• 1 The most comprehensive studies are still Dalman 1908, 1912 on the Felsheiligtümer of Petra. The man (...)
• 2 The term is used by Plutarch when describing how Horus, angry at his mother Isis’ release of Typhon (...)
• 3 Malaise gives an overview of the use of the symbol in the Egyptian context. In Egypt, the associati (...)
• 4 Analysis of the cult of Isis in Petra has generated a considerable amount of literature. See Milik (...)
• 5 This has been most extensively analysed by McKenzie, who has shown that several Alexandrian archite (...)

1The unique iconography of Petra has puzzled scholars since the city was rediscovered in the 19th cent., and we are little further forward today in our understanding of the unrecognisable forms of the hundreds of geometric idols that line the wadis leading to the town centre1. These played a central part in the patterns of worship that were played out by Petra’s inhabitants, and of which the city’s rock-cut monuments provide only a lifeless shadow. Occasionally, however, a recognisable form or inscription appears that may allow us a glimpse into the minds of those worshippers. The Isis basileion is one such example2. This familiar symbol, consisting of the disc of the sun surrounded by two horns, has a long history in Egypt, where it is also associated with several other deities3. Of these, however, only Isis is attested at Petra, and so the discussion surrounding the basileion has focused on her4. Petra was the only place in Nabataea where the deity was certainly present, and she had a cult following there by the end of the 1st cent. bc. The city’s close connections to Ptolemaic and Roman Egypt, most visible in the tomb architecture5, make it likely that she was introduced from the west, where some of the caravans that passed through Petra were undoubtedly destined.

• 6 I am deliberately refraining here from the use of ‘betyl’ to describe these monuments. Gaifman has (...)

2My focus here, however, is not the origin of her cult, nor do I intend to concentrate in great detail on the numerous instances where Isis has been identified in the city, although some of these will necessarily be revisited. I will rather use our evidence for Isis as an illustration of how we ought to approach Petra’s religious monuments, and in particular its geometric idols6. We are fortunate in this case that we can follow the iconography, and to some extent the cult, of Isis across the city, which supplies an easily identifiable reference point among the largely ‘aniconic’ religious landscape, and this provides an example of how symbols, forms and motifs were employed by worshippers. It will first, then, be necessary to track the use of the basileion across Petra, and discuss its association with the cult of Isis, before moving to some more general conclusions about Petra’s religious art.

• 7 The site was first published in Milik and Starcky 1975, p. 120-124, but they concentrated chiefly o (...)
• 8 Nehmé classifies the site as a ‘private and collective’ cultic space (Nehmé 1997, p. 1047). We know (...)
• 9 Merklein and Wenning 2001, p. 424.

3Our only certain attestation of Isis in Petra comes from a narrow ledge high up on the bank of the Wadi Siyyagh before it opens out into the town centre7. Here, four niches are arranged together, with a basin/niche set into the rock face further to the right (fig. 1). The ‘sanctuary’ could not contain large groups of worshippers, and it has been suggested that it was used by a private religious association, a mrzh≥’, devoted to Isis8. It seems likely that the layout was conceived of as a whole, or at least that the monuments were incorporated together into rituals. The basin would provide water for lustrations or offerings, as was common in the city. Merklein and Wenning believe that there was a connection between this ledge and a larger platform below, where rituals involving larger groups of worshippers could have been held9.

• 10 See Milik and Starcky 1975, p. 121, for the dating of the text. The decision is between Obodas I (c (...)

4The first niche, slightly apart from the rest, is empty. The second niche is of most interest to us here and contains the figure of a seated draped goddess (fig. 2). An inscription surrounding this niche tells us that this is the image of Isis that was made in the fifth year of King Obodas III (26/25 bc)10. The text is inscribed on both sides of the image, arranged as shown below, and badly eroded in parts :

• 12 Merklein and Wenning 2001, p. 426.
• 13 For Isis Lactans, see Tran Tam Tinh 1973. The figure is well attested in the art of Pharaonic and R (...)
• 14 Milik and Starcky proposed that the figure would have worn the ‘Isis knot’, and so the right hand c (...)
• 15 Merklein and Wenning 2001, p. 426.
• 16 Donner 1995, p. 15. We should note that the motif of a goddess seated on rocks has a place in the o (...)

6The carving has unfortunately suffered considerable defacement ; there is nothing left of the head and the arms are also missing. Indeed, Merklein and Wenning remark that were it not for the inscription we would not be able to identify her with any certainty12. We cannot, therefore, know whether Isis here was wearing the basileion, although parallels with figurines, which we will see later, make it more than a possibility. We also cannot discern with certainty her pose. The right arm seems to be held across her breast, as in the Isis Lactans type, but the exact configuration is too obscured to be certain13. Some have seen rather Isis in mourning, no doubt keeping in mind the many figurines from Petra that show a goddess in this pose, but her right hand does not seem to be held to her cheek here14. Merklein and Wenning propose that we have the “Palliata type”, arguing that the “figure follows none of the common Isis types, but Hellenistic types of draped women in general.”15 What can be discerned is the very peculiar form of the throne Isis is seated on. This is made up of an asymmetrical series of spherical objects grouped on either side of Isis’ legs. The most convincing explanation is that we have here a representation of the precipitous surroundings of Isis’ location. Indeed, the intent may be to show here Isis presiding over the whole of Petra and its wadis16.

• 17 Donner 1995, p. 12 ; Merklein and Wenning 2001, p. 426 ; Janif 2004, p. 124.
• 18 Milik and Starcky 1975, p. 123. See Donner 1995, p. 15, for further arguments as to why we should n (...)

7There are three niches to the right of Isis. The furthest away is described by most as a basin, and so would have provided the water needed for certain rituals17. We should note also that there is a square recess in the base of this, clearly carved as part of the ensemble, which may have held a portable idol. The two niches to the right of Isis hold geometric forms (fig. 3), one with a triangular head and the other consisting of two recesses, one inside the other. It was initially thought that the first may have also originally been an anthropomorphic figure that was also defaced, but the smooth surface suggests that this has survived largely as it was carved18.

• 19 For what follows see ibid.
• 20 See Lightfoot 2003, p. 446-449 and 540-547 for a discussion of the Semeion and its iconography.
• 21 Janif 2004, p. 123.
• 22 Ibid., p. 124. Janif was no doubt working from earlier suggestions here. Milik and Starky had descr (...)
• 23 Janif 2004, p. 124, n. 32.
• 24 Coins from Hierapolis show this arrangement of the two gods on either side of the Semeion, with Ata (...)
• 25 Janif 2004, p. 126, fig. 4.
• 26 Drijvers 1980, p. 80-81.
• 27 Lightfoot 2003, p. 542.
• 28 I am following here the reading of Healey 2001, p. 50. He notes the unusual spelling of Atargatis, (...)
• 29 Janif 2004, p. 121.

8It is worth pausing here to consider Janif’s peculiar analysis of the ensemble, as it is symptomatic of some of the problems with the interpretation of Petra’s religious art19. His aim is to demonstrate that scholars place too much emphasis on inscriptions when interpreting Petra’s monuments, and that more emphasis needs to be placed on the iconography. He argues that these two niches plus the Isis figure are in fact based around the iconography of the trio of Atargatis, Hadad and the Semeion, as described in Lucian’s DDS (31-33)20. Attention is drawn to the peculiar iconography of Isis which, it is true, does not seem to easily conform to any of the usual poses of the goddess21. Janif then claims that the objects to the sides of Isis’ legs are lions, and that there is therefore a link here with the representations of Atargatis enthroned between two lions22. No such features, however, can be discerned today, and there is also no sign that that part of the sculpture underwent the same kind of defacement that other parts suffered. The theory does also not take into account that the fourth niche could have held a geometric idol, although Janif shows an awareness of it23, thus adding a fourth member to the ‘triad’. We also have the difficulty that the Semeion is considerably larger than Hadad, which is not the case in other representations of the trio24. His comparison between the form of the central idol and a conical block shown in between two seated deities on a relief from Edessa is equally erroneous25. This was originally interpreted by Drijvers as representing Atargatis and Hadad seated on either side of the Semeion, which is represented as a conical block26. If the Semeion is represented in block form here, it is argued, then there should be less difficulty seeing it in the Wadi Siyyagh. Firstly, however they are clearly a different shape, the Edessan block being a plain narrow cone. Secondly, the closest parallels from Petra’s iconography for such a form are the numerous nepheshes carved throughout the city. The block on the Edessan relief is certainly not a nephesh, showing that iconographical parallels alone sometimes carry no great significance. In any case, the Edessan block has now been dated to the 3rd millennium, making any comparison with the Hierapolitan trio redundant27. A further difficulty is that a block further along the Wadi Siyyagh, a c.30 mins walk from the town centre, is explicitly identified as Atargatis with the inscription : ’tr‘t’ mnbgyt’ - Atargatis Manbigitess28. Janif does not explain why the figure here is not identified similarly, although he shows an awareness of the inscription29.

• 30 Parr was the first to publish the site, although he was helped by earlier reports from many others (...)
• 31 Parr 1962, p. 21.
• 32 Ibid., p. 22 only makes a brief mention of these graffiti, with no attempt to decipher them. The si (...)

10Another statue of an enthroned goddess was found on the side of the Wadi Abu Olleiqa, near the Jebel Harun, to the southwest of the town centre (fig. 4)30. This monument was again carved on a terrace above a ravine, but in this case there seems to have been room for a greater number of worshippers. Parr also notes evidence for permanent settlement at the site, including pottery, building blocks and even a column drum31. The site should therefore be seen as one of the many smaller satellite settlements that surrounded Petra’s town centre. Numerous graffiti lined the rock faces nearby, suggesting that the site was frequented by travellers moving westwards from Petra32.

• 33 Ibid., p. 22. Several examples of similar ‘panels’ can be seen in El-Khouri’s catalogue of figurine (...)
• 34 Roche 1987, p. 218.
• 35 Milik and Starcky 1975, p. 123.

11In this case there is no inscription to identify the figure, but the similarities in dress and pose have led most to identify her as Isis. Parr suggested that a rectangular recess beneath her feet may have held a dedicatory inscription, but similar shapes on figurines suggest rather that this was intended to represent a pedestal33. Again she sits on a throne, although this time the seat is represented in a more regular fashion. The head and arms have again been defaced, so there is uncertainty as to her pose and we cannot discern whether she was wearing the basileion or not. The position of the arms is hard to discern from the available photos. Roche remarks that “La pose est hiératique, les mains sur les genoux”, but goes on to comment that many details have been lost as a result of later iconoclasm34. It seems, in any case, that, like the statue of the Wadi Siyyagh, the pose of this goddess does not easily conform to any of the well-known Isis types. Milik and Starcky, however, remarked that the ‘Isis knot’ can be seen on her breast, and so suggest that the same symbol adorned her statue in the Wadi Siyyagh, although this has now disappeared35. There is a more specific link beyond the dress and pose, then, with the wider iconography of the goddess, but without an inscription we can have little certainty as to how this figure was understood in Petra.

• 36 El-Khouri 2002, p. 11 and p. 52-54.
• 37 The mythical reference here is to Isis mourning the death of her husband Osiris. See Bricault 1992.
• 38 El-Khouri 2002, p. 53. A clear parallel can be seen in a figurine now in the Louvre which shows Isi (...)
• 39 Those dated examples are nos. 33 (second century ad), 44 (fourth century ad), 51 (dated by coin to (...)

12A number of terracotta and stone figurines have been found during excavations at Petra that show an enthroned goddess wearing the basileion. In her catalogue of Nabataean terracotta figurines, El-Khouri lists twenty examples from Petra which she identifies as Isis36. These can be broadly split into two types : the first shows the goddess in the characteristic pose of ‘Isis in mourning’, with her right hand raised to her cheek37 (El-Khouri nos. 33-39, 43 and 45), the second has her hands held together on her lap (nos. 40-42, 44, 46-52). Although the goddess is similarly draped, none of these poses seem to correspond exactly to the statues of the Wadi Abu Olleiqa or the Wadi Siyyagh. One noteworthy similarity with the Wadi Abu Olleiqa figure is that a few of these figurines have under the feet a similar rectangular plaque (El-Khouri fig. 7, nos. 41-42). El-Khouri makes no mention of markings within the space and interprets it as part of a pedestal, which seems the most likely suggestion38. Although the provenance of many is unknown, the figurines seem to come mainly from the town centre. Examples of El-Khouri’s Isis types were found in excavations at Ez-Zantur, El-Katute, Zurrebah and the Temple of the Winged Lions. Apart from the two examples from the Temple of the Winged Lions, then, El-Khouri can classify all of these as coming from a domestic context. Information as to their date is very scarce, mainly relying on the finding of coins in the same context. Of El-Khouri’s Isis types, only four can be loosely dated by associated finds, spanning 1st–4th cent. ad. They were, then, being produced in the Nabataean period39.

• 40 Zayadine 1991, p. 297.

13A few more specific points must be made before we move on. Of El-Khouri’s twenty ‘Isis’ figurines six wear the basileion, and these are split into two different styles : in the first, the headdress consists of a circular dish with three shapes protruding from it (fig. 5), which Zayadine identifies as a “plante trèflée”40. In the second type (fig. 6), the disk is surrounded by two stylised horns and a feather or palm protrudes from the top. In all these cases the goddess holds a palm or a feather in her left hand, and the ‘Isis knot’ is clearly visible on her breast. Two further figurines have the arms in the same position and a feather/palm in the left hand, but no basileion (nos. 43 and 45). The rest have neither the basileion nor a similar pose ; their association with the Isis types seems to rest either on their similar hairstyle or the fact that they are draped and seated. El-Khouri’s unqualified identification of all these figurines as Isis, then, may need to be somewhat tempered by cautioning that we have very little idea of who the figurines were intended to represent in Petra. They do, however, seem to form a broad typological unity within Petra’s figurines.

• 41 El-Khouri mentions that two moulds are included among her catalogue of figurines (El-Khouri 2002, p (...)
• 42 Ibid., p. 45-47. She sums up : “In general, the terracotta figurines of Petra are evidence of the w (...)

14We should also stress here some of El-Khouri’s conclusions about Petra’s figurines as a whole, as they will have relevance later. While the pose of ‘Isis in mourning’ has many parallels from Egypt and elsewhere, there can be no doubt that these figurines were produced in Petra. Beyond the fact that moulds and kilns have been found there41, it seems that there was a recognisable style to Petra’s figurines. El-Khouri lists lack of attention to anatomical details, disproportionately sized body parts, rounded faces and asymmetrical limbs among the distinctive features of Petra’s figurines42, and her Isis types are more or less in line with these. They are, then, first and foremost a local expression of piety.

• 43 Wenning provides an overview of the place of eye idols within the discussion of Petra’s religious a (...)
• 44 For a discussion of the variety of forms of this type of monument, see Gaifman 2008, p. 56-58. She (...)
• 45 Hammond 1980.
• 46 See Hammond, Johnson and Jones 1986. The temple had been dated on the base of stratigraphic data to (...)

15‘Eye-betyls’ or ‘Gesichtsbetyls’ have formed a distinctive category within Petra’s geometric idols since their first discovery43. They vary greatly in the detail and manner in which facial features are represented, some showing just a pair of thinly incised square eyes, others including a nose, mouth and even a headpiece44. Perhaps the most famous, and the most detailed, example was uncovered in excavations at the Temple of the Winged Lions in 1975 (fig. 7)45. The face here is set within an architectural frame, with a decorated column on either side and a frieze with dentils running along the top. Below the frame the carefully carved letters ’lht h≥yn br nybt identify the block as “The goddess of Ḥayyan son of Naybat”. On either side of the inscription there are carved two symbols, perhaps palms or feathers, and we are reminded of the numerous figurines showing ‘Isis in mourning’ that have the goddess holding a similar object in her left hand. The face itself is the most anthropomorphic example of the eye idols so far discovered in Petra. The shape of the lips is clearly outlined, and the mouth within this. The eyes are oval shaped, not the usual rectangular form, and are covered by thick curving eyebrows. Similarly, the long nose is a more rounded form than is usual. The inscription does not include a date ; the best we can do is to recall that the temple had been built by ad 27/28, and that the block must have been installed either at this point or sometime afterwards46.

• 47 Zayadine 1991, p. 289. Healey, for example, identifies the block as an ‘Isis Figurine’ (Healey 2001 (...)
• 48 Hammond 1990, p. 123-124.
• 49 Hammond brings in architectural decoration, a ring-seal, fragments of figurines, a lamp, pieces of (...)

16For our purposes, however, it is the wreath or diadem that runs across the block above the eyes that is of most interest. In the centre of this there is a deep round recess, which may have originally held a precious stone, surrounded by two crescent shaped symbols. Zayadine has suggested that we have here a stylised version of the Isis basileion, with a disk surrounded by two horns —a suggestion which seems to have gained a general acceptance47. It is one of the pieces of evidence, rather confusingly, that led Hammond to identify Allat as the chief deity of the Temple of the Winged Lions48. His analysis is based on extremely fragmentary evidence, much of which does not have a direct relevance to cult practice inside the temple, and involves drawing connections between deities and pieces of evidence that come from very different contexts49. In any case, the Goddess of Ḥayyan did not stand on the podium at the heart of the building but was probably set into the wall at some point in the interior. The basileion symbol takes an unusual form : setting it inside a laurel is peculiar to Petra’s block idols, as we shall also see in the next example, and the ridges on the horns are not found elsewhere in the city. However, the basic components are still there, and we can accept that the inspiration for this design is likely to have been another example of the basileion from Petra.

• 50 Lindner 1988, p. 84.
• 51 Ibid., p. 85. The sanctuary was first published in detail by Savignac (Savignac 1932, 1933 and 1934 (...)
• 52 The two ‘star-eyed’ idols from Ain es-Shalaleh were argued by Starcky to represent the stellar aspe (...)
• 53 Zayadine’s position is a little more complicated than this. He sees ‘al-‘Uzza-Aphrodite’ as the pri (...)
• 54 Savignac 1934, p. 574-574, no. 17, pl. XXXVIII. Savignac did not recognise the exact form of the di (...)

17The second example shows the motif a lot more clearly (fig. 8). This smaller block was found in the domestic area of Ez-Zantur and labelled by Lindner “Eine al-‘Uzzā-Isis-Stele”50. Again a wreath runs along the top of the block, and at the centre of this we have another version of the basileion. This time the disk is surrounded by a ring, a feature that may just be visible in some of the figurines, and the two horns protrude upwards. The facial markings are more in line with other eye idols. There is a rectangular nose set in between the two eyes which are not completely square, but curve inwards a little on either side. Lindner draws a parallel here with a block from the Ain es-Shalaleh sanctuary in Wadi Rumm51. This has a similar set of eyes, although these seem to be even further away from square, leading some to suggest that they should be seen as stars and that they refer to the astral aspect of the deity represented52. Absent from the Wadi Rumm example, however, are the inner details on the eyes, perhaps intended to represent pupils. Lindner and Zayadine see the block as confirmation that Isis and Al-‘Uzza were assimilated in Petra, but we should avoid such conclusions53. The Wadi Rumm block is identified as Al-‘Uzza in an accompanying inscription, but it appears alongside another idol with the same form of eyes that is identified as Al-Kutba (figs. 9-10)54. The form, then, was clearly not particular to one deity. The block cannot be dated with any certainty.

• 55 See, for example, Lyttleton 1990 and Stewart 2003.
• 56 McKenzie 2005, p. 40. See Farajat and Al-Nawafleh 2005 for the results of the excavations.

18Every visitor to Petra is greeted with a view of the basileion at the centre of the city’s most famous façade (fig. 11). The series of female figures along the top row clearly take their inspiration from Hellenistic models, although there has been found no parallel of the whole arrangement and attempts to decipher a meaning behind the whole scene have not been successful55. The date of the monument has also been a source of considerable debate, with suggestions ranging from the early 1st cent. bc to well after the Roman annexation in ad 106. There now seems to be consensus that the structure is a product of the Nabataean kingdom. McKenzie’s detailed analysis of the architectural features gave it a terminus ante quem of the beginning of the 1st cent. ad, but recent excavations in front of the façade have suggested a date in the first half of that century56. The prominent position of the Khazneh has given rise to the idea of it serving as a royal tomb, and in this context the most likely occupant is Aretas IV (9 bc–ad 40).

• 57 See Thompson 1973, although the figure there usually appears without the Kalathos. See p. 51-62 for (...)
• 58 See Meshorer 1975. The coins of Aretas III (87–62 bc), based firmly on Seleucid models, show on the (...)

19For our purposes, it is the central figure above the door and the symbol below, perched on the top acroterion of the pediment, that are of most interest (fig. 12). The symbol was quickly recognised as the Isis basileion with two horns curving around the central disk, and an ear of corn on either side. This led early commentators to suggest that the building was a temple to Isis, but the layout of the interior has now shown beyond doubt that it was a tomb. The female figure clearly takes its inspiration from Ptolemaic models, the closest parallels coming from figures on faience vessels dating from the 1st cent. bc. There, such figures are explicitly identified as Isis, Tyche, the Queen, or any combination of the above57. No inscription identifies the figure on the Khazneh, but this has not stopped scholars trying to give her a name, with most suggestions surrounding Isis, Demeter, Tyche or various combinations of the three. It seems more likely that she could be interpreted in numerous different ways by different viewers. There is, however, a clear link with royal iconography. She holds a cornucopia in her left arm, which also appears on Nabataean coins from the earliest designs to the very end of the kingdom58. We shall see that coins make a more explicit link between the basileion and the queen.

• 59 Schwentzel 2005, p. 162. The symbol was first identified by Parlasca on an unpublished coin of Huld (...)
• 60 They are Meshorer 1975, nos. 163 and 164 respectively, although Meshorer did not recognise the exac (...)

20A later development in Nabataean coinage, starting under Obodas III (30–9 bc), was the inclusion of the bust of the queen alongside the king, or sometimes alone on the reverse. From the time of Aretas IV, she could be shown wearing a stylised version of the basileion. Schwentzel has tracked this phenomenon, suggesting that it appeared on both the queens of Aretas IV, Huldu and Shaqilat, and those of Rabbel II (ad 70–106), Hagiru and Gamilat59. Those of Huldu (fig. 13) and Shaqilat (fig. 14) take different forms, the first showing the two horns and the disk and the second showing a disk with four protrusions, perhaps intended to show two horns with two ears of corn, as on the Khazneh. Those of Hagiru and Gamilat have not been published in enough detail for us to discern the exact form of the symbol60.

• 61 Schwentzel 2005, p. 162.
• 62 Schwentzel 2008, p. 276, although in this case it is the role of “Isis-Tyché” that he sees the quee (...)
• 63 Ibid., p. 271.
• 64 . Versnel 1990, p. 39-95.

21Schwentzel sees the symbol as evidence that “les reines nabatéennes à partir de Huldu étaient sans doute assimilées à Isis et Al-Uzza”61, basing himself on the idea of an assimilation of Isis and Al-‘Uzza in Petra. A link is then made with the figure on the Khazneh, and the suggestion is made that we have here a queen, perhaps Huldu herself, assuming the role of the deity in Petra62. While the evidence cannot take us this far, and we should avoid ascribing names or characteristics to figures unless they are explicitly identified, Schwentzel is more convincing in his analysis of Ptolemaic parallels. It has long been established that the Nabataeans looked to the neighbouring Seleucids and Ptolemies for inspiration in their coinage, and this clearly extended to the imagery and titles of the queens63. The connection between the Ptolemaic queen and Isis was well established, culminating in Cleopatra VII (51–30 bc) adopting the title Nea Isis. The supreme nature of the deity complemented and reinforced the position of the Hellenistic sovereign, as has been analysed by Versnel64.

• 65 . Gawlikowski remarked of Isis in Petra : “C’est le seul cas certain d’une divinité étrangère adopt (...)
• 66 For an overview of the material from the Decapolis and surrounds, see Lichtenberger 2003, p. 313-31 (...)

23There is one characterisation of Isis that has persisted throughout the examination of her in Petra : she is a ‘foreign’ goddess, somehow distinct from those deities more at home in the city65. It is argued that the location of her cult sites, in the Wadi Siyyagh and Wadi Abu Olleiqa, are evidence of the importing of Isis from Egpyt, and that her cult may have been established and maintained by merchants moving between Petra and Egypt. The Wadi Abu Olleiqa site, as we have seen, is on one of the main routes out of Petra to the west, and it is also remarked that the Wadi Siyyagh leads westwards out of Petra. Both these sites were on the fringes of the town and they could not hold a large number of worshippers. Added to this, Isis is attested nowhere else in Nabataea, although she was known in other areas of the Roman Near East66. This is in contrast, of course, to her considerable following in Egypt. Those who introduced Isis to Petra, and she must have been introduced at some point, had most likely then first encountered her in Egypt, as the cultural and mercantile connections between Petra and Egypt are well documented. She then seems to have failed to make any further progress within Nabataea. Furthermore, Isis is represented anthropomorphically, certainly in the Wadi Siyyagh and in figurines and perhaps in the Wadi Abu Olleiqa. This is unusual against the background of Petra’s hundreds of geometric idols, and the anthropomorphic Isis stands out in the ‘aniconic’ religious landscape.

• 67 The earliest being the inscription from the Aṣlaḥ triclinium before the entrance to the Siq which m (...)
• 68 Milik and Starcky restored Brhbl, Qymw’ and Tym’, but later commentators have not been able to foll (...)
• 69 The exact form Qywm’ is not attested elsewhere in Nabataea, but it is very similar to the forms Qwm (...)

24The classification of Isis as ‘foreign’ to Petra, however, may be more damaging than helpful to our understanding of the goddess in the city. We should first draw attention to the date of the dedication, 26/25 bc, which makes this the second earliest certain attestation of a deity in Petra. If this was the date when Isis was first introduced to the city, then as far as we know she was one of the earliest deities to have received a cult in Petra67. Secondly, we should take note of the names of the dedicants, which unfortunately do not survive well68. All commentators, however, have agreed on Qywm’, a name which is not at all out of place or ‘foreign’ in Nabataea69. We can certainly take issue, then, with the idea that the cult was established or maintained by Egyptians who moved through Petra. It is the manner of Isis’ representation, however, that most significantly sets her apart in the eyes of the modern observers. Anthropomorphic sculptures are very rare among Petra’s rock-cut monuments, although they are more common in the architectural sculpture of the city centre, and so they have often been considered a separate category to the rest of Petra’s religious monuments. In the Wadi Siyyagh, however, we should not lose sight of Isis’ surroundings. She is carved on a narrow ledge, immediately alongside three other niches, two of which certainly contained geometric idols. While Janif goes too far in classifying the group as a triad, it seems likely that the four niches, and perhaps the empty niche to the left, were employed together in ritual practice. Those who worshipped here, then, must have seen no conflict between the anthropomorphic and ‘aniconic’ ; these modes of representation were used in unison and not seen to be in opposition, and this has important implications for how we should view the wider religious art of Petra.

• 70 Donner 1995, p. 15.

25Isis should therefore not be viewed as a foreigner to Petra’s religious landscape, or at least in no way more ‘foreign’ than Petra’s other deities. She had worshippers with personal names common in Nabataea, was one of the earliest deities we have evidence of, and she appears at ease alongside Petra’s characteristic geometric blocks. Indeed, her representation here, carved into the rock, suggests rather a deity just as ‘at home’ in the city as any other. Her throne is the strongest confirmation of this. There has been so much difficulty in establishing its exact nature because no parallels exist ; the design is unique to Petra. Although the common elements of Isis’ many representations throughout the Mediterranean, a seated draped goddess, are reproduced here, a local interpretation is also added : “ein Bergthron also, der Isis als Herrin des Gebirges erkennen läßt, in welchem Petra liegt und in dem sich auch das Nischenheiligtum befindet.”70 The cult of Isis in Petra, then, is not merely a reflection of the iconography and rituals of the cult elsewhere, but just as much the product of local religious ideas and practices.

• 71 Patrich 1990 and 2007 ; Mettinger 1995.
• 72 Both Mettinger and Patrich consider that there was an injunction against graven images among the Na (...)
• 73 Gaifman 2008, p. 57, n. 57 cites Healey 2001, p. 156, Mettinger 1995, p. 63 and Patrich 1990, p. 86 (...)

27This should be put into the context of the debate over ‘aniconic’ representations of the gods in Petra and Nabataea as a whole. Mettinger and Patrich have amply illustrated this phenomenon, and shown that there was a longstanding tradition of such modes of representation in Nabataea and surrounding areas71. The minority of anthropomorphic representations are treated as the product of foreign influences, most notably as the product of Graeco-Roman artistic traditions that are seen to have moved into the Near East during the Hellenistic Period72. This explanation is able to also conveniently explain the eye idols, like those we have seen above, that show anthropomorphic features on geometric blocks. They are the result of an intermediary stage, where these outside anthropomorphic influences have started to leak into Petra’s religious art73. We therefore have a scale with entirely ‘aniconic’ rectangular blocks, the product of a true ‘Nabataean’ culture, at one end and anthropomorphic representations like Isis in the Wadi Siyyagh, the product of a ‘foreign’ culture, at the other end.

• 74 Patrich 1990, p. 113.
• 75 Mettinger 1995, p. 27.

28This model must now be challenged in Petra, and not only from the evidence of the Wadi Siyyagh and the use of the basileion. Figurines and geometric idols of a similar size have too often been overlooked in the debate. El-Khouri’s catalogue shows that such figurines have been found not only in Petra but all over Nabataea. Patrich treats them very briefly and concludes that miniature geometric blocks “may have been used as cult objects in conservative circles” whereas “the goddess figurines were used by those who either did not strictly adhere to the desert tradition or, not having been the descendants of the earlier settled agricultural inhabitants of the land, never inherited it at all.”74 Mettinger downplays the importance of figurines as religious artefacts, arguing that they are only the “quasi-subject” of a cult, and so should not be included within his definition of aniconism75. However, they are found in considerable numbers all over Nabataea, and El-Khouri has demonstrated how they formed a distinctive local style and so were the product of local religious ideas. We can also comment that, as far as we know, anthropomorphic figurines were a lot more common than geometric, but we should caution that the anthropomorphic are far more likely to be identified and recovered in excavations. Mettinger is correct to draw a distinction between cult images and figurines, but he is not correct to downplay the significance of the latter. While they were not the direct object of veneration in temples, they were undoubtedly representations of deities and had an important function in the household, where most examples from Petra were found. It is difficult to see how worshippers with religious objections to anthropomorphic representation could include them in their rituals.

• 76 Stewart ends with a revealing passage from Pausanias (2.9.6) describing statues of Zeus and Artemis (...)
• 77 Gaifman 2008, p. 53-62.
• 78 Ibid., p. 54.
• 79 Ibid., p. 58.

29This should be placed in the context of recent discussions of the aniconic image in the Near East and the Roman World, which have shown that the categorisation of all kinds of geometric idols as ‘betyls’ has provided a false sense of opposition between these and anthropomorphic idols. Stewart examines the reception and presentation of non-anthropomorphic images both inside and outside the Near East, and concludes that the modern emphasis on the difference between these modes of representation may not be an entirely accurate reflection of ancient attitudes : they were seen as different, and indeed non-anthropomorphic images were chosen exactly because of their ‘otherness’, but they were not so fundamentally different that they could not be represented side by side76. Stewart unfortunately does not examine the position of the aniconic image in Nabataea, but Gaifman does pay more attention to this77. Her central point is that the use of the term ‘betyl’ oversimplifies what is a very complex category of monuments, and assumes “that the importance of the form is simply the fact that that object is non-figural”78. She goes on to show how the use of geometric forms to represent facial features on eye idols should not be seen as a concession to Hellenistic influences, but rather a product of “local taste and preference”79. The linear scale, then, with Hellenistic anthropomorphism and ‘Nabataean’ aniconicism at opposite ends, has clearly skewed our perception of this kind of monument and oversimplifies the different artistic influences visible in Petra.

• 80 Hammond had first considered the relief to be feminine, and suggested that Atargatis was represente (...)
• 81 Patrich 1990, p. 109.
• 82 The evidence for any chronological development in Petra’s rock-cut religious monuments is unfortuna (...)

30We should finish this discussion with an examination of another of Petra’s unique monuments, the so-called ‘Medallion and Block Relief’ on the way down from the El-Madbah high-place (fig. 15). This shows the bust of a deity within a medallion surmounting a rectangular geometric idol. No inscription accompanies the sculpture but there have been attempts to identify the deity, inevitably resulting in the conclusion that we have here Dushara in some form or other80. The iconography cannot tell us his identity with any certainty, but it does show better than any other monument how aniconic and anthropomorphic representation could be combined in Petra. Patrich’s conclusion is that we have here a somewhat confused artisan “caught between two worlds, between two polar iconographic conceptions of how to represent the deity”, who ended up making a compromise81. We can see now that this was probably not the case. Patrich and Mettinger see a need to provide a social explanation for conflicting anthropomorphic and aniconic representations, but the conflict is mostly a creation of modern observers. The geometric blocks of Petra are too often approached against the background of ‘normal’ Graeco-Roman anthropomorphism, with an interest in emphasising and explaining their ‘otherness’, and not taken on their own merits. Hellenistic anthropomorphism is seen as the superior cultural force, inevitably encroaching on and replacing Nabataean traditions even though the meagre evidence does not support this82. On the contrary, the evidence suggests that Petra’s worshippers saw no conflict between anthropomorphic and aniconic. We can perhaps imagine a situation in reverse to that which Stewart outlines for the Mediterranean more generally. The various forms of geometric representation were clearly the norm in Petra, and anthropomorphic representation would have stood out as unusual against this background, but there is no evidence of conflict or competition between the two modes of representation. Rather, it seems that worshippers could and did draw on geometric and anthropomorphic forms to express their interpretations of Petra’s deities.

• 83 “Θεὺς ῎Αρης· τουτέστι θεὸς ῎ Αρης, ἐν Πὲτρᾳ τῆς ᾽Αραβίας. σέβεται δὲ θεὸς ῎ Αρης παρ᾽ αὐτοῖς. τόνδε (...)
• 84 See, most conveniently, Nehmé 1997, p. 1044-1045. As well as Dushara, al-‘Uzza, Atargatis, the ‘God (...)

31There is one other longstanding debate that our examination of the basileion of Isis is able to shed light on. The identity of Petra’s geometric idol blocks has long puzzled scholars. They were mostly first identified as Dushara, and so sometimes designated ‘Dushara-blocks’, after an often quoted passage from the tenth century Suda that describes the chief god of Petra, Theos Ares, as a square block83. This designation has now been abandoned after the discovery of several inscriptions identifying the blocks as different deities in Petra84. It seems clear that Petra’s worshippers could choose to represent a number of deities in this manner. Discussions therefore revolve around the shape, position and setting of the blocks in an attempt to uncover their identity. Parallels are brought in from other Nabataean sites, and the hope is that, as long as all the evidence is collected and analysed correctly, it will become apparent that certain iconographical features were peculiar to particular deities, and we will therefore be able to understand who was represented where in Petra.

• 85 See Wenning 2001, p. 85, for his typology. He concludes that a “typological approach is of limited (...)
• 86 Wenning 2008.
• 87 One of the clearest examples of this is the identification of the eye idols. For a long time it was (...)
• 88 Avner 1999-2000.
• 89 Ibid., p. 106-107.
• 90 Raymond 2008.
• 91 Ibid., p. 145. As regards orientation, Raymond concludes that although this was largely defined by (...)

32There have been several recent attempts to ‘solve’ the problem of Petra’s geometric idols. Wenning is in the process of creating a new catalogue of the monuments, and has published several times on how the idols should be approached. He has constructed a typology of the monuments and the methods of carving, following the examples of Dalman and Patrich, but cannot produce a scheme for recognising their identities, and concludes by hoping that further research will lead to a solution85. On another occasion, Wenning summarises the different ways the idols can be approached, concentrating either on the inscriptions (“The Epigraphical Approach”), the shape (“The Typological Approach”) and the context (“The Functional Approach”)86. Presumably any solution would have to explain all these frames of reference. The epigraphy is of little help, as Wenning himself highlights, as those few examples that are identified do not allow us to attach any particular forms to particular deities87. Avner has recently taken the typological approach in comparing the idols to standing stones in the Negev and Sinai88. These stones have no inscriptions to accompany them, but Avner nevertheless attempts to decipher their gender and identity. This produces a number of interesting parallels, but the system Avner constructs is incomplete for Petra’s idols and does not take into account the variety of forms represented there. When he comes to compare his system against the named idols of Petra, it does not find much support, and he acknowledges that there are differences between the arrangements of blocks found in the desert and those carved in relief at Petra89. Other typologies also suffer from the same issues : they inevitably reduce the complexity and variety of the monuments and cannot be satisfactorily used to determine the identity of the idols. The third approach, which takes into account the surroundings of the idol, has been partially addressed by Raymond who has made a study of the form of the idols, the niches containing them, and their orientations90. The niches are just as varied as the idols themselves, ranging from entirely blank recesses to those framed with elaborate architectural features. She concludes : “This suggests that in Nabataean society, individual deities did not necessarily require a particular niche shape or type”, and also argues that niche orientation is determined largely by the topography and not any wider religious considerations91. There is no need to go into any more detail here ; the point is that attempts to explain Petra’s geometric idols in this way do not work : they cannot be satisfactorily categorized to produce a solution as to their identity.

• 92 In her analysis of the cultic space of Petra, Nehmé includes ‘private and individual’ space as one (...)
• 93 Patrich makes this point succinctly : “At the same time there are deviations from the fixed forms, (...)
• 94 Healey provides a useful summary of these personal or family deities (Healey 2001, p. 151-152).

34We should begin to take the evidence at face value, and examine rather what this variety and lack of consistency can tell us about Petra’s religious practices. It suggests numerous different interpretations of the deities, and this seems to be the result of the personal or familial nature of these monuments. Beyond those figurine-sized geometric idols, which were certainly personal objects, the isolated nature of many geometric idols, and the small numbers of worshippers that they could accommodate, suggests that many of these monuments were the responsibility of small groups, families or even individuals92. They were therefore the product of their interpretation of the deity they chose to represent93. The personal significance of the gods is well-attested in Petra and Nabataea, one of the best examples being the ‘Goddess of Ḥayyan’ mentioned above94. The modern reaction has been again to reduce the individuality of the deity in an attempt to link her to one of Petra’s better known goddesses : Isis, Al-‘Uzza, Allat or Atargatis. She is, however, first and foremost the goddess of Ḥayyan, and there is no need to see another deity in the background here. Again the assumption is that she must fit into a wider coherent system of Petraean deities with static identities, and as long as we recognise the iconographical clues properly we will be able to put her in her correct place, and get back to her ‘true’ identity. There is no evidence, however, that she was understood in this way. We should rather take the name at face value : the worshipper has chosen to stress the personal significance of the deity, and similarly the iconography would have been the result of his personal interpretation.

• 95 The mōtab can be understood as the seat or throne of the deity. Two inscriptions from Petra mention (...)

35It is, then, precisely the variety of Petra’s geometric idols that ought to be stressed. There are certainly some normative tendencies within this. Most, for example, are plain rectangular blocks, and we can see certain motifs, like the base or mwtb beneath the idol, appearing with some consistency95. Undoubtedly, the idols were to some extent mutually understandable among Petra’s worshippers, but this may have been limited to certain areas or social groups that we have little understanding of today. It seems clear, however, that they do not form part of a universally recognisable system of iconography where different symbols were understood consistently. It is likely that the same shapes, attributes and motifs had different meanings to different worshippers and could be employed to represent different deities.