The archaeological interpretation of the geophysical results allowed us to distinguish the characteristic urban features, including the remains of walls and entire buildings erected from mud and red bricks, and identify the presence of furnaces and kilns, heavily burned structures, and layers of ash.
The areas with fired brick rubble on the surface were recorded on the magnetic map as high-amplitude noise, making it impossible to identify any well-preserved structures below. The case was visible in 5-x-5-m trench 1/O located on Mound O, covered with red-brick debris. On the magnetic map, the area was marked with a strong reading of the surface materials and had no signs of the architectural remains below. Red-brick mounds have been a target of brick extraction in the past (Welsby, 1998, p. 17–18); thus, the architecture in such places can be heavily damaged and had mixed materials. In trench 1/O, small fragments of a red-brick floor with faint traces of occupation above and below were discovered, making it possible to determine the absolute chronology of the feature. Radiocarbon dates from a sample taken from under the brick floor show that the floor was built after AD 773–906 (with 72% probability). Because no traces of walls were found, it was impossible to distinguish the specific construction phases that this building may have undergone. However, the radiocarbon date obtained from the layer located above the brick floor gave a date of AD 1210–1281 (95%), the period of the general decline of the town. Other layers (except for one walking level) in the trench were mixed and contained red-brick debris, creating magnetic noise that obscures the identification of the brick floor remains in the survey.
In trench 1/OS, located at the top of a gravel hill, a complex of regular N-S and E-W oriented mud-brick walls were discovered. Excavations conducted in a central room (feature 17) yielded radiocarbon dates suggesting that the mud-brick construction took place after AD 561–651 (95%). For around 300 years after that, the area west of the eastern wall was probably an outdoor space. In the next short phase, other walls were constructed, and the area was covered with a roof resting on palm logs (a negative of one log was recorded in debris filling of the room). Radiocarbon dates from the layer below the southern wall of feature 17 suggest that the wall was built after AD 940–1021 (74.5%). Radiocarbon dates from a layer related to the use of feature 17 indicate that the building existed in AD 857–986 (83%). The area became a burial ground in the second half of the tenth century AD, with some of the grave pits cutting through the remains of the mud-brick walls (see Table 1; Fig. 8).
The magnetic survey recorded complete and extensive remnants of urban structures in the OS, FN, F, and HW areas, giving a glimpse of the density of buildings in the medieval capital. Only a few anomalies interpreted as walls have a different orientation, which may be a clue to distinguishing changes in the settlement pattern. Trench 1/CW (west of Church C; see Fig. 3) was located on one of the low mounds where magnetic survey indicated the presence of walls in both orientations—true north and slightly off north. It was expected that the dig would be approximately one meter deep since the mound was approximately 0.8 m high. To our surprise, at the level where we were expecting to find virgin soil, well-preserved NE-SW-oriented mud-brick architecture appeared. Due to time restrictions, we were not able to reach the virgin soil level in that trench and had to stop excavations 1.54 m below the surface (Fig. 9, top right).
In trench 1/CW, four occupation phases have been identified. The lowest is comprised of a mud-brick wall complex with small and elongated rooms (Fig. 10). Apparently, the organization of space was changing with time since some walls were found to be high and we could uncover their full preserved height, while others were set deeper and their foundations were not reached. Radiocarbon dates obtained from the foundation of wall no. 33 (see Fig. 10, spot no. 1) indicate that it was built after AD 527–615 at 68.6% probability (see Table 1). The second phase of occupation in 1/CW is connected with timber structures built after the mud-brick complex fell into ruin. Due to the limited size of the trench (5 × 5 m), only several posthole cuttings into a compacted mud floor level were identified.
Phase three, with radiocarbon date AD 890–1013 (95%), sees the return of mud-brick architecture parallel with timber constructions (the date comes from a sample taken from beneath wall no. 31, see Fig. 10 spot no. 2). Unlike the mud-brick architecture from the first phase, the younger structure was oriented N-S. The poor state of preservation does not allow us to specify the function of the building. The entire structure was heavily damaged by modern digging (phase four), which went deep and cut through phase two, ending at the upper levels of phase one. This feature, containing mixed materials, was probably the reason why magnetic readings were so obscure.
When analyzing the amplitude of magnetic anomalies, it is possible to distinguish the building materials indirectly. Two major types that can be interpreted by reading the magnetic map are red-brick and mud-brick. Red-brick debris is marked on the interpretive map, most probably reflecting the destroyed buildings. Vast disturbances have been recorded not only at Mound O but also in areas N, HN, and around Church C. In one of the areas with dense mud-brick architecture (area F, see Figs. 3 and 7), two distinct types of anomalies were recorded. The readings are interpreted as walls and a linear feature. Both are likely the magnetic image of alleyways between buildings. The mutual spatial correlation of walls and passages indicates that these mud-brick buildings and alleyways were contemporary. The GPR mapping performed in this area seems to confirm these observations.
Hearths and kilns are notable characteristic features from the magnetic survey perspective. The high temperature within the kiln causes the material to acquire specific magnetic properties and, through creating high-intensity magnetization, can be easily identified on a magnetic map (Fassbinder, 2017, p. 504). Features interpreted as kilns and hearths are placed on the interpretative map in high correlation with other features related to the mud-brick constructions. Between them, often along the walls, there are legible anomalies interpreted as ash dumps. Trench 2/OS was set on one of the “kiln anomalies” on Mound OS. This area was a lime production zone, consisting of remains of mud-brick constructions and the residue of a lime kiln. Radiocarbon dating originating from the bottom of the lime deposit and another date (from the deposit above the lime layer) suggests that the lime deposit was formed in the first quarter of the eleventh century. After the abandonment, the southeastern part of Mound OS (like the central part) was used as a burial place. Although geophysical surveys have provided evidence that the development of Soba extended over an area much larger than that of the mounds, many areas appear to be devoid of or containing only minor remains of the settlement. This may be due to the looting of building materials from the site, but these areas may also have remained undeveloped during the medieval period occupation, or the remains could simply be eluding geophysical methods.
The survey did not identify the third type of building material used in Soba—timber. Such buildings were discovered during previous research (Welsby, 1998; Welsby & Daniels, 1991) and in two of our trenches (1/SH and 1/CW). In the latter, they were documented as layers with multiple post holes and, in some cases, simultaneous with mud-brick structures. Timber architecture in trench 1/SH was erected after AD 892–1014 (95%). A charcoal sample related to the final phase of the timber architecture provided a calibrated date of AD 986–1052 (63.8%). Small circular (up to 0.8 m in diameter) and rectangular pieces (full size unknown) were recognized among the timber structures. Mud-brick architecture coexisted with the late phases of timber architecture. The topmost preserved remains were dated to the eleventh century. Such architectural features might have covered at least part of the area that is “empty” on the magnetic map, especially in the areas where it was possible to distinguish features interpreted as pits.
The magnetic properties of the archaeological structures are recorded in Soba, and the geological conditions in the Nile valley area make magnetometry a suitable tool for recognizing the spatial structure of Soba (Herbich, 2019, p. 197). However, attention should also be paid to the significant degree of destruction of numerous parts of the site and the presence of disturbances caused by modern infrastructure (e.g., power line poles and barbed wire iron fences). Geophysical research is also limited by the presence of modern houses, arable land, and the boundaries between them, as well as bushes and trees that cover part of the site. It should be understood that simple graves and timber structures are not represented on the map. Despite these limitations, the use of magnetometry has allowed for a completely new view of Soba’s urban layout.
A large number of small finds were recorded. Pottery is the dominant category and the most commonly reported by the modern residents of Soba. Pottery examined so far comes only from trenches 2/OS and 1/O (Figs. 11 and 12). The main types of vessels recorded in these locations are presented below in the context of previous research carried out in the 1980s and 1990s (Welsby, 1998; Welsby & Daniels, 1991). Three leading pottery groups have been identified: pots, jars, and bowls. Kitchenware/cooking pots are classified in groups A-D; storage jars are classified in groups E–H, and J; specialized vessels, qawadis, are in group J; large bowls (kitchenware), known as dokat, are grouped as category K; and small bowls are classified in groups L-P.
The assemblages are similar and consist mainly of coarse wares. Handmade pottery is the largest group in the pottery assemblage (Fig. 13). They include the following:
Cooking pots (A) generally have wide straight or slightly out-flared rounded rim, short and broad neck, and globular body, mainly black (sometimes brown/red) burnished surface and sometimes slipped, and with grooved and incised decoration
Large handmade pots (B, C, D) (storage jars?) with rounded inturned rim, globular body, with grooved and sometimes incised decoration on the external surface
Storage jars (E, F), characterized by a wide out-flaring rim, gradually develop into a narrow neck or jars with outflaring rim and short and broad neck
The RCW-HM (red fired, coarse ware, handmade) nondiagnostic handmade sherds probably belong to jars (E, F, G). The vessels recorded in small quantities are handmade jars (G), with a broad mouth and black slipped and burnished external surface, often referred to in the literature as “beer jars,” and handmade bowls and pots with mat-impression on the body (J); and dokat (K) (cooking ware)—large handmade bowls characterized by thick walls and rounded rims, with burnished internal and plain external surfaces (Fig. 14).
Wheel-made pottery is represented by jars (Fig. 13H) with broad orifices and rounded straight or slightly out-flared rims. Qawadis (Fig. 13I) are vessels used for drawing water in conjunction with a waterwheel characterized by a knob-ended base, and the bottom part of the qadus is broader than the top. RCW-WM (red fired, coarse ware, wheel-made) presented here correspond to nondiagnostic wheel-made sherds, which probably belonged to qawadis (Fig. 13I) or storage jars (Fig. 13H) or even to drainpipes which were discovered in one layer.
Black and red burnished wares dominated the repertoire of small handmade and wheel-made bowls, such as handmade carinated bowls (Fig. 14L,M) with pointed inturned rims, black or red slip, and sometimes incised decoration. There are also black or red slipped bowls burnished on both surfaces (Fig. 14M-P), with plain rims and sometimes with grooved or painted decoration.
Imported, wheel-made pottery (Imp.) consists of very small sherds. In some cases, these were reused as disks or as plugs for closing off vessels (see Fig. 14, bottom 1–3). In trench 1/O (see Fig. 11), the pottery assemblage in the earliest layers (layers 8–5 dated to the ninth century) is not abundant and was found together with large quantities of animal bones. The most significant occupational phase was identified in layers 1–4 dated to the ninth through thirteenth century (Fig. 15). Here, the pottery assemblage, mainly cooking pots (D), dokat (K), handmade jars (E–F), black and red burnished bowls (M, N), and wheelmade jars (H), was much more extensive and also contained animal bones and artifacts, such as stone tools, beads, and water-draining pipe fragments.
In trench 2/OS (Fig. 16), in the occupational layers 10 and 12 dated to the tenth and eleventh centuries (OS2/117 and OS2/119), mainly cooking pots (D), dokat (K), and qawadis (I), were recorded. Small black burnished bowls (M) were more common among the fine wares. In these layers, animal bones were also recorded. Some layers were connected with the occupation and production phase (see Figs. 13, 14, and 16). However, only two reused (nondiagnostic) sherds with traces of lime were recorded. The main occupation started with layers 8 and 7 dated to the eleventh century (OS2/093, OS2/090, OS2/092, OS2/091). Still, cooking pots (A, D) and dokat (K) were dominant. Among the fine wares, black and red burnished bowls are significant in number. In these layers, animal bones were also noted. The most plentiful pottery assemblages were discovered in layers 2–5, postdating the eleventh century (OS2/084, OS2/089, OS2/088, OS2/083). Here, the high predominance of cooking pots (D) is notable. Also, the number of pots (A) and pots with mat-impressed patterns increased (J). The quantity of black burnished bowls (M) also increased in comparison to the earliest occupation layers. These contexts yielded animal bones as well as stone tools and other artifacts.
The coarse and fine wares described above fit well into the pottery collection recorded during previous research (Welsby, 1998; Welsby & Daniels, 1991). Handmade pots (A) correspond probably to Class L, types 37, 39, 43, 69, 77 (Welsby & Daniels, 1991, p. 179, 185, Figs. 98, 100). Large handmade pots (B-D) correspond probably to Class I, types 1–11 vessels (Welsby & Daniels, 1991, p. 179, Fig. 91). Handmade jars (E–F) are the same as Class A vessels (Welsby & Daniels, 1991, p. 165–66, figs. 82–83), while Dokat (K) corresponds to Class K (Welsby & Daniels, 1991, p. 175–78, Figs. 89–90). The small wheel-made and handmade bowls, black or red slipped and burnished on both surfaces (M, N, O), belong to Class N and types 76, 121, 172–174, 201 of pottery (Welsby & Daniels, 1991, p. 193, Figs. 106, 110, 112, 113). Qawadis (I) are comparable to Class G (Welsby & Daniels, 1991, p 175–78, figs. 89–90).
The two trenches 1/O and 2/OS did not yield the most distinctive fine ware, Soba ware, characteristic of the early Medieval occupation in the capital (sixth through the eighth or ninth century AD). However, since the layers of the first centuries of the period do not always contain such material, radiocarbon dating resolved the issue (see Table 1), indicating that the features discovered in 1/O and 2/OS should be associated with the later phases of Soba (ninth century and later). The homogeneity of the ceramic assemblage from the site has been noted before (Welsby, 1991, p. 12; 1998, p. 87–177), and the situation is the same for the most recently excavated materials. The vessels are similar in terms of execution, style, and form, showing that the inhabitants of medieval Soba were using quite uniform sets of ceramics. This observation can be used to study the assemblage collected as random finds by the residents.
Some of the roughest coarse wares may have been homemade. However, the fine wares (black and red polished wares) required expertise and controlled conditions to be manufactured. Do they come from a single workshop? The only pottery kiln discovered so far was in the western part of Mound B and might be associated with the production of qawadis and dokat, indicated by large quantities of these types of vessels recorded in the vicinity (Welsby, 1991, p. 12; Welsby & Daniels, 1991, p. 103–106).