The current excavations were undertaken in three squares near the fortress’s southwest corner (Figs. 2, 3): one square was opened outside the fortress, to its south, and two—inside the fortress, along its west wall. The first of the two excavations aimed at determining the depth of the archaeological remains inside the fortress and at planning a pathway into the fortress. The second excavation concentrated on uncovering the fortress’s entrance gate, whose jambs were unearthed during conservation work at the site.
The two excavations yielded ancient remains directly beneath the surface. Inside the fortress, its gate and the foundations of its west wall were unearthed. Outside the fortress were a few remains from the Late Bronze Age and the Iron Age I, as well as remains of a building that probably dates from the late Byzantine or Early Islamic periods (sixth–seventh centuries CE).
 
Late Bronze Age and Iron Age I. The north part of a large refuse pit (L116, L117) was uncovered the square opened to the south of the fortress. The pit yielded numerous potsherds, some bearing signs of burning, fragments of burnt mud bricks, and ash (Figs. 4, 5). The pit cut into a thick layer of fill (L100, L110) that contained a few Late Bronze Age potsherds, including a handle of an imported Cypriot milk bowl (Fig. 6:1). The refuse pit yielded a few Early Bronze Age potsherds, including a holemouth jar (Fig. 6:2), and folded jar handles from the Intermediate Bronze Age (not drawn), but most of the pottery dated from the Late Bronze Age (Fig. 6:3, 4) and Iron Age I, including a burnished flask decorated with red concentric circles (Fig. 6:5) and a bichrome jar fragment (Fig. 6:6). A fragment of a leg of a basalt vessel, possibly a chalice (Fig. 6:7), was also retrieved from the pit. The limited excavation hampers the interpretation of the remains, but the refuse pit probably dated from the Iron Age I. The ash, the scorched pottery and the burnt mud bricks suggest that remains from a building destroyed by fire were discarded in the pit. This clearance operation probably also raked up fragments of earlier vessels.
 
Roman Period. A foundation trench (L118) for a late Byzantine – Early Islamic building (below) yielded a body fragment of a glass bowl with raised ribbing, dating from the Early Roman period, and part of a glass juglet handle from the Late Roman or early Byzantine periods (not drawn). Additional glassware sherds from these periods were found at several locations in the area (Y. Gorin-Rosen, pers. comm.). These represent settlement periods on the tell, of which no architectural remains were unearthed in the current excavation.
 
Late Byzantine or Early Islamic Period. The square opened to the south of the fortress yielded sections of the foundations of three walls (W105, W107, W124), which were set into the refuse pit and the adjacent fill. The walls, preserved only one course high, seem to belong to the south part of a room (width c. 4.5 m). The date of the building’s construction is unclear, A few fragments of ribbed bag-shaped jars bearing white painted marks (not drawn) were found in the foundation trench of W107 (L118). Such jars are characteristic of the late Byzantine or Early Islamic periods (sixth–seventh centuries CE). The poor state of preservation of the walls makes it impossible to determine the length of the building’s use and when it was destroyed. Its walls are truncated in the north, and thus seem to have been destroyed when the fortress was built.
 
Mamluk Period. The fortress on the summit of the tell was erected during this period. The excavation unearthed additional parts of the fortress, including the lower part of the outer walls and an entrance gate that was set mid-way along the west wall. The walls to the south and north of the gate (W101, W127, respectively; Fig. 7) were built of large fieldstones. Their lower part, up to about half the height of the entrance gate, is remarkably massive (width 2.2 m), and is constructed of two rows of large fieldstones with an inner core of small stones; the upper part of the walls is narrower (width c. 1 m).
Sarcophagus panels are incorporated in the fortress wall in secondary use, and they seem to mark the transition between the wide foundations—rising above the level of the entrance threshold—and the narrower walls of the building’s superstructure.
The entrance gate (Fig. 8) was built as a stepped opening. Its outer jambs delineate a recess in the wall (width 3.2 m, depth c. 0.8 m); in the center of the recess is a square opening (width 1.7 m). The gate and the jambs were built of finely dressed ashlars. The entrance jambs were placed on a broad threshold (L141; Fig. 9) consisting of two wide stones. It is abutted on the west by a tamped earth surface (L142) with a bedding of small fieldstones (L144). The surface was probably part of the road that led to the gate from the west. Due to the limited excavation area, it is impossible to fully ascertain the stratigraphic sequence. However, Stone Bedding 144 appears to be related to the gate’s construction, as it probably served to level the ground when preparing the location of the entrance in the west wall prior to the construction of the gate itself, i.e. prior to laying the threshold and installing the jambs. On both sides of the gate, behind the jambs, were two socket depressions (Fig. 9), indicating that the gate had a double door that opened inward. A long, narrow channel (L146; Fig. 10) was set inside the fortress wall on the north side of the entrance, about a meter above the level of the threshold; it probably served as a groove to insert a long wooden beam that bolted the doors from the inside. Opposite it, on the gate’s south wall, is a square hollow, in which the end of the beam was probably slotted.
Potsherds from the earthen surface and the floor’s bedding date from the Mamluk period (thirteenth–fourteenth centuries CE; Fig. 11:1, 2), confirming that the gate was constructed during this period. Nevertheless, their presence is not sufficient to determine when the entire fortress was built; it may have been built earlier, in the Crusader period, and only the gate may have been renovated in the Mamluk period.
The gate was repaired in a later phase, still within the Mamluk period. Two narrow east–west walls, built of a single row of roughly dressed fieldstones (W131, W136; Fig. 8), were constructed outside gate flanking a surface of small fieldstones (L135, L137, L138). The walls and surface, which led to an opening, covered the stone threshold, cancelling it and creating a ramp-like feature that led directly into the gate.
Above Surface 137 was a collapse of fieldstones and large ashlars (L133, L134, L140) that blocked the gate and spilled out slightly beyond it. The stones lay haphazardly, interspersed with a fill of soil and small stones that probably resulted from the collapse of the fortress walls. A few fragments of basalt grinding stones (not drawn) and potsherds from the Mamluk and Ottoman periods were retrieved from between the collapsed stones.
Among the Mamluk potsherds found in the surface blocking the gate and in the collapse were handmade bowls (Fig. 11:3, 4), wheel-made jugs (Fig. 11:5, 6), cooking pots (Fig. 11:7) and glazed bowls (Fig. 11:8–12). Body fragments of decorated wheel-made jugs (Fig. 11:13) and decorated handmade jugs and jars (Fig. 11:14–16) were also discovered. Another find was a mold-made flask (Fig. 11:17). This ware dates from the end of the thirteenth and the fourteenth centuries CE. The dating is based mainly on the cooking pots, the glazed bowls and the flask, as it is difficult to date the handmade and painted pottery with any precision. One fragment of a glazed bowl (Fig. 11:7) might be earlier, from the late twelfth or early thirteenth centuries CE. The handmade bowls and jars were very common in the Mamluk period, but the painted jugs and handmade cooking pots were less common, and the glazed cooking pot and the flask were relatively rare (E.J. Stern, pers. comm.).
Evidence of several construction phases from these periods was also discovered inside the fortress. Remains of two rooms were found while excavating to the south of the gate, near the fortress’s west wall (Fig. 12). Two walls built of small fieldstones were uncovered near the surface: an east–west wall (W106) abutted to the south by another, north–south wall (W123), lying parallel to W101, the fortress’s outer wall.
The north face of W106 was plastered, and the plaster at its base seems to have curved northward, probably at the point where the wall met the room’s floor. A row of three large stones (W119) was unearthed beneath the floor level to the north of W106; the middle one is a crushing stone from an extraction installation. To the south of W119 was a small patch of a white plaster floor that abutted the crushing stone and the stone to its west, sloping gently southward and continuing beneath W106. Due to the limited size of the excavation, it was impossible to determine whether this was an in situ installation, or whether the crushing stone was incorporated in secondary use; in any case, this construction phase appears to predate W106. In the soil fills to the north of W106 (L111, L120), which covered the extraction installation, were Ottoman potsherds, including jars (Fig. 13:1), coffee cups (Fig. 13:2), jugs (Fig. 13:3), cooking pots (Fig. 13:4) and clay tobacco pipes (Fig. 13:5) from the eighteenth and nineteenth centuries CE. The floor of the room that originally abutted W106 may have been cut into in this phase.
Both the white plaster floor and the floor abutting W106 were c. 1.5 m higher than the gate threshold, suggesting that there may have been a few steps leading from the gate to the interior of the fortress.
 
Animal Bones
Lee Pri Gal
 
The excavation of the entrance complex of the fortress at ‘Afula retrieved bones from a small area. The bone sample is limited (N=35), but it is sufficiently large to enable a general understanding of the site and its economy. The loci in which the bones were found belong to two main phases of the fortress’s occupation: the Mamluk and the Ottoman periods. The number of bones identified from each period is insufficient to substantiate a quantitative analysis of the data or to comparison the two periods.
The bones were collected by hand and were not washed. The bones were identified using as reference the comparative collection of the archaeozoological laboratory at the University of Haifa. The finds were quantified, and the bones were classified according to biological species and skeletal part, following a targeted protocol based on diagnostic areas (Davis 1992). Epiphyseal fusion and dental erosion were determined based on the skeletal part (Haber 2001) and were used to estimate the age of the animals at death. Taphonomic observations were recorded, such as butchering marks (Seetah 2006), pathologies (Baker and Brothwell 1980; Bartosiewicz 2008), gnawing by predators (Fisher 1995), burning (Stiner et al. 1995), weathering (Behrensmeyer 1978) and the way in which the bone was broken (Villa and Mahieu 1991). Metric measurements were collected for indicative bones where possible (Driesch 1976; Table 1).
Table 1. Classification of skeletal parts by taxon and bone parts by association with period
Period
Locus
Basket
Taxon
Bone
GB
Bd
Bp
Glpe
GL/L
Mamluk
135
1134
Sheep/goat
Horn
 
 
 
 
 
Mamluk
135
1134
Sheep/goat
Metapodial bone
 
 
 
 
 
Mamluk
145
1174
Sheep/goat
Tibia
 
32.6
 
 
 
Mamluk
145
1174
Sheep/goat
Digit-1
 
 
 
 
53.9
Mamluk
143
1146
Sheep/goat
Humerus
 
 
 
 
 
Mamluk
139
1141
Sheep/goat
Mandible
 
 
 
 
 
Mamluk
130
1118
Equidae
Digit-1
 
27.6
34.5
 
61.8
Mamluk
130
1118
Equidae
Digit-2
35.5
27.1
31.8
 
29.9
Mamluk
130
1118
Equidae
Digit-3
 
 
 
 
 
Mamluk
130
1118
Sheep/goat
Mandible
 
 
 
 
m3=38.2
Mamluk
132
1126
Sheep/goat
Mandible
 
 
 
 
 
Mamluk
132
1126
Sheep/goat
Digit-1
 
 
 
 
 
Mamluk/
Ottoman
133
1129
Sheep/goat
Digit-1
 
 
 
 
43.9
Mamluk/
Ottoman
140
1145
Sheep/goat
Humerus
 
 
 
 
 
N/D
111
1043
Sheep/goat
Mandible
 
 
 
 
 
N/D
111
1054
Sheep/goat
Mandible
 
 
 
 
Dp4=24.6
N/D
111
1054
Sheep/goat
Radius
 
 
 
 
 
N/D
114
1052
Cattle
Metapodial bone
 
 
 
 
 
N/D
120
1051
Sheep/goat
Metatarsus
 
 
23
 
 
N/D
120
1051
Cattle
Digit-1
 
 
29.5
 
 
N/D
120
1051
Cattle
Digit-2
 
 
 
 
 
N/D
103
1021
Sheep/goat
Femur
 
 
 
 
 
N/D
103
1021
Sheep/goat
Metapodial bone
 
35.4
 
 
 
N/D
108
1054
Rooster
Tibiotarsus
 
22.6
17
 
92
N/D
110
1036
Cattle
Digit-3
 
 
 
 
 
N/D
110
1036
Rooster
Femur
 
 
16.2
 
74.6
Surface level
128
1105
Sheep/goat
Digit-2
 
 
 
 
 
Surface level
128
1105
Sheep/goat
Digit-1
 
 
 
 
 
Surface level
125
1128
Cattle
Mandible
 
 
 
 
 
Surface level
125
1128
Cattle
Astragal
 
 
 
 
 
Surface level
125
1128
Sheep/goat
Mandible
 
 
 
 
m3=28.5
Surface level
125
1128
Sheep/goat
Radius
 
41.6
 
 
 
Surface level
125
1128
Sheep/goat
Horn
 
 
 
 
 
Surface level
125
1113
Equidae
Digit-1
 
 
53.8
 
82
Surface level
125
1113
Cattle
Digit-1
 
 
 
 
25.1
Of the 35 identified animal bones, nine were surface finds; these belong to sheep/goat, cattle and equines; twelve bones were dated to the Mamluk phase, including sheep/goat and horse/donkey; and two were sheep/goat bones from Mamluk/Ottoman loci. The other bones could not be attributed to one of the periods, but they included the variety of fauna listed above, as well as the bones of a rooster. The livestock from the Mamluk and the Mamluk–Ottoman periods includes a high percentage of sheep (79%) and a low percentage of equines (21%; all representing a single individual); cattle are absent from this assemblage. The bones from the Mamluk and Mamluk–Ottoman periods include long bones with high nutritional value (humerus, tibia), as well as parts of the skull and digits that are regarded as butchering remnants (Hellwing and Gophna 1984). All the equine bones are digits; the three digits (L130, B1118) almost certainly belong to the same young horse/donkey (Fig. 14). The size of the sample does not provide a clear picture of the age distribution at death. However, as 50 percent or so of the sheep bones in the assemblage are unfused—thus belonging to young individuals that were not fully grown—and the dental sample from sheep includes milk teeth (dp4) and molar teeth with little enamel wear, the sheep assemblage can be assumed to represent a flock comprising young individuals, some of which were butchered before they were two years old.
The most pronounced taphonomic feature of the animal bones in the Mamluk assemblage is the percentage of bones that were broken while fresh (N=4, 80%)—i.e., shortly after the animal was butchered and probably as it was prepared and consumed—compared with the percentage of bones that were broken after drying out (N=1, 20%). The outer surface of two digits, one of an equine and one of cattle, exhibited exostosis-type paleopathologies characteristic of beasts of burden that are caused by carrying heavy loads. A cut mark was found on a single bone: a distal humerus of a young sheep from a Mamluk–Ottoman context (L140, B1145). Only three bones from a Mamluk context exhibited signs of weathering, with one instance of a dog-bite mark on a sheep’s proximal radius.
The assemblage of animal bones from the site is too small to allow for reliable conclusions about the meat economy and the livestock management pattern during the Mamluk and/or Ottoman periods at the fortress. Nevertheless, it can be stated with relative certainty that the meat-production economy was based mainly on sheep, and that cattle were not a major part of it, since nearly all the cattle remains came from surface contexts, and they were entirely lacking in the Mamluk assemblage. The presence of equines in the assemblage is evidence that they were used for work, transportation and riding; the pathology on some of the bones supports this conclusion. The relatively few signs of weathering and the single bone broken after drying indicate that the bones were buried in the ground soon after use, probably while the site was still occupied.
 
The southern peak of Tel ‘Afula yielded remains from the Late Bronze Age, the Iron Age I and the Roman, Byzantine, Early Islamic, Mamluk and Ottoman periods. Remains from these periods were also unearthed in previous excavations on the tell. The earlier remains were encountered in a single square; this limited area added little to our understanding of the settlement’s nature during these periods.
Most of the finds are associated with the fortress constructed at the top of the tell. The excavation uncovered parts of the fortress, and numerous details of its original plan were discovered. The fortress’s gate was unearthed mid-way along its west wall. The plan of the structure and its substantial foundation walls indicate that it was a square tower fortress that reached a considerable height. However, as the interior of the gate was not excavated, it is unclear whether the thick walls uncovered in the excavation rose above the level of the floor, or whether a staircase led up from the gate to the first story of the fortress.
The limited excavation did not answer the question of when the fortress was built. Mamluk potsherds found beneath the floor of the gate date its construction to this period, but the fortress itself may have been built earlier, in the Crusader period, and it is possible that only the gate was renovated at a later stage.
Mamluk fortresses are known from various places outside Israel, such as the Lions’ Tower in Tripoli, but no fortresses built by the Mamluks have yet been found in Israel. Only secondary use of earlier fortresses built by the Crusaders has been identified, as at Zefat (Damati 2003:144–145) and Qaqun (Dar and Mintzker 1990). The plan and appearance of the fortress—a square tower fortress that incorporated Roman sarcophagi in secondary use for construction and ornamentation purposes—are very reminiscent of the Crusader fortress at Zippori. Comparison with the gates of other Mamluk and Crusader fortresses—the Crusader fortresses of Zippori and Hunin, or the Mamluk Lions’ Tower in Tripoli—suggests that the wide gate was surmounted by a pointed arch that continued over the outer entrance jambs. In contrast, the inner entrance lintel was straight, made of a single monolithic stone. The lintel was presumably surmounted by a flat relieving arch (Fig. 15).