BACKGROUND

Towards a Prehistory of Islam

There exist today in Islamic studies two paradigms for the rise of Islam. The traditional paradigm sees Islam as arising in the Hijaz of the early seventh century A.D. in essentially its final form, to take its place in a world populated by discrete monotheistic religions: Judaism, and several varieties of Christianity. The problem with this view is that it denies Islam a prehistory: a period of development prior to the formulation of orthodoxy and scriptural canonization, such as is attested for other religions. Moreover, as has often been noted, the sources upon which this paradigm are based are all very late, so that

what they do not, and cannot, provide is an account of the "Islamic" community during the 150 years or so between the first Arab conquests and the appearance, with the sira-maghazi narratives, of the earliest Islamic literature. (Wansbrough, SM 119).

This is no problem for the traditional view, which accepts the accounts in the late sources regarding what happened during the period from which sources are absent, maintaining that they are absent only because they have not survived.

The new paradigm, based upon the work of Wansbrough and previous worksl sees the Muslim sources as representing a late stage in the development of Islam, reflecting the stage (and process) during which orthodoxy was imposed and scriptural canonization achieved. According to this view, we have no orthodox Muslim sources dating from the first 150 years of the Hijra era simply because none ever existed during that period. Rather, we should expect to find, in those first 150 years, at least remnants of evidence for the development of the Arab religion from some point of origin towards Islam. Wansbrough has demonstrated that the point of departure is to be sought in inter-sectarian polemic, in which Judaic and Judaeo-Christian notions are much more prominent than Christian ones. And Pines has attested the importance of Judaeo-Christianity as a factor both in al-Sam and Mesopotamia, and in the Christology of the Qur'an. From this it would appear that Islam arose in the areas conquered by the Arabs, not in the Hijrz, that the starting point of its development was some form of Judaeo-Christianity, and that this process began considerably later than the date of the Arab conquest.

This viewpoint does indeed give us our missing Islamic prehistory, but it has until recently lacked enough evidence on which to amplify and base its conclusions. What it needs, in short, is the missing link between the religion of the Muslim texts from the late second century A.H. on, and the (apparently Judaeo-Christian) religious environment of the first century A.H., attested by non-Muslim sources.2

Just such a series of missing links is provided by the Arabic rock inscriptions scattered all over the Syro-Jordanian deserts and the Peninsula, and specifically the Negev, where most of the work of collecting and analyzing them has been carried out to date. Moreover, in view of these inscriptions, the official texts already long familiar-such as those in the papyri, the Dome of the Rock and the Arab-Sassanian coins-should be seen in a new light. It is my intention in the present paper to point out the existence of a considerable corpus of Arab religious texts dating from the first century and a half of Arab rule, exhibiting an Arab monotheistic creed which is demonstrably not Islam, but from which Islam could have developed. It is my opinion that Islam did in fact develop from some variant of this creed, mainly during the first century A.H., but this paper will not be concerned with the process of that development. Here I wish only to indicate that the very existence of this corpus of texts poses considerable problems for the traditional Muslim account of the history of Islam, whereas the alternative paradigm finds in them some, at least, of the evidence it needs to confirm its hypotheses.

I start with a fact to which too little attention has been paid. A distinctive feature of all the Arab religious institutions during the Sufyani and on into the Marwanid one is the complete absence of any reference to Muhammad. Neither the Prophet himself nor any Muhammadan formulae appear in any inscription dated before the year 71/691. This is true whether the religious content of the inscriptions is its main purpose, as in supplications ('ad`iya, sg. du a'), or whether it is part of a commemorative inscription with a religious emphasis (Mucawiya's dam inscription near Ta'if). When an inscription is religious, the absence of any Muhammadan locution is an argumentum e silencio of considerable weight. It is a fact that the name of the Prophet Muhammad and the use of set phrases or formulae which can be shown to be connected with his name occur in the Arab inscriptions only after ca. 70/690. We can, in fact, be more precise: The first occurrence of "Muhammad" and the Triple Confession of Faith, viz. 1. tawhid (with or without ladam al-`irk but always with the word wahdahu ["alone"], and not, for example, the words illy huwa ["but He"]); 2. Muhammad rasul Allah; 3. the human nature of Jesus (Isa) as merely rasul Allah wa-`abduhu is in 'Abd al-Malik's inscription in the Dome of the Rock in Jerusalem, dated in the inscription itself A.H. 72/691-92. And the first dated occurrence of the phrase Muhammad rasul Allah is from just one year before that, from an Arab-Sassanian coin of Xalid b. 'Abdallah from the year 71/691 which was struck in Damascust3 Xalid could perhaps have brought it with him when he returned to Basrah from Damascus after the Marwanids succeeded in defeating the Zubayrites in Iraq. And after 72/691-92 and all through the Marwanid dynasty, Muhammad's name occurs as a rule whenever religious formulae are employed: on coins, milestones and papyrus "protocols."

If no religious text dating from before 72/691-92 had been discovered, we might have concluded that it took time for any religious notions to penetrate the Arabic inscriptions. But the first Arabic papyrus, an entaqion (receipt for taxes paid) from Egypt dated 22/642, is headed by the basmala in both Greek and Arabic (in that order). While the basmala is a monotheistic preamble-formula, it betrays no specific creed: it is unquestionably not Christian, but neither is it Muslim per se. The religious content is more pronounced in the rock inscriptions, which start in about the 40s of the first century A.H., and on the Arab-Sassanian coins prior to 71/691-92. These bear religious texts, but with no mention of the Prophet or of Muhammadan formulae. This means that the official Arab religious confession did not include Muhammad or Muhammadan formulae in its repertoire of set phrases at this time. I will refer to this religious stage as "pre-Muhammadan."

The pre-Muhammadan texts exhibit what I call "indeterminate monotheism": a monotheistic form of belief, but one which contains no features specific to a known monotheistic religion. The Muhammadan texts differ greatly from them. The introduction of Muhammadan variables indicates the imposition of new confessional notions on a preexisting stratum of belief. Muhammadanism is a previously unattested Marwdnid introduction. And the result became-overnight!-the state's only form of official religious declaration, to be used in certain kinds of formal documents and inscriptions, as, for example, the papyrus "protocols." In short, the state decided, as a political act, to adopt Muhammadanism as its official creed.

CLASSIFICATION OF THE TEXTS

The inscriptions found in the Negev may be divided into several classes on the basis of religious content:4

1) The common texts. These occur all over the Middle East from the mid-first century A.H. (40-60 A.H.) through the later Islamic periods. Their most prominent feature is the lack of any indication of a specific creed-Islam, Christianity or Judaism-and their lack of any hint of Muhammadanism. The tawhid is not mentioned, and they do not engage in the polemics which is the heart of the later, Muslim texts. The term janna appears, but the concept it expresses is unclear. There is no mention of a prophet, such as Moses, Jesus, Aaron or Abraham. The only deity mentioned is Allah, who is also referred to as Allahumma ('allhm) or rabb/rabbi, though no other trait is assigned to Him. I call this class "common texts" both because of their wide distribution, and because their vocabulary continued to be used into the Muhammadan period.

2) The basic class. These are inscriptions from the mid-first century A.H., currently attested only from the Sde Boqer site. They are distinguishable from the common texts by their style, and in fact this class was discerned by text analysis. All the texts of this group exhibit a characteristic language and content, and many of them were inscribed by, or on behalf of, a small number of individuals whose names recur. They may contain references to Moses and Jesus, but nothing at all that is definable as Muslim. Inclusion of inscriptions in this class was made on the basis of the unmistakable occurrence of certain formulae in them, or because the author of the inscription was also the named author of other inscriptions which were unmistakably of this class. It is quite possible that this class is in fact only a subset of the common texts. There is one recognizable subclass of basic class inscriptions, distinguished by their reference to Allah as "Lord of Moses and Jesus" (rabb Musa wa-,Isa).

3) The second class, which immediately followed the basic class (through the end of the first century A.H.).

4) The Muhammadan inscriptions, which are Marwanid but cannot be more precisely dated.

5) The Marwanid Islamic inscriptions, which seem to be somewhat later than the Muhammadan texts, and continue through the second century A.H. They are definitely Islamic in language and content, without necessarily reflecting orthodox Islam.

6) Muslim (Abbasid) inscriptions from the second century A.H. on. This class, though Muslim, is still not in perfect harmony with Muslim orthodoxy.

THE PRE-MUHAMMADAN TEXTS

The basic class texts are recognizable by the use of rabb Musa wa-`Isa in their opening phrase, and/or by a certain set of phrases and allusions in the text itself, such as hayyan wa-mayyitan, and gayr halik wa-la mafqud. (Examples of these and others may be seen in the attached translations and notes to them.) I would be inclined to read Old Testament (and not New Testament) connotations in at least some of them. Though some phrases and formulae are found in the Qur'an, these texts do not seem to me Qur'anic; rather, they appear to belong to a certain body of sectarian literature which developed Judaeo-Christian conceptions in a particular literary style. The resulting creed, basically indeterminate monotheism with the addition of Judaeo-Christian variables, is scarcely identifiable as that of a defined sect, but rather is the expression of belief of one group of indeterminate monotheists among many.

PROPHETICAL LOGIA

Wansbrough introduced the term "prophetical logia" to designate subcanonical versions of scripture,5 i.e., the components from which the canonical text was composed. These may have been produced by any Arab group affiliated to one or other of the Christian or Judaeo-Christian sects in the area. Wansbrough's "emphatically provisional conclusion"6 was that

the canonical text of scripture exhibited separate [prophetical] logia collections which had for some time prior to their final redaction been in liturgical and homiletic use in one or several related communities (SM 57 para. 3).

However, an attempt to investigate this thesis was, he found, hindered by a lack of texts for comparison:

The practical quest for prototypes of Qur'anic diction is hindered both by the transmission history of the document and the absence of trustworthy comparative material (ibid. 69 para. 2, emphasis added).

Nevertheless, with the aid of Wansbrough's terminology, it became easy to see that the hundreds of rock inscriptions scattered over the Near Eastern deserts are in fact sub-canonical texts, providing in some measure those prototypes of Qur'anic diction which Wansbrough found so sorely lacking. The inscriptions of the first century and the early half of the second century A.H. include logia and formulae which are prototypes of Qur'anic diction, exhibiting what Abbott referred to as "Koranic flavour." 7 They are, of course, pre-Islamic, within the framework of the Schacht-Wansbrough hypothesis concerning the Qur'an's creation and canonization. Thus now, some ten years after Wansbrough's work, we may regard as overly pessimistic his conclusion that

we are no closer than we have ever been to the actual forms of preIslamic Arabic and ... a change in these conditions seems unlikely (Wansbrough, loc. cit.).

For not only are the religious texts of the rock inscriptions pre-Islamic, but the language of the chancery papyri can also furnish examples of "actual forms of pre-Islamic Arabic."8

Sde Boqer is to date the only site to yield texts with Judaeo-Christian variables in Arabic. But Judaeo-Christian Christology is inscribed in Arabic on the octagon of the Dome of the Rock. All that is written there, with the exception of several repetitious Muhammadan formulae, is a declaration of the official state opinion (i.e., decision!) concerning Pauline Christology. Muhammad is introduced merely as a messenger of Allah, with no further clarification. This is also the treatment Muhammad receives in the Qur'an. The Judaeo-Christian content of the Dome of the Rock inscription is no novelty to theologians of the seventh century A.D., and Pines's work9 has shown that the Christology of the Qur'an is also Judaeo-Christian. The Christological text of the Dome of the Rock is composed of verses which occur in the Qur'an verbatim (the Muhammad text is not Qur'anic). I see the Christological text as composed of quotations of prophetical logia, i.e., of "sub-canonical versions of scripture"; but the very fact that such logia were inscribed on the royal shrine of the newly emerged state creed may be considered a form of canonization of this text. One would not expect this text to be altered by those who refer to it as being the declaration of the official state faith. This canonical status is demonstrated in the papyrus "protocols" and on the coins, where formulae/logia are reproduced in great numbers invariatur (though not always in full). This is not, of course, evidence for the existence of "Umayyad scripture," but for a process of ongoing determination of the formulae to be included in an official state confession, expressed in authorized texts whose formulae must not be altered after promulgation. This process is what I call "canonization of state logia," which were, to use Wansbrough's phrase, "separated from an extensive corpus of prophetical logia." 1 o

The Muhammadan texts were not accepted by the public as promptly as they were officially employed. For years after their appearance in state declarations, and not withstanding the exclusive use of Muhammadan formulae by the state in official contexts, people continued to include non-Muhammadan legends in personal inscriptions. In fact, Muhammadanism was not in common use even for routine chancery writing. Thus Qurra b. Sarik fails to include set Muhammadan phrases in his Arabic and Greek correspondence, though he obeys the rule of inscribing Muhammadan formulae on papyrus "protocols" bearing his name and title. Muhammadan formulae began to be used in the popular rock inscriptions of the central Negev only in Hisam's days (105-125/724-43). For instance:

Allahumma, salli `ala Muhammad al-nabi' wa-lala man yusalli 'alayhi (inscr. HS 3154(6), not reproduced here, dated 117/735).

And there are other (dated) Muhammadan inscriptions from these years, in the Negev and also in Jabal Usays, east of Damascus. i I

It thus seems that the period 71-105/690-723 is marked by the use of Muhammadan formulae for official purposes, but not by the general population. But by Hisam's days, about thirty years-a generation-after the introduction of Muhammadanism by 'Abd al-Malik, it had taken root in some circles, apart from its official use by the state, and was perhaps gathering momentum. However, while these texts are Muhammadan, they cannot yet be called Muslim.12 Texts which qualify as Muslim started to appear in the Negev only at the end of the second century A.H., approximately coinciding with the appearance of the first written traditional Muslim accounts.

Having given the general background of stages of development, I would like to take the two ends of the process-the pre-Muhammadan and the Muslim texts-and make a preliminary attempt to compare their language and religious concepts. My intention in so doing is to highlight the differences between them: to demonstrate that by any criteria of literary context analysis, the pre-Muhammadan texts cannot have been produced by the same religious conceptual system that gave rise to the Muslim ones. In other words, it is very difficult to define as Muslim the religion found in popular expressions of faith dating from the first century of Arab rule.

THE MUSLIM CLASS

The Muslim class of epigraphical texts can be distinguished from the Muhammadan class by the occurrence of various terms and concepts, and in general by their idiom. I am still hesitant concerning the precision of my typology of texts as Muhammadan and Muslim, but the two dated groups of Muhammadan texts from Hisam's days found in the central Negev so far (one dated 112/730, the other 117/735) do not appear to use the same diction as the Abbasid texts from the same area.13 I shall refer to the latest dated inscriptions we possess from the central Negev (second century A.H., facs. 1.25-26) as examples of Muslim texts, and compare their, with the Judaeo-Christian texts from Sde Boger.14 The latter are a category of basic class texts with Judaeo-Christian variables, which we tentatively date to the mid-first century A.H. or a little after.

I draw my examples of Muslim texts from three inscriptions. The first two, reproduced in facs. 1.25 and 1.26, belong to Said b. Yazid and are dated 164/780 and 170/786 respectively. They were found in two wadis in close proximity to each other, Wadi-Hafir/Nahal Lacanah and Wadi Idtar/Nahal Yeter in the central Negev. The third inscription is the lower part (lines 5-12) of facs. 1.12, which was made by one 'Abdallah, the son of the author of lines 1-4 (as he indicates in the last line). Facs. 1.12 is thus a composite text. The father's part is classifiable as a basic class inscription, while the son's is Muhammadan or Muslim. All three inscriptions are Muhammadan; two are also typically Muslim, while the third is less so. A careful reading of them will enable us to distinguish between such Muslim texts and the basic class, with or without Judaeo-Christian variables.

Chart 1 presents a content analysis of concepts in these three inscriptions, and the phraseology in which they are expressed (see p. 152). The verbs of entreaty are significant, and serve as identifying attributes:

Table 1

As Table 1 shows, in the basic class g.f.r is the only verb of entreaty found in an opening phrase;15 it is still frequent in the Muhammadan texts, but very uncommon in the Muslim ones. In the Muhammadan texts sly starts to occur, although infrequently, and it is very frequent in the Muslim texts (those in the table are from ca. A.H. 300/912). The verb radiya (r.d.w.) (not included in the table) is not common in an opening phrase, and has not been found in basic class inscriptions, but it does occur in the Muhammadan/Muslim ones. The fact that g.f.r is so uncommon in the Muslim texts means that they are devoid of the "sinners' load" which is the major-in fact almost the only-notion governing the basic class forms of entreaty and supplication.

SOLICITING GOD'S FAVOUR

In considering whether a text is Muslim or basic class, both content and form of expression must be considered: both what the text says, and the words it uses to say it. As I noted at the Third International Colloquium on "From Jahiliyya to Islam" in 1985, the Muhammadan (there called "Islamic") supplication texts exhibit a new attitude,

that of approaching God to settle personal affairs. This is a new type of supplication, aimed at securing God's benevolence in order to procure a very personal list of desiderata, temporal as well as spiritual. The janna is only one of the desired items ... and recorded in the Negev inscriptions there are still others which together form a rather impressive list of requests.

Gone are the stereotyped expressions of pleading for forgiveness for undisclosed and general transgressions; instead there is optimism, an expectation to receive God's grace: 'urzughu min fadlika ("provide for him from your bounty"); 'atimma 'alayhi ni'mataka/ni`amaka ("Bestow Your favour/s wholly upon him"); ij`alhu min al-muflihin ("make of him one of the prosperous") and more.

The verb sa'ala, "requested, asked for" is common as a soliciting verb in the Muhammadan and Muslim texts, e.g., 'anni 'as'aluka ni'mataka: "I request of you your favors upon Bigr b. Tamim" (inscr. HR 522, ca. 300/912). In addition, in a Muhammadan (or Muslim) text from Sde Boqer we find: 'as'al Allah al-janna: "I request of Allah al-janna" (here facs. 1.12:8). In the basic class, where janna is not yet found, a supplicant noted wa-huwa yas'al Allah al-majanna: "and also he entreats Allah [to grant him] this sanctuary (majanna)" (facs. 1.9:5).

Like sa'ala, the verb daxala ("entered") is also common in Muhammadan and Muslim texts which solicit God's favour in accepting the supplicant unto Allah's rahma or, again, letting him enter al-janna. This verb is almost absent from the basic class; there is only one instance of it, in connection with an interesting occurrence of a Qur'anic topos (the jannat al-na am): 'adxil [Xalid b. Humran] jannat al-naiim: "let [PERSONAL NAME] 16 enter jannat al-na am" (facs. 1.19:3-35). The terms janna and jannat al-naiim in these texts are, I suspect, not yet well understood; the "obvious" translation, "paradise," is very doubtful.

GIVING TESTIMONY

Verbs from the root §.h.d. are not known from basic class inscriptionsor common texts in general-or from Muhammadan texts. 17 I therefore consider their occurrence to be a Muslim trait. The very notion of testimony-giving witness as the mode of announcing one's faith-is Muslim, and is a significant aspect of the polemical features of the Muslim texts. It seems not to occur earlier than the Abbasid period, or late in the Marwanid period. Thus in the Muslim texts, the claim "I wit ness" (aThadu or a similar expression) is very common. This is seen in inscriptions 1.25:2-4 and 1.26 passim, which routinely incorporate 9.h.d. into the text. The basic class and Muhammadan texts, on the other hand, seem to lack it entirely. With this may be compared the fact that words derived from the root ..h.d., with the meaning of "to give witness," are not rare in the Qur'an. In fact, the idea of giving witness is Qur'anic:

'a-'innakum la-taThaduna 'anna ma'a Alldhi 'alihatan 'uxrd, qul: Id' aThadu, qul: 'innama huwa 'ilahun wdhidun (Q 6:19)

Can ye possibly bear witness that there are other gods ('dlihat) associated with Allah? Say: I shall not/cannot bear witness. Say: but in truth (innama) He is God, one and only (translation following Yusuf All and Ben-Shemesh).

ni'ma al-mawla wa ni'ma al-nasir

This expression is found in the Muslim text 1.25. This exact phrasing is not in the Qur'an, but the concept of Allah as overlord (patron) and helper is well attested there, e.g., 'i'tasimu bi-llahi huwa mawlakum fa-ni`ma almawla (Q 22:78): "Strengthen yourselves with Allah, He is your Overlord (Patron), a Gracious Patron and a Gracious Helper" (translation following Ben-Shemesh). There is also wa-ma lakum min duni-iidhi min waliyyin wa-ld nasirin (Q 2:107): "Beside Him ye have neither Overlord (Patron) nor Helper." In Syria, in a text from Hisam's time, we have also Allah waliyy [PN],18 and Allahumma kun 'anta waliyy Yazid b. `Abd al-Wahid al-'Asadi.19 This concept is not found in the basic class.

kafa bi-llahi §ahidan ["And Allah is sufficient as a witness"]

This expression occurs in 1.26:8-9. The expression kafa bi-llahi, bi-rab- bika is commonly attested in the Qur'an, e.g., Q 4:6, 4:45, 4:69-70,4:132, 4:78-79; 48:28. But the phrase kafa bi-lldhi does not occur in the preMuhammadan texts: it is Qur'anic and Muslim only. In its occurrences in the Qur'anic and Muslim texts, I would distinguish between the different kinds of patron referred to: the one who counts his deeds or money (Q 4:6); the helper (Q 4:45); the one who knows (Q 4:69-70); the overlord, who is in charge of someone (Q 4:132); and finally, the phrase kafa billahi . ahidan: "and Allah suffices as a witness":

It is He who has sent His Messenger with Guidance and the true din, to proclaim/to exalt20 it over the entire (other) din-and Allah suffices as a witness (Q 48:28).

The notion of giving testimony to divine truth is another monotheistic concept well-attested in Rabbinic Judaic and Christian sources, where it is connected with the concept of self-sacrifice for the sake of such divine truth.21 In the Qur'an there is no question of bringing witness from scripture, canonical or otherwise. Allah himself gives witness, through his mes- senger(s), with no recourse to a book. Those who bear witness from scripture are not the true believers, but the 'ahl al-kitab. This, I think, reflects the inter-sectarian polemic, current before the Qur'an was put together from the existing sectarian logia, in which the adherents of a prophet, who lack a scripture of their own, maintain that under the new dispensation of their prophet such a scripture is in fact unnecessary. It is, in these circumstances, ironic that these phrases found their way into the canon.

la hawla wa-la quwwata ...

The famous expression la hawla wa-la quwwata illa bi-llahi al-`ali al- `azim, "there is no strength or power but through Allah, the High, the Great," is not Qur'anic, but a saying in hadit,22 and various meanings have been proposed for the word hawl in this context.23 A very similar phrasing occurs in facs. 1.12. This inscription is undated, but its Muhammadan idiom and its genealogical context (the continuation of a basic class text written by the owner's father) lead me to date it to Hisam's days (r. 105-25/724-43). Again, this expression does not occur in the basic class.

jami'u-l-xala'iq

The expressions jami`u-l-xala'iq, "all the created beings" (1.26:6-7); jami' xalqihi, "all His created beings" (1.25:2); and al-nas 'ajma`in, "all humankind" (1.13:4-5) have the same referent as `alamin in rabb al-'alamin, "Lord of Creation" (1.12:3-4; 1.13:4-5), an expression attested in basic class texts as well a I as later ones. We should consider this concept, Lord of Creation, one of the constants of monotheism. In the basic class it is expressed only via the phrase rabb al-'alamin, while in the Muslim texts the religious vocabulary has widened to included the phrases given above.

al-nar

This is a Qur'anic concept, the opposite of janna:

bald man kasaba sayyi'atan ... fa-'ula'ika ashabu al-nari hum fiha xaliduna; wa-alladina 'dmanu wa-`amilu al-sdlihati 'uld'ika ashabu aljanna hum fhd xaliduna.

"They who accumulate evil (deeds on their account) ... such are inmates of the fire, where they will remain forever; but the faithful who perform good deeds, they are the inmates of al janna, to remain there forever" (Q 2:81-82, translation following Ben-Shemesh).

It would also seem to be one of the constants of the monotheism of late antiquity, yet it is not attested in the common texts, including, of course, the basic class. There are at least two possible reasons for this. One is that not all indeterminate monotheists of the late sixth and seventh centuries A.D. shared this concept, although it of course existed in the other monotheistic religions of the area. The other is that our basic class authors did not mention either janna or nar when entreating Allah for forgiveness of transgression, because of their specific beliefs or religious attitudes. For instance, one of the most striking features of these texts is their heavy atmosphere of dread because of undisclosed transgressions; it is clear that the writers regarded themselves as "sinners" in a general sense. It could therefore be that they did not mention al-janna because they believed that a sinner should not request such a great favour for himself.24 And perhaps nar was not mentioned in order to avoid naming the unspeakable horror.25 Janna is, however, requested in the second class.

Although al-nar is, as stated, a Quranic concept, its context in these texts is somewhat different from that in the Qur'an. The writer of our inscriptions mentions al-nar in the context of seeking protection from it: 'a'udu bi-hi/bi-llahi min al-nar, "I seek protection through Allah from the fire" (facs. 1.12:9). In Hisam's days we find, 'as'alu al-janna wa-'a'udu bi-hi min al-nar in Syria,26 but in the Qur'an, '.w.d. (a`udu), "I seek protection," refers mostly to protection from Saytan, "Satan," and the fire is not mentioned in this connection.

Allah

The texts consider Allah in various ways, which I shall first list and then discuss.

1. Hypostatic definitions27 lie at the root of the Trinitarian Christological controversy.

2. Subordination is the doctrine that everything is Allah's creation and subordinated to Him, including angels and prophets, and among the latter, Jesus. This, of course, contradicts any Pauline Trinitarian confession.

3. Allah's qualities are distinguished from His predicates, and both are detailed in the text.

1. The nature of Allah

In the earliest Muhammadan confession of faith, namely the Dome of the Rock, S:A4-5,28 we find the maxim that God was not born and does not give birth. However, this is not attested in any other early (i.e., Muhammadan but non-Muslim) inscriptions, either from the Negev or elsewhere. It occurs in the Qur'an (Q 112), and in our Muslim texts (e.g., facs. 1.26:9-11). It therefore seems most likely that it signals the Muslim layer of confessional phraseology. Our understanding of how it came to occur in the Dome of the Rock but in no other Muhammadan text must await further research.

God's singularity, expressed by the tawhid in full, concluded by the `adam al-girk, appears in the Dome of the Rock and is attested in the later (ca. 300/913) texts. It is not rare, either, in earlier Muslim texts from the central Negev. And in the Qasr al-Hayr, lines 1-3, we have basmala, the tawhid in full, and Muhammad rasul Allah (DKI 211, by the people of Hims, dating from 110/728-29). This is an official construction text:

'amara bi-sina`at hada al-madina [= fortress(?)] 'Abd Allah Hi.am 'amir al-mu'minin:

"This fortress (?) was constructed by order of `Abd Allah (title!) Hi.am, 'amir al-mu'minin."

But the tawhid does not appear in dated Muhammadan popular inscriptions, nor do we find it in the group of Marwanid texts published by Muhammad al-cUss in JUL I have thus concluded that the official Marwanid convention was to inscribe the tawhid (with or without the concluding 'adam al-.dirk) in official texts, but that the writers of the popular texts did not employ this formula.

The idea of God's singularity appears in the Muhammadan -Muslim tawhid in a form slightly different from that found in the Qur'an. In the tawhid, the reference to God's singularity ends with the word wahdahu, "him alone," e.g., in the Dome of the Rock, S:A1-2 and passim. In the Qur'an, singularity is expressed by the words wahid and illy huwa. This would appear to be a minute difference, but I consider it to be significant, for the following reason. Most of the text of the Dome of the Rock is familiar verbatim from the Qur'an, but the locutions that are not Qur'anic are precisely the conspicuously Muhammadan ones, which make their first appearance in the Dome of the Rock: tawhid; `adam al-.dirk; wa alsalam 'alayhi wa-rahmat Allah; wa-tagabbal Jafa`atahu yawm al-giyama ft ummatihi; and finally, sallaAllah `alayhi. We also find Allahumma salli `ala rasulika wa-'abdika 'Isa b. Maryama: "Allahumma incline toward Your messenger, Jesus the son of Maryam" (Dome of the Rock, NW-W:B 11).29 This verse is not only non-Qur'anic, but also foreign to the context of the Qur'anic verses relating to Jesus. And the expression salla Allah 'ala does not occur in the Qur'an. In fact, the idiom Allahumma salli 'ala/salla Allah `ala, so common in the sal'am, is not Qur'anic but Muhammadan-i.e., Marwanid-in origin, appearing first in relation to Jesus. I would therefore ascribe importance even to small linguistic differences between Qur'anic and non-Qur'anic locutions.30

To sum up the question of the description of Allah's nature in our texts: the source for the details of the hypostatic definition found in our Muslim texts is the Qur'an; the same verbatim description is also found in the Dome of the Rock. But in the latter there occur a few formulae which cannot be found in the Qur'an. These same formulae also occur as official Marwanid texts, to be used on papyrus "protocols," coins, milestones and construction texts. They continued to be employed under the 'Abbasids and have been used ever since, becoming commonplace Muslim locutions. But these official formulae did not, it would seem, take root at a more popular level, for in our Muhammadan popular rock inscriptions they seldom if ever occur.

2. Subordination

This was a question of great ideological significance: Are all creatures subordinated to God, or are there exceptions? The idea that all of creation is subordinate is explicit in rabb al-`alamin, "Lord of Creation," and in the phrase rabb al-nas ajma`in, "Lord of the people, all of them" (facs. 1.13:4-5). The absolute superiority of Allah over all seems to be another monotheistic constant, though in the form in which it appears in the Arabic inscriptions it cannot be accepted by Trinitarian (Pauline) Christians. It is therefore interesting that one of our Muslim inscriptions (facs. 1.26:4-7) finds it necessary to declare that Muhammad, Jesus, Ezra (,Uzayr) and (or: as well as) all creation jamicu al-xala'iq-are "subordinated worshippers"-'ibad marbubin (for marbubun). I read this as an addendum to Q 9:30:

wa-galat al-Yahud: `Uzayr 'ibnu Allah, wa-galat al-Nasara: al-Masih 'ibnu Allah

The Jews say: 'Uzayr (= Esdras/Ezra) is the son of Allah, and the Christians say: al-Masih is the son of Allah.31

Our inscription (lines 4-7) pursues an internal Muslim controversy regarding the status of Muhammad. It denies that anyone, Muhammad included, can be so close to Allah as to merit the description "son of God," which implies a share in the Godhead. In the late first-early second centuries A.H., there were those who considered Muhammad to be qualitatively different from the messengers of the past, and awaited his second coming, another Judaic and Christian monotheistic constant. Nor was Muhammad the only one whose position was open to doubt. There are Muslim inscriptions from the Negev in which the supplicant entreats not only Allah but also His whole retinue to bestow favour upon him:

Allahumma salli anta wa-mala'ikatuka al-mugarrabin wa'anbiya'uka al-mursalin wa-`ibaduka al-salihin 'ala PN

Allah! You and Your angels who are nigh unto You, and Your prophets who were sent and Your righteous worshippers, be inclined (in favour) unto 'Imru al-Qays b. Tamim (BR 5117, from the Negev, 300/912; not reproduced here).

This phrasing does not state that Allah's entourage is subordinate to Him; whether such was understood to be the case has yet to be proved. By contrast, a common text inscription from Iraq (near Karbalah) dated 64/684 (the interregnum; Muxtar in Iraq) has:

Allahumma rabb Jibril wa-Mika'il wa-Israfil igfir li PN

Allahumma, Lord of Gabriel, Michael and 'Israfil,32 forgive PN (DKI 163:5-8; source GAP fig. 45).

This is a subordinationist declaration regarding the three archangels: Allah is their rabb, i.e., they are subordinate to Him. And from the same period, there are a few basic class (i.e., in my opinion, Judaeo-Christian) inscriptions from Sde Boqer which contain a similar subordinationist formula regarding Moses and Jesus: Allah is rabb Musa wa-'Isa. This is not a variation on the theme of Q 9:20, but a reference to a different layer of Judaeo-Christian concepts. Coupling Jesus with Moses, the lawgiver who spoke to God face-to-face, gives to Jesus the highest rank attainable by the most pious of human beings, and has him sharing with Moses the honor of serving as an identifier of God. Nonetheless, it makes Allah's supreme position, and Jesus' subordination, quite clear.

3. God's Qualities and God's Predicates

Eulogizing God is a common feature of religious discourse, and an essential element of doxology. The wording of the descriptions of God in our texts can indicate, to some degree, the author's confession. This method of classifying texts is, however, at a very rudimentary stage; as more information is gathered and sorted in future research, it will, it is hoped, be possible to improve our discernment of different groups of texts, and to penetrate their laconic language to an understanding of their authors' practical conception of God.

In the basic class God's qualities are expressed succinctly, via predicates composed of single words, as in the Beautiful Names: 'al-'asma' al- husna. There may be minimal elaboration of the predicate phrase, e.g.:

'arham al-rahimin wa-'ahkam al-hakimin

the most loving among the loving ones and the best ruler among the rulers (facs. 1.3).

These predicates are also found strung together in basic class inscriptions, to form a chain which itself constitutes the main content of the inscription (e.g., facs. 1.9). Such a string of predicates implies that liturgical use was the main avenue of development; elaboration such as "the most loving among those who love" (facs. 1.3) or "the one who knows what is hidden" (facs. 1.20) may be seen as similarly rooted in liturgy. For such phrases must be placed in their proper literary context, the communal, liturgical environment of a particular sect, which cannot be understood by a study of the texts that confines itself to thematic classification or philological and literary comparisons.

In the Qur'an these and similar predicates also occur, but their usage differs. They occur as pausal phrases, e.g., fa-' 'lamu anna Allah 'aziz hakim (Q 2:209); fa-'inna Allah .adid al-`igab (Q 2:211). We may see here the employment, for literary ends (to demarcate the component logia and point out their moral, whether actually implicit in them or attributed to them), of the phrases already familiar from liturgical use. Thus again, the use of these phrases in pausal function in the Qur'an implies that they were part of the liturgical, not just the conceptual, environment of the sect that used them. It was the phrases thus familiar from liturgical use that were employed for literary purposes in the formation of Scripture.

CONCLUSION

The real meaning of the many inscriptions from all over the Near East lies in the insight they can give us into the spiritual and social phenomena which eventually matured into Islam. Insofar as we may judge from the evidence so far presented, it would appear that the common texts represent the writings of a (probably Judaeo-Christian) Arab sect, but not the sect whose logia were adopted to form the Qur'an. The subset I call the basic class were certainly left by a Judaeo-Christian group, whose logia similarly were not adopted into the Arab Scripture. Apart from linguistic evidence, such as that given here, regarding the absence of various words and phrases in their inscriptions versus their occurrence in the Qur'an and Muslim texts, there is also the fact that the logia of the basic class are not attributed to a prophet specific to that sect, or presented as the words of such a prophet, in sharp contradistinction to the logia which form the Qur'an. Nonetheless, the texts (prophetical logia, in Wansbrough's terminology) out of which the Qur'an was canonized reflect a Judaeo-Christian environment lacking in specifically Muslim concepts. The basic class texts reflect, similarly, a Judaeo-Christian environment and attest its existence, in the Negev at least, in the late first and early second centuries A.H., while the common texts reflect the most widespread form of Arab monotheistic belief during the first century, the stratum upon which, as it were, the more complicated religious edifices were built. From the fact that the Qur'an exhibits a "prophetical" Judaeo-Christianity and the basic class does not, I conclude that the general Judaeo-Christian sectarian environment was widespread, including at least one group defined by adherence to a prophet, whose corpus of logia form the basis of the Qur'an. From the fact that the Qur'an contains many phrases present in the Muslim inscriptions of the late second century A.H. and later, but absent from the inscriptions of Hisam's days or earlier, I would conclude that it was canonized quite late, i.e., after these phrases had entered the religious vocabulary. Our texts demonstrate that this scarcely happened before the Abbasid era.

ACKNOWLEDGMENTS

The survey I have been conducting is part of the ongoing Negev Archaeological Project: Early Arab Period (Research of Rural Settlement), headed by Professor Myriam Rosen-Ayalon of the Institute of Archaeology of the Hebrew University of Jerusalem. This work, which includes excavations and an epigraphical survey as well as a survey of sites of cultic activity and inhabitation, has been possible due to the continuous assistance unselfishly given since 1983 by Kibbutz Sde Boqer, many of its members and its three secretaries during that period: Mr. Ze'ev Zivan, Mr. Razi Yahel and Mrs. Brenda Habshush. Professor Emmanuel Marx, head of the Centre for Social Studies at the Blaustein Institute for Desert Research, appointed me an honourary fellow of the Institute, thereby enabling me to continue my research uninterruptedly in close proximity to the site of excavation and the area surveyed. The funds for the ongoing research were supplied by the Ministry of Science, the Hebrew University and the Israel Academy of Sciences and Humanities. To all of these people and institutions, my warmest thanks.

The survey could not have been carried out, nor its results made known, without the devoted efforts of a selected few who gave unreservedly of their time and energy: Nurit Tsafrir, who helped me to trace the inscriptions over an extended period of time and no less extended area of the Negev; Zmirah Cohen, who helped us decipher the results, which in many cases were apparently unreadable scribbles and might have remained so but for her aid; Amnon Rothenberg, who together with me excavated the site of Sde Boqer, and recorded and computed the findings; Mali Borenstein, whose research assistance was invaluable; Dalia Heftmann, who persuaded the computer to analyze the results and handled all the technical drawing and graphical work of producing the survey; and finally Judith Koren, who translated ideas into English. Our work is not yet finished, and I know how inadequate are the phrases of an acknowledgment to describe their ongoing contributions.

Chart 1: Content analysis of concepts in three Muslim inscriptions

Sources: Inscriptions from central Negev, nos. MA 4205(14), here face. 1.12:5- 12 (the father), no date: EKI 261, here face. 1.25, A.H. 164/780/81; ST 640(34), here face. 1.26, A.H. 170/786.

Selected facsimiles

TRANSLATION

Note: The following are translations of only some of the texts. Repeated locutions are translated once only.

Facsimile 1.2. (related to 1.17).

Facsimile 1.3

Facsimile 1.9

Facsimile 1.10

Facsimile 1.12

Facsimile 1.14

Note: This recalls Isaiah 66:1: "Thus saith the Lord: the heavens are my throne and the earth is my footstool." We may ask whether the phrase was not a part of the common stock of monotheistic logia, see 1.19 for another `ar..

Facsimile 1.20

Facsimile 1.24

Facsimile 1.25

Facsimile 1.26

BIBLIOGRAPHY

Abbot, Nabia (1946). "The Kasr Khranah Inscription," Ars Islamica 11-12 (1946), 190-195 + Fig. 1.

Ben Shemesh, A. (1971). Ha-Qur'an: Sefer ha-sefarim shel ha-Islam. Ramat Gan, 1971.

DKI = Corpus of Dated Kufic Inscriptions, collection of Yehuda D. Nevo.

GAP = Grohmann, Adolf. Arabische Palaographie, Band 2, Teil 2. Graz-Koln, Hermann Bohlaus, 1971.

JUI = al-'U99, Muhammad Abu al-Faraj (1964). "Kitabat `arabiyya gayr mangura fi Jabal 'Usays" (Arabic Inscriptions from Jabal 'Usays). Al-Abhat 17 (1964), 227-316.

Lampe, Lexicon = A Patristic Greek Lexicon, ed. G. W. H. Lampe. Oxford University Press, 1961.

Lane, Edward William. An Arabic-English Lexicon. London, 1863.

ODCC = The Oxford Dictionary of the Christian Church. Oxford University Press, various editions.

Pines, Shlomo (1966). "The Jewish Christians of the Early Centuries of Christianity According to a New Source," Proceedings of the Israel Academy of Sciences and Humanities, vol. 2, no. 13 (1966).

Pines, Shlomo (1984). "Notes on Islam and on Arabic Christianity and JudaeoChristianity." JSAI4 (1984), 135-152.

Pines, Shlomo (1990). "Jahiliyyah and 'I1m." JSAI 13 (1990), 175-194.

Schacht, Joseph. The Origins of Muhammadan Jurisprudence. Oxford, Clarendon, 1950.

Al-IUgs, Muhammad Abu al-Faraj (1964). "Kitabat `arabiyya gayr mangura fi Jabal 'Usays" (Arabic Inscriptions from Jabal 'Usays). Al-Abhat 17 (1964), 227-316.

Yusuf 'Ali, 'Abd Allah (1946). The Holy Qur'an. Text, Translation and Commentary. New York, 1946.

Walker, John (1941). A Catalogue of the Arab-Sassanian Coins. London, Trustees of the British Museum, 1941.

Wansbrough, John. QS = Quranic Studies: sources and methods of scriptural interpretation. Oxford University Press, 1977.

Wansbrough, John. SM = The Sectarian Milieu. Oxford University Press, 1978.

NOTES

1. Schacht (1950).

2. Pines (1984, 1985).

3. Walker (1941) lviii. The early coins of al-Sam are not dated, and in any case do not bear religious legends.

4. This classification is the fourth so far attempted, and like its predecessors will no doubt prove to be temporary.

5. Wansbrough, SM 2.

6. Ibid., 57.

7. Abbott, N. (1946).

8. Wansbrough, referring to epigraphical linguistic evidence, notes that:

"It is important to remember that from the beginning of [Classical Arabic's linguistic] evolution to its end the evidence is always that of a written language, whatever its pre-history may have been. The material is thus always witness to formal and/or formalized communication, for which emergence of a standard of excellence was not only its organizing principle but also its logical conclusion" (Wansbrough, QS 92, emphasis added),

I accept his point, but I would argue that the documents of the first and early second centuries A.H., both chancery and popular inscriptions, provide us with evidence and examples of the (written) idiom of the pre-Islamic period-what we may call the prehistory of Islam.

9. In several articles, listed in the bibliography.

10. Wansbrough, QS 1.

11. Al-'U99, JUL

12. I distinguish between Muhammadan and Muslim texts, at least at present, according to the conceptions expressed and the idiom employed, not merely by whether the Prophet's name and the Triple Confession formulae are mentioned, for these are common to both the Muhammadan and the Muslim texts.

13. Those from Hisam's days are recorded under the site codes SC and HS, but are not reproduced here. The `Abbasid texts (al-Mahdi and first year of Harun's reign) are reproduced here as facs. 1.25-26; they were found at two different sites, YT and ST respectively.

14. Reproduced here as facs. 1.1-1.20.

15. Two verbs of entreaty have been found in the basic class texts at Sde Boqer, construed as a request as part of the supplication, not as its opening phrase (here facs. 1.9:5 and 1.19:3).

16. Henceforth PN.

17. In one Muhammadan text this root does occur, but in the form 'istiThad "to take part in battle," not with the meaning "to give testimony" (inscription HS 3155:5, not reproduced here).

18. JUI no. 103, 108/729.

19. JUI no. 105, undated.

20. I am not happy with either translation, though they do not matter much here, since the excerpt given is intended only to demonstrate the context of the kafa phrase.

21. Compare the terms martyred, martyrion, martys in Lampe, Lexicon, 828 col. 1-830 col. 2; and the various forms of bearing witness regarding particulars of faith, common in the Rabbinic literature.

22. Lane, Lexicon 2, 675 col. 3.

23. Idem, 676 col. 1.

24. Even in this case I would still argue that janna did not mean "paradise."

25. In the Muslim texts of ca. 300/913 we have, similarly, cases where the word farik of the 'adam al-.dirk was replaced by a dash: la-lahu (HR 507); la-k (BR 5135). Neither is reproduced here. This I also understand as avoidance of pronouncing the unspeakable.

26. JUI no. 72, from ca. 113/732.

27. The term hypostasis (literally "substance") "also came to mean individual reality and, from the middle of the fourth century onwards, especially in Christological contexts, a `person' " (ODCC, s.v.).

28. S = the south face, first strip (S' indicates the south face, second strip); A = the first subdivision of the text of the first strip, and similarly B = the second subdivision, etc.

29. See note 22 for an explanation of the divisions of the Dome of the Rock text.

30. The phrase "there is no God but Him" is known from Christian intersectarian polemic; cf. Pines (1984).

31. Esdras/Ezra: i.e., he who, according to anti-Jewish tradition, rewrote the Torah from the memory of the Elders after the copy transmitted by Moses had been lost when the First Temple was destroyed. This allegation is the "fact" upon which was based the claim that the book which the Jews call the Torah is not that which Moses transmitted to the Children of Israel.

32. 'Israf il: the angel who is to blow the horn on the day of resurrection (Tafsir al-Jalalayn on Q 6:73; Lane, Lexicon 4, 1351 col. 3).

33. I suspect here a scribal blunder, and that the first line should be reconstructed thus: [This is the testimony of] Said b. Yazid: Allah [is] my Lord, and He is an excellent patron (etc.).

34. I understand irk as the Greek synthetos, equal to murakkab or mu3`akkal in meaning. It belongs to the vocabulary of polemic: synthetos was used derogatively to denounce inter-confessional opponents (see Lampe, Lexicon).

35. Smd: I propose that it means an integer, an indivisible whole, a substance which cannot be split apart or compounded.

If you find an error please notify us in the comments. Thank you!