Figure 1. After R.W. Hamilton, Khirbat Al Mafjar: An Arabian Mansion in the Jordan Valley, Oxford, 1959, fig. 258, copyright Oxford University Press. Threshhold lozenges marked with blue dots, central lozenges with red.
An electronic publication ISBN 0-944940-17-X
Table of Contents
Since I wrote the first version of this article fifteen years ago (as “A Lozenge in the Bath Pavement at Khirbat al-Mafjar”) good photographs of the bath pavement have been published, showing the existence of a second lozenge near the center of the bath, complementing the one I remarked on. I have also studied the bath further. Consedquently I have rewritten that article as the present one and withdrawn the old one.
The conclusions are largely the same.
While preparing for a vacation in Sicily I noted William Tronzo's discussion of the lozenge motif in the opus sectile pavement of the twelfth-century Cappella Palatina in Palermo (a lozenge is a rhombus one axis of which is longer than the other, oriented so that its axes are parallel to the sides of the enclosing field). Tronzo believes that the pavement of the south aisle includes designs related to the route the visitor was intended to take through the building. The designs of the two westernmost bays appear significant, as they break the otherwise consistent symmetry of the pavement. In the westernmost bay, the one a visitor enters today (and on Tronzo's showing the original public entry point), there is an arrangement of five circles flanked by checkered bands. Tronzo argues that this design marks a threshold.1 The next bay toward the altar in the south aisle is decorated with a lozenge filled with circles and developed into circles in the outer corners of the rectangle enclosing it. This is the only lozenge in the pavement and, so far as I can tell, the only lozenge in the building's entire decoration. Tronzo interprets this decoration, too, as a threshold, and contends that “taken together, these patterns may be seen as inscribing a path of movement through the chapel, from a point of entry at the southwest, down the south aisle one bay and into the nave.” The need for two thresholds comes about because the platform at the west end of the nave blocked access to the nave from the entry bay.
Tronzo cites as comparative evidence the occurrence of such “threshold pavements”, characterized by “central elements and narrower flanking bands”, in Antique and medieval buildings at entrances. One of his examples, from Pompeii, has as its central element a bichrome lozenge.2 He also draws attention to the use of patterns to mark significant points in a processional route in “a group of ecclesiastical pavements from Rome and Monte Cassino from the later Middle Ages” and the use of porphyry disks for the same purpose in the Great Palace in Constantinople.3 Perhaps the lozenge in the second bay in the south aisle of the Cappella Palatina, which has no flanking bands, might be interpreted slightly differently: not as a threshold but as marking a place to stand, stop, or wait.
It occurred to me to look for lozenges on thresholds in Umayyad architecture, and the pavement of the bath of Khirbat al-Mafjar was the first place I looked.
The main complex of Khirbat al-Mafjar, in Palestine near Arīḥā (Jericho), was built in the second quarter of the eighth century A.D. and consists of main courtyard with a centered pavilion, a palace, a mosque, and a bath that had been completed and used before the whole complex was destroyed in a mid-century earthquake. The bath has an outsized main hall with a remarkably preserved mosaic floor that has been the subject of much discussion. The structure was composed of nine bays on sixteen piers surrounded by aisles the bays of which were developed into apses. The public entry was in the center of the east side, a private entry from the palace was in the southwest, the three apses on the south side and the aisle in front of them were enclosed by a raised pool, the hot rooms and latrine were on the north, and at the northwest corner was a richly decorated audience hall. The center apse on the west side was marked by a special design in a border strip in its pavement, interpreted variously as a picture-puzzle or a rebus, and a curious stone chain that hung from its frontal arch: this was evidently a place for the owner to sit in audience.
The fields of the mosaic pavement, except those in the apses and the audience hall, were designated “panels” by the excavator, Robert W. Hamilton. Very generally and disregarding the border designs, each bay and each space between piers contains one panel, or rectangular field of geometric decoration, while the outer aisles are treated as continuous fields; each apse has its own decoration, as does the audience hall. The central bay and those on axis with it are larger than the others, and the pavements of the central bay and the four bays in the corners of the nine-bay zone, all of which were domed (see below), are marked by large circular designs, the central one being outsized. There are two exceptions to this scheme, which I shall discuss below.
One need not look far for lozenges. The threshold between the portal and the interior (immediately east of panel 1, which stretches the length of the eastern aisle) is decorated with a lozenge formed of loops of rainbow cable intertwined with each other.4 Another lozenge appears on the threshold to the audience hall (which Hamilton called the dīwān).5 It too is composed of intertwined loops of rainbow cable. As in Palermo, the significance of the lozenge was to mark a spot to stand, stop, or wait. (The occurrence of these lozenges in these locations has led me to accept Tronzo's remarks on “threshold pavements” without further investigation.)
Surprisingly, two lozenges are floated on the patterns of panels 18 and 16, just north and south, respectively, of the central circle and in line with the center of the large pool (shown in Figure 1, crudely redrawn after Hamilton's plan). I find no other occurrence of such a floating motif.6 The lozenge floated on Hamilton's panel 16 is yet another construction of intertwined loops of rainbow cable; the one on Hamilton's panel 18 is angular, its rainbow cables forming swastikas with loops at the corners. It is with these two lozenges that I am concerned.
No circulation pattern that can be derived from the location of the bath's doors and features of its elevation leads to or through either of the central lozenges. From the private entry the owner could have entered the pool immediately, reached the central western apse with a few steps, or proceeded to the audience hall in a straight line. From the public entry, facing the central western apse, guests could have headed to the right to the latrine (entered through a door east of the easternmost apse on the north side) or the hot rooms (entered through the western apse on the north side); or they could have headed to the left to the pool. would have disrobed in the main hall, leaving their clothes in any of the niches that lined it.7 Guests coming to meet the owner could have proceeded due west to the central western apse or by some undefined route to the audience hall. There is no door in the central apse on the north side, so there was no completely axial approach to the pool.
There are other clues to the circulation pattern in the pavement. The continuous carpeting of the three outer aisles strongly suggests they were intended to be walked along. The two exceptions to the scheme outlined above by which the pavement was divided into fields are panels 15 and 25. Panel 15 lies just north of the pool, covering what would otherwise be three fields: this feature seems to be a matter of formal correspondence to the pool8
More importantly, panel 25 lies west of the public entry, just beyond panel 1, which covers the eastern outer aisle. Panel 25 covers what would otherwise be two fields: the space between the two central eastern piers and the next bay to the west, on axis with the central circle and the central western apse. The double size of this panel defines what may be thought of as an entry zone: a guest entering from the west who did not turn immediately left or right along the eastern aisle instead stepped forward onto panel 25. Having done so, he would then have three choices: continue west on axis, or turn left to the pool or right, perhaps to traverse the northern bays of the nine-bay zone to reach the hot rooms or the audience hall (there may have been some water feature in the westernmost apse on the north side, where the entry to the hot rooms was, which would have made the northern aisle unsuited as a route for clothed guests headed to the audience hall).9
Such routes—or others—could have been defined by curtains stretched between the piers. The Umayyads often seated themselves behind curtains before receiving guests.10 Hamilton thought there could have been curtains around the pool and in front of the apse of the audience hall.11 I can imagine a curtain hung between the piers at the west edge of panel 25, blocking a direct view of the western apse and directing guests to one side or the other. But this curtain is pure speculation on my part; because of the destruction of most of the height of the piers there is no direct evidence for it.
Variations of complexity in the design of the panels of the mosaic pavement provide further evidence for the circulation pattern. Hamilton categorized them in seven groups on formal grounds, and it so happens that these groups are nearly uniformly of different degrees of complexity.12 Working from outside in and north to south, I arrive at the following observations:
The three full-length aisle panels (1, 2, and 3) all belong to the category of “interlacings on rectangular grid,” which is the most complex group; the three panels are of equal complexity.
The apses, aside from the central western apse and apse XI, are all fairly simple, exhibiting “sprig patterns” (IV and IX) and “rainbow matting patterns” (VI, VII, VIII, and X).
The central western apse (apse V) is filled with a “basketry pattern”, which is actually a spectacular radiating pattern, perhaps not geometrically complex but visually showy; it occurs elsewhere only in the central circle (panel 17). Uniquely, this apse also has a formally distinct border strip along the front, in which the picture-puzzle or rebus is set.
Apse XI, located south of the eastern entry, at the east end of the aisle of bays in front of the pool, is divided similarly, but somewhat incoherently. The main part of the apse bears a pattern of “concentric interlacing” (complex), while along the front is a strip of interlacing on a rectangular grid (also complex), joined up to the first pattern somewhat uneasily.13 I cannot understand this design except as the result of a change in plan involving an originally intended border strip of some different design that was either never executed after the mosaic in the main part of the apse had already been laid or was removed and replaced. (Had the change been made before any mosaic was laid the main pattern could presumably have been extended.)
The panels between the piers on the north side (6, 10, 12, 14, 18 [the one with the lozenge], 20, 22, 26, 28, and 30) are all simple, falling in Hamilton's categories of “rectilinear diapers or reticulations” and sprig patterns (panel 20 only). The former are coffering patterns, very likely reflecting the decoration of the ceiling above these areas; in panel 20 sprigs replace rectilinear coffering. All of these patterns are relatively simple.
The three panels between the piers that stood south of the central axis (8, 16, and 24) are also rectilinear diapers or reticulations (coffering patterns).
Panel 4, between the two piers at the west end of the aisle in front of the pool, is an interlace on a rectangular grid (complex, not a coffering pattern), while panel 29, corresponding to it on the east, is a rectilinear diaper or reticulation (simple, a coffering pattern).
The four panels in the bays at the corners of the nine-bay zone (7, 11, 23, and 27) have circular concentric interlacings (complex) with corner fillers. The concentric interlace, which is closely related to interlacings on a rectilinear grid, occurs elsewhere only in the curious apse XI and here should reflect domes above. Hamilton thought that these corner bays were cross-vaulted, but only because he found no pendentives; the pavement tells another story.14
As noted above, the central circle (panel 17) had a showy basketry pattern, reflecting its dome.
Panel 15, in front of the pool, has a pattern of interlacing on a rectangular grid (complex), as does the double-length entry panel 25.
Finally, there are the remaining panels along the two main axes of the building. Panel 19, in the bay on the north-south axis north of the center, is on the simple side; Hamilton made a special category for it, “overlapping circles”, but it is not so complex as the interlacings on a rectangular grid. Panel 9, in the bay west of the central circle, is another interlacing on a rectangular grid (complex), as is panel 5, between the two piers directly east of the central western apse. Panels 13 and 21, which flank the central circle on the east-west axis, are classified as sprig patterns but are notably complex compared to the other members of this group.
To summarize, there is a clear and systematic distribution of the more complex patterns in the three long aisles, along the central east-west axis leading to and including the central western apse, and in front of the pool, including the panel between piers closest to the private entry (for which I have no explanation),15 and in the apse on the other side of the hall, the design of which indicates that it either had some significance (which escapes me) or was originally planned to.
Still, none of these observations on circulation patterns yields a direct route to either central lozenge. To reach one of the lozenges a guest would have to have wandered around to it or have been directed to it by attendants or some means such as an arrangement of curtains. Standing at the lozenge, the guest would not have faced the central western apse but the center of the pool or the central northern apse, which is apparently of no special account. So the lozenge may be connected not with circulation but with static location: it may be a place-marker. But for what purpose?
Hamilton interpreted the bath variously and in combination as a “ceremonial apartment with provision, and extravagant provision, for bathing parties” or “Frivolity Hall”; as a majlis al-lahu or “play-room … designed with the spatial, acoustic and decorative properties requisite for the entertainment of a numerous company with music, bathing, perhaps ball games and … other relaxing pastimes”; and as a “music room”.
The main hall was not a ceremonial apartment: the ceremonial apartment—if that is what it was—is the separate, finely decorated chamber entered from the northwest corner of the main hall. The central western apse, as it has been interpreted, is a travesty of a formal throne setting. Nor is the plan of the main hall suitable for an audience hall with the central western apse as its focus, as the piers block the view of that apse from much of the hall. Known Umayyad audience halls at Khirbat al-Minya and Mushattā are instead basilical.
Then again, the main hall may have been intended to accomodate functions such as were associated in Antiquity with large unheated rooms in major baths, including athletic competitions such as wrestling; or performances by the acrobats and dancers who may appear in the frescoes of Quṣayr ʿAmra.16 But an audience for such performances would have found its view obstructed by the piers, and the performers would have had to work around them. As for Hamilton's suggestion of ball games, it would be a strange ball game indeed that could be played among so many obstacles.
In Roman baths the large unheated hall was usually a single unobstructed space, as known in Rome from the Baths of Titus, Caracalla, and Constantine, and reflected in the European and North African provinces.17 The Baths of Diocletian in Rome offer only a partial exception, but even there the subsidary spaces screened by columns were dwarfed by the vast central space. Among Umayyad baths, both Quṣayr ʿAmra and Ḥammām al-Ṣarakh have unheated halls of three aisles opened by large arches to form as unified a space as possible using the architectural devices the builders seem to have had at hand. At ʿAnjar there is a bath at the north city gate with a main hall of three aisles on four piers, but these are slim compared with the massive piers at Khirbat al-Mafjar, and again the intent must have been to create a relatively unobstructed space.18 Thus there is no parallel in bath architecture for an unheated hall filled with piers.
It may be, of course, that the Umayyad builders of Khirbat al-Mafjar simply could not vault so large a space in stone and brick, and that the form of the main hall is an unhappy compromise sacrificing sight lines and open space for overall size. But Hamilton's last suggestion, that the main hall was intended for musical performances (including recital of poetry), is of a function that required no line of sight between performers and most of the audience. Based on anecdotes of the life of the Umayyad caliph al-Walīd II, whom Hamilton believed (with reason but not beyond dispute) to have built it, he emphasized what he felt was the acoustic suitability for singing of the many curved surfaces of the elevation.19
Music and oral recitation were activities also performed in churches. Consequently churches may be considered as possible prototypes, especially as Hamilton notes that al-Walīd spent much time—drinking—in monasteries, although none of the anecdotes he collected mentions any performance in a church.20 The central nine bays of the main hall, plausibly reconstructed by Hamilton with a central dome on a drum (to which I add four corner domes), recall cross-in-square churches. But this is a deceptive similarity. Cross-in-square churches are not ringed with aisles and apses on all sides. The main hall would not have impressed a visitor as being like the sixth-century Ghassānid building north of the city walls of Rusafah, which has a nine-bay plan and has recently been reinterpreted as a church.21 Syrian churches of centralized plan, such as the cathedral of Buṣrā, could perhaps be more likely as inspiration for a design full of curved surfaces, but they have large central spaces.
It is possible to conclude that the main hall would not have been unsuited to musical and poetic performances because of its hypostyle interior, and if the curved surfaces of its elevation—and the breaking up of space with the many piers—in fact enhanced sound, or at least were expected to do so, it would have been positively suited to auditory performance.22 In short, despite the lack of clear formal prototypes, the peculiar form of the main hall of the bath suits it better to musical and poetic performance than to anything else.
Hamilton thought that a reciter or singer would have stood or sat on the central circle, which he labelled “the performer's circle.” 23 If so, the lozenges may have marked its thresholds, so to speak, signalling places to stop, out of sight from the central western apse, before stepping into the central circle.
Or, as the design of the central circle corresponds to the central western apse, which is one of the special locations presumably reserved for the owner on some (other) occasions, the performers might have stood or sat on the lozenges and the patron in the central circle, marked as his special spot, at the point where echoes of the music converged and in the center of the most magnificent element of the mosaic floor.
In a partial survey of floors in other Umayyad palaces I found few other lozenges, and none on thresholds. At Quṣayr ʿAmra a lozenge similarly developed with knots and with a circle in its center is set in the back of the mosaic pavement of the apse of one of the two small rooms behind the three-aisled hall.24 This paucity of threshold mosaics and place-markers may show that the meanings borne by the mosaic pavement of Khirbat al-Mafjar were drawn from Late Antique practice in a particularly idiosyncratic and conscious way that was not paralleled in other Umayyad palaces. If not, the lozenge north of the central circle in the Khirbat al-Mafjar bath pavement may represent a tradition of using place-markers that has gone unnoticed in early Islamic architecture.
1. The Cultures of His Kingdom: Roger II and the Cappella Palatina in Palermo, Princeton, 1997, pp. 99–100 and figs. 20 and 22.
2. Ibid., fig. 138.
3. As cited in De ceremoniis aulae Byzantinae of Constantine VII Porphyrogenitus; Tronzo, op. cit., p. 100. For one of these disks, see Cyril Mango, The Brazen House: A Study of the Vestibule of the Imperial Palace of Constantinople, Copenhagen, 1959, p. 84.
4. Hamilton, Khirbat al Mafjar: An Arabian Mansion in the Jordan Valley, Oxford, 1959; pl. 79; the numbering of floor mosaic panels is given in fig. 258 on p. 328. Also by Hamilton on Khirbat al-Mafjar: “Who Built Khirbat al Mafjar?” Levant, v. 1, 1969, pp. 61–67; “Khirbat al Mafjar: The Bath Hall Reconsidered”, Levant, v. 10, 1978, pp. 126–38; and Walid and His Friends: An Umayyad Tragedy (Oxford Studies in Islamic Art, v. 6), Oxford, 1988.
5. Hamilton, Khirbat al Mafjar, pl. 84.
6. See Hamilton, Khirbat al Mafjar, pl. 78, no. 18; for panel 18. In the recent book by Ḥamdān Tāhā and Donald Whitcomb, The Mosaics of Khirbet el-Mafjar, Hisham's Palace, Chicago, 2015 (here, Tāhā), all elements of the floor except the apses were renumbered silently, confusingly using Hamilton's term “panel”. This numbering is shown on Tāhā's p. 40; the plates that follow have numbers that do not correspond to it, so it is necessary to use the table of plates, pp. 36–37. As identical plate numbers were used for three different sections of plates it is necessary to supply context. The two central lozenges are illustrated in the “Audience Hall” (main hall) section. The large central circle and adjacent panels are shown in Tāhā's pl. A1 (p. 41), which is oriented with up as north. Hamilton's panel 18 is Tāhā's panel 13, shown in his pl. A11 (p. 48). Hamilton's panel 16 is Tāhā's panel 14, shown in his pl. A13 (p. 50). These lozenge are centered not in their fields but between the central circle's edge and the opposite edge of their fields.
To these floated lozenges compare Hamilton's panel 22, Hamilton, Khirbat al Mafjar, pl. 77 (Tāhā's panel 15, pl. B5, p. 53), which has a central emblem anchored in the geometry of the design.
7. Hamilton thought there were wooden benches for guests everywhere (Khirbat al Mafjar, p. 51):
Hamilton seems to have generalized based on partial evidence. The photographs of the main bath hall he published (pl. 76–84) were intended to illustrate the floor mosaics, but they reveal that 1) in many places the walls were not preserved up to a level that would include the “mortices” (square holes), 2) those holes occur irregularly and inconsistently, and occasionally at other heights, and 3) there are fewer holes at 15 cm. than at 60 cm. These observations are borne out by Tāhā's plates (discounting the restored masonry shown in them, which of course has no such holes). I have no suggestion for the purpose of these holes, but I would want to find similar evidence in other baths before concluding that they accomodated wooden benches, which the floor mosaics do not seem to take into account.All round the hall there were wooden benches. The benches themselves had disappeared, but we can infer their existence from mortices in the walls of the exedrae and between them. These mortices were in two rows, 15 cm. and 60 cm. above the floor. Similar mortices, grouped also in pairs, three to a side, were to be seen in the plinths of the piers; these were set at 10 cm. and 40 cm. above the floor.
8. Hamilton called it “a broad gesture of invitation”, Khirbat al Mafjar, p. 327.
Ignazio Arce has argued that the pool on the south side of the main bath hall is secondary, having been installed after the mosaic floor (“A New Umayyad Mosque at Khirbet el-Mafjar, and Other Discoveries: A New Understanding through Architectural Stratigraphy and Landscape Archaeology”, Rachael Thyrza Sparks et al., eds., Digging Up Jericho: Past, present and future, Summertown, 2020, pp. 151–71). Despite his illustrations, it is unclear just how the masonry of the pool is related to the mosaics. At one point Arce wrote that “some sections of the mosaic carpets were cut away after they were laid” (p. 161), but elsewhere he wrote variously “the mosaic carpets are cut or covered by these steps” (p. 161) and “red arrows show where mosaic floors have been covered and cut by the walls and stops of the pool” (caption to fig. 12). Either the mosaic was removed (cut) or built on top of (covered); it cannot be both, at least not in the same place.
Arce analyzed the mosaic design in support of his view that the pool was secondary:
As a normal procedure in mosaic production, the decorated motifs (the ‘carpet’) are set aside from the walls and pillars by means of a plain band of white tesserae, in order to better adjust the decorative pattern to the available space, as defined by the architecture. These bands can be seen around all the perimeter of the building, and also around the pillars. Surprisingly, no white perimeter band can be seen in front of [the] flight of steps of the pool (natatio) located in the souther[n] bay of the building. Here the mosaic carpets are cut or covered by these steps, indicating that they were added once the mosaics had already been laid and been in use for a certain period.
In fact, the mosaics are adapted to the presence of the pool. Panel 15, directly in front of it, extends across not only the central bay of its intermediate aisle but also out to the outer edges of the piers flanking it, unlink the corresponding Panel 19. There is no trace of anything corresponding to Panel 2 on either side of the pool (a corresponding panel would have butted up against Panel 1, but the south end of that panel is surrounding by a white border). The continuous border that outlines all the apses is routed around the pool and special borders lie directly against the steps—these are not fragments of larger panels cut or covered by the pool. Correspondingly, the white border with zigzag edging that surrounds all the piers except where space was tight at the north end of the hall is pared down to accomodate those borders. It was a design decision not to run a white border in front of the pool.
9. Hamilton, “The Bath Hall Reconsidered”, p. 133.
10. For example, in Hamilton, Walid and His Friends, p. 110, is an anecdote in which al-Walīd is said to be behind a red curtain, which apparently the guest walked around, as it is not said to have been drawn back.
11. Hamilton, Khirbat al Mafjar, p. 50; and Walid and His Friends, p. 38, on the basis of another anecdote rather than archaeological evidence.
12. Hamilton, Khirbat al Mafjar, p. 329.
13. Op. cit., pl. 81, center.
14. Op. cit., pp. 74–75.
15. This was not the owner's route to the pool; there was a small stair built for him in the western aisle next to the pool, even closer to the private entry; Hamilton, Walid and His Friends, p. 42 and fig. 10.
16. For these frescoes see Garth Fowden, Quṣayr ʿAmra: Art and the Umayyad Elite in Late Antique Syria, Berkeley, 2004, best consulted through the index.
17. Representative examples in J. B. Ward-Perkins, Roman Imperial Architecture, London, 1981.
18. K. A. C. Creswell, Early Muslim Architecture, 2 v., Oxford, 1932–40; v. 1 revised and published in 2 parts, 1969, pt. 1, p. 480 and fig. 542.
19. Hamilton, Khirbat al Mafjar, pp. 105 and 346; “The Bath Hall Reconsidered”, pp. 126–27, 133; Walid and His Friends, pp. 22–32. I think the attribution to al-Walīd II is confirmed by the rich and inventive decoration of Khirbat al-Mafjar's palace and bath, which are paralleled at the comparatively nearby sites of Khirbat al-Minya, Qaṣr al-Ṭūbah, and Quṣayr ʿAmrah, all works of al-Walīd II,
20. Walid and His Friends, pp. 86–91.
21. It is also thought to have had a pyramidal wooden roof over its central bay; Elizabeth Key Fowden, “An Arab building at al-Ruṣāfa-Sergiopolis”, Damaszener Mitteilungen, v. 12, 2000 [2001], pp. 302–24; and Gunnar Brands, “Der sogenannte Audienzsaal des al-Mundir in Resafa”, Damaszener Mitteilungen, v. 10, 1998, pp. 211–35.
22. This expectation could be tested by building a replica of the bath, for which adequate guidance is available from Hamilton's detailed reconstruction.
23. Hamilton, Walid and His Friends, caption to fig. 9.
24. Martin Almagro et al., Qusayr ʿAmra: Residencia y baños omeyas en el desierto de Jordania, Madrid, 1975, fig. 10 and pl. 13, c; nicely drawn in Creswell, op. cit., figs. 458 and 460. These are reproduced by Christel Kessler, “Die beiden Mosaikböden in Quṣayr ʿAmra”, Studies in Islamic Art and Architecture In Honour of Professor K. A. C. Creswell, ed. C. L. Geddes, Cairo, 1965, pp. 105–31, figs. 3 and 5. (Kessler also noted some aspects of the distribution of mosaic patterns at Khirbat al-Mafjar but with attention to the interlace ornament. The design in the other small room corresponding to the lozenge is a vase with vine tendrils emerging from it; I do not see a correspondence with the lozenge, but perhaps there is one). This lozenge was also noted by Julian Raby, “In Vitro Veritas. Glass pilgrim vessels from 7th-century Jerusalem”, Bayt al-Maqdis: Jerusalem and Early Islam (Oxford Studies in Islamic Art v. 9, pt. 2), Oxford, 1999, pp. 113–90, p. 127, in another context. In that context Raby sees the lozenge enclosing a circle as a symbol (my word) for Jerusalem as the omphalos of the world (pp. 184–90). That particular symbolism cannot apply to Quṣayr ʿAmra and Khirbat al-Mafjar; how the omphalos-symbolism of the lozenge and its use as a place-marker are connected is a fascinating topic I cannot explore here.