• Fig. 5.19 - Phase 3.1
  Figure 5.19
  
  Phase 3.1 plan the foreground.
  
  McQuitty et. al. (2020)
  plan from McQuitty et. al. (2020)
• Fig. 5.28 - Phase 7
  Figure 5.28
  
  Phase 7 plan the foreground.
  
  McQuitty et. al. (2020)
  plan from McQuitty et. al. (2020)

• Fig. 5.19 - Phase 3.1
  Figure 5.19
  
  Phase 3.1 plan the foreground.
  
  McQuitty et. al. (2020)
  plan from McQuitty et. al. (2020)
• Fig. 5.28 - Phase 7
  Figure 5.28
  
  Phase 7 plan the foreground.
  
  McQuitty et. al. (2020)
  plan from McQuitty et. al. (2020)

• Fig. 4.12 - Phase 3.1
  Figure 4.12
  
  Phase 3.1 plan the foreground.
  
  McQuitty et. al. (2020)
  plan for Far II from McQuitty et. al. (2020)
• Fig. 4.20 - Phase 3.2
  Figure 4.20
  
  Phase 3.2 plan the foreground.
  
  McQuitty et. al. (2020)
  plan for Far II from McQuitty et. al. (2020)

• Fig. 4.12 - Phase 3.1
  Figure 4.12
  
  Phase 3.1 plan the foreground.
  
  McQuitty et. al. (2020)
  plan for Far II from McQuitty et. al. (2020)
• Fig. 4.20 - Phase 3.2
  Figure 4.20
  
  Phase 3.2 plan the foreground.
  
  McQuitty et. al. (2020)
  plan for Far II from McQuitty et. al. (2020)

• from McQuitty et. al. (2020:73)

This phase includes the continued use of the Khan and its paved courtyard; the conversion of the courtyard to the south of the Khan into a roofed chamber (G2004) by reusing elements from its earlier external phase; and the partial collapse of G2004 and its adaptation and reuse. The main façade of these two adjoining units would have opened to the west.

• from McQuitty et. al. (2020:73-74)

(Figure 4.12)

During this sub-phase, the open paved area to the south of the Khan (described in Phase 2) was converted into an internal space (G2004) (4.50 m × 4.40 m). The south wall was created from the wall (SG2152) of an earlier building to the south of the courtyard. This wall blocked both the doorway (SG2150) and the alleyway in the southeast corner of the courtyard (SG2144) shown graphically in Figure 4.14. The north wall was formed by the Khan (SG2074) and the east wall comprised a narrow rubble stone wall (SG2145) built against the southern extension of the Khan’s east wall (SG2061). The new wall face (SG2145) incorporated a springer for one of the arches supporting the roof. The archspringer (SG2156), recorded as context 1718 in Figure 4.15, survives as three squared limestone blocks bonded into the wall (SG2145). Together with the arch-springer in the western wall (SG2158), recorded as context 1719 in Figure 4.4 and clearly visible in Figure 4.16, this formed part of the southern roof arch (SG2138), contexts 257 and 495 in Figures 4.15 and 4.18, which collapsed in Phase 5. There is no trace of the northern arch, which was probably removed in a later phase (see below Phase 2, Sub-phase 3). The total reconstructed height of the room would have been approximately 3.50 m, which is comparable to the main building of Phase 2 at The Highest Point, (G5002). It is assumed that there was an entrance in the western wall, between the two arches (SG2126) as shown in Figure 4.16. Whether this entrance originally opened into a building to the west of the excavation trench or simply led to an outside area is unknown. The presumed doorway is filled with rubble, including stone rafters resting on an in situ threshold stone (SG2118).

The floor of the room (G2004) comprised irregular stone slabs, the southern part of which (SG2127) were later robbed, and recorded as contexts 788 and 790 in Figures 4.4, 4.15, 4.17 and 4.18. Patches of ash and occupational material accumulated on the surface and these were particularly rich in artefacts (SG2128). A two-compartment stone feature (SG2124) was built against the west wall and contained large amounts of ash and fish bones. Its purpose is unclear — perhaps the compartments were originally used as stands for large water jars (ziyar, sing. zir² ) or even as mangers, if the room was reused as a stable. There was a similar feature in the northwest corner of the room (SG2119), shown as context 805 in Figure 4.21, which was also used in Phase 4, Sub-phase 1.

To the north, the Khan continued to be used, with a rough, patchily paved and cobbled interior (SG2047, SG2051, SG2058, SG2052 and SG2049), some of which may have survived from the earlier phase. A central hearth (SG2048) was built into this cobbled floor. Outside, to the west, lay an even and substantial flagged surface (SG2090), shown in Figure 4.19 and as contexts 1091 and 1123 in Figures 4.7 and 4.8.

• from McQuitty et. al. (2020:74-77)

(Figure 4.20)

Two sub-phases of Phase 2 were distinguished in the occupation of the Khan but the same two sub-phases could not be distinguished in the excavation of G2004, the room to its south. During this sub-phase, in the Khan, the rough, patchily paved and cobbled floor was replaced with another beaten earth surface (SG2041), shown as contexts 71 and 72 in Figure 4.1. A stone platform — possibly a mastaba for sitting, sleeping or storing bedding — was built along the south wall (SG2044). A substantial stone feature of unknown function (SG2045) located in the northwest corner, shown as contexts 461, 616, 619, 683, 684 and 685 in Figure 4.20, makes its first appearance in this phase and it continued to be used in later phases. Its curving walls are two courses high and as a whole, the feature is well made. The ashy lenses found within it are interpreted as ephemeral hearths. The doorway and original threshold led up and out to a paved area (P2-3.1: SG2090) which was irregular at the south end (SG2097), Figure 4.20. The paving continued southwards, beyond the southwest corner of the Khan (SG2117), which is recorded as context 103 in Figure 4.17. This corner appears to have been rebuilt during this phase.

• from McQuitty et. al. (2020:74-77)

During this period yet another beaten earth surface (SG2040), recorded as context 69 in Figure 4.1, was in use inside the Khan. It was cut by various pits and post holes (SG2038, SG2039 and SG2042). To the south, room G2004 collapsed and was no longer in use, although pit-digging and ephemeral hearth activity (SG2120 and SG2123) was evident within the ruins. The pits were dug into structural collapse which included faced and squared ashlar blocks and stone rafters (SG2118) from the assumed northern arch (SG2125). The southwest corner (SG2116) of this southern room G2004 also collapsed. Destruction debris (SG2132 and SG2133), shown as contexts 1025 and 1030 in Figure 4.5, was also found east of room G2004 in an area that is presumed to have been an open courtyard.

• from McQuitty et. al. (2020:77-)

The ingenious conversion of the courtyard south of the Khan into an internal area, recorded as G2004, with its main façade towards the west, suggests that G2004 was conceived as part of a larger complex. The function of the complex as a whole is not clear although the excavated rooms appear to have had a domestic purpose. The artefacts found within the use, abandonment and collapse deposits were overwhelmingly Early Islamic. Large numbers of dateable lamp fragments were found and steatite vessels make their first appearance. Tile fragments and glass tesserae presumed to have come from a church in the area of The Highest Point were found in the deposits associated with the use and abandonment of the chamber G2004. While its collapse was not nearly as dramatic as building G5002 at The Highest Point — in that here the southern arch remained standing — the sequence of collapse is similar. On those grounds, it is suggested that Phase 3 of Far II, Far IV and House 2 was contemporary with Phase 3 of Far V, which lies immediately to the west.

• from McQuitty et. al. (2020:104)

This phase represents the collapse of architectural elements within G5002 (Phase 3, Sub-phase 1) and the dumping of debris in the areas to the north, east and west of G5002 (Phase 3, Sub-phase 2). Outside the east wall of G5002 (SG5109) there was an accumulation of rubble, silt and debris collected in even bands (SG5121, SG5124 and SG5108) that sloped up against the window (SG5125). To the west levelling layers of destruction debris and rubble (SG5065, SG5066 and SG5067) were revealed.

• from McQuitty et. al. (2020:104-105)

The remains of the arches and stone rafters that once supported the flat roof of G5002 were spectacularly and graphically preserved (SG5087, SG5100, SG5104 and SG5174), seen in Figures 5.6 and 5.20. Large amounts of loamy material, lumps of clay and small rubble were found in amongst the stones, which presumably related to the thick earthen layer that would have topped the roof; no tiles were found. Rubble of worked and unworked limestone, and chinking stones, represented the packing of the arch spandrels. The arch stones were light marly limestone, while the spandrel stones and keystone were of harder limestone. The roof rafters were of both basalt and limestone and appeared to have fallen directly downwards as opposed to sliding off the arch. A few of the limestone rafters showed signs of burning. Amongst the collapse were fragments of marble, which presumably had been used for wall cladding and paving. One basalt meta [SF 3357] shown on Figure 5.20 was preserved under the collapse. The collapse had all the appearance of being sudden rather than gradual. However, the room seems to have been empty or abandoned at the time: no concentrations of single ceramic vessels were found and the interpretation of the previous phase had suggested that the room was no longer permanently occupied and was already subject to some transient use, before it finally collapsed.

• from McQuitty et. al. (2020:105)

While the architectural collapse in G5002 was left, literally, where it fell, the areas to the north and west were cleared out and levelled with destruction debris moved to prepare for later construction. These deposits (SG5065, SG5066 and SG5067), shown as contexts 1128, 1420, 1445, 1446 and 1448 in Figures 5.11 and 5.15, included rubble, charcoal flecks, oven fragments, glass and burnt bone. The areas to the south of G5002, and G5004 itself, were empty of architectural debris. As always, the area to the east of G5002 presented something of an enigma: the window or door (SG5125), shown as context 1397 in Figure 5.7, seems to have been filled with debris (SG5124) and material was dumped against the wall (SG5121).

• from McQuitty et. al. (2020:105)

The collapse in G5002 was almost certainly the result of an earthquake and support is given to this hypothesis by the similar architectural collapse of G2004 seen in the Central Area Phase 3. The chronological parameters for the date of the collapse are set by the Late Byzantine to Early Islamic ceramics of Phase 2, Sub-phase 3, and the 9th-century and later ceramics of Phase 4. A convincing candidate would be one of the three sizeable events that took place in the period A.D. 746–757. It is most likely to have been the earthquake of 746 A.D. (Ambraseys 2009: 235). After the structural collapse, the debris was cleared, over a period of years if not decades, and this is what the deposits of Phase 3, Sub-phase 2 represent. The material associated with these deposits gives some insight into the character of the buildings in the immediate vicinity. Some glass and stone tesserae were found in the collapse, as well as fragments of marble wall cladding, coloured marble inlay ‒ possibly from an opus sectile floor ‒ a capital fragment, and part of a funerary plaque with a few words of a Greek inscription [SF 3392]. The dump heaped up against the eastern wall of G5002 (SG5121) was particularly rich including a huge number of glass and stone tesserae. A sample of the glass tesserae are pictured in Figure 7.15. In general, the other artefacts from this deposit were domestic in nature, such as grinders and steatite cooking vessels.

The trench was located here to explore the possibility of finding a public building, such as a church. While there is no specific architectural evidence to support the identification of G5002 as part of a public building, if it was not itself part of an ecclesiastical complex, the related finds indicate that it lay very near to a church.

• from McQuitty et. al. (2020:110-112)

(Figure 5.27)

In essence this phase describes the abandonment of G5001 and G5003. In G5001, architectural collapse (SG5044, SG5045, SG5079, SG5080, SG5081 and SG5173), silt accumulation (SG5049 and SG5158) and dumping of material (SG5046 and SG5051) were all encountered and can be seen in Figure 5.25. The roof rafters and associated roof rubble lay jumbled amongst the G5001 arches (SG5048 and SG5071). The area to the south of this structural collapse was filled with a variety of dumps of varying compactness (SG5047 and SG5051). G5003 showed a similar pattern of structural collapse (SG5007), silt accumulation (SG5008) and dumping (SG5012).

The area east of G5001, known throughout the phases of this trench as the courtyard, did see some activity. The courtyard surface was beaten, swept earth (SG5023), including cobble dumps (SG5029). An oven- or tabunhouse, dug into the southeast corner, utilised the walls of earlier phases and also involved the construction of a new, irregular wall (SG5147). The location of the entrance to the oven-house is not clear and it is presumed to have related to a structure or tent to the south. The oven-or tabun-house (SG5043), housed at least three successive ovens or tawabin (SG5040 and SG5026) and was filled by levels of an ash dump (SG5025) and recorded as contexts 320, 663 and 1428. Its walls were flimsy, one stone wide and two courses high, and were dug down into the earlier outside area shown in Figure 5.22. A straight-sided, shallow pit (SG5021) was also dug from this level ‒ its purpose is unclear.

• from McQuitty et. al. (2020:112)

The deposits and features of the courtyard produced significant concentrations of 11th- to 13th-century ceramics, in addition to post-13th-century Middle Islamic ceramics. A Crusader billon denier in the name of King Amaury, datable to c. 1169–87 [SF 3069] was found within the collapse of G5001 along with a fragment of tobacco pipe [SF 385] dated to the 19th century. Abandonment clearly took place over several centuries. The tobacco pipe fragment provides both a terminus ante quem for the start of this process of collapse and ‒ at 1.80 m below the trench surface ‒ a convincing terminus post quem for the following Phase 7 and final abandonment of the whole area.

• from McQuitty et. al. (2020:189-193)

(Figures 4.25, 9.15 – 18, Appendix 7 Table 5.43)

... The non-rotary querns are the earliest form and within these the saddle quern is the earliest. These have a long use history and are paralleled over a wide time and area. The one definite example, SF 4272, was unstratified. Three other possible saddle querns are SF 1273, SF 2214 and SF 3248, all from early phases of the excavated occupation levels. The circular quern, SF 2000 (Figure 9.15F), of which only a fragment is preserved, is an anomaly in form but the concave, smooth face with traces of polish leads to its inclusion in the non-rotary querns.

The rectangular quern is widespread throughout the Levant and also has a long history of use, as can be seen from the following parallels. An example from Pella (McNicoll et al. 1992: 68 RN 70120) is longer than ours at L33.00 cm and dates to the Bronze Age. From the Madaba Plains are two examples (Platt 1991: 247 1118/A 7K60.12 fig.10.1 and 777/D 6K07. 2 fig.10.3) dated to the early Persian and Late Ottoman periods respectively. From the Red Tower (Pringle 1986: 174 11) is an unillustrated example, from the final abandonment and destruction dated to the 20th century, in Area A Phase F.

The earliest example at Khirbat Faris is a fragment, SF 2305, from The Central Area in a dump associated with a complex of walls and dated to the Roman period. The only complete example, SF 3225 (Figure 9.15H), comes from accumulation and clearance levels in an outside courtyard area at The Highest Point, dating to the 7th and 8th centuries AD.

The use of the hand-operated rotary quern is believed to have become fairly widespread by the Hellenistic period. (Forbes 1965: Volume III 146). Many of the hand rotary querns found at Khirbat Faris are in contexts which suggest they were in use in earlier phases; a good example is SF 2166 (Figures 4.25, 9.16C) from The Central Area, which was reused in a paved floor in G2003, the Transverse Arch House. The floor is 9th–13th century and so, presumably, the quern dates from before this period.

The two Type 002 rotary querns, the Pompeian metae, are from G5002, the Late Antique House at The Highest Point. SF 3421 appears to have been dumped inside this room during a period of squatter-like occupation dated to the Byzantine period. SF 3357 was found beneath the roof collapse associated with the 8th-century AD earthquake. It is thought that the room had already been abandoned prior to the destruction and that the quern had been dumped inside. The date for the actual use of these querns is difficult to assess. Their upper stones, catilllae, were not found which suggests that they were not actually being used in G5002.

• from McQuitty et. al. (2020:199-200)

For ease of reference, a rigid taxonomy has been adopted to categorise the architectural types encountered at Khirbat Faris. The names given to the various types are based on their principal architectural features and their date. The types are discussed in chronological order. Much use is made of ethnographic analogy, particularly when considering the Transverse-Arch House, the Barrel-Vaulted House, the Arch-and-Grain Bin House and the Oven-House. Although such types have been noted previously, there are few detailed archaeological descriptions despite the wealth of ethnographic information regarding their construction and use in the late 19th and early 20th centuries. For the Kerak Plateau, much information has been collected and summarised by the anthropologists of the Faris Project, Fidelity and William Lancaster (1999: 262–270) and the Vernacular Architecture Survey. All of the houses, in all periods, were single storey, and most seem to have been single units. Apart from the Phase 5 complex on the Western Edge, it is impossible to say from the excavation whether these units opened onto an alley or courtyard, or were part of a multi-room complex: comparison with known parallels suggests that they are more likely to have opened onto exterior spaces. At Fihl (Pella), a mid-8th-century house was built around a central courtyard (Walmsley 1988: 149); at Hisban, a suite of barrel-vaulted rooms, similar to that uncovered in Far. I, Phase 5, clustered around an open court (Walker 2003: 251); and, at Dhiban, single unit, Middle Islamic houses opened onto a courtyard (Tushingham 1972: 83–84). Further afield, Hirschfeld illustrates many examples in his comprehensive study of rural settlement in Byzantine Palestine (1997: 33–71).

• from McQuitty et. al. (2020:200-201)

(Figure 10.7)

The use of barrel-vaults is well attested in the Eastern Mediterranean from the 3rd century BC and became extremely popular in the Roman period, as the constructional demands of public monuments were met by this highly functional building technique. Barrel vaults were used primarily as a structural element ‒ for roofing subterranean cisterns, to support terraces or tiers of seating, or in series for a row of shops or as part of a larger complex, such as a bath house. Such vaults are very rarely found in isolated, free-standing, singleunit buildings. However, the Khan at Khirbat Faris is paralleled in Nabatean structures in the Negev, dated to the 1st century BC/AD At Oboda, the antechamber of a Nabatean burial cave had walls of well-dressed ashlar masonry and was roofed by a fine barrel vault built of smaller marly limestone blocks (Negev 1997: 84). In subsequent centuries, this antechamber seems to have been in almost constant use by the local Bedouin as a foul-weather shelter. Another example of the vault as mausoleum is Khirbat ‘Ain near Jerash. The Roman mausoleum was located close to a farm, a reservoir and quarries and, like Oboda but unlike the Khan at Khirbat Faris, had slots dug into the floor to hold the bodies of the dead (Kennedy and Bewley 2003: 5 and personal communication). While such parallels might suggest that the Khan was built as a funerary monument, the location of the structure might indicate a different function. Near to Khirbat Faris on the flanks of the Wadi Ibn Hammad stands Qasr Al Himma (Johns et al. 1989: 70, where it is referred to as Burj Al Majdalayn; Miller 1991: 49, Site 55). This small rectangular tower is of similar dimensions to the Khan (4.00 m × 5.00 m) and its exterior is remarkably similar. However, it is roofed not with a barrel vault but by stone rafters supported on corbels; this may, or may not, be a later alteration. As part of the preliminary survey of the area surrounding Khirbat Faris, the remains of two other possible towers, built of massive blocks, were also identified (Johns et al. 1989: 70). All these towers are situated close to the edge of the plateau and may perhaps have been part of a chain of watchtowers guarding the east-west route through the Wadi Ibn Hammad, near to its intersection with the north-south route of the King’s Highway. A similar interpretation was initially offered by Hirschfeld for a chain of structures found in a survey of the eastern edge of the Hebron hills: here, the towers were almost twice the size and surrounded by a talus (1997: 50). On the basis of subsequent research, he now classifies such structures as towered farmhouses. At least one such tower, from the west of Sa’adon in the Central Negev, is comparable in size to the Khan at Khirbat Faris (Hirschfeld 1997: Fig. 63) and, like it, has no talus. These towers are part of more extensive complexes including dwellings, sheep folds and other agricultural features (Hirschfeld 1997: 59). If the Khan belonged to a similar complex, this could explain the eastern wall (SG20161) which would have acted as a courtyard wall, enclosing the paved area, SG2148. In Palestine, such towers are interpreted as being multi-storey: there is no evidence for this in the Khan, either from incomplete stonework or in an access route to an upper floor. Given the lack of specific knowledge about rural settlement in this area during the Late Antique period and more particularly the lack of plans of sites found during the Kerak Plateau Survey, it is impossible to do more than suggest that a type of farmhouse identified in Byzantine Palestine may also occur further east. Perhaps this was the original function of the Khan at Khirbat Faris, rather than a funerary monument or military tower.

• from McQuitty et. al. (2020:201-202)

(Figure 10.8)

The techniques used in the construction of this type of house have both a long history and an exceptionally wide distribution in the Eastern Mediterranean in rural and urban, commercial, domestic, ecclesiastical and military contexts. The construction relies either upon arches springing from the house walls at a high level or upon arches resting on pilasters buttressed by the house wall. In either case, the arches support a flat roof of stone rafters or wooden beams. The width of the room is determined by the span of the arch, an average of approximately 5 m. The arches are spaced according to the length of the stone rafters or wooden beams available and may occur in any number. At Khirbat Faris, no more than two arches were used and the average area covered is 20.25 square metres. In the area close to Khirbat Faris, examples are known from the Negev (Negev 1997: 28–31), Petra (Kanellopoulos 2001: 11–13; Kolb 2000: 283–296), Khirbat edh-Dharih (Villeneuve 1994: 762–764) and Dhiban (Tushingham 1972). This roofing technique does not appear to have become common in Southern Jordan and the Negev until the 3rd to 4th century AD (Kolb 2000: 295) – a pattern that is repeated in Northern Palestine (Hirschfeld 1995: 243), although here the earliest examples appear in the 1st century AD. At Khirbat Faris, the period of use lies within the Late Byzantine to Early Islamic range, but a similar technique of supporting the roof with arches can also be seen in the later Transverse-Arch House. At Khirbat Faris, the two rooms built according to the Late Antique House type (G2004 and G5002) are further distinguished by the placement of the original entrance perpendicular to and between the arches. The original construction of the walls of these rooms is characterised by regular courses of large, roughly squared, calciferous limestone blocks, surrounded by smaller stones. In rural domestic contexts, these houses are almost always part of a courtyard complex. Kolb (2000: 295) has drawn up a check-list of the most distinctive features of this type in the architecture of the Petra and Negev ‒ one is tempted to say Southern Jordanian ‒ region, and the examples from Khirbat Faris correspond to this. The most striking similarities are found in the roof construction and in the doublefaced, rubble-filled walls, with a particularly well-built outer face.

• from McQuitty et. al. (2020:202-206)

(Figure 10.9)

The Transverse-Arch House was the most common type of rural house in Jordan during the 19th and early 20th centuries. The type is distinguished from the Late Antique House by two features. First is the technique of supporting the arches which, rather than being bonded into the house walls themselves as in the Late Antique House, are sprung either from the base of the house walls, or from arch-walls that effectively act as buttresses; it is the introduction of such arch-walls that is the most distinctive feature of both the TransverseArch House and the Arch-and-Grain-Bin House. The second distinguishing feature of this type is the position of the arches relative to the entrance of the house, which is almost always in a wall running parallel to the arches; it is the position of the arches transverse to the main axis of the house, as determined by its entrance, which gives the type its name.

Examples are increasingly well documented as the recording and analysis of traditional rural architecture gains popularity (Ayoub 1982; Biewers 1992, 1993, 1997; Biewers and Kana‘an 1993; Fakhoury and Sweiss 1995; Kana‘an and McQuitty 1994; Khammash 1986; Lancaster and Lancaster 1997: 260–273; Mershen 1992; Noca 1985; Kana‘an and al-Rifai 1988). This type has also been recorded in 20th-century villages in the Lebanon, Palestine and Syria (Canaan 1933; Fuchs 1998; Hirschfeld 1995; Thoumin 1932). Similar roofing techniques were used in rural architecture throughout the Eastern Mediterranean, including Crete (Rackham and Moody 1996: 167), Cyprus, Rhodes and Turkey (personal observation). From both Jordan and further afield, examples are known dating back to at least the 14th century: Dhiban, Jordan (Tushingham 1972), Tell Tuneir, N. Syria (Fuller and Fuller 1987–88: 279) and el Qdeir, N. Syria (Aurenche 1990: 46). The examples at Khirbat Faris are dated to the 12th to 13th centuries.

Distinct regional variations applied in the 19th century and may well have been evident in earlier centuries. In both Palestine and Northern Jordan, different zones of use are distinguished by variation in the height of the floor: the lowest zone by the entrance is used for storage, penning animals at night, leaving shoes, etc., and a step of 50 cm up to a higher level demarcates the family area used for socialising and sleeping. The space beneath this raised platform may often be used for storage of household and farming articles (Fuchs 1998: 158). Such a differentiation of zones of use by floor height is not common in Southern Jordan and is not encountered in the Early to Middle Islamic examples at Khirbat Faris (G2003, G5001, G5003).

The method of construction has been described well by Canaan (1933: 29), Dalman (1932–33), Hirschfeld (1995: 125–126) and Khammash (1986: 76–78), so that only a few points need be repeated here. These houses have an average covered area of 64–100 square metres. The normal span of the arch, and thus the width of the room, is increased from 5 m (as in the Late Antique House) to 10 or even 15 m (as in the Arch-and-GrainBin House), because the arches are sprung not from the house walls themselves, but from an arch-wall built perpendicular to the house walls and extending for 3m or more. The arch-walls, rather than the house walls, are built in deep foundation trenches (e.g. context 1082 on Figure 4.2), because they are the elements that bear the weight of the roof and support the thrust of the arches. At Khirbat Faris, the arch stones are often reused from earlier structures. In more recent examples from elsewhere, the arches are occasionally two stones wide, with considerable amounts of plaster holding the two edges together. There is no indication from any of the published ethnographic examples that stone rafters were used to support the roof: rather, a combination of wooden beams, brushwood and soil are reported. The original wall construction of these rooms is characterised by irregular courses of roughly shaped stones that are larger at the base of the wall and at the corners, and smaller at the top. The walls are built with an inner and outer skin and bonded with large amounts of mud mortar.

Ethnographic parallels show that while these houses were used as dwellings, an important subsidiary function was as stores for the annual harvest. Grain and pulses are more commonly stored in free-standing mud and wicker storage bins, known as kuwaʾir (sing. kuwara), rather than in grain-bins constructed between the arches, known locally as rawiyat (Figures 6.1, 10.10– 12). Kuwaʾir are often found in a range known as sanduq (Figure 10.13). Occasionally, semi-secret rooms are inserted behind a range of kuwaʾir, or a house may be built on to a cave at its rear or side; both features are used for storage. This is particularly characteristic of Southern Jordan (Biewers 1990: 193; 1993: 29, Fig. 43) and this is how the southern end of building G5001 at Khirbat Faris is interpreted (see discussion of Phase 5 in Chapter 5).

A few architectural features are built into the houses. The thick walls often include recesses or niches in which household possessions were kept. Larger niches were used for storing bedding and are often located in the wall opposite the entrance (Figure 10.14). The amount of bedding and seating owned is a sign of the owner’s wealth; the bigger the household flock, the greater the surplus of wool used to stuff the mattresses and quilts used for bedding and seating.

These houses are not lit by windows. There are usually a few ventilation holes pierced through the wall above the entrance, both to let smoke out and to allow access to pigeons, kept for their meat and eggs. The entrances of the houses were simple, using single-leaf doors mounted on a door post that swivelled in a simple socket (Figures 10.15 and 16). Frequently, a stand for a water jar was placed near the entrance. On the exterior, the surround of the entrance was often picked out with reused decorated stones or with simple ablaq (blackand-white masonry). The general impression given by the exterior, however, is of a dour, fortress-like building. Most houses opened on to a courtyard that, over time, came to be filled on two or three sides by similar units built for each successive new generation.

G2003, in the Central Area, is a hybrid, in that the southern arch and walls have been reused from the earlier Late Antique House on the same site; consequently, its smaller room dimensions are reproduced. However, the northern arch has been built with an arch-wall and storage areas have been inserted between the eastern ends of the arches. At the Highest Point, G5001 also reuses the walls of a Late Antique House but, here, additional arches and new walls on every side except the north, plus a storage room, have been inserted to produce the familiar plan of the Transverse-Arch House.

• from McQuitty et. al. (2020:206-209)

(Figure 10.17)

The occupation of the Western Edge in Phase 5 was represented almost exclusively by Barrel-Vaulted Houses (see discussion in Chapter 3). This type of building did not occur anywhere else in the excavated area. The description of the architecture is based upon these examples and upon two 20th-century examples from the nearby Christian village of Hmoud (Figures 10.18 and 10.19). The houses in Hmoud were built by Christians from the nearby town of Kerak in the first half of the 20th century, not by the present villagers (Jacob al-Odat: personal communication). The earliest building was used as a madafa or guest-house for Christian tribes from Kerak and was part of a cycle of seasonal use of the land: in the spring they would come with their flocks for the spring pasture and to sow grain; they would return in the late summer to harvest the grain and to graze their flocks on the stubble. In the early 20th century, this seasonal use gradually developed into permanent settlement (Médebielle 1961; 1987).

The Barrel-Vaulted Houses excavated at Khirbat Faris are characterised by their relatively small dimensions (average internal measurements 4 m × 3 m × 2 m high) and massive side-walls (average 1 m thick), which either reuse earlier walls or are constructed from two skins of stone with a rubble and mud mortar core. The ground plan is often trapezoidal,¹ with the entrance at the wider end. In the ethnographic examples, no differentiation in height within the covered space was reported. In the examples from both Khirbat Faris and Hmoud, the entrance wall was not bonded to the vault itself and, in the examples at Hmoud, the entrance was marked by a rectangular antechamber, roofed with reeds and branches (Figure 10.19). The wall stones are roughly shaped and squared, but the calciferous limestone blocks used for the vault itself are singularly shaped and look almost like loaves of leavened bread, although they show no sign of having been worked. Huge amounts of mortar were used to secure the vault, which was built over a form of sacks filled with earth that was later removed. The end walls, which bear no load, are less substantial and, in the example of Vault V (G1002), were merely slapped up against a pre-existing earthen bank (context 766 in Figure 3.27 and also pictured in Figure 3.45). The roof was flat and consisted of substantial layers of earth and brushwood built up on the exterior of the vault and then sealed with a final layer of mud mortar. In neither the archaeological nor ethnographic examples is there any indication of builtin storage facilities.

Parallels for this architectural type are mainly confined to Jordan. However, Hirschfeld reports a 13th- or 14th-century example from Horvat Berechot in the Hebron hills (1995: 43–44). A complex of barrel-vaulted structures, surrounding a central courtyard, were excavated at Hisban and dated to the Mamluk period (Bird 1968: 200–202; de Vries 1993: 157–162; LaBianca 1990: 220–221). They have recently been interpreted as belonging to the governor’s residence (Walker 2003: 251). The latter were not only similar in size to those at Khirbat Faris, but shared many other features in common, including details of construction. De Vries refers to ‘many low-ceilinged vaults’, while Bird describes a vaulted room with a window and a plastered mastaba (compare with Far. I, Vault VI). Many of the houses surrounding the walled town and ecclesiastical complexes at Umm Al Rasas are of this type, but are not arranged around courtyards. A nearly complete barrelvaulted structure, containing artefacts of late Ottoman date, was excavated at Raʾs Al Qabub in Northern Jordan (Banning et al. 1989: 54). Khammash comments that such Barrel-Vaulted Houses were a common sight in Jordanian villages of the recent past, when they were used as stables or oven-houses (1986: 43) and Petersen observed the same phenomenon in Palestine at Dayr Al Shaykh near Jerusalem (1996: 99). At the present state of our knowledge, it would appear that such BarrelVaulted Houses were built from the late 13th to the mid-20th century.

While it was not fully excavated, Vault IV (G1007) is also interpreted as being barrel-vaulted (Figures 3.34, 3.35 and 3.40). It certainly predates the excavated examples and was on a much larger scale. Similar structures are visible on the ground to the north of the Western Edge trench (Figure 2.1) and measure on average 5 m × 12 m. Large barrel vaults have a long history and have been part of major military installations in the region for many centuries e.g. Kerak Castle, (Deschamps 1939: 35–98). Here they often serve as multi-purpose and multi-storey galleries. Free-standing examples are less commonly reported but in his survey of secular and domestic buildings in the Crusader Kingdom of Jerusalem, Pringle identifies several types (1997: 11– 12). In the same way as the vaults in castles are used as sub-storeys, they often serve as basements within non- or semi-fortified hall-houses. More interestingly for parallels with Khirbat Faris, such vaults are found, free-standing in rural settlements (1997: 19, 27, 30, 43, 77), or as part of courtyard complexes (1997: 88–89) that Pringle interprets as curia or the centres of rural estates governed by the Crusader population. Pringle dates these vaults as 12th–13th century (1997: 1) in line with the dates of Crusader occupation but also comments that the type of architecture may well be earlier. Standalone vaults are adaptable to many uses: to house an olive press at Bait Mahsir in Palestine (Pringle 1997: 29) and as the late-12th-century Castle Chapel at Kerak (Pringle 1993: 289). This type of construction is also seen in the luxury residential complex at Khirbat Al Dusaq near Shobak Castle (Figure 10.20), which has most recently been interpreted as 13th–14th century in date (Vigourox et al. 2015: 209).

Closer to Khirbat Faris is the large vault at Hisban that was interpreted as the ‘Mamluk Governor’s Storeroom’ and was found ‘full of sugar jars, glazed wares with military-style inscriptions, and javelin points’. The vault sustained ‘earthquake collapse and fire in the mid-14th century’ (LaBianca and Walker 2001: 10). Based on stratigraphic evidence at Khirbat Faris, it is suggested that Vault IV had ceased to be used by the mid-14th century. Its function is unknown but the character of material – wheel-made ceramics and fine imported glassware (Figure 3.49) found in the nearby cistern and postulated to have come from the clearing out of Vault IV – suggests that the people who used this building were certainly still part of the wider regional economy (Sarley-Pontin 1997–98: 189). It should also be noted that much of the material was obviously burnt.

• from McQuitty et. al. (2020:209-211)

(Figure 10.21)

Two examples of this architectural type ‒ House 1 (G3001) and House 2 (G2001) – were the most prominent standing remains at Khirbat Faris when excavation began. The plan of a similar structure immediately behind and to the west of House 1 could be seen on the surface (Figure 2.1). The following description of the type is based upon these and on examples from the villages of the Kerak plateau (Kana‘an and McQuitty 1994). In addition, 19th-century examples are reported from the north to the south of Jordan, including Umm Al ‘Amad (Lewis 1987: 231) and Sumiya (Khammash 1986: 50–52) near Amman and in the ‘plains villages’ east of Irbid (Mundy 1996: 91). Their distribution is largely confined to areas that historically produced grain on a large scale.

Although the overall construction techniques are similar to those employed in the Transverse-Arch House, the Arch-and-Grain-Bin House can be distinguished by its greater size, by the fact that it stands alone and is not part of a courtyard complex, and by the provision of rawiyat (built-in grain-bins). Examples of this type with up to five transverse arches have been seen on the Kerak Plateau. Local oral sources suggest that such buildings date from the 19th and early 20th century, although they are often described as ‘Mamluk’ in the literature (Worschech 1985: 43–45).

The covered area in this type falls within the range of 135–225 square metres and the average internal height is 5 m. The sensation of standing inside one is like that experienced in a dark, lofty cavern (Figure 10.21). The spaces between the arch-walls are enclosed, plastered and used as grain-bins (rawiyat) or as general storage chambers (Figures 10.10–12). The rawiya was filled through a hole in the roof and houses of this type are often equipped with an external ramp up which sacks of grain were dragged, or carried by donkeys, to be poured into the bin (Figure 10.22). Apart from the grain-bins, there are generally few built-in features, except niches of various shapes recessed into the thick walls. As with the Transverse-Arch House, the exterior is severe and fortress-like with no windows.

While the Transverse-Arch House was regularly used as a dwelling by sedentary villagers, it could also be used as a seasonal dwelling or storehouse by semi-sedentary farmers. The Arch-and-Grain-Bin House, however, was only rarely used as a dwelling and was built by semi-nomads or townsmen to store their harvest and equipment while they resided and socialised in tents pitched nearby. On the rare occasions when residence in the Arch-and-Grain-Bin House became protracted, the spaces between the arches that were not used for storage could be employed as private sleeping areas: during the drought of the 1940s, in nearby Al Qasr, Archand-Grain-Bin Houses similar to those at Khirbat Faris were used by the Majali in precisely this way (Thayil ibn Shlash ibn Faris: personal communication).

• from McQuitty et. al. (2020:211-213)

These buildings have been described as ‘houses’ – a term that is often synonymous with the amorphous concept of ‘home’ and that implies dwelling or residence. A house can also be a place of work, a place for socialising and recreation, a place for storage: a house can function in all these ways and, clearly, the function can change within the same form, and the form can change while the function remains the same. Function may change from season to season, or even from day to night. Function may even be gender-specific. Every house has its own life-story and the interpretation of the architecture at Khirbat Faris attempts to tell such stories of the houses excavated. Here form is more durable than function and the bywords are flexibility and variability (for further discussion of these themes see Bailey 1990: 19–48; Rapaport: 1969). In vernacular architecture, the constraints and opportunities of the local environment are often seen as crucial in determining the form of a building. The local environment may not only determine what materials are available for construction but also influence the very form of the building so that, for example, house openings are generally placed away from the prevailing wind. At Khirbat Faris, major shifts in architectural form are evident and the assumption underlying this interpretation of these shifts is that they reflect similarly major shifts in the needs of the communities which built and occupied the structures. These changing needs may be linked with environmental factors, political and governance factors such as land-ownership and taxation, changing traditions and a combination of all these. The current analysis of the architecture very much concentrates on the link between architecture and the economy.

Four distinct architectural phases, and three radical shifts, may be discerned from the evidence excavated.

1. The Classical Vault and the Late Antique House reflect a culture in contact with the urbanised world of the Eastern Mediterranean. In this wide area, both architectural types were used in a variety of contexts – rural and urban; commercial, domestic, ecclesiastical and military – and exhibit neither the regional variation nor the functional specificity that later appear. The original function of the Khan (Classical Vault) at Khirbat Faris is unknown, but by far its longest use was as part of a complex which incorporated the Late Antique House of Far II (G2003). Both buildings were oriented in the same direction and opened westwards onto a shared paved area. The character of the architecture and the artefactual evidence alike suggest that these houses belong to an ‘urbanised countryside’(Roberts 1996: 70). Although in the countryside and far from any major city, there is not a vast difference between the character of the domestic architecture, glass and ceramics found at Khirbat Faris and that found in larger cities such as Jerash or towns such as Madaba. To paraphrase Hirschfeld (1997: 33), urban elements infiltrated into the rural settlement. Based on archaeological parallels from Palestine and Jordan, Late Antique Houses are seen as indicative of a sedentary life-style.


2. The conversion of the Late Antique House into the Transverse-Arch House marks the first radical shift in architecture and occurred during the Early to Middle Islamic periods (9th–13th centuries). The conversion was effected by the addition of arches that used a different technique of construction (G2003), by imposing a new spatial relationship between the entrance and the roof arches (G5001), and by the insertion of storage areas – both bins and small rooms (G2003 and G5001). The orientation of the houses remained the same, with the door opening westwards, and yet not only the technique of construction but also the entire impression of the houses is altered. External factors, and not necessarily a change of population, may have prompted the shift to the Transverse-Arch House, such as an alteration in the pattern of land-holding, a shift in the balance of pastoralism and agriculture within the mixed-farming economy, or a change from communal to family-based storage. However, what is clear is that these houses belong to an architectural repertoire that is rural rather than urban. In recent times, the Transverse-Arch House has been constructed and occupied by both sedentary and semi-sedentary populations and this same fluidity probably existed in the past.


3. The appearance of the Barrel-Vaulted Houses on the Western Edge marks the second major architectural shift and occurred during the Middle to Later Islamic periods (14th–16th centuries). These single-celled houses are clustered together around a set of courtyards and alleys. Although they frequently made use of the walls of earlier structures, neither the form of the earlier structures nor the spatial organisation of the earlier settlement was followed. Their architecture, the spatial arrangement of the complex and the associated artefactual assemblage, all indicate that the Barrel-Vaulted Houses were occupied by a sedentary, mixed-farming community. The absence of built-in grain-storage facilities might indicate that households stored the grain they produced elsewhere, for example in caves, cisterns or clamps. The Project anthropologists, William and Fidelity Lancaster, report that in the late 19th and early 20th century, the share-croppers who farmed the land around Khirbat Faris lived on the site in small houses, while the large Arch-and-Grain-Bin Houses were used by their Majali landlords for storing the surplus harvest. It is possible that the excavated Barrel-Vaulted Houses housed sharecroppers or agricultural labourers and that other unidentified and unexcavated structures, belonging to their landlords, were used to store the surplus that they produced. If so, then the shift from the Transverse-Arch House, with built-in grain storage, to the Barrel-Vaulted House without provision for domestic storage of grain, might indicate a significant change in the pattern of landholding.


4. The appearance of the Arch-and-Grain-Bin House is the third major architectural shift and, on the evidence available, seems to have occurred at Khirbat Faris in the late 19th or very early 20th century. The Arch-and-Grain-Bin Houses were built de novo. The constructional techniques are those used in the Transverse-Arch House, but they are applied on a far larger scale. There was clearly the need for vastly increased storage. Another factor pointing to the totally different character of the housing of this period is that the orientation totally changed: the doorways open eastwards. The prevailing moisture-bearing wind is from the west and to this day tents are pitched with their opening to the east. The Arch-and-Grain-Bin Houses are distinguished by their isolation – they are not part of a courtyard complex. The earlier structures of the Central Area and the Highest Point, which had their entrances to the west, probably opened into an enclosed courtyard and not directly out into the settlement. The courtyard and the dense cluster of surrounding housing would have provided protection from the wind. The isolated Archand-Grain-Bin Houses were not sheltered in this way but exposed to the elements and therefore their doors opened away from the wind. Through their prominence and their function as stores of their owners’ accumulated agricultural wealth, the Arch-and-Grain-Bin Houses proclaimed their ‘ownership’ of the land. What is more, they were built at a time when the Ottoman government was seeking to introduce land registration (Lancaster 1995: 114–115). In many instances – Al Qasr, Smakieh – the clustering of such houses has marked the beginnings of more permanent settlement. Ethnographic parallels show that they were used in tandem with tents: they are not skeuomorphs of tents. Such buildings are more accurately described as barns and storehouses.

Based on the excavated evidence at Khirbat Faris, a rough calculation gives some idea of the difference in storage capacity of the two types of house.² The TransverseArch House in the Central Area, G2003, possessed 18.90 cubic metres of built-in storage capacity; G5001 at the Highest Point had 42 cubic metres, plus the suggested annexe. By contrast, House 1 provided 559 cubic metres of storage and House 2 had a maximum capacity of 262 cubic metres. Of course, such calculations do not take into account storage in perishable containers such as sacks or mud kuwaʾir and in places other than the excavated structures. Nevertheless, the massive difference in built-in storage capacity provided by the two types of house is clear enough. But what did it mean in human terms? Let us assume that the average adult male consumed 2850 calories per day and obtained the bulk of these – say 75% – from grain. Given that 1 kg of wheat produces approximately 3000 calories, he would therefore have consumed 0.7 kg (850 cm³) of wheat daily, or its equivalent in bread.³ On that basis, and assuming that the grain stores in the four houses from Khirbat Faris were filled to maximum capacity (which they almost certainly never were), Table 10.1 illustrates how many men each could have fed for a year:

Although the figures in the table above are certainly not realistic, they nonetheless make a strong case that the storage capacity of the Transverse-Arch House was intended to provide for a single family or household, but the Arch-and-Grain-Bin House served a group much larger than the immediate family or household, perhaps larger even than the ‘ashira or extended family. Much more research needs to be done upon the socioeconomic role of the Arch-and-Grain-Bin Houses in the Ard Al Kerak, but it seems probable that, except in the leanest years, they would have stored grain surplus to the requirements of their Majali owners and their families. Whether that surplus was transferred to the Majali’s urban base in Kerak for consumption there, or sold locally to supply the annual hajj-caravan, or exported to Palestine and even to the Hijaz, cannot now be said and it may be that all these cases and others applied. In any case, the appearance of the Arch-andGrain-Bin House seems to be intimately linked to the political and economic dominance of the Majali and to the closely related change in the pattern of landholding.

That the Arch-and-Grain-Bin House is more properly conceived of as a barn or storehouse than as a dwelling raises the question of where those who produced the harvest lived. From at least the mid-19th century, there was a sharp distinction at Khirbat Faris between landowners and share-croppers. The latter are reported by oral sources to have lived in small houses on the edge of the settlement. Such houses were not identified during the excavation, although the small TransverseArch House (G3002) attached to the southwest corner of House 1 is a potential candidate. During the years of excavation at Khirbat Faris, the Banu Hamida families from the nearby village of Faqu‘, who farmed the land around Khirbat Faris as share-croppers of the Majali landlords, lived seasonally in tents on the edge of the site. De Saulcy, writing in 1853, records the same phenomenon of tent-dwelling Banu Hamida farmers (1853: 311–315).

• from McQuitty et. al. (2020:227-228)

... At Khirbat Faris, the AD 746 earthquake is manifested in the form of broken and collapsed stone lintels dramatically recalling the destructive power of nature. Whilst the date of the AD 746 earthquake is confirmed by the ceramic sequence of the site as well as comparison with neighbouring archaeological sites it is important to remember that there were both earlier and subsequent earthquakes which may have left less of a physical impression on the archaeology of the site (Ambraseys 2009). For example, there were three large earthquakes in the period AD 746–757, and it is known that the Mamluk citadel at Hesban was destroyed by an earthquake in the mid-14th century (Walker et al. 2017). Also it is possible that changes to the architecture of the buildings may have made Khirbat Faris more resilient to earthquake damage, thus the survival of the barrel vaulted Khan may have been influential in the adoption of the barrel vault as a structural technique in other parts of the site, such as on the Western Edge. It is also noticeable that when transverse arches were used as a roofing method later on in the site’s history (19th/20th century), instead of being spanned by stone lintels the space between the arches was covered with wooden beams covered with a layer of earth. This method would have been both cheaper to build, easier to repair and, during an earthquake, less dangerous.