide, are built of stones approximately 0.50 x 0.40 x 0.30 m large, carefully arranged and fit together; the four upper courses (preserved at the northern end) are built ofsmaller stones (0.30 x 0.20 x 0.15 m), carelessly placed and not fitted together. The original eastern wall is wider-0.30 to 1.00 m-and built of two rows of stones with a fill in between. This wall, too, has a narrower, shoddily built top section. The northern wall is not uniform. The part east of the doorway is wider, but the part to the west resembles the western wall. The doorway is 0.65 m wide and includes two steps descending to the level of the room. The western doorpost is a monolith, standing 1.05 m high. The eastern doorpost (or perhaps the lintel), measuring 1.05 x 0.40 x 0.25 m, was found fallen at the foot of the steps. The room west of Locus 9 is a mere cell, its rounded wall a row of stones one course high; it may belong to the secondary stage.

Locus 1 + 10. This locus is a cell leaning on the southern outer wall, which is 1.20 m wide at this point. A few isolated patches of beaten earth may be all that remains of a floor. By the eastern wall is a hearth. The eastern wall, one stone thick (0.40 to 0.50 m), survives to a height of four to five courses (1.10 m). The western wall, two stones thick (0.70 to 0.80 m), is bonded with the outer wall of the site and has been preserved to a height of three courses. The meager finds included three Negev sherds, a few jar fragments, two burnished sherds, a flint mallet, and a flint tool (Mousterian).

Locus 12. Locus 12 is another open cell leaning on the wall. At a height of -7.36 m is a layer of

dark ashes, overlying the bedrock, which slopes steeply down to the south. The finds included a few sherds of cooking pots, sherds of jars, and a sherd of a pithos (no Negev ware). Also found here were four flint mallets, a bead made from the head of a cone shell, bone fragments (mostly from the ash layer), and a Roman Byzantine sherd.

Locus 2. The walls of this room show clear signs of the secondary stage. It was originally a room adjoining the wall, which is 1.50 m wide at this point. After the wall had been founded on bedrock, a fill of earth and small stones, about 0.20 m thick, was deposited along it, thus leveling the area. The eastern wall was built on this layer, abutting the outer wall but not bonded with it. Its width is ca. 0.50 m, the width of the row of stones comprising it. Only the eastern third of the room has been excavated. A thin layer of ashes overlying the fill represents the floor. In the late stage, after the upper parts of the original walls had collapsed, the remains of the walls were raised by heaping up stones, and the room was divided into three. These walls are poorly and irregularly built: they are not straight and their courses are irregular. The finds included a burnished sherd of the rim of a bowl, a few jar fragments and two flint mallets.

Locus 11. Locus 11 probably is an open courtyard, adjoining Locus 2. The entrance is at the northeastern corner, between an upright monolith at the end of the eastern wall and a large monolith placedlengthwise; the latter is part of the northern wall, of which a single course of large stones has survived. Within the entrance itself is a stone threshold, with a stone step descending to the south. The stone step was placed on a local layer of ashes. Found near the southern wall was a large group of sherds: a few dozen fragments of jar bodies, a handle and sherds of a cooking pot, fragments of the base of a pithos and a jug, fragments of a decorated jug, and a flint mallet. No Negev sherds were found.

Locus 15 (Figure 13). This was a casemate with a stone bench along its walls, similar in plan to Locus 9. Of the southern outer wall, which was built here directly over the slope, there remains only the inner face, built of stones on the average 0.55 x 0.35 x 0.25 m large. Bonded into this face are the walls of the casemate, each of them one stone wide (ca. 0.55 m). The lowest course of the western wall is built of two consecutively laid monoliths, each more than 1 m long. The northern wall is wider-0.75 m and massive, with an obvious entrance at its corner, 0.75 m wide, with threshold stones. Here, too, a large monolith (0.75 m) is built into the lowest course. This wall has been preserved to a height of more than 1.5 m. The bench is built of a single course. A few patches of ash may represent the floor. At that level, and in the debris above, we found a few fragments of jars, a fragment of a decorated jug, a jug base, and a fragment of a juglet. There were no Negev sherds. Near surface level, above the location of the

Fig.14. Locus 3. Looking west. Notice the monolites.

threshold, were a few ribbed sherds from a single vessel, dating to the Roman Byzantine period.

Locus 3 + 14 (Figure 14). This locus probably was an open courtyard surrounded by a wall, whose western part is built of two monoliths laid lengthwise and its northern part of three monoliths standing vertically, with a fill in between. Bone fragments, a few jar sherds, a sherd and a handle from a jug, a sherd from a cooking pot, and a bead were found near the southern wall at altitude 7.13, in an intermittent conflagration layer overlying the rock and running up to the entrance to Locus 15. Near the entrance were a few sherds from a jug decorated with a brown band. In the northwestern corner were a few green scraps of some cuprous mineral, as well as two sherds from the Roman Byzantine period. No Negev ware was found. Locus 3 + 14, together with Casemate 15 and the adjacent casemate on the east, constitute a single unit. This unit may have had some special function, e.g., a "commander's house." Some corroboration for this speculation may be provided by the remains of imported wood found here: cedar of Lebanon and cypress (Appendix 1). Even if we cannot determine their precise use roofing, facing for the wall, furniture, etc. they nonetheless were undoubtedly used for some special purpose.

Locus 4 + 5. This locus was a casemate adjoining the northern outer wall and the wall that probably cut across the courtyard. The width of thenorthern outer wall here is 1.33 m, and it is built differently from the southern outer wall (below). The other walls of the room are irregular; it cannot be determined whether they were bonded with the outer wall or only built up to it (the same is true of the above-mentioned dividing wall, which is 0.85 m wide). The width of the room is 2.09 m, and its walls are built on the southward-sloping rock. No floor could be detected, other than a few patches of ashes that contained jar fragments, a sherd of a juglet, and a few bone fragments. The large amount of rubble cleared from Locus 5 indicates the original height of the dividing wall. It was built along a stone ledge, 0.75 high, and therefore collapsed.

Locus 6 (Figure 15). Locus 6 was another casemate adjoining the northern outer wall; the wall is 1.25 to 1.40 m wide here. Its faces are built of large and medium-sized stones, and the fill is of small stones, gravel, and earth. Its composition distinguishes Locus 6 from the other walls at the site, recalling Byzantine masonry in the Negev rather than Iron Age masonry. Nevertheless, we believe that the northern wall undoubtedly belongs to the original building stage.6

Both the outer wall and the other walls are built on bedrock, which slopes southward and is fissured into several large blocks. Built into the southern wall are several monoliths; the entrance was probably to their east. A fragmentary layer of ashes containing sherds and a few pieces of ostrich

Fig. 15. Locus 6, plan.

egg shell represent the floor. Most of the sherds were from the body of a single jar. Also found here were two thick sherds of pithoi, the rim of a jug, and the shoulder of a juglet. There was no Negev ware.

Locus 16 (Figure 16). This locus comprised the northeastern corner of the courtyard, adjoining the gate. As mentioned previously, the area was buried under deep nibble, probably fallen from Wall 10. After the locus had been cleared, it was suggested that the area in the shadow of the wall was the kitchen of the site. This was indicated by the complete circumference of a circular clay stove, a concentration of ashes and charcoal indicating a hearth (between the stove and the northern wall) and a layer of ashes overlying most of the area. The finds included a few bone fragments, many sherds from jars, including two rims, a burnished sherd, a few pithos sherds, and four Negev sherds.

The pottery in general resembles that found in other fortress sites, as thoroughly demonstrated by Cohen (1986: 155-63). The bowl base, slipped on the outside and with band burnishing inside (figure 17:1), is similar to Cohen's figure 155.10. The cooking pot (figure 17:2) is a broad, low pot with carinated body; rounded base; two handles; and a broad, thickened lapped rim, either straight or slightly incurved. A few examples are illustrated

Fig. 16. Locus 16.

by Cohen (1986: figures 79:9, 89:5, and, in particular, the assembled figure 157:3, 6-8).

The jar rims are of three types (figure 17:3-5). The variety of types is characteristic of the finds at other sites, too (see Cohen 1986: figure 160).

Four sherds from jugs are shown (figure 17:912). Two are slipped, burnished, and decorated with a brownish-black painted band. The ring base no. 12 recalls a base from Mesora (Cohen 1986: figure 83:13); no. 12 recalls a sherd from Nahal Sirpad (Cohen 1986: figure 116:25). Parallels to the decorated body sherds may also be found (Cohen 1986: 158:23). The pithos (figure 17:7) recalls one from Nahal Sirpad (Cohen 1986: figure 117:6).

The flint tool (figure 17:15) has been defined as Upper Palaeolithic or Mousterian, but since it also bears late flakes it may have been in use during the occupation of the site.7 Perhaps it was used then as a scythe blade; this is a common feature of these sites, perhaps indicating some practice of agriculture (Cohen 1986: figures 121:13, 140:6-13, 145:15). In this connection one should note the large quantity of pollen from domesticated cereals in three of the four samples (Appendix 2).

Among the other finds (which were so few that they have already been listed in our description of the loci) were a few mallet stones of limestone and flint; a bead made from the head of a cone shell (figures 17:13,18, possibly indicating some contact with the Red Sea to the south); a clay weight, probably from a spindle (figures 17:14,18; cf. Cohen 1986: figure 137:12); a few small beads;pieces of ostrich egg shells; bone fragments; and remains of wood (for the definitions see below, Appendixes 1, 3).

According to the general plan of the site and the finds, it may clearly be classified as one of the "Israelite fortresses in the Negev." As far as construction is concerned, the nonuniformity is typical, as it is in the other "fortresses."8 The walls vary in both width and construction; they do not generally run straight, the doorways are different, and so on. Unlike most of the similar sites, the casemates are not consecutive; another special feature of the site is the indirect gate complex, well built in accordance with a preconceived plan, which could be sealed by a door, probably also with a bolt. As in other sites, the masonry is made of the local stone, much use also being made of large blocks as monoliths and columns.

The site is particularly large in fact by far the largest of its type. The location is also unique: it was apparently of crucial importance to erect the fortress at this particular site. Most convincing is the way the building was built to conform to the topography: the outer wall encircles the entire summit, and it was built at the very edge of the cliff. This determined not only the shape of the building but also its size, including the area of the courtyard.

Fig. 17. Pottery types and other finds.

Fig.18. (a) a clay spindle; (b) a bead of cone shell.

The very sparse finds which are nevertheless characteristic and the lack of any conflagration of destruction layer indicate that the site was not destroyed in battle, but deliberately abandoned. As stated, the configuration of the debris in some loci gave the impression that the walls collapsed in an earthquake. The few finds came to light mainly between the lowest debris layers and the floor; in most loci the latter was fragmentary, poorly built, or difficult to identify. Could this indicate that the building functioned for a short time only? Among the sherds were only a few samples of Negev ware less than 10 percent, in stark contrast to the situation in the other fortresses (Cohen 1986). If this ware does indeed indicate the presence of local nomads, its relative paucity is striking.

As to precise dating, the excavation revealed no new data. To our mind the finds themselves do not permit dating the site more precisely than somewhere between the end of the 11th and the 10th centuries B.C. (below). Some poorly built walls showed signs of a second building stage. Our impression was that this stage represented occasional occupation by nomads. It was difficult to decide whether this could be dated to the Roman Byzantine period, on the basis of the few sherds from that period, or whether it was much later and was unrelated to those sherds, which can in fact be found almost everywhere in the Negev.

The botanical and zoological finds make an interesting contribution to the study of the site and its inhabitants' occupations. The botanical finds indicate landscape of flora similar in the main to that of today; if this be considered evidence of climate, the climate was also similar to that of today. A few samples of nonlocal wood cedar and cypress were discovered in the adjacent Loci 3 and 15 of the same building, which included two casemates with a forecourt, adjoining the dividing wall of the courtyard. Perhaps this is not accidental: the finds may represent special furniture, roofing, or wooden ornamentation, and their presence in a building standing apart from therest of the fortress might hint at some special function. At another site, such a distinctive unit was called the "commander's house" (Meshel 1974: 40).

The most interesting palynological find is the high percentage of pollen from domesticated cereals that is, grain at the site. This is probably evidence of agriculture in the area. This is not surprising, as the oasis of Qadesh Barnea is not far off. There, too, as we know, a fortress of the same period was discovered (Cohen 1983). The prospects for agriculture there may be inferred from the present-day situation and from the surviving remains. In good years, the Bedouin still cultivate grain in the loess plains northeast of the site. This does not necessarily mean that the inhabitants of the site were themselves farmers. No scythe blades were found at the site.

The bones found at the site provide no evidence of the wild life, other than ostriches, broken shells of whose eggs were unearthed. Does this indicate that the occupants did not hunt or purchase game from their neighbors? Most of the bones were of domesticated animals, mainly goats and sheep, as well as an ass, cattle, and fowl of various species. No camel bones were found. While the shell bead may hint at a southern contact, the cedar and cypress point to the north, as does the wheel made pottery. These contacts bring us back to the controversy over the function of these sites their builders, and their occupants.

At this point in the research, lacking clear-cut finds that point to the identity of the sites' builders, their inhabitants, and their precise dates, we must concentrate on one main problem: Were the "Negev fortresses" the result of a central initiative, directed by a single authority from the outside,9 or do they represent an independent effort on the part of the desert nomads themselves, whether for reasons of security or as part of the transition from nomadism to sedentarization?10 There are two conflicting theories, which may be summarized as follows:

Initiative: The nomads themselves, by virtue of a socio-economic process.

Causes: Climatic change, economic boom because of control of Arabian trade and mining at Timna, or eco-nomic pressure due to cessation of mining at Timna.

Process: Initial sedentarization = large flocks = oval (el-liptical) enclosures + large court. Consolidation = small flocks + agriculture = smaller enclosures + small court. Peak = smaller flocks + more agriculture = settlements + farmhouses.

Abandonment: Due to climatic change or military occu-pation by King David; or the Kingdom of Israel as-sumes control of trade and establishes hegemony over the Negev by political means.

Theory of Imposed Settlement

Initiative: From the outside Kingdom of Israel aimed at securing the southern border of the kingdom or gaining control of the roads through the Negev and to Eilat.

Causes: Occupation of the Negev by some king of Israel. Process: Establishment and manning of outposts (for

tresses) and small settlements by desert nomads and

Israelites. Occupations of inhabitants: pastoralism +

agriculture, defense of borders and roads in the king's

service.

Abandonment: Nomads' agreement to Israelite rule + cessation of flow of royal funds + departure of Israelite settlers; or decline of Israelite control and dis-continuation of initiative, due to political crisis or military catastrophe (e.g., Shishak's campaign).

Before discussing the two theories, we must clearly stress (see also below) the two unavoidable conclusions to be drawn from the finds. First, the desert nomads were deeply involved in the building and manning of the sites. There is no other explanation for the presence of quantities of Negev ware and the obvious nonuniformity of the masonry. The second conclusion is that there are several hundred small sites, of different types and sizes. The fact that some 50 of them, which are very prominent and possess certain well-defined features, have been defined as fortresses is rather misleading, although some in particular, the Aharoni fortress are built on more suitable topography for that purpose. The term fortress is generally applied to a strongly fortified architectural structure, such as the tower fortresses (below). That characterization is not valid for these 50 sites, which might more aptly be termed "casemate-ringed enclosures." If I continue, following others, to call them fortresses, I do so only out of habit and in the interests of brevity. Besides the casemate-ringed enclosures, some 350 sites have been discovered in the Negev Highlands, mainly during the Emergency Survey (Haiman 1989: 40-51; he also discusses the types of sites and the question of whether they were permanent or temporary settlements). Most are very small residential sites, isolated buildings, farmsteads, and concentrations of potsherds.

As to the central question, I still favor the theory of externally imposed settlement, despite the two aforementioned conclusions and a few weak points, some of which I have specified in the past (Meshel 1974: 87-96; 1979b); other points have been discussed comprehensively in recent papers of Finkelstein, Herzog, and others (Eitam 1988; Finkelstein 1984; 1985; 1988a; Herzog 1983; 1984; 1990; Finkelstein and Perevolotsky 1990), who support the alternative theory. In what follows I shall try to point out some of the weak features of the latter model, demonstrating that the last word has by no means been said and so spurring further research.

Both Finkelstein and Herzog base their conclusions on the following points: Comparison with sites outside the Negev Highlands, which themselves lack a clear-cut definition and plan; // adoption of the model of the socioeconomic process nomad sedentarization which, they believe, explains the plan, architecture, and pattern of the sites.12 They adhere so closely to this model that they adapt the archaeological facts to it, while it would be preferable to derive the model from the facts. The facts as I see them are as follow:

Casemate-ringed Enclosures, Not "Enclosed Settlements"

The basic architectural element, common to all the so-called fortresses, is an outer ring of casemate rooms.13 They were built as a continuous, preplanned ring, not as houses or separate structures adjoining or attached to one another. It is true that in some sites rooms or courtyards were built up against the casemates on the inside; but there, too, the casemates are the basic element. This is the case, contra Herzog (1983: 46-47),14 even in the Hatira fortress. One cannot, therefore, accept the suggestion, first put forward by Herzog and seconded by Finkelstein, that some of the fortresses, including the Aharoni fortress, should be compared with the "enclosed settlements" that he has reconstructed in Beersheba VII, Tel Masos II, Tel Esdar, and Arad Level XII, and which he dates accordingly to the 11 th century B.C. (Herzog 1983: 44-45; cf. 1990: 228-31).15 To my mind, one must reject not only the comparison but even the proposed reconstruction (figure 19). Herzog, like Finkelstein, adopts a sedentarization model of people surrounding themselves with an enclosure, and accordingly draws many buildings of which, in

Fig. 19. The proposed reconstruction of Beersheba Stratum VII (after Herzog).

fact, less than five percent have actually been discovered. In those five percent there is not even one proven doorway. These reconstructions may be good "working ideas," around which and to prove which one might plan and develop a specific excavation. To present them, however, as a fundamental thesis, in which one suggested reconstruction is supposed to support and to be supported by another, itself hypothetical, seems inconsistent with sound interpretation.

A further objection to Herzog's model is that, while attributing the need for enclosures to the settlers' desire "to secure their safety and property," he is forced to reconstruct his enclosed settlements in poorly defensible positions on hill slopes, covering only part of the summit, while buildings left outside the enclosing perimeter testify, he writes, to "the settlers' feeling of security" (Herzog 1990: 228, 232).16 One should not be surprised, of course, that Finkelstein unreservedly accepts the reconstruction, as well as the comparison (Finkelstein 1988a: 242-43),17 as they seem to agree with his own sedentarization model, based on his excavations at `Izbet Sarta, in particular his reconstruction of Stratum III.

Size, Shape, and Area of Courtyard in Most of the Fortresses Due to Topography Rather than Planning or Size of Flocks

Finkelstein's and Herzog's central thesis is that "elliptical sites [represent] the transition to sedentarization" (Finkelstein 1988a; 238; Herzog 1990: 228-29). As I have pointed out, this view is due, in Herzog's case, to his reconstruction of Beersheba Stratum VII, and in Finkelstein's case, to his interpretation of `Izbet Sarta Stratum III. The latter points out that some of the sites of the Negev Highlands are almost identical to it in shape: "They are very similar in overall shape, in size, in the shape of the casemate-like rooms, and in ratio of courtyard to built-up area" (Finkelstein 1988a: 242). Both Herzog and Finkelstein suggest that the large courtyard was necessary to accommodate the settlers' flocks and herds during the early stages of sedentarization (Finkelstein 1985: 373; 1988a: 244; Herzog 1990: 232).18 I have shown elsewhere, however (Meshel 1992a: 294-95), that the general shape and size of most of the fortresses, defined by others as "oval" or "rectangular" (Cohen 1986: 328-29, 331-39),19 are directly attributable to the

Fig. 20. General plan of 'Aqrab Fortress (after Cohen).

topography. The Aharoni fortress is a perfect example, as is the `Aqrab fortress20 (figure 20).

The builders generally chose raised ground, a hill, a spur, sometimes even a steep mountain, and built their fortress on the summit. The casemates generally encircled the entire summit; and the outer enclosing wall was built at the very edge of the summit, directly overlooking the slope. If the summit was oval-shaped, so was the enclosure, as in the `Ein Qadis fortress and elsewhere. If the level ground of the summit was approximately rectangular, so was the fortress, as in the Aharoni fortress. These factors also determined the general size of the enclosures and the area of the courtyard. Sincein most cases the casemates account for most of the built-up area, one needs no higher mathematics or complicated tables (Finkelstein 1984: 191) to understand that the larger the enclosure, the larger the area of the courtyard; the same holds true for the ratio of courtyard to buildings and the courtyard area per capita. But these calculations are meaningless when both size and shape depend on the features of the terrain, not on prior planning or on the size of the flocks. If the statement that "the elliptical sites of the Negev Highlands represented an early stage in the sedentarization of desert nomads . " (Finkelstein 1992: 244) is based on their "elliptical" shape alone, it illustrates how facts may be distorted to fit a preconceived model. If it derives from a functional approach to the structure, which is based on an "analysis of the ratio between the courtyard area of the site and its built-up area" (Finkelstein 1984: 294),21 then it is based on a prior conception and on fortuitous data rather than on deliberate design.

Why were most of the "fortresses" built to conform to the topography? Though here we are knowingly entering the area of interpretation rather than facts, I suggest that the goal was to increase security by perimetric observation and contro1.22 The principle is well demonstrated by the Aharoni fortress; as we have seen, the builders chose a high hill, rising to a considerable height over the surroundings, whose edges afforded a clear view in all directions of the lower land and roads at the foot of the hill. Had they built on the slope, according to Herzog's prescription, the personnel within could not have had an unobstructed view in all directions.

To summarize: if the first stage of sedentarization may be characterized by houses built on a slope around the perimeter of a large oval courtyard, that is not the situation in this case.

The explanation for the courtyard/built-up area ratio variability is to be found, in our opinion, in different social and economic patterns. "Courtyard sites" (most of them oval), were most probably the habitations of people occupied mainly with pastoralism, the flocks being gathered in the courtyard at night, while inhabitants of settlements with relatively small courtyards were less involved in pastoral activity and must have had other sources of livelihood (Finkelstein 1984: 195).

These statements aptly sum up Finkelstein's general approach and his interpretation of the courtyard. To the best. of my knowledge, however, no archaeological evidence for such a function has yet come to light. This is true even for the sites that gave rise to the idea: Beersheba Stratum VII, Esdar, Masos, Arad Stratum XII, and even `Izbet Sarta. Did the flocks gather at these sites on the slope, among the underground granaries? Neither at these sites nor in the fortresses has excavation of the courtyard revealed a layer of dung or any other find that might testify to its use as a stockyard. Even if one doubts the very possibility of discovering such finds, and therefore does not consider the lack of finds to disprove the thesis, one cannot ignore the evidence of livestock pens that have indeed been discovered; some are attached to fortresses (Meshel 1992a: 300) and many have been found "in almost all settlements, sometimes three pens or more in one settlement. Their diameter is 10 to 50 m and they are built of stones of all sizes" (Haiman 1989: 46-47, 53; Cohen 1986: 353).

It should be noted that I have no intention of rejecting the idea which I, too, accept that nomadic pastoral inhabitants of the Negev did indeed inhabit the fortresses. The point of dispute is the thesis that large enclosures, ringed by case-mates, are invariably, sites of sedentarization, and that their courtyards were used as pens.23

The fortresses are not all built in the same kind of position (Finkelstein 1984: 203-6). Some, like the Aharoni fortress or the fortress on Sheluhat Qadesh Barnea, for example, occupy a typical fortress site, affording a view to great distances but not convenient for normal settlement. These sites may be considered military outposts. This in itself does not preclude the possibility that they were built by nomads perhaps as tribal strongholds; but neither does it clash with the thesis that they were built as such by external initiative. Some fortresses, such as the Harocah Fortress, were built on low-lying land, which was more suitable for normal settlement. We do not always have an explanation for the precise location. For example, why is the fortress on Sheluhat Qadesh Barnea, north of the Barnea plateau, located in a typical fortress position while the `Ein Qadis fortress, south of the plateau, was built on a low hill rather than on the nearby Mount Horshah or above the spring itself? The failure to build next to permanent water sources (except for Qadesh Barnea`) and the recourse to cisterns also present problems; but this is the case with regard to both state-initiated and nomadic settlement. Thus, the locations of the sites do not necessarily corroborate either theory.

As for the distribution of the sites in the region, although many of the fortresses guard routes that might have been highways, one cannot postulate clearly drawn lines between identical way-stations built at regular intervals, as in the Nabataean spice road.24 On the other hand, all the fortresses have been found in a certain limited geographic region. Natural conditions here do indeed offer the best prospects for settlement of any part of the Negev, but even here the site distribution is far from uniform. Nevertheless,

one cannot deny the logic of the argument that the limits of the settled area, which coincide approximately with the southern and western limits of the Negev Highlands, constitute not only a topographical, climatic, ecological, and habitational frontier, but also a political and ethnic one, which required a network of strongholds of some sort for its defense.

To summarize this point: judging from the distribution and variety of sites in the Negev Highlands, I believe that they were ordinary settlements, not necessarily military outposts. This is not to say that some external agency could not have initiated the establishment of a network of sites along and near major routes, at key points and along the border of the region controlled by that agency.

The inhabitants of the sites subsisted on both animal husbandry and agriculture; however, the latter was probably the less important occupation (cf. Haiman 1989: 54-55; Eitam 1988: 319). The quantity of precipitation and the large deviation from the multiannual average permitted smallscale cultivation in shallow riverbeds, which receive extra water from runoff; cultivation of large areas, however, was impossible. The great majority of agricultural dams found in the Negev seem to be late. Tests in Nahal Qadesh and Nahal Mitnan have dated such dams to the Roman Byzantine period at the earliest; the dams at Horvat Haluqim are built over an Iron Age structure, so they are later (Goldberg 1984; Bruins 1986: 71-85). If we assume that such installations in the Iron Age were simple, low walls damming small tributaries, such as those in the Matred plateau, it follows that crop yields were relatively limited. Agriculture was at most an auxiliary branch of the economy, inadequate in itself to provide a basis for permanent settlement.

As to the suggestion that the underlying cause of the settlement and its subsistence were due to control of Arabian trade in the area, we shall see later that there is no evidence to that effect.

Finkelstein, faithful to his socioeconomic model, draws the expected conclusions, stating that the first stage consisted of the oval structures with large courtyards. As agriculture grew in importance, a shift occurred, represented in architectural terms by the smaller fortresses. At that stage, isolated houses became more common. As security increased, lone farmsteads were built (Finkelstein 1985: 376).

Herzog takes advantage of the lack of clearcut data on this count, making even bolder proposals. In his first paper he happily unties what he calls the "Gordian knot" created by the categorization of all these sites as Negev fortresses, stating, "This is a rare example of the use of architectural analysis in solving a chronological problem for which the ceramic data were inconclusive" (Herzog 1983: 47). But in a later paper he reties the knot, rejects his own division into "several stages representing a logical, continuous evolution" (Herzog 1984: 32), and ascribes all the sites to one stage of occu-pation (Herzog 1990: 238).25

Clearly, this question of "stages" is also heavily influenced by preconceived notions. The fact is, however, that, except for some signs possibly indi-cating two stages in some of the casemates in the Har Boger fortress (Meshel 1974: 35-44; 1979b: 13, 20) and there, too, the second stage probably dates to the eighth or seventh century B.C. the fortresses have revealed no stratigraphic evidence that they are not from a single period. That is the situation in the small settlements and isolated buildings as well (Haiman 1989: 55). The question of stages was one of the reasons for our excavations of the fortresses at Hatira and Refed, which are very close to each other (Meshel and Cohen 1980). If they are contemporary, why were they built in such close proximity? But if they belong to different stages, why did the builders of the later fortress not erect it on or inside the earlier one, or at least make use of the latter's remains? The same question applies to other pairs of fortresses and buildings. However, as we have already stated, none of these sites show any stratified remains. Neither is there any argument that the pottery, which constitutes the bulk of the finds in all the sites, is not the same.

There is still a possibility that, despite appearances, the sites are not single stage, and that the buildings in different sites are not contemporary. This would require the assumption that brief time intervals elapsed between different stages, perhaps to be discerned in the numerical proportions of pottery of different types rather than in their actual appearance at the sites; only statistical analysis can, we hope, determine the truth. Lacking such a study, the burden of proof is upon those who argue for several stages.

Pottery thus far has been the only evidence for dating of the sites, as against proposals based on historical considerations, which themselves depend on subjective interpretations or preconceived notions. Cohen and others have dated the wheelmade pottery, which was probably imported to the sites (contra Eitam 1988: 327-28), to the tenth century B.C. If we agree with their theory that the sites were destroyed during Shishak's campaign (Cohen 1986: 315, 316, 368, 374, 410, 472), we may point to the last quarter of that century. Others have tried to show that at least some of the pottery types date back to the end of the 11th century B.C. (for a summary of the literature, see Cohen 1986: 402-8). The sites have produced no Midianite ware or other types that might imply a much earlier date. The exact date of construction could be anywhere between the late 11th and the mid-10th century. This wide margin enables some scholars to base their conjectures on historical interpretation or models of some sort (e.g., King Solomon as builder and Shishak as destroyer), rather than on typological considerations. Thus, Herzog proposes an 11th century date (Herzog 1990: 238); and Finkelstein, true to his multistage theory, dates the whole process to the end of the 11th and beginning of the 10th centuries (Finkelstein and Perevolotsky 1990: 78). This means that the oval fortresses were built in the 11th century, with the smaller fortresses and settlements coming somewhat later. Dating is, of course, crucial for both theories: a date earlier than the establishment of the kingdom of Israel would settle the argument once and for all. But as long as no finds permitting an unambiguous dating have surfaced, the controversy will continue.

Another common denominator of the Negev fortresses besides the casemate structure, is Negev ware (Cohen 1986: 385-94). One may dispute its significance and the identity of the makers, but one cannot ignore the fact that Negev ware is a distinctly regional phenomenon, never appearing in the Beersheba Valley sites, whether early (Arad, Esdar, Beersheba, Masos) or late (Arocer, Ira, Uzza, Kitmit). I agree with the attribution of Negev ware to the southern desert nomads, and consider this quite adequate for our purpose. The attempts to narrow down the identification to a specific ethnic group seem unreasonable (Eitam 1988: 333 attributes it to the Hurrians). Herzog, too, accepts the general identification (Herzog 1983: 44). That being the case, it is hard to understand how he can agree with the identification of Tel Masos, where no Negev ware has been discovered, with "the city of Amalek," since the later element is also associated in the Bible with the southern wilderness. Moreover, as no Negev ware has been discovered at Beersheba Stratum VII, Arad Stratum XII, and Tel Esdar, the parallel drawn between those sites and the fortresses is highly questionable (Herzog 1983: 46-47). On the other hand, the large quantities of Negev ware unearthed at all the relevant sites probably indicate that a considerable part of the inhabitants, if not the great majority, were nomads from the southern deserts.26 The obvious nonuniformity of the plans of buildings, as well as the sizes and shapes of walls, rooms, doorways, gates, etc., and the lack of attention to the finer points of construction (angles, straight lines, etc.) clearly point to the prominence of the nomads among the builders, although they need not have been the only builders. The wheelmade pottery, however, may indicate contacts with Judah the probable source of that pottery. The large quantity of such ware at the sites, as well as certain architectural elements four-room houses, use of monoliths as columns, etc., and the use of imported wood (Aharoni fortress) imply that some of the inhabitants may have come from the north.

These data do not necessarily dispose of the self-initiated sedentarization model, but the contacts with the north seem sufficiently strong to hint at some involvement from that part of the country.

Finkelstein, too, ignores the evidence of the Negev ware, stressing the link between the Negev Highland sites and Masos Stratum II. This link is necessary to support his theory that the people of the Negev Highlands engaged in commerce, a theory I find untenable.

Motives for Construction of the Sites: Assumptions, Not Facts

We do not have facts that might indicate the motive for the establishment of the fortresses. Accordingly, it is not surprising that even the champions of the self-sedentarization theory are divided as to the underlying reasons, and also to its end.

Herzog ascribes everything to climatic changes, which he believes were responsible for both the settlement and its end (Herzog 1990: 217, 239). As we have seen, his point of departure is the Beersheba Valley, where he ascribes the initial stage, in the 1 1 th century B.C. to climatic change. He extends this argument to the Negev Highlands.

Eitam attributes the initiation of the process to the cessation of copper mining at Timna, which severed the economic lifeline of the tribes who frequented the area Edomites, in his view and forced them to settle down. Extreme climatic conditions drove them north, he believes, to the Negev Highlands. He does not explain the subsistence base of the tribes before Timna, or what happened to them in the interval between the discontinuation of work at Timna and the mid-11th century, his date for the beginning of sedentarization (Eitam 1988: 333-34).

Finkelstein's proposal is somewhat similar. He believes that the impetus for sedentarization was not the cessation of work at Timna, but on the contrary, the wave of prosperity due to the successful copper mining. This was the reason for the economic prosperity of the southern deserts in the 12th-1 1 th centuries. . . . the desert tribes may have taken part in a possible revival at Timna. . . . Moreover, the share of the local population in the southern trade may even have increased. The strongest evidence for the economic prosperity of southern Israel in this period is to be found at Tel Masos, Stratum II. . . . [The finds there] point to copper trade and close ties between the head of the Gulf of Eilat, the southern Arabah, Transjordan, and the Mediterranean coast, with the Beersheba Valley as a venue of transit . . . . (Finkelstein 1984: 200).

Such proposals not only exaggerate the significance of a few Midianite sherds and a few other finds connected with copper, which were indeed found at Masos,27 but also ignore the fact that there is no proven connection between Masos and the Negev Highland sites. There is therefore no basis for the statement that "the desert dwellers undoubtedly participated in this intensive commerce" (Finkelstein 1984: 200). This reference is to the network of Arabian trade, of which he writes, "the desert nomads eventually seized the monopoly" (Finkelstein and Perevolotsky 1990: 28). This has become the central pillar of Finkelstein's model (1988b). I question the factual basis of the statement (as does Eitam 1988: n. 61). To the best of our knowledge, there has been no significant find at Negev sites to attest either trade or the loss of a dominant role in that trade. Hence the idea that the nomads were encouraged to settle down by such factors is also untenable.

The excavations at the fortresses have revealed no evidence of raising of floors, sealing or piercing of openings, structural changes, or any other data that might point to stages or phases in their occupation. Hence all excavators have classified them as single period, briefly occupied sites (Cohen 1986:410, 433; Haiman 1989: 55). How long they were occupied and whether they were all abandoned at the same time remain open questions. Why they were abandoned is also still a mystery; the answers that have been suggested also are based on assumptions rather than on facts. Herzog blames climatic changes (Herzog 1990: 217, 239; contra Horowitz 1980; Danin 1985). Eitam, for his part, suggests military pressure of some sort, "undoubtedly [by] King David, who instituted a new order in the Negev" (I Chron. 18:12-13; Eitam 1988: 333-34).

Finkelstein, adhering to the theory of commercial motivation, suggests that the decline was due to a loss of control over trade and resumption of a nomadic lifestyle in the Negev Highlands (Finkelstein 1984: 202). The newly consolidated kingdom of Israel seized control of trade in the region, thus destroying the nomads' economic mainstay. Were it not for the intervention of the kingdom of Israel, Finkelstein believes, the process would have continued "and caused the creation of a solid political entity in the desert," somewhat similar to the Nabataean kingdom in a later age (Finkelstein and Perevolotsky 1990: 79).

Scholars who attribute the settlements to the royal Israelite initiative attribute their abandonment to Shishak's campaign, as stated above (Cohen 1986: 316, 413-18, 472; Haiman 1989: 56). But despite theories of war and conquest, no evidence has come to light of large-scale conflagration and total destruction. Although evidence of burning and ashes has been discovered in most of the fortresses, that evidence is of limited extent (Cohen 1986: 335, 338-39) and cannot testify to destruction by war. The scarcity of finds in general, and of pottery in particular, also points rather to a process of abandonment that enabled the departing inhabitants to take their portables and valuables with them. Even if we agree that the sites were indeed abandoned, the reasons are still unknown: harsh climate, interruption of commercial contacts, war, cessation of support by the initiating authority any of these would be sufficient. On this count, therefore, both theories gain equally because we are still unable to gauge the length of the abandonment process, which may have differed from site to site.

In his last paper, Eitam (1988: 331-33) devotes considerable attention to the identity of the inhabitants of the sites. He rejects previous identifications with the Amalekites, Israelites, or Shimonites (the latter being his own), suggesting instead that they

were Edomites. Based on a geographical division of the Negev Highlands and a classification of the sites into nine groups, he identifies "three to five distinct tribal units, in which a secondary division can be distinguished . . . " (Eitam 1988: 329, 333, map on p. 340). This identification, writes Eitam, may also provide a "geographical historical basis for deciphering the Edomite genealogical lists. . . . " In that connection Herzog's discussion of the concept of ethnicism is interesting. He rejects the possibility that cultural evidence in general, a fortiori archaeological finds, which represent a limited cultural sector, can be used to identify ethnos (Herzog 1990: 240). At the current stage of knowledge, however, the finds do not furnish sufficient data to determine the ethnic identity of the inhabitants of the fortresses.

Israelite control of the Negev in the Iron Age II was based on a network of tower-fortresses, such as those excavated at Uzzah, Arad, Qadesh Barnea, and lately at `Ein Hazevah (Cohen 1988; 1991). It is generally believed that these structures were built "to carry out an administrative function and to defend the southern frontier of the Judaean monarchy" (Finkelstein 1984: 183; Eitam 1988: 318).28 If, as Finkelstein also believes, "the Qadesh Bar-nea fort became the focus of rule over the Negev Highlands" (Finkelstein 1984: 202), where were the intermediate fortresses, between this focus and Arad? How can we explain the lack of way-stations along the routes to Qadesh Barnea and `Ein Hazevah? And "above all, south of Qadesh Barnea along important Darb Ghazza, the main route to the Gulf of Eilat, not even one fortress has been found" (Finkelstein 1984: 191-92). One should note that these are two of the weak points Finkelstein detects in the theory of royal initiative, in his discussion of the Israelite fortresses; these points are even more crucial in regard to Horvat Teman (Kuntillet `Ajrud), even farther away in the desert, but its links with Judah and Israel are unquestionable (see, e.g., Meshel 1992b).

What, then, was the function of the tower fortresses? If they were indeed Judahite administrative centers, how were they reached? If some kings of Judah indeed sent delegates to Eilat and Ezion Geber (via Horvat Teman during its existence, or by some other route), where were the way-stationsand the chain of defensive fortresses so necessary to protect the roads from the hostile desert and its nomads? It would seem that Finkelstein's comparison with the Nabataean highway or the Egyptian road along the Sinai coast (Finkelstein 1984: 190) is not always valid.

The tower fortresses, which existed from the ninth or eighth century B.C. on, testify to Israelite control of the Negev Highlands, which was maintained without the need for a dense network of forts. Such control was first achieved under Saul, David, or Solomon, who conquered the region and achieved a state of cooperation and coexistence with the inhabitants: the nomads maintained Israelite control of the region, its borders, and roads (like the desert tribes in the Byzantine period, Finkelstein 1984: n. 7), and were rewarded with food, grain, or some other commodity. The authorities directed them to establish settlements in the interests of the king, which they built and largely populated. In time, they began to develop some agriculture, resulting in the construction of various small settlements and farmsteads scattered over the Negev Highlands.29

This settlement boom, entirely the fruit of external (Israelite) initiative, was short-lived. With the cessation of royal backing it declined rapidly, and the settlements were abandoned. What caused the cessation? This is one of the weak points of our reconstruction, but as I have tried to demonstrate, it is also a weakness of the other models.

Elsewhere I have suggested that "when the local inhabitants recognized the existence and permanency of this authority, they accepted it, and either integrated themselves into it or lived quietly on its borders. As the need for these Negev strongholds diminished, they were abandoned or deserted" (Meshel 1977: 133). At any rate, it is clear that in the next period, that of the tower fortresses, one fortress was adequate to secure Israelite control, "for the Negev Highlands were ruled, not with the sword, but by virtue of political wisdom, backed up by military power ......(Finkelstein 1985: 376).3°

Our main aim has been to reveal the weak points in the theory that the settlement of the Negev Highlands was a process of nomadic sedentarization. An arid or semiarid zone is not a natural focus for sedentarization, and a model that may be adequate in other zones is not necessarily valid in the desert. Let us hope that additional facts, rather than theories, will some day bring us closer to the solution of one of the most interesting problems in the history of the Negev.

NOTES

1In detailed 1:100,000 maps.

2The excavations took place in January 1981, under the auspices of the Tel Aviv University Department of Archaeology and the Israel Defense Forces Officer for Archaeology in Sinai. They were directed by Zeev Meshel and Avner Goren. The associate director was Eitan Aialon. The area supervisors were Moshe Sadeh, Adar Avisar, Ittai Hadari, and Ronit Steinberg, then Tel Aviv University students. The excavations were carried out by students of the School of Field Studies at Sedeh Boker and local Bedouin. Surveyors and draftsmen were Israel Watkin and Michael Feist. The finds were drawn by Rudika Menashe and Ada Peri. Finds were photographed by Montana Billings. The field photographer was Zeev Meshel, except for the gate, which was photographed by Yoram Weinberg. Simultaneously, brief exploratory excavations were conducted at two nearby sites at the east-ern foot of the spur, near the ancient road, which is still in use today. One of these sites is a small fortress, con-temporary with the Aharoni fortress (map ref. 0883 0121, no. 1 in Figure 1); the other is a Nabataean way station (map ref. 0886 0118, no. 2 in figure 1).

31 shall continue to use the term "fortress," purely for convenience, despite the dispute as to the nature of the sites (see below). Cf. Meshel 1992a: 294.

4Cohen would have classified it, in his terms, as rectangular; (Cohen 1986: 325). I have already argued (Meshel 1992a: 294-95; see also below, n. 20) that his classification method, which ignores the relationship between the shape of most of the fortresses and the topography, is schematic, technical, and superficial.

5In only four fortresses Hatira, cEin Qadis, Nahal `Aqrab and Nahal Sena` were gates proper, possibly planned in advance, discovered. All of these, however, afford direct access and are much simpler (Meshel 1992a: 296; Cohen 1986: 343-44).

6Similar masonry may be found in the fortresses of Sheluhat Qadesh Barnea and Mishor ha-Ruah (Meshel 1992a: 300).

71 am indebted to Avi Gofer for the definition.

8Meshel 1992a, where I tried to show that the non-uniform style of masonry is a characteristic feature of the fortresses.

9The leading exponent of this view in recent years is Cohen (1986: 471).

10The major representatives of this view, with some variations, are, in this order: Rothenberg (1967: 88-92), Eitam (1988: n. 71), Finkelstein (1990) and Herzog (1990).

11 "The existence of an enclosed settlement in Stratum VII at Beersheba is a point of departure for the analysis of the phenomenon of the Negev fortresses" (Herzog 1984: 28). Herzog's reconstruction of Beersheba Stratum VII is extremely dubious, as is the comparison itself (see below). Finkelstein touches on the subject in connection with his excavation at cIzbet Sarta, which leads him to a comprehensive discussion of sedentarization (Finkelstein 1988a). This should be noted to emphasize that the point of departure for his discussion (Finkelstein 1988a: 237) is his own model, which relies essentially on his reconstructed plan of `Izbet Sarta III (Finkelstein 1988a: 73/80; 238-39). Here, again, some scholars reject the reconstruction, the stratigraphy, and the dating (Cohen 1986: 428-35). Especially interesting is the sharp criticism of Bunimovitch (1989: 93), which is aimed not only at the reconstruction but at the whole model relating sedentarization to oval structures with courtyards.

12Finkelstein's two articles (Finkelstein 1988b; Finkel-stein and Perevolotsky 1990) proposing a "historical eco-logical model for human activity in Sinai and the Negev" betray a basic tendency to interpret historical processes in the light of models.

13For the plans of the sites see Cohen 1986; represen-tative examples may be found in Meshel 1992a and Haiman 1989. In a very few cases the ring of casemates is incomplete.

14Since Herzog wrote his article, additional fortresses have been discovered in addition to `Ein Qadis in the region he calls the "Wilderness of Beersheba," such as cAqrab, Horshah, Elah, Sirpad. In all of these the basic element is the casemate. At any rate, Hatira is the only one with so many other rooms abutting the casemates.

15See also Herzog 1992: 233, 239. The publication of these reconstructions in that book grants them a degree of legitimation at least, in the editors' view, which I find rather surprising.

16See here his reconstructions of Beersheba, figures 6, 7; and Tel Masos, figure 5. The latter, in particular, breaks all records. Cf. also Miriam Aharoni's plan of Arad Stratum 12, as published in Aharoni 1981: 181. Herzog is so fascinated with the idea of locating the settlements on hillslopes that he counts it as one of the "characteristics" of the "enclosed settlements," and accordingly corrects Kochavi's "erroneous" statement that the settlement at Asdar was built in a circle on the hilltop. Herzog's explanation of the "considerable relative size of the central courtyard" at Asdar a courtyard that is nothing but a theoretical invention is untenable. For a differently reasoned critique of the interpretation of these sites see Cohen 1986: 420-21. Elsewhere Herzog suggests that the settlements were built "preferably on a slope, probably for protection against high winds .....(Herzog 1992: 233). We must reiterate that none of the fortresses were built on a real slope.

17Elsewhere he goes even further, to my mind, com-paring the fortresses to Early Bronze Age sites (Finkelstein and Perevolotsky 1989: 31).

18Thus, Herzog (1990: 232) writes: "This peculiar ar-chitectural design therefore reflects the inhabitants' sub-sistence base and the social framework of their lives." Eitam, too (1988: 317), stresses the central role of the courtyard, which is "the nucleus of such a habitation . . . among herd-rearing cultures." He even draws comparisons with Africa and England (which are, to my mind, invalid, as the conditions in Africa and England are quite different from those obtaining in the Negev, and the livestock bred there were cattle). Eitam believes that the courtyard was used to protect communal property and was therefore of crucial importance; he therefore suggests calling the sites "courtyard structures." Unlike Finkelstein and Herzog, he does not associate the sites specifically with a sedentarization process, and does not consider the size of the courtyard to reflect the stage reached in such a process. At one point, however, he argues that the size resulted from the adaptation of the buildings to the terrain (Eitam 1988: 318), while elsewhere he relates the size of the courtyard to the "village's property" (Eitam 1988: 328), that is, the size of the flock, contradicting himself.

19Finkelstein agrees with Cohen's classification, but gets needlessly entangled by his insistence on the special significance that he ascribes to "ovalness." In his tables and maps (Finkelstein 1985: 368, figure 1, and in less detail in 1984: 191-92), he lists 17 oval sites, 7 square and rectangular, and 16 "other shapes." He goes on to show that some of them, including, of course, some of the rectangular sites, are rather large and have large courtyards and large courtyard/built-up area ratios, while some of the oval sites have relatively small areas, courtyards, etc. By his own criteria, do these sites not "represent the traditional stage to sedentarization"? And if so, why emphasize the oval shape? Should the reason for his insistence be sought at `Izbet Sarta and the need to make the latter's shape a rule?

20 we particularly mention `Aqrab as one of the few sites whose plan Cohen presents with contour lines, indicating the topography (Cohen 1986: pl. 125). In most cases Cohen does not indicate topographical features.

21In light of his assumption concerning courtyard size and the axiom that the size was deliberately preplanned, Finkelstein draws statistical distribution maps and divides the Negev Highlands, in what he himself considers an arbitrary way, into two units, whose common boundary runs along "the latitude of Avdat." This border is at variance both with the terrain and the precipitation chart (see, e.g., Katsnelson 1979: 56-57). Incidentally, as far as rain is concerned, the more significant element is its variability or instability, which reaches more than 40 percent in this region (Katsnelsen 1979) 56-57. Moreover, Finkelstein's statement that the geographical pattern of Iron Age sites in the Negev resembles that of earlier sites in the Negev is only true in a broad sense. Updated distribution maps have been made in recent years and should be taken into account.

220n this point, cf., Haiman 1989: 42; we reject the camouflage factor.

23Such casemate-ringed enclosures appear in different periods and have various functions. Suffice it to mention Masada or the khans along the Nabataean Spice Road these sites, of course, had nothing to do with sedentarization or concentration of flocks. We have taken the comparison to an absurd extreme, to emphasize the absurdity of the thesis.

240n the Nabataean road see Meshel and Tsafrir 1974. The situation described by no means implies that there was no connection between the fortresses and roads (Finkelstein 1984: 190). The location of some fortresses, such as the Aharoni fortress and Nahal Loz, cannot be explained otherwise.

25His reason was that "at any rate, the civilian-agricultural nature of the settlement seems obvious and it differs from the royal-settlement pattern of the fortress at Arad." In other words, he derives everything from the comparison with the Beersheba Valley.

260f the 569 vessels published by Cohen in his study, 39 percent are Negev ware (Cohen 1986: 36364). Cohen believes that Negev ware was made by the Kenites. He, too, somewhat ignores the significance of the lack of Negev ware in the Beersheba Valley. He has his own way of comparing the fortresses with the Beersheba Valley sites, suggesting that the inhabitants of the "fortresses" may have come from there (Cohen 1986: 434, 471).

27These were mainly several sherds of small melting bowls and two pieces of copper ingots, first thought by the excavators to have come from Feinan and now be-lieved to have originated in melted down vessels (Kempinski, personal communication). It is important in this context to cite Cohen (1986: 395): "A prominent phenomenon in the Iron Age finds is the very few metal articles: one copper axe, two arrowheads, one sword, one copper needle."

28The tower fortress at Qadesh Barnea was built on an oval fortress of the fortress family; perhaps this is no accident, and the fact that the later structure was built by the monarchy hints that the same may be true of the earlier one.

29Finkelstein answers his own rhetorical question, "What then causes nomads to adopt a sedentary life style?" by attributing the process to "an improvement in security conditions, influence of nearby cultures and, above all, creation of external economic resources," and he adds another question: "What were the external fac-tors?" (Finkelstein 1984: 198-200). Does our model pro-vide a better answer to these questions and conditions? Finkelstein refers to his own doubts on this count (1984: 202, n. 6).

"He rejects our view, described above, and citing only the last sentence, as being "simplistic and un-explained" (Finkelstein 1984: 197). Is his view any different?

APPENDIX 1. THE BOTANICAL MATERIAL OF

THE AHARONI FORTRESS (QUSEIMA)

By Nili Liphschitz (1985)

Institute of Archaeology, Tel Aviv University

Tel Aviv, Israel

Botanical material, including charred pieces of wood, was gathered during the excavations at the site. Samples of 0.5 to 1.0 cubic cm, taken for wood identification, were aspirated in absolute ethyl alcohol, dipped in methyl-benzoate-celloidin for 24 hr, rinsed in benzene, transferred to 55° C paraffin. Blocks were made in paraffin, and cross as well as longitudinal, tangential, and radial sections were prepared. The identification of the wood species, based on the three-dimensional structure of the wood, was made microscopically from these sections, by comparison with reference sections prepared from systematically identified living trees. The results are presented in Table 1.

As can be seen from the results, the nine samples gathered were made of six species of trees and shrubs: four samples were of Pistacia atlantica (Atlantic pista-chio), and a single piece of each of the following species: Tamarix (x 5; five organs in each whorl in the flower) (tamarisk), Retama reotam (whitew broom), Populus (poplar), Cedrus libani (cedar of Lebanon) and Cupressus sempervirens (cypress). The first four species

TABLE 1. Botanical Remains from the Aharoni Fortress

Locus Reg. no. Item

16 25 Pistacia atlantica

4 2 Pistacia atlantica

2 7 Pistacia atlantica

7 14 Pistacia atlantica

10 5 Tamarix (x 5)

15 30 Cedrus libani

3 1 Cupressus sempervirens

7 10 Populus

5 3 Retama reotam

still grow in the region, and probably grew there in the past as well. The last two were brought either from the Mediterranean region of Israel or from Edom, while ce-dar of Lebanon was imported from either Lebanon or Cyprus. Cypress and cedar of Lebanon are characterized by a straight, tall trunk, suitable for the roofing of large-size rooms.

APPENDIX 2. POLLEN ANALYSIS OF FOUR SAMPLES

FROM THE AHARONI FORTRESS

By Aharon Horowitz (1985)

Institute of Archaeology, Tel Aviv University

Tel Aviv, Israel

Plant

% pollen Sample

No. 36,

Locus 3

alt. 7.13 No. 34

Locus 10

alt. 7.41 No. 35

Locus 12

alt. 7.95 No. 23

Locus 7

alt. 2.90

Gramineae (wild) 45 14 11 3

Chenopodiaceae 17 17 4 4

Compositae 8 58 22 7

Tamarix 2 1 28 7

Cereals 20 4 31 60

Typha 1

Acacia 1

Populus 2

Malvaceae 2

Pinus 1 1 4

Olea europaea 1 1

Astragalus 1

Artemisia 1 3 5

Ephedra 4

Plantago 1

Liliaceae 1

Cruciferae 1

Spores 3

The most significant find is the relative quantity of pollen from domesticated cereals, possibly evidence of grain fields, i.e., practice of agriculture.

1994 THE "AHARONI FORTRESS" NEAR QUSEIMA 65

APPENDIX 3. ANIMAL BONES FROM THE AHARONI FORTRESS

By Shlomo Hellwing and Y. Agiman (1982)

Institute of Archaeology, Tel Aviv University

Tel Aviv, Israel

Bone fragments (Table 1) from 11 baskets, weighing a total of 10.5 kg, were analyzed at the laboratories of the Tel Aviv University Institute of Archaeology. The bones were classified by species and skeletal elements. The identifications were based on comparison with os-teological material in the Institute's collection. Some of the samples were extremely small bone fragments, sometimes too small to be identified. Others were poorly preserved. Nevertheless, 27 samples were identified, as presented in Tables 2 and 3.

TABLE 1: Animal Bones by Locus TABLE 2: Minimum Number of Individuals (MNI)

Locus Reg.

no. Identification Animal No. fragments MNI Age

Ovis/Capra

20 1 Adult

4 2 Ovis aries/Capra hircus (sheep/goat) Bos

1 1 Adult

7 39 Equus asinus (ass), Ayes (fowl); 37: Ovis/ Equus 2 1 Adult

Capra; Ayes 4 1 Adult

13 Ayes

11 6 Ovis/Capra; 19: Ovis/Capra; Bos taurus Total 27 4

(cattle)

12 31 Ovis/Capra; 18: Ovis/Capra; 33: Ovis/

Capra

15 29 Ovis/Capra

16 25 Ovis/Capra

TABLE 3: Skeletal Elements

Skeleton

element Ovis/

Capra Bos Equus Ayes Total

Mandibula (lower jaw) 1 1

Skull fragments 4 4

Scapula 1 1

Humerus (forearm) 1

Metacarpus 1

Pelvis 1

Femnur (thigh) 2

Tibia (shin) 2 2

Astragalus (heel)

Molars 2 2

Metapods 2 2

Costae (ribs) 6 7

Vertebrae 2 2

Total 20 2 4 27

Conclusions

1. Fauna at the site consisted of: sheep, goat, cattle, ass, fowl. The group consisted of domesticated animals only, to the exclusion of wild animals, implying that hunting played no part in the inhabitants' economy. In-terestingly, only in a few of the 13 fortresses excavated by Cohen have bones of wild animals been found, and even these represent a relatively small number of spe-cies: ibex, gazelle and hare (Cohen 1986).

2. Classification of samples and minimum number of individuals: The number of samples was 27, minimum number of individuals 4. Bones of sheep and goats con-

stituted the bulk of the finds, being found in 9 of the 11 baskets. The most common skeletal elements were limbs. The ass was represented by teeth only.

3. The most commonly used animals were small ruminants, such as sheep and goats. Most of the remains were of adult animals, indicating that they were bred for meat, milk, and wool. Other mammals and fowl played a smaller part as a source of protein in diet. The absence of wild animals, such as gazelles, is surprising and deserves study. The finds at the nearby Nabataean site, which included, besides the species found here, a pig's mandibula and two camel phalanx II, also included no wild animals' bones.

REFERENCES

Aharoni, M.

1981 The Pottery of Strata 12-11 of the Iron Age Citadel at Arad. Eretz-Israel 15 (Yohanan Aharoni volume): 181-204.

Bunimovitch, S.

1989 Review of I. Finkelstein, The Archaeology of the Period of Settlement and Judges (1986), Archeologia 2: 91-94 (Hebrew).

Bruins, J. H.

1986 Desert Environment and Agriculture in the Central Negev and Kadesh Barnea During His-torical Periods. Nijkerk: Midbar Foundation.

Cohen, R.

1979 The Iron Age Fortresses in the Central Negev. Bulletin of the American Schools of Oriental Research 236: 61-79.

1983 Qadesh Barnea. Jerusalem: Israel Museum (Hebrew).

1986 The Settlements of the Central Negev in the Light of Archaeology and Literary Sources during the Fourth to First Millennia B.C.E. Un-published doctoral dissertation, The Hebrew University, Jerusalem (Hebrew and English).

1991 Hadashot Arkheologiot 96: 38-39 (Hebrew).

Danin, A.

1985 Palaeoclimates in Israel: Evidence from Weathering Patterns of Stones In and Near Archaeological Sites. Bulletin of the American Schools of Oriental Research 259: 33-43.

Eitam, D.

1988 The Settlement of Nomadic Tribes in the Negeb Highlands During the 11th Century B.C. Pp. 313-40 in Society and Economy in the Eastern Mediterranean (c. 1500-1000 B.c.), eds. M. Heltzer and E. Kipinski. Leuvan: Peeters.

Finkelstein, I.

1984 The Iron Age "Fortresses" of the Negev Highlands: Sedentarization of the Nomads. Tel Aviv 11: 189-209.

1985 The Iron Age "Fortresses" of the Negev-Sedentarization of the Desert Nomads. Eretz-Israel 18 (Nahman Avigad volume): 366-79.

1988a The Archaeology of the Israelite Settlement.

Jerusalem: Israel Exploration Society.

1988b Arabian Trade and Socio-Political Conditions in the Negev in the Twelfth Eleventh Centu-ries B.C. Journal of Near Eastern Studies 47: 241-52.

Finkelstein, I., and Perevolotsky, A.

1989 Sedentarization and Nomadization in the Southern Deserts in Early Periods. Cathedra 52: 3-36 (Hebrew).

1990 Processes of Sedentarization and Nomadization in the History of Sinai and the Negev. Bulletin of the American Schools of Oriental Research 279: 67-88.

Goldberg, P.

1984 Late Quaternary History of Qadesh-Barnea. Zeitschrift fair Geomorphologie 28: 193-217. Haiman, M.

1989 Shepherds and Farmers in the Qadesh Barnea Region. Sedeh Boker: Sedeh Boker College (Hebrew).

Hellwing, S., and Agiman, Y.

1982 Internal report, School of Archaeology Labo-ratories, Tel Aviv University.

Herzog, Z.

1983 Enclosed Settlements in the Negeb and the Wil-derness of Be'er Sheva. Bulletin of the Ameri-can Schools of Oriental Research 250: 41-49.

1984 Enclosed Settlements in the "Negeb of Judah" and the Wilderness of Beersheba. Ca-thedra 32: 23-34 (Hebrew).

1990 The Beersheba Valley from Nomadism to Monarchy. Pp. 215-41 in From Nomadism to Monarchy Archaeological and Historical Aspects of Early Israel, eds. I. Finkelstein and N. Na'aman. Jerusalem: Yad Yizhak Ben-Zvi (Hebrew).

1992 Settlement and Fortification Planning in the Iron Age. Pp. 231-74 in The Architecture of Ancient Israel, eds. A. Kempinski and R. Reich. Jerusalem: Israel Exploration Society (Hebrew).

Horowitz, A.

1985 Internal Report (15.8.85), Institute of Ar-chaeology, Tel Aviv University.

1980 Palynology, Climate and Settlement Pattern. Qadmoniot 51-52: 80-86 (Hebrew). Katsnelson, J.

1979 Rains of the Negev. Pp. 51-73 in The Land of the Negev, eds. A. Shmueli and Y. Grados. Tel Aviv: Ministry of Defense (Hebrew).

Liphschitz, N.

1983 Mehqarim Dendroarcheologiim 121. Quse-ima, Institute of Archaeology, Tel Aviv University.

Meshel, Z.

1974 History of the Negev in the Time of the Kings of Judah. Unpublished Ph.D. disserta-tion, Tel Aviv University (Hebrew).

1976 Sites in the Vicinity of Quseima. Hadashot

Arkheologiot 57-58: 45-46 (Hebrew).

1977 Horvat Ritma An Iron Age Fortress in the

Negev Highlands. Tel Aviv 4: 110-35.

1979a History of the Roads in the Negev. Pp. 297-307 in The Land of the Negev, eds. A. Shmueli and Y. Grados. Tel Aviv: Ministry of Defense (Hebrew).

1979b Who Built the "Israelite Fortresses" in the Negev Highlands? Cathedra 11: 4-28 (Hebrew).

1981 The History of "Darb el-Ghazza" The Ancient Road to Elat and Southern Sinai. Eretz-Israel 15 (Yohanan Aharoni volume): 358-71. Jerusalem: Israel Exploration Society (Hebrew).

1992a The Architecture of the Israelite Fortresses in the Negev. Pp. 294-301 in The Architecture of Ancient Israel, eds. A. Kempinski and R. Reich. Jerusalem: Israel Exploration Society (Hebrew).

1992b Kuntillet `Ajrud. Pp. 103-9 in The Anchor Bible Dictionary, vol. 4, ed. D. N. Freedman. New York: Doubleday.

Meshel, Z., and Cohen, R.

1980 Refed and Hatira: Two Iron Age Fortresses in the Northern Negev. Tel Aviv 7: 70-81. Meshel, Z., and Tsafrir, Y.

1974, The Nabataean Road from `Avdat to Shear 1975 Rammon. Palestine Exploration Quarterly 106: 103-18; 107: 3-21.

Rothenberg, B.

1967 Negev, Archaeology in the Negev and the Arabah. Ramat-Gan: Massada (Hebrew).

By Steve Rudd: Contact the author for comments, input or corrections.

Go To Start: WWW.BIBLE.CA