Journal of Semitic Studies doi: 10.

1093/jss/fgaf012
© The author. Published by Oxford University Press on behalf of the University of Manchester.
All rights reserved.

ANCIENT ALLAH:
AN EPIGRAPHIC RECONSTRUCTION

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

AHMAD AL-JALLAD
THE OHIO STATE UNIVERSITY

Abstract

The identity of the pre-Islamic Arabian deity, Allāh, has been the sub-
ject of much scholarly debate. Opinions vary from regarding him as the
primary creator deity of the pre-Islamic Arabs to seeing him as a minor,
peripheral superbeing with no associated cult. While the epigraphic
evidence confirms that Allāh was worshipped across pre-Islamic Arabia,
the texts known until now offered limited information about his role
in ancestral Arabian religion. This article presents the discovery of a new
archaeological site from the northeastern Jordanian Ḥarrah (MH09),
which comprises a mortuary installation and a considerable number of
Safaitic inscriptions. One of these carvings provides our first glimpse
into Allāh mythology among the nomads east of Ḥawrān roughly two
thousand years ago. This complex text proffers strong evidence for
Allāh’s role as a light-giving creator deity, called upon to act against
death, represented by darkness. After a detailed analysis of this inscrip-
tion, key to interpreting a number of other challenging Safaitic expres-
sions and terms, the article concludes with a discussion of how this
text informs our understanding of ancestral Arabian cosmology and
how the ancient Arabs might have received biblical ideas about creation
in late antiquity.

1. Introduction:1 Allāh between the chisel and the pen2

The identity of the pre-Islamic Arabian deity, Allāh, has been a matter
of considerable debate in recent scholarship. While some have sug-
gested that the divine name is a loan from Syriac, alāhā (allāhā) (Jeffery,
2007, 66), Kiltz has demonstrated that this is extremely unlikely and,
1 l wqṭr w zmz ḥbby ʾb-hm l-ʾbd w l-hm h-sfr.
2 I owe my gratitude to dear friends and colleagues for reading various drafts
of this paper, encouraging and restraining my imagination. Thank you, Michael,
C.A. Macdonald, Alessia Prioletta, James Moore, Sean Anthony, Marijn van Putten,
Benjamin Suchard, James Bejon, and the two anonymous JSS reviewers. The tran-
scription of Ancient North Arabian inscriptions follows the OCIANA conventions.

1
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

if there is a relationship between these two theonyms, the influence

went in the opposite direction (Kiltz, 2012).3 The opinions of the
medieval Arabic lexicographers are divided with regard to its etymol-
ogy. Some consider it a proper name, treating the initial al- as an
inseparable part of the word. This is supported by the fact that allāh can

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

be preceded by the vocative particle yā, which is otherwise not permit-
ted before definite nouns: yā ʾallāhu (‘O Allah’). Others, however, view
it as a contraction of the definite article al- and the generic term for
celestial superbeing, ʾilāh, producing ‘the god’.4 Modern scholarship
seems to have settled around the latter view. Brockelman drew atten-
tion to a similar sound change in other words, e.g. an-nās < *al-ʾunās
‘the people’, indicating that it is a semi-regular phenomenon in Ara-
bic (Brockelman, 1908, 123–124, §58c). This observation was refined
by Testen (2005), who regarded it as a sound change characterizing
the old Ḥijāzī dialects. Kiltz expanded the geographic scope of the
sound change to encompass a ‘western’ dialect continuum of Old Ara-
bic (2012, n. 24). This judgement would appear to be correct as the
contracted form allāh is attested in the Nabataean realm as far north
as Bostra.5
The contraction of al-ʾilāh to allāh is mirrored in the development
of the feminine divine name, allāt, from al-ʾilāt. The uncontracted
form appears first in Herodotus’s Histories (c. 425 BCE) as αλιλατ,
whereas ʾallāt is the sole form found in Nabataean inscriptions from
the 2nd century BCE onward, suggesting a chronological progression
from the former to the latter.6 So then, if allāh simply means ‘the god’,
to whom does it refer? Is it the proper name of a particular superbeing,
or a generic term applicable to any deity during worship? Kiltz sum-
mons epigraphic evidence supporting both interpretations. In inscrip-
tions where Allāh is invoked alongside other divine beings, it is most
naturally read as the name of a specific deity. On the other hand, its
3 Kiltz discusses in some detail the geminated Syriac form allāhā, especially its
resemblance to the Arabic allāh, but George Kiraz informs me that, despite some
grammarians reporting the l as geminated, the Eastern reading traditions of Syriac that
have retained the gemination of consonants do not pronounce the l as geminated.
4 See Lane, 83, for a comprehensive discussion on the various opinions.
5 While the name hlh is attested in East Arabia, it appears to be drawn from a

non-local dialect, as the language of the local inscriptions employs a suffixed ʾ as a

definite article; see Al-Jallad (2024).
6 Herodotus, lib. iii., cap. viii. See Hämeen-Anttila and Rollinger (2001) for an

attempt to interpret the transcription in other terms, but there remains a consen-
sus among specialists that it reflects the Arabic definite article + ʾila(:)t ‘goddess’.
This epithet is found elsewhere in Arabia as hnʾlt and also hlt, so the contraction
is certain.

2
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

frequent use in personal names like wahballāh and ʿabdallāh could sug-
gest it was a general appellation. In the Arabian context, deities that
were frequently called upon in inscriptions rarely appear in theophoric
names. Instead, one encounters ʾl and (ʾ)lh as the commonest divine
elements in the onomasticon (Kiltz, 2012, 39).

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

Ch. Robin (2020) has recently studied the use of the divine name
Allāh before Mohammed and argued that the label was nothing more
than an Arabic ‘translation’ of the West Semitic creator god ʾilu. His
argument is based on the fact that the Qurʾān implies that the super-
beings allāt, al-ʿuzzē, and manōh were regarded by the audience as the
daughters of Allāh, paralleling the ‘daughters of ʾil’ mentioned in the
inscriptions of Yemen, the Negev, and Palmyra.7 Indeed, the Central
Semitic generic term ʾilāh ‘god’ does appear to ultimately derive from
ʾilum through the addition of a stem expander, -āh. Huehnergard under-
stands this as part of an original plural construction: singular *ʾilum >
plural *ʾilāh-ūna. He then argues that the singular ʾilāh was backformed
from the expanded plural (2005, 191). Even if the Arabic allāh does
derive from ʾilu through ʾilāh, it is not a straightforward reflex of the
expanded Central Semitic form, as it also incorporates the definite
article.8 Moreover, the presence of ʾil in ancient Arabic personal names,
along with both ʾil and ʾilāh as divine names/titles in Sabaic, indicates
a more complex situation (Robin, 2020, 98–99). If we wish to maintain
an equation between the Central Semitic creator deity ʾilu and allāh,
then the latter cannot be regarded as simply a translation. It must have
been a divine title that was applied to ʾilu, but is there good reason to
assume a definitive equivalency in the first place? Is the rough etymo-
logical congruence along with siring divine daughters enough to estab-
lish this connection?
A close reading of the Qurʾān indicates that its Meccan antagonists
regarded Allāh as a creator deity as well. Indeed, the theological inter-
vention of the Qurʾān does not seem to have concerned the role of
Allāh in the cosmos, but rather the power and veneration of other

7 See in particular this paper and its treatment of ‘bnyt-h … his construction’;
and see Qurʾān 53: 19–20. See also Crone (2010, 55–58) for a judicious presentation
of the facts.
8 The Arabic form would then parallel the development of Aramaic elāhā and

Hebrew hāʾĕlōhîm, both of which contain the definite article and can be translated
as ‘the god,’ referring to the deity presently being worshipped. In other words, these
are not translations of ʾilu as such, but rather should be regarded as divine titles. It is
important to note that ʾilu, when he appears in theophoric names, never occurs with
the article, e.g. mîḵāʾēl ‘Michael’ and even in the Ancient North Arabian and South
Arabian inscriptions, ʿḏrʾl (Dadanitic), krbʾl (Sabaic), whbʾl (Safaitic), etc.

3
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

superbeings (Crone, 2010, 59–61). Passages such as Q 39:38 and

Q 43:87, where the ‘pagans’ respond to the question posed by the
Qurʾān: ‘who created the heavens and earth?’ with an unequivocal
‘Allāh’ have motivated scholars, both medieval and modern, to under-
stand the deity as a ‘high god’.9 Another clue as to the mythologi-

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

cal background of the pre-Islamic Allāh lies in the etymology of the
verb ‘to create’ in Arabic, found in the question posed in Q39:38, for
example: laʾin saʾalta-hum man ḫalaqa s-samāwāti wa-l-ʾarḍa la-yaqūlunna
llāh ‘if you ask them who created the heavens and the earth they une-
quivocally reply ‘Allāh’. The verb ḫalaqa with this meaning is unique to
Arabic. Its cognates in other languages mean ‘divide, separate’: Syriac
ḥəlaq, Hebrew ḥālaq, Sabaic ḫlq, and Minaic ḫlq.10 The Arabic mean-
ing of the word must have developed from the ancient Near Eastern
cosmological myth, in which the creation of the world was accom-
plished through the ‘dividing’ of pre-existing matter.11 This concept
likely drove the semantic shift through metaphorical extension. Given
that this term was understood by Mohammed’s interlocutors as well,
it indicates that the semantic shift happened in pre-Islamic times and
was not something introduced by the Qurʾān. To sum up, there are
three points that speak to the equation of Allāh with ʾilu:
1. Etymology: ʾilāh ultimately derives from ʾilu, but the inclusion of the
definite article prevents understanding it as a simple translation;
2. The Arabs of Pre-Islamic Mecca, as depicted in the Qurʾān, under-
stood Allāh as a creator deity;
3. The generic verb of creation in Arabic, ḫalaqa, which is applied to
Allāh’s cosmic creative capacity, etymologically derives from the sense
of ‘dividing’. This is consistent with ancient Near Eastern myth.
None of these points is decisive, and indeed these similarities may
be the result of a rather chronologically shallow process of syncretism
between ancestral Arabian religion and biblical monotheisms. In the
popular imagination, pre-Islamic Arabia was hermetically sealed off
from outside influences. The Meccans would have, therefore, had
little to no exposure to biblical ideas and would not have known or
9 For a full survey of the relevant passages, see Sinai (2005, 68–69). For a dis-

cussion of the earliest opinions in modern scholarship regarding Allāh as a high god,
see Watt (1971), and more recently Crone (2010, 77–82), who rightly demonstrates
that the category has no explanatory value when it comes to understanding poly-
theistic religious environments.
10 The term ḫlq was recently attested in Sabaic where it seems to mean ‘com-

pensation’, < *portion, *lot; see Robin et al. (2014, 1073–1077).

11 See recently Baasten (2009).

4
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

acknowledged the biblical god. Recent scholarship, however, increas-

ingly underscores the interconnectedness of all regions of Arabia with
the world around them. Archaeological, epigraphic, and literary evi-
dence demonstrates the presence of monotheistic ideas and communities
throughout the Peninsula from the fourth century CE onward.12 This

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

revised understanding of late antique Arabia better contextualizes
Crone’s characterization of the pre-Mohammedan Meccans as Judaiz-
ing ‘pagan monotheists’, who recognized the biblical god while still
venerating ancestral Arabian deities as angelic figures (Crone, 2016).
Hawting went further and argued that the entire drama of the Qurʾān
was between competing understandings of monotheism, and that that
polytheism was retrojected into the text by later readers (1999). Sinai’s
recent study of the divine name Allāh in pre-Islamic poetry is generally
consistent with its usage in the Qurʾān and its interlocutors, particu-
larly regarding creation, the provision of rain, and control over worldly
affairs (2019, 57). One could interpret these facts to suggest that the
identification of Allāh with ancient ʾilu may be attributed to the impact
of biblical ideas on ancestral Arabian religion following the fourth cen-
tury CE, and the subsequent application of the west Arabian divine
epithet al-lāh ‘the god’ to refer to the biblical creator god. In other
words, the similarities between Allāh and ʾilu are not due to a direct
link between the two. Instead, they arise from the fact that the bibli-
cal god himself is a monotheistic development of ʾilu (Wilson-Wright,
2019). These shared qualities and characteristics were then integrated
into ancestral Arabian religion through a process of syncretization with
biblical monotheism(s) and do not speak to the continuation of ʾilu
mythology into the seventh century CE. The same reasoning would
then explain the semantic development of ḫalaqa from ‘to divide’
to ‘to create’, a term used to describe the biblical god’s initial crea-
tive act.
While this line of thinking has become popular as of late, it has not
achieved a consensus. Robin maintains that Allāh’s role as creator orig-
inated within ancestral Arabian religion. Monotheistic Arabs of late
antiquity, particularly Christians, referred rather to their god as al-ʾilāh,
a transparent calque of the Greek ho theós and Syriac ʾalāhā. He goes
on to argue that the pagan Meccans opportunistically leveraged the
etymological relationship between their pagan creator god allāh and
the Christian al-ʾilāh to establish a universal cult center to the deity at

12 For an in-depth treatment of the matter in South Arabia and its relevance to
other parts of the Peninsula, see Robin (2021), and also Robin (2020). See recently
Lindstedt (2023) on the penetration of biblical monotheisms in the Ḥijāz.

5
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

Mecca in reaction to Abraha’s conquest of Arabia Deserta in 552 CE

(2020, 102). Al-Azmeh, however, rejects all of the aforementioned
interpretations of the material (2014, 283). For him, Allāh lacked any
‘theological or mythological profile’. He was an obscure being with no
cultic center, only occasionally called upon in pre-Islamic times. The

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

promotion of Allāh to a ‘high god’ and ‘creator’ was directly the result
of Mohammed’s theological project. Accordingly, Al-Azmeh under-
stands Sinai and others to have anachronistically misread earlier mate-
rial, assuming that the cosmic functions attributed to Allāh were solely
his and not simply the result of rhetorical ‘nomenclature, as epithets,
and as hyperbolic terms of exultation by elevation and amplification’
common in pagan liturgical language (2021, 425). Allāh was a title that
could be applied to any god at the moment of worship.
While the epigraphic evidence remains difficult to interpret, it ten-
tatively suggests that Allāh was widely regarded as a remote deity, not
generally invoked to intervene in the affairs of humans. Most of this
material has been surveyed in Robin’s 2020 article so there is no need
to repeat all of the examples here.13 Robin argues convincingly that the
way the divine name was used in the pre-Islamic epigraphy suggests that
it was not simply a generic term for any deity. Rather, it referred to
a specific, well-defined superbeing within the mythology of the ancient
North Arabians. This is clearly illustrated in the Rbbl bn Hfʿm grave
inscription of Qaryat al-Fāw, where Allāh is called upon alongside two
other gods to protect the grave of the deceased man: ʾʿḏ-h b-khl w-lh
w-ʿṯr ʾ-śrq ‘he put it (the grave) under the protection of Kahl, Allāh, and
ʾAṯṯar of the east’.14 Allāh is invoked more frequently in the west and
north, where the theonym originates. At ancient Dadān, we encounter
him first in personal names, such as ʿbdlh /ʿabdallāh/ (AH 78). A single
unprovenanced text at the Saudi National Museum attests to a priest, ʾfkl,
of hlh, which Robin equates with Allāh via the h-article, and records a
dedication to the god of an obscure place/people rmʿt, called ʾhlh, which
Robin corrects to Allāh, though the matter remains unclear (Robin 2020,
54–55). It is impossible to date these texts, precisely, but the writing
tradition associated with this oasis seems to begin sometime in the first

13 The following paragraphs will focus on attestations of Allāh in northwest Ara-

bia as they are directly relevant to the text under investigation. For cognate divine
names in East and Central Arabia, the reader is encouraged to consult the appendix
of Robin’s 2020 article.
14 For the original discussion of this text, see Beeston (1979). The text is

quoted by Robin (2020, 107) and discussed in further detail by Macdonald in his
contribution to Fiema et al. (2015). Most scholars date the text to around the first
century BCE based on the paleography.

6
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

half of the first millennium BCE and ends sometime around the Nabataean
takeover at the end of that millennium (Kootstra, 2022, §1; Rohmer,
2021). Allāh is only called upon directly in one invocation, JSLih 08:
ʿbdmnt / ʾṣdq / f rḍ-h / h lh / w sʿd-h ‘ʿAbdmanōt performed the ṣdq so
satisfy him, O Allāh, and help him’.15 This infrequency, however, is not

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

entirely unexpected as the Dadanitic inscriptions are overwhelmingly
henotheistic and call almost exclusively upon the tutelary god of the
oasis, ḏ-ġbt. Further to the north in the Ḥismà, Allāh is invoked more
frequently, but still not in great numbers. In the Hismaic inscriptions,
which span from Northwest Saudi Arabia into central Jordan, Allāh
takes a unique vocative form, anticipating late pre-Islamic and Quranic
diction, allāhumma. This is, at present, attested seven times:
Jacobson D.12.4: h lhm l-qn bn mskt
‘O Allāh, (grant) to Qn son of Mskt
Jacobson D.3.A.7 b: h lhm l-bn ʿry bn ʿklm bʾs
‘O Allāh, may the sons of ʿry son of ʿklm experience
misfortune’
KJC 47: h lhm l-ḥbbʾl
‘O Allāh, (grant) to Ḥbbʾl’
The form ʾlh, with the initial glottal stop, however, is attested in one
Hismaic text from Sakākā.
WTI 14 ḏkr ʾlh slm
‘May Allāh be mindful of Slm’
The Hismaic inscriptions overlap geographically to a great extent with
Nabataean, where the divine element ʾlh(y) is frequent in personal names,
although it is surprisingly absent in religious invocations.16 The latter
situation can be interpreted to support Al-Azmeh’s view, namely, that
Allāh was a generic term referring to ‘the specific god whom I wor-
ship’, which Healey suggests understanding as a reference to Dusares
(2001, 23–24). If this is true, then the Hismaic inscriptions beginning
with the invocation lhm /allāhumma/ can be compared to the late pre-
Islamic talbiyah, which Al-Azmeh understands as applying the vocative
appellative to various deities at their specific time of worship (2021,
426). Still, WTI 14 suggests that ʾlh is the proper name of a deity,
15 See the commentary in the OCIANA card for a discussion of the interpretation
of this text, in particular, the verb ʾṣdq
16 There is, for example, no lemma associated with this deity in Healey’s impor-

tant conspectus of Nabataean religion (2001). The closest possibility is an invocation

to ʾlhʾ literally ‘the god,’ and the Aramaic equivalent to Allāh, in a late and uncertain
context (55–56).

7
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

especially since it should be parsed as ʾ-lh ‘the god’ with the ʾ-definite
article, while Hismaic lacks any means of morphological definition.17
In other words, this divine name must originate in another dialect of
Arabic, namely, Nabataean, and is not a Hismaic-internal attempt to
simply say ‘the god (whom I am now worshipping)’.

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

In 2018, H. Hayajneh published what is possibly the longest Ancient
North Arabian inscription discovered to date: a Hismaic text begin-
ning with a description of difficult circumstances and concluding with
a blessing for 72 of the author’s loved ones. Hayajneh interpreted the
first two lines as follows:
HHM 2 (Hayajneh, 2018, 98)
l brd bn ddʾl w sqm
‘By Brd son of Ddʾl and he was ill’
f tśwq {f} mʿṣ l-h ḥnn-h
‘and he was full of longing, {and so} became angry {because of} his
yearning’
Hayajneh’s interpretation was not sensitive to the ritualistic/cultic
meanings of the lexemes, sqm, tśwq, and ḥnn. Both sqm and ḥnn appear
in the monumental Madaba inscription published first by Graf and
Zwettler in 2004. The text’s penitential nature is evident: it confesses
a transgression against the god ṣʿb, which is signalled by the verb sqm,
and follows this with a vow (nḏr) intended to absolve the sin. The god
is asked to show the transgressor compassion (ḥnn).18 Thus, the col-
location of sqm and ḥnn in HHM 2 must be regarded as significant,
indicating that it belongs to the same genre. Moreover, the term tśwq
need not only signify longing for humans. Two stone tripods held by
the Bible Lands Museum bear Hismaic inscriptions recording longing
for the gods Dśr and Lt.19 I believe this phenomenon is linked to per-
ceived divine abandonment and the subsequent withdrawal of divine
17 On the grammar of the variety attested in the Hismaic inscriptions, see Al-
Jallad (2020) and King (1990).
18 sqm f tḍrʿ w tʿny…f yhb l-h hṇṇ f ygzy nḏr w yzd ‘he transgressed, so he humbly

supplicated and suffered (for the god Ṣʿb) … so may he show him compassion in
order that he fulfil his vow and do more’. For a full philological commentary, see
Al-Jallad, 2020.
19 I studied these artifacts in Jerusalem in 2019 and produced the following read-

ings, which were sent to Dr. Yigal Bloch, the curator of the collection: BLMJ 00968.c:
l zhmn w tśwq l-dśr w l-lt ‘by Zahmān and he longed for Dusares and Allāt’;
BLMJ 00969.b: l lb w tśwq ʾl-dśr ‘by Lobb and he longed for Dusares’. I have given
the photographs that I took of these pieces to Michael C.A. Macdonald, who is now
preparing a comprehensive study of Ancient North Arabian inscribed objects.

8
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

favour as a consequence of transgressions against a deity.20 Putting all

three of these concepts together, we are motivated to reparse HHM 2
to produce the following:
HHM 2 (re-interpreted)
l brd bn ddʾl w sqm

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

‘By Brd son of Ddʾl and he transgressed’
f tśwq {f} m ʿṣ lh ḥnn-h
‘and was filled with longing (because of divine abandonment), so show
compassion to the one who had disobeyed Allāh!’
My new reading takes the third clause as a conditional. The m should
be construed as the indefinite relative pronoun *man, with the com-
mon assimilation of the n. This is followed by the verb ʿṣ, which I would
connect with Classical Arabic ʿaṣā ‘to disobey’, offering a near parallel
to the common Quranic phrase man yaʿṣi llāha wa-rasūla-hū ‘whoso-
ever disobeys Allāh and his messenger’ (Q 4:14; 33:36; 72:23).21 If
this interpretation is correct, then it offers another attestation of the
direct worship of Allāh and indicates the existence of certain ritualistic
obligations concerning him.
We finally come to Safaitic, the corpus to which belongs the inscrip-
tion that forms the main concern of the present study. Allāh is com-
paratively rare in Safaitic, invoked under fifty times, a tiny number
when compared to the more than 1,400 invocations attested to Allāt
(Al-Jallad, 2022, 93–94). The divine name is written in two ways: lh
and ʾlh. Safaitic orthography does not graphically represent vowels of
any quality or length; thus, the first spelling clearly renders allāh, begin-
ning with a vowel, while the second spelling records what is known in
Classical Arabic grammar as alifu l-qaṭʿ, implying the pronunciation
ʾallāh. A similar phenomenon is occasionally registered in verbal stems
that begin with a consonant cluster: tẓr /ettaṯara/ (passim) <*intaṯara
(cf. Classical Arabic intaḏara) ‘he awaited’ vs. ʾtgnn /ʾVtgannana/
(BES15 857) ‘to go mad’ (cf. Egyptian Arabic itgannan); the latter spell-
ings are much rarer than the former. Greek transcriptions conclusively
demonstrate that the Safaitic spelling lh renders allāh: the same man
inscribes his name as whblh in Safaitic and then as Ουαβαλλας in Greek
(Kiltz, 2012, 37; Al-Jallad, 2017b, 167–168).
It is important to note at the outset that the Safaitic spelling com-
pels us to regard Allāh as a proper name and not a generic title. This

20 On this phenomenon, see Al-Jallad (2022, 69–71).

21 On defectively spelled final weak verbs in Hismaic, see Al-Jallad (2020).

9
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

is because lh and ʾlh are not the way to express ‘the god’ in the most
commonly inscribed Safaitic dialect, which instead makes use of the
h- definite article.22 Thus, we would expect something like h-ʾlh, or
even h-lh if the initial glottal stop were lost, both of which are attested.23
Thus, lh and ʾlh must be regarded as the name of a specific superbeing,

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

originating in a western dialect that employed in a more regular fashion
the al-definite article, like Nabataean Arabic. The importation of a
‘foreign’ divine names is well attested in Safaitic, where some of the
most commonly called upon deities originate in languages and dialects
from beyond the Ḥarrah, such as dśr /diśar/ and bʿlsmn /baʿal-šamēn/,
both of which appear to have an Aramaic source.24
The Safaitic corpus as presently known does not advance our knowl-
edge of Allāh’s mythological profile by much. Most of the prayers to
him are rather generic and overlap with prayers to other deities. He is
most commonly invoked to provide security, slm, and also relief, rwḥ,
and is attested in prayers alongside other gods. To illustrate:
AAWHA 130 l skrn bn śmt bn zkr bn ġyr{ʾ}{l} w ʾśrq f h lh w śʿhqm
slm w ʿwr l-ḏ ḫbl
‘By Skrn son of Śmt son of Zkr son of Ġyrʾl and he set
off for the inner desert so, O Allāh and Shayʿ haq-
Qawm, may he be secure, and blindness be to the one
who effaces (this inscription)’
C 2816 l rbʾl bn ʾʾs()d bn rbʾl bn ʾnʿm w ḥl h-dr w ḫr(ṣ) h-śn(ʾ) w
h-ḫsf f h bʿlsmn rwḥ w ś(ʿ)hqm w h lh w lt ʿrg w kmh l-ḏ
yʿwr h-ḫṭṭ
‘By Rbʾl son of ʾʾsd son of Rbʾl son of ʾnʿm and he
camped here and kept watch for enemies and the rain
clouds so, O Baʿal-Samīn, send the winds and Shayʿ
haq-Qawm; and O Allāh and Allāt, lameness and blind-
ness be to the one who effaces this writing’

22 While the ʾal-article is attested in Safaitic, it is much rare than the h- one

(Al-Jallad, 2019, 351–352). If the divine name should truly be understood as ‘the
god’, reflecting the grammatical frequency of article forms in Safaitic, then we should
find a preponderance of h-ʾlh and h-lh, but the opposite is true.
23 For example, WH 3923, which Macdonald reads as: l {{b}}{{d}}{{ḥ}} w

ʿw{{ḏ}} {b-} {{h}}{{ʾ}}lh, trans. By {Bdḥ} and {he sought refuge} {in} {h- ʾlh}, i.e.
‘the god’. Macdonald (2024, 346–347) cautiously proposes that the Safaitic h-ʾlh may
refer to the Christian god, corresponding to the Paleo-Arabic ʾal-ʾilāh (‘the god’),
which is found almost exclusively in Christian Arabic inscriptions. Attestations of hlh
may reflect rare strategies of rendering al-lāh in the Safaitic dialect.
24 In both cases, Safaitic speakers occasionally localize the divine names by ren-

dering them into the dialect of the Ḥarrah, ḏśr(y) and bʿlsmy, respectively (Macdonald,
2000, 46).

10
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

A key difference between the Safaitic Allāh and the portrayal of the
Quranic interlocutors’ understanding of the deity concerns the parentage
of Allāt. In the Qurʾān, the antagonists appear to regard Allāh as Allāt’s
father (Q:53), whereas Roḍaw is identified as her father in two Safaitic
texts (Al-Jallad, 2022, 56–57). Although one might appeal again to the

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

understanding of Allāh as simply a title, which was sometimes applied to
Roḍaw, there is a possibility that they are invoked separately within the
same inscription. The text, however, is known only from a poor-quality
hand copy, making its reading and interpretation highly conjectural.
C 3051 l rbḥ bn mlk [b][n] ʾys¹(r) (w) rḍw [w] l(h) by w ʿwr m ʿwr
‘By Rbḥ son of Mlk [son of] ʾysr and so Ruḍaw and Allāh be
with me; and blind whosoever effaces’
More decisive, however, is the fact that Allāh is invoked alongside Roḍay,
which is simply a dialectal variant of Roḍaw (Al-Jallad, 2022, 97).
SIAM 31 l wdm bn ʾḥd bn ʾqwm h-ḫṭṭ w h rḍy w lh ʿwr m ʿwr
‘By Wdm son of ʾḥd son of ʾqwm are these carvings and, O
Roḍay and Allāh, blind whosoever effaces’
C 3712 l ʾnʿm bn sʿd bn ʾdʿgt w h lh slm w rḍy ʿwr ḏ yʿwr h-sfr
‘By ʾnʿm son of Sʿd son of ʾdʿgt and, O Allāh, may he be secure,
and, O Roḍay, blind him who would efface this writing’
Thus, lh and rḍy must represent two distinct superbeings. One Safaitic
inscription gives a slight hint that Allāh had specific properties, as evi-
denced by the unique invocation associated with him:
SIJ 293 l msk bn bdbl bn msk bn rfʾt w ʾṣly w ʾqsm b-ʾlh ḥy l-hdy ʿẓm
‘By Msk son of Bdbl son of Msk son of Rfʾt and he made a
burnt offering and swore by Allāh, who is living, that he shall
lead bravely’
The inscription is a unique record of a religious ritual associated with
the appointment of a military commander, hdy (SafDict, 77).25 Its sub-
ject swears an oath and dedicates a burnt offering to Allāh, a sequence
of activities not associated with other deities yet. But perhaps more
significant is the epithet the author applies to the deity: ḥy ‘living’.
I have previously interpreted this as an adjective, which is the most
common grammatical category in this syntactic position, but it is also
possible to take it as a verb, as Safaitic permits both grammatical cate-
gories in this position to form asyndetic relative clauses (Al-Jallad, 2019,

25 See also Macdonald (2014) on the cultural context of military commanders

among the nomads of the Ḥarrah.

11
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

362). Its meaning, however, is more challenging. The most direct

interpretation is to take it as ‘living’ or ‘alive’. But what would that
mean exactly? Is it a contrastive epithet—Allāh who is always living,
unlike other deities who die, even momentarily, such as Baʿal? Or does
it indicate that Allāh is the source of life? Or perhaps both? The epithet

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

parallels the common description of the biblical god, e.g.: Deut. 5:26
‫ֹלהים ַח ִ֜יּים‬
ִ֨ ‫‘ ֱא‬the living God’, Ps. 42:3 ‫ים ְל ֵ ֪אל ָ ֥חי‬ ִ ‫‘ ָצ ְמ ָ֬אה נַ ְפ ִ֨שׁי | ֵל‬My
֮ ‫אֹלה‬
soul thirsts for God, for the living God’, Hos. 2:1 ‫ל־חי‬ ֽ ָ ‫‘ ְבּ ֵנ֥י ֵ ֽא‬children
of the living God’.26 If we opt for the former understanding, we would
more readily expect a D- or C-stem verb, so ḥyy or mḥy. It seems impos-
sible to understand the two as constituting a construct-genitive con-
struction, Allāh of life.27 The meaning of this epithet will become clearer
in the next section, but for now it suffices to say that it is compatible
with the identification of Allāh as a creator deity, or at the very least,
it speaks to the fact that he possesses a specific role in the cosmos.
Let us summarize the epigraphic evidence collected to this point:
while Kiltz observed that the Nabataean onomasticon is the only body
of inscriptions that could suggest that Allāh was a generic appellation,
a contextual examination of the distribution of the divine name in the
Nabataean realm does not support such an interpretation. Elsewhere we
encounter the divine name Allāh, it must be construed as the proper
name of a distinct figure, not only because it is invoked besides other
deities, but also because the name is not analyzable as ‘the god’ in the
linguistic contexts in which it is couched. And because the name
of this figure must have been borrowed into surrounding areas,
such as the Ḥarrah, from Northwest Arabia, it is difficult then to inter-
pret the appearance of Allāh in Nabataean as simply a generic title
either; Nabataean culture is likely its point of origin. Instead, his
absence in Nabataean invocations must be explained in some other
way. Generally speaking, the patterns of invocation to the deity are
consistent with the idea that pre-Islamic Allāh was a remote god and
not the primary object of veneration (Robin, 2020, 102), and this
may explain why the Nabataeans focused their prayers on more proxi-
mal members of the pantheon. The only glimpse into ancient Allāh
mythology comes from the Safaitic inscription SIJ 293, which pre-
sents us with an interesting epithet, living, but falls short of providing
a definitive identification.
26 On the use of the epithet ‘the living god’ more widely in the ANE, see Met-

tinger (2004). In Hismaic, the ‘living’ epithet appears to be applied to Allāt: ʾḥy-lt.
The grammatical construction of this epithet remains to be worked out.
27 Equally impossible is its understanding as ‘god of life’, as the noun is attested

as ḥyy, and more rarely as ḥyw and ḥywt (SafDict, 88).

12
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

2. Inscriptions: Allāh, the light giver

The inscription under investigation here was discovered in 2024 by

the author in the area of Marabb al-Hilāl in the northeastern basalt
desert of Jordan at Survey Site no. 9.28 The site is located on a basaltic

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

hill within the marabb, a large open tract of land where water gathers
and herbage is abundant during the rainy seasons. MH09 is character-
ized by a unique stone installation (see images 1–3 below): a rectilinear
open structure with inscribed stones facing towards the center. This
construction sits before a medium-sized cairn, which is likely a burial.
The structure is roughly 182 × 365 centimetres. While the construction
bears a superficial resemblance to the open-air desert mosques, impro-
vised structures created by faithful Muslims to perform their prayers in
the desert, several facts demonstrate that it is an ancient installation
associated with the Safaitic inscriptions. Desert mosques are sometimes
constructed with previously inscribed stones, but their builders are
never so careful to arrange them such that the inscribed surfaces face
towards the center. If the construction were built by those who had no

Image 1: Site MHL 09, stone enclosure 1 (© BES24)

28 The text was documented during the Missing Link/Badia Epigraphic Survey
(June, 2024), directed by the author, Ali al-Manaser, and Mr. Zuhayr al-Qadi (rep-
resenting the al-Ḥuṣn research center). The project was generously supported by a
grant from the Al-Ḥuṣn research center.

13
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

Image 2: The stone enclosure with the marabb in the background (© BES24)

Image 3: The southern wall of the enclosure (© BES24)

regard for the texts or their meaning, it is hard to understand why they
would have gone through the effort of placing them in this manner.
Moreover, desert mosques contain a niche (miḥrāb) and are oriented
south towards Mecca (Jarrar, 2024). The present construction lacks
this feature, and its orientation, if judged by the small opening into
the enclosure, actually faces west.
The inscribed stones describe the construction of a funerary instal-
lation. The rocks contain at least two sets of inscriptions, one set of
14
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

which is thematically connected. These are hammered in large letters

in variant 3 of the Safaitic script.29 They commemorate grieving and
the building of a structure called a ṯyt ‘resting place’,30 for a man
named ʿqrb bn ḥgg, as reflected in BES24 4 (Image 5). We can be sure
this is the name of the dead man, as the author of BES24 2.1 (Image 4)

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

records grieving for his brother and undertaking the building.
The structure itself is composed of twenty-two relatively large, con-
tiguous stones. Eight of these, on the east and south walls, are inscribed
to commemorate the dead man. The stones on the other sides are rela-
tively rough and not ideal for writing. The following list provides the
reading and interpretation of the associated inscriptions, which were
(nominally) written by the burial party consisting of fourteen men,
including two brothers of the deceased.
BES24 2: l sʿn bn ḥgg w wgm ʿl-ʾḫ-h w bny
‘By Sʿn son of Ḥgg and he grieved for his brother and built’
BES24 3.1: l qymt bn trṣ w wgm w bny
‘By Qymt son of Trṣ and he grieved and built’
BES24 3.2: l ʿhn bn qsy w bny
‘By Hnʾ son of Qsy and he built’
BES24 3.3: l ḏbb bn ʾṣlḥ w bny
‘By Ḏbb son of ʾṣlḥ and he built’
BES24 4: l ʿqrb bn ḥgg w h-ṯyt
‘By ʿqrb son of Ḥgg and the Ṯyt-memorial [is his]’31
BES24 5: l ḥrb bn ... [w][w][g]m w bny
‘By Ḥrb son of …. [and he grieved] and built’
BES24 6: l nyr bn tmlh w wgm w bny
‘By Nyr son of Tmlh and he grieved and built’
BES24 7.1: l whb bn hnʾ w bny
‘By Whb son of Hnʾ and he built’

29 Variant 3 of the Safaitic script is the neutral nomenclature used on OCIANA

for what J. Norris has described as the Safaitic-Hismaic script (Norris, 2018).
30 This is a nominal form derived from the root ṯ-w-y ‘to rest’, ‘settle down’, but

also gives rise to meanings to do with death and burial, Classical Arabic ṯawwà /
ṯawiyyatun / taṯwiyatun ‘he was slain and remained where he was; he remained in
his grave’ and ṯuwiya ‘he was buried’. (Lane, 366a). Another ṯyt installation of a com-
parable composition, also with numerous building and commemorative inscriptions
arranged in a circuit, has been discovered in 2019 by the present author and was
excavated during the 2024 campaign. The report is currently in preparation. The
present installation type is, therefore, comparatively rare. On previously described
Safaitic mortuary structures, see Kennedy (2012); Al-Jallad (2022, 26–40, 78–81).
31 An identical formula is attested in IMA.Saf 1.

15
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

Image 4: The enclosure with the inscription panels marked (© BES24)

Image 5: BES24 4 detail (© BES24)

BES24 7.2: l hnʾ bn ḥgg w wgm w bny

‘By Hnʾ son of Ḥgg and he grieved and built’
BES24 7.3: l sbq bn sʿd w bny
‘By Sbq son of Sʿd and he built’
BES24 8.1: l ʿdy bn tmlh w bny
‘By ʿḏy son of Tmlh and he built’
BES24 8.2: l ʿbdmk bn rdfn w bny
‘By ʿbdmk son of Rdfn and he built’

16
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

BES24 9: l mqtl bn ʿrhz w bny

‘By Mqtl son of ʿrhz and he built’
BES24 10: l gdy bn zbln w bny
‘By Gdy/ʿdy son of Zbln and he built’
BES24 12: l ʾs bn ʿmr w wgm w bny

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

‘By ʾs son of ʿmr and he grieved and built’
In addition to the texts carved as part of the building inscription, the
same stones were inscribed by later passersby. Some wrote typical
Safaitic inscriptions containing prayers for wellbeing during their desert
enterprises while others, perhaps moved by the memorial installation,
took the opportunity to mourn their own loved ones. Given that the
secondary texts outnumber the original carvings, precluding individual
analysis of each, the ones from BES24 3 will serve to illustrate this phe-
nomenon. Almost all of the secondary inscriptions are carved in differ-
ent script variants. BES24 3.5, for example, is carved in Variant 2, which
is the most common variant represented in this category.32 This variation
indicates that the secondary texts were carved by later passersby and had
nothing to do with the creation of the installation or its burial party.

Image 6: BES24 3 detail (© BES24)

32 Variant 2 is the neutral term for what Clark (1979) first described as the Fine
Script. Della Puppa (2022, ch. 4) provides a detailed paleographic overview of the
script type and demonstrates that it is most frequently used by the tribe of Ḍayf.

17
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

BES24 3.4:33 l bnḥr bn ḫlṣt bn ʾys ḏ ʾl [ḍ]{f} <<n>> [w] wgm

ʿl-kzn w ʿl-glḥn w ʿl-srʿt w ʿl-ġṯ w ʿl-ḫlṣt w rʿy {f} {h}
lt ġnyt
‘By Bnḥr son of Ḫlṣt son of ʾys of the lineage of [Ḍf]
[and] he grieved for Kzn and for Glḥn and for Srʿt

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

and for Ġṯ and for Ḫlṣt and he pastured so, O Allāt,
let there be abundance’
BES24 3.5: ---- bn whblh bn gls ḏ ʾl ḍf w rʿy mdbr w ---- ʿl-fnyt f h
lt slm
‘---- son of Whblh son of Gls of the lineage of
Ḍf and he pastured in the inner desert ---- the
edge of Fnyt (toponym) so O, Allāt, may he be
secure’

About 15 meters to the north of the enclosure, a small boulder com-

memorates two men killed by the enemy Ḥawīlat tribe. Their territory
spanned from Ḥegrā to Taymāʾ and the tribe frequently appear as adver-
saries of both the Nabataeans and some of the nomadic tribes of the
Ḥarrah in the Safaitic inscriptions.34

Image 7.1: BES24 11 (recto) (© BES24)

33 Secondary inscription carved in between the letters of the building texts.

34 On these conflicts, see Norris and al-Manaser (2018).

18
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

Image 7.2: BES24 11 (© BES24)

BES24 11 (recto): l ḥmy bn mʿn ḏ ʾl ʿmrt w wgm ʿl- rḍwt ʾḫ-h w ʿl-ʿqrb
dd-h mqtln
BES24 11 (verso): ʾl ḥwlt f h lt ṯʾ{{r}}
‘By Ḥmy son of Mʿn of the lineage of ʿmrt and he
grieved for Rḍwt, his brother, and for ʿqrb, his paternal
uncle, both killed by the tribe of Ḥwlt so, O Allāt, may
he have vengeance’
The text commemorates the killing of a man called ʿqrb, the paternal
uncle of the inscriber, and the inscriber’s brother, rḍwt. Although his
father’s name is not given, it is likely that this ʿqrb is one and the same
as the man commemorated by the ṯyt construction; this hypothesis is
bolstered by the fact that this text is carved in the same script variant
as the primary texts of the ṯyt construction. The circumstances sur-
rounding the carving of BES24 11 are unclear. Ḥmy son of Mʿn was
not mentioned among the men who constructed the ṯyt, so it is possible
that he came to this area at a later point, found the funerary installa-
tion of his uncle, and then carved this text in his memory. He also
took the opportunity to commemorate his brother who was killed by
the same tribe, although it is not mentioned if they were killed during
the same conflict.
It is clear that this prominent point in the landscape, and the struc-
ture itself, attracted inscribers, many of whom used the installation to
memorialize their own dead. Let us now return to the ṯyt. Two Safaitic
inscriptions are carved onto a rather rough stone at the enclosure’s
southwestern corner, facing slightly outward rather than directly into
19
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

the open area. The stone was not originally part of the inscribed por-
tion of the ṯyt and is separated from the inscribed sequence of stones
by a rough, uninscribed stone to its east. Behind both of these stones is
BES24 9, which does contain a building inscription. The stone is not
well positioned and seems to have been moved to this place. Unlike the

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

memorial inscriptions, the texts carved on BES24 1 are lightly incised
in the variant 2 script, indicating that these belong to the secondary
inscriptions category and were produced after the construction of the
installation.

Image 8: BES24 1 (© BES24)

The first of these, positioned at the top of the panel, is a simple prayer.
BES24 1.1: l ṣʿd bn ʾdm bn ṣʿd w h śʿhqm rwḥ m wqr
‘By Ṣʿd son of ʾdm son of Ṣʿd and O Shayʿ haq-Qawm send
relief to whosoever carves (?)’
Prayers for rwḥ are extremely common in Safaitic, but none have been
attested in this exact wording. The term wqr occurs for the first time
here in Safaitic although it is frequent in the Himaitic inscriptions.35
The meaning at present is suggested based on the Himaitic unders-
tanding, but it is possible that the term signified something else.36
35 See Prioletta (2018) on the verb wqr and its derivatives in Himaitic.
36 The term wqr in Classical Arabic can refer to ‘a load’ (burden) or ‘deafness’
(Lane, 2960), both of which are reasons to ask for relief.

20
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

The second inscription is our primary concern, and our key to advanc-
ing our understanding of pre-Islamic Allāh.

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

Image 9: Tracing of BES24 1.2

I will present the reading first, then provide a grammatical commentary

on each section in what follows, and conclude with an attempt to inter-
pret the unformulaic text.
BES24 1.2: l ʾws bn ʾdm bn ṣʿd w tʾmr rdf nfs wdd ʿṯrt f h lh nr w ḥmm{ġ}
my[[]]bnyt-h w whb-nh ht ḏ ḫlq

2.1 Genealogy
ʾws bn ʾdm bn ṣʿd
This three-generation genealogy repeats in ten inscriptions; he is
the brother of the author or BES24 1.1. Seven of these appear to be
authored by the man himself,37 which is an extremely rare occurrence
in the Safaitic corpus, and three texts were carved by the sons of ʾws
/ʾOways/, SIJ 152 and ASWS 185 by ʿdn and AAWAB 56 by rgl. One

37 See: SIJ 2, KRS 338, GS 65, C 4490, AWS 246, AbaNS 283, AAWHA 107;

Map 2 shows their distribution.

21
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

of the inscriptions of ʿdn traces the genealogy back four further genera-
tions beyond ṣʿd, terminating with an ancestor called ysmʿl */yesmaʿʿel/.38
ASWS 18539 l ʿdn bn ʾws bn ʾdm bn ṣʿd bn ʿlyn bn mrwn bn s¹ʿd bn
ys¹mʿl w rʿy h-ḍʾn f hy lt slm w tẓr h-smy f h bʿlsmn rwḥ
b-mṭr w nqʾt b-wdd ḏ ʿwr mʿl- ḥwq

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

‘By ʿdn son of ʾws son of ʾdm son of Ṣʿd son of ʿlyn son of
Mrwn son of Sʿd son of Ysmʿl and he pastured the sheep
so, O Allāt, may he be secure; and he awaited the rains,
so O Baʿal-Samīn, send the winds with rain, and may he
who would efface (this writing) out of jealousy be thrown
out (of the grave) by a loved one’.

2.2 Narrative
w tʾmr rdf nfs wdd ʿṯrt
The interpretation of the narrative is extremely challenging. All but
one of the terms has previously appeared in the corpus, but not in this
exact combination.

tʾmr to fear, to watch out for, to be manifest (?)

The exact meaning reconstructed for this Safaitic term is not found in
the Classical Arabic lexica, but its context in the inscriptions strongly
supports understanding it along the lines of ‘looking out for’ and by
semantic extension ‘fear’ and ‘anticipation’ cf. Gəʿəz taʾamra ‘to show
one’s self, be manifest’, Ugaritic, ʾmr ‘see’, Akkadian amāru ‘see’. The
meaning I am suggesting here, based on new inscriptions and the one
under discussion, replaces Winnett and Reed’s (1978) translation ‘to
be made ʾamīr’, and improves upon the meaning ‘to be manifest, wide-
spread’, found in the SafDict.40 While tʾmr almost always takes abstract
nouns as objects, the primitive sense of ‘to look out for’ is supported
by the two occurrences in which it takes the substantive ʾḫ (lit. brother)
as an object and one case, in an unpublished inscription, where it takes
a proper name as an object as well. However, to fully understand the
development of the semantics of this verb, let us look to another verb
of ‘seeing’ that frequently takes brother(s) as an object, ḫrṣ.

38 This name is rather common in the Safaitic corpus, attested some 199 times.
Variants of it include ysmʿʾl and ʾsmʿl. On the final spelling, see Al-Jallad (2025).
39 The OCIANA 1 edition left mʿl ḥwq untranslated. For the present meaning,

see SafDict, 87-88.

40 This meaning was first suggested in Al-Jallad (2015, 300); see also SafDict

(130–131).

22
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

As I have argued before, the primitive meaning of ḫrṣ was ‘to look,
keep watch’, which then developed metaphorically to ‘anticipate’, and
then finally to ‘guess, estimate, speculate’, which is the sense attested in
the Qurʾān.41 The visual sense of keeping watch is highlighted by the
following prayer in SIT 1: w h lt slm l-ḏ ḫrṣ w ʾbn ‘O Allāt, security be

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

to him who keeps watch and make manifest (that which he is looking
out for)’. The verb roughly overlaps with nẓr as well, C 892 ḫrṣ h-s¹my
‘he kept watch for the rains’.
This etymological interpretation of ḫrṣ means that it shares with
tʾmr its core semantics. If we compare the contextual overlap of these
two verbs, a remarkable pattern emerges—ḫrṣ and tʾmr have a nearly
identical distribution.

ḫrṣ ‘to anticipate’ (= to fear) tʾmr ‘to fear’

śḥṣ ‘scarcity’ AbWH 1: ḫrṣ h-śḥṣ f h śʿhqm WH 1001: tʾmr h-śḥṣ f h lt ġnyt
ġnyt ‘he anticipated scarcity so, ‘he feared scarcity so, O Allāt, let
O Śayʿ-haq-Qawm, let there be there be abundance’
abundance’
śnʾ ‘enmity’ C 4261: ḫrṣ h-śnʾ f h lt slm ‘he Al-Mafraq Museum 1: tʾmr h-śnʾ
anticipated enmity/enemies so, O snt qṣr w h-mḏy f h lt w gdḍf slm
Allāt, may he be secure’ ‘he feared enmity/enemies, the
year of Caesar and the Persians,
so, O Allāt and Gadd-Ḍayf, may
he be secure’
wḥd C 3258: ḫrṣ h-wḥd f śʿnr {l-}{h} RWQ 333: tʾmr h-wḥd f h gdḍf
‘loneliness’ s²hrt ‘he anticipated loneliness ‘he feared loneliness, so, O Gadd-
so, O Śʿnʿr, grant him compan- Ḍayf, may he be secure’
ionship (?) (lit. conspicuousness,
fame)’

Table 1: A comparison of the contexts of ḫrṣ and tʾmr

The verb tʾmr also occurs in the following contexts:

RWQ 340: tʾmr h-mśʾt f h lt ʾqdn ‘he feared want so, O Allāt, let there
be abundance’
AMSI 31: tʾmr h-wrdt f bʾlsmn ġyrt ‘he feared the need to go to water
(drought), so, O Baʿal-Samīn (the storm god), let there be
abundance’

41 This etymology was first proposed in Al-Jallad (2015, s.v.), see also SafDict,

82; see Q 6: 116, 148, etc.

23
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

Ḫrṣ also takes abstract objects that have not yet been attested with tʾmr:
C 4430 records w ḫrṣ h-mḥl ‘he feared the dearth of pasture’, which is
in a sense comparable to śḥṣ. What is important here is that other verbs
of keeping watch, like nẓr and tẓr, do not seem to take abstract nouns
as objects; there are no cases of nẓr h-śḥṣ, h-wḥḍ, h-mḥl, etc. So while

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

ḫrṣ can have the sense of ‘keeping watch’, as a concrete activity overlap-
ping with nẓr, it seems also to have developed a specific semantic exten-
sion meaning ‘to anticipate’ or more precisely ‘to fear’ when used in a
negative context.
The distribution then suggests that tʾmr could be a near synonym,
developing along the same semantic lines: ‘to see, look out for’ > ‘to
anticipate (something negative)’ > ‘to worry/fear’. While one cannot
disprove the ‘manifest’ meaning, which works well in other circum-
stances, the parallels with ḫrṣ makes the present understanding more
economical and therefore preferable. The proposed semantic scope of
these verbs is underscored in constructions where they take ʾḫ as an
object.

ḫrṣ tʾmr
AMSI 13: ḫrṣ ʾḫ-h m-mdbr f h gdḍf w h BES17 1326: w tʾmr ʾḫ-h mśrq mdbr f h
lt w h dśr w h śʿhqm qbll slm gdḍf slm
‘he kept watch for his kinsman (lit. ‘and he looked out for his kinsman,
brother) from the inner desert, so, O migrating towards the inner desert, so,
Gadd-Ḍayf and O Allāt and O Diśar O Gadd-Ḍayf, may he be secure’
and O Śayʿ-ha-Qawm, may there be a
secure reunion’
KRS 29: w ḫrṣ ḏ bʿd m- ʾs²yʿ f h s²ʿhq[m] KRS 27: w tʾmr ʾḫ-h bʿd f qṣf f h lt qbll
w ds²[r] w ʿr qbll
‘and he kept watch for those who were ‘and he looked out for his kinsman, who
far from among companions so, O Śayʿ- was far, and was sad so, O Allāt, may
ha-Qawm, and Diśar, and ʿr, may there there be a reunion’
be a reunion’

Table 2: ḫrṣ and tʾmr with kinsman (lit. brother)

The primary sense of ‘to look out for’, seems to be supported also by
the prayer in the following unpublished inscription:
AAWAB 47:42 l mnʿm w rʿy h-mʿzy w ʾśrq {h-}mdbr f h lt slm w ġnmt l-ḏ
dʿy w ʿ[w][r][m][ʿ][w]{r} h-sfr w tʾmr ḥnn wdt-h ḫld

42 There are two new terms in this text that require a short discussion. The term

wdd ‘beloved’ is frequent in Safaitic, a synonym of ḥbb. I would suggest that the word

24
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

‘By Mnʿm and he pastured the goats and set off for the inner
desert so, O Allāt, may he be secure and spoil be to him who
would read (this) and [blind whosoever effaces] this writing;
and he looked out for ḥnn, his beloved forever’

rdf to follow

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

This verb is previously attested in Safaitic, where it has the simple
meaning of ‘to follow’, when used in a concrete pastoral context: w rdf
h-ḍʾn ‘and he followed after the sheep’ (ASWS 127); w ḫrṣ ʿl-ʾḫ-h rdf
ʾbl-h ‘he worried about his brother who was following after his camels’
(HCH 102).43 The term occurs once as an adjective in a military con-
text, gś rdf ‘rear guard’ (LP 146).
But perhaps the most informative attestation for our present context
is found in LP 679, where the author uses the verb to describe what
appears to be a funerary procession.
LP 679: l ḥny bn sḫr bn ʿbd bn ʾdm bn msk bn srb w tʾs rdf bn ʾḫ-h qtl
ʿrḍt f h lt ṯʾr m ʾslf ----
‘By Ḥny son of Sḫr son of ʿbd son of ʾdm son of Msk son of
Srb and he despaired during the procession of his nephew,
who was killed by/in a troop so, O Allāt, those who remain
will have vengeance’
If we take rdf literally, we produce ‘he despaired following after his
nephew…’. This of course does not make sense in terms of following
his tracks, as his nephew had died, nor does it suggest that the intended
meaning was that the author died as well, as he demands vengeance
later in the text. Thus, I would take the term in this context as an
abstract noun meaning: the following of the corpse of the dead man
being taken to his grave, that is, a funerary procession. This process
probably saw the dead man carried on a bier to his resting place.44

wdt is a feminine form, cognate with Classical Arabic waddun ‘a friend, beloved
person’ (Lane, 2931a). The term ḫld should be understood as the cognate of Clas-
sical Arabic ḫuldun (Lane, 783c-784a) ‘eternity’, ‘longevity’, here to be taken as
an adverb, paralleling the similar construction, ḥbb-h l-ʾbd ‘his beloved forever’
(WH 2051).
43 Given the previous discussion on the precise meanings of ḫrṣ, I would suggest

that when it takes an object introduced by ʿl- it should be understood along the
lines of ‘to worry about s.o.’.
44 For evidence of this, see Al-Jallad (2022, 35), where a dead man is carried

deep into the Ḥarrah on a bier from Ḥawrān for burial. While rdf occurs as a per-
sonal name, verbs of despair always introduce their object, the person for whom the
author grieves, with the prepositions, ʿl- or rarely m(n)-. The absence of a prepo-
sition here requires us to understand rdf differently. In Rabbinic Hebrew, rdp can

25
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

Although rdf is attested as a personal name in the Safaitic onomas-

ticon, the site contains no commemorative structures or secondary
inscriptions on ʿqrb’s mortuary installation mentioning anyone by that
name. Furthermore, our survey of the entire hill revealed no inscrip-
tions by or about an individual named Rdf.

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

nfs wdd beloved
This expression is previously attested in C 2031, where it is best under-
stood to refer to ‘one who is beloved’, as it patterns in the same for-
mulaic context with wdd ‘beloved’, ḥbb ‘ ‘beloved’, and ṣdq ‘friend,
confidant’.45 The term nfs should be compared with Classical Arabic
nafsun, meaning ‘the very thing’ or ‘self’, a sense already grammati-
calized in Safaitic (SafDict, 103). The literal sense is probably the life-
force of a beloved, referring to intentionality in C 2031, but here to the
deceased. It is preferable to take both terms as plurals, as we shall see
below, probably /nofūs wedād/ ‘lifeforces of the ones beloved’, but a
singular reading is also permissible grammatically.
It is not possible to interpret nfs as an inscribed funerary stela,46 as
nothing of this sort was discovered at the site. Moreover, such stelae
have only been discovered in settled areas and not in the desert. The
nomads seem to have sometimes commemorated their dead with an
architectural feature called a nfst (pl. ʾ(n)fs), the exact identification of
which is unclear (Al-Jallad, 2022, 33–37). Finally, it would be most
odd to mention a funerary monument in an anonymous manner such
as this.

ʿṯrt fallen
The word is attested for the first time in this inscription. The basic
meaning of the G-stem in Classical Arabic is to fall, ʿaṯira, ʿaṯura ‘he fell
upon his face’ or simply ‘he fell’ (Lane, 1952a). The adjective ʿaṯīrun
is also recorded; the term attested in the present inscription is likely
the feminine equivalent of that, with nfs as its antecedent, meaning

mean ‘to yearn for’ (Jastrow 1903, 1453; I thank G. Khan for this reference), which
would be more appropriate in the present context, paralleling the common verb tśwq.
However, in this case we would expect a prepositional complement. Moreover, verbs
of yearning take either divinities or living people as their objects, while the subject
matter of the present inscription deals with the dead.
45 The term is attested in the common curse requesting that an effacer be ‘thrown

out (of the grave)’ nqʾt by a beloved person, ḥbb (AMSI 10) and wdd (22 times in
OCIANA, e.g. AAWAB 1).
46 For an example of inscribed Safaitic funerary stelae, see Hayajneh (2017).

26
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

simply ‘fallen’, a euphemism for ‘death’. The term parallels the much
more common rġm, meaning ‘struck down’.47 Again, this is unlikely
to refer to a fallen funerary monument, for the reasons mentioned in
the paragraph above, and also that the other elements in the clause
would not be compatible with such an understanding.

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

The task now is to attempt an interpretation of this unique nar-
rative. The verb tʾmr offers us two meanings, ‘to fear’ and ‘to look out
for’, where the former meaning takes abstract objects while the latter
refers to persons. If we wish to understand rdf as a person, then we
might produce something along the lines of ‘and he looked out for rdf,
a beloved, who fell’. While grammatical, the sentence is not meaning-
ful as it is unclear what sense looking out for a fallen, i.e. dead, person
would have. Indeed, when tʾmr is used with this meaning, the persons
are always living, as they are engaged in activities like migration and
the prayer that follows is for ‘reunion’. The same applies to ḫrṣ. For the
same reason, understanding rdf in the literal sense of following after
the beloved is incongruent as its object is described as fallen.
We should then turn to the understanding of ‘to fear’, which prefers
an abstract complement. This leads us to interpret rdf as a noun mean-
ing ‘following’, not in the sense of tracking someone, but rather similar
to the interpretation suggested for LP 679, referring to the procession
of a deceased person. This reading is further reinforced by the follow-
ing phase which describes ‘fallen loved ones’, producing ‘the procession
of fallen loved ones’. Putting these elements together, we produce the
translation: and he feared the procession of fallen loved ones. Now, the
true sense of this phrase will only become clear as we move onto the
prayer. But it is important here to establish that the narrative does not
refer to any specific person: 1) tʾmr refers to events that have not yet
come to pass. It is an expression of fear and anxiety about the future,
such as the possibility of enemies, drought, solitude, as evident by the
apotropaic prayers that follow; 2) This text was not produced by some-
one who participated in the burial of ʿqrb and the construction of his
ṯyt so it cannot be construed as a building inscription; 3) no specific
dead man is named; 4) the inscription does not record grieving for any
individual. There is no other cluster of mourning inscriptions at the
site besides that of ʿqrb. Rather, taking into account the physical con-
text of the inscription, the narrative should be regarded as a reflection
on the site itself: a mortuary installation established to commemorate
the murder of ʿqrb and subsequently marked by passersby expressing

47 See Al-Jallad (2022, 73–75) on this term and its connection to funerary inscrip-

tions and the commemoration of the dead.

27
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

their sorrow for lost loved ones. As such, the narrative is a poetic
expression of the fear of death. This is not a unique example of such a
phenomenon, as we shall see later.

2.3 The Prayer

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

f h lh nr w ḥmm {ġ}my[[]]bnyt-h w whb-nh ht ḏ ḫlq

f h lh So O Allāh
As discussed previously, this is the expected spelling of the divine name
allāh after the vocative in Safaitic orthography, indicating the pronun-
ciation /hā llāh/.

nr light
While the n and r are clearly legible, there is a curvilinear scratch between
them, with similar, yet not identical, patina; however, it does not form
a letter. The noun nr is attested for the first time here and is undoubt-
edly the cognate of Classical Arabic nūrun ‘light’. It may either be con-
strued as an unmarked existential clause, ‘let there be light’, or the
object of an implied verb, ‘(shine) light’. Safaitic prayers are usually
tangible and concrete, invoking the gods for spoil, vengeance, security,
abundance, rain, etc. This is the first instance in which a deity is invoked
to perform an abstract, supernatural act.

w and/when
The conjunction can function simply to link two clauses together, or
it can introduce a temporal clause, the so-called coordinate circum-
stantial construction (Fisher, 2012, 209–211).

2.3.1 Reading 1
ḥmm darkness
The basic range of this verb encompasses both ‘heating’ and ‘to blacken’
or ‘darken’, with the latter set of meanings deriving from the former.
Ḥamamun is a word meaning ‘black’ and ḥumamun can refer to charcoal
or the darkness of complexion (Lane, 635a).48 Treating this word as a
noun suggests that the clause follows an SVO (Subject-Verb-Object)
word order, which is rather rare in Safaitic, but can be used to signal

48 The root meaning of this word in Semitic languages is ‘to be hot’ or ‘heat’,

see CDG 233, for cognates.

28
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

topicalization or focus (Al-Jallad, 2015, 181). The same parsing can

also produce an a D-stem verb /ḥammama/. While its general range of
meanings in Arabic lexica relates to making something dark or black,
one specific meaning is noteworthy: ‘to give a wife a gift after divorce’
(Lane, 635c). This specific sense no doubt reflects a process of semantic

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

narrowing from a more general verb of ‘gifting’ or ‘offering’. I would
suggest that such a meaning ultimately derives from the use of the verb
to denote a burnt offering, which is then semantically bleached to
mean ‘offering’ in general, and then finally narrowed to the specific
offering presented after a divorce in Classical Arabic.

2.3.2 Reading 2
ḥm m- protect from
If we choose to parse the sequence as ḥm m-, we produce an imperative
verb from the root ḥmy ‘to protect’, /ḥVm/ or /eḥm/, followed by the
preposition */men/ ‘from’, with the assimilation of the n. This reading
has consequences on the grammatical identity of the following word.
The defective spelling m can also represent the indefinite relative par-
ticle, */man/.

{ġ}my to cover / darkness

The verb ġamà in Classical Arabic means ‘to cover’, and can be used to
mean ‘a roof’ or as a verb ‘to roof’, in addition to signifying the cover-
ing of celestial bodies at night to produce darkness, ʾuġmiyat laylatu-nā
‘our night was one whereof the new moon was veiled, or concealed’ or
to the covering of the sky during the day by clouds, ʾuġmiya yawmu-nā
(Lane, 2298c). In Aramaic, ʿəmā has a similar range of meaning: ‘to
grow dim, faint’, ‘to die out (of ember), to fade away’(CAL, s.v.). As a
noun, the meaning ‘a cover’ in a concrete sense is obvious and in a
more abstract way, ‘darkness’ of colour is attested in Arabic.
The area on the stone following the y is scratched in such a way that
it creates an illusion of a stroke, slightly curved at the top with a rather
imperfect fork at the bottom. This could produce a malformed <h>,
but certainly not an <l>. It clearly stands out from the other glyphs, dif-
fering in shape, stroke weight, and patina. These facts indicate that the
feature is not a deliberate part of the text. The most likely interpreta-
tion is that the author mistakenly carved a stroke rather than a b and
attempted to fix it, thus leading to its different constitution and the
slight curvature at the top and an attempt to curve it at the bottom.
Realizing that it could not be salvaged, he moved on to carve the b cor-
rectly next to it, almost contiguous with the erroneous stroke.
29
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

Image 10: Erroneous stroke

bnyt-h his construction

The term derives from the basic root bny ‘to build’, which is abun-
dantly attested in Safaitic. This specific sequence appears only twice
elsewhere. The first instance is in AWS 215, where it has previously
been translated as ‘little daughter’: wgd ʾṯr ḥrb f ḥdṯ h-mẓl l-bnyt-h
‘he found the trace of Ḥrb and then renewed the shelter for his little
daughter’.49 The second occurrence is in KRS 1515: l slḥ {{h-}}bnyt.
While its context is unclear, the formulaic structure of the inscriptions
supports interpreting the term as referring to a structure, ‘this con-
struction belongs to Slḥ’. The latter example makes it more likely that
the term in AWS 215 should also be understood as ‘construction’,
implying that the inscription should be reparsed as: ḥdṯ h-mẓll bnyt-h ‘he
renovated the mẓll, his bnyt (lit. construction)’. Now, when we exam-
ine the inscription in its totality, the sense of bnyt in a funerary context
reveals itself.
AWS 215: l mġny bn zbʾ bn zbʾl bn ḥny bn ḥḍg w wgd ʾṯr ḥrb f ḥdṯ
h-mẓll bnyt-h
‘By Mġny son of Zbʾ son of Zbʾl son of Ḥny son of Ḥdg
and he found the trace of Ḥrb and renewed the mẓll, his
construction’

49 This is the translation presented in OCIANA 1 and has been corrected in

OCIANA 2.

30
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

The ‘trace’ of Ḥrb to which AWS 215 refers to is the inscription

AWS 167, which states l ḥrb bn dʾyt bn ḍhd and AWS 166 l brd bn
dʾyt bn ḍhd w wgm ʿl-ḥrb ‘By Brd son of Dʾyt son of Ḍhd and he
grieved for Ḥrb’. These inscriptions were documented at a cairn, which
is clearly funerary, and so AWS 167 should be interpreted as the grave

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

marker and translated as ‘For Ḥrb son of Dʾyt son of Ḍhd’, expressing
ownership of the burial cairn. Thus, the ‘renewal’ of the mẓll must
therefore refer to the renovation of the burial, which is then described
as ‘his (i.e. the dead man’s) construction’, i.e. his grave complex.
The interpretation of mẓll as a burial makes good etymological
sense as the verbal root ẓll means generally ‘to cover, to shade’: Ara-
bic ẓallala/ʾaẓalla ‘to give shade’ along with ẓullatun / miẓallatun ‘a
thing that covers or protects one, overhead; anything that protects
and shades one’ (Lane, 1915–6); Aramaic ṭll ‘to roof over, protect,
cover’ (CAL, s.v.). The term mẓl(l) is previously attested in Safaitic
and was translated faute de mieux as ‘shelter’ in the Safaitic Dictionary
(SafDict, 144). However, this new context suggests that the term refers
specifically to a burial, perhaps the burial pit inside the cairn which
is covered by stones and earth. A related term ẓlt / ẓllt is attested in
the corpus as well, but it is impossible to define precisely its meaning.
Constructions such as that found in CEDS 398: l nr bn tm ḏ ʾl ʿmrt w
l-h h-ẓlt follow the pattern of funerary inscriptions syntactically but
give us no information on the function of this structure. In fact, one
inscription, C 3221, only known from a hand copy, would suggest that
the man associated with the ẓlt was still alive as the text then records
wrd h-nḫl ‘going to water in the valley’. It is, therefore, possible that
ẓlt refers to an actual shelter one can use to shield themselves from
the elements while mẓl(l) would refer to an architectural feature of the
burial. However, strong conclusions should be avoided when the only
data come from poorly copied texts over a century old. The burial mẓll
is likely a locative noun to be vocalized as /maṯlal/, translating literally
as ‘place of darkness’.
The sequence m ẓ l l is attested in a recurring mourning expres-
sion, bʾsmẓll. The first three glyphs form the noun ‘misery’, ‘evil’, or
the stative verb of the same meaning. However, the parsing of the
remaining glyphs remains debated. It was common to understand them
as forming a single word, a participle mẓll meaning ‘overshadowing’
or ‘overshadowed’ with bʾs ‘misery’ as the agent. But the attestation
of the variant bʾs mn ẓll in AbSWS 79 would seem to indicate that
the m is not a performative prefix of a passive participle but rather the
indefinite pronoun, ‘whosoever’. A closer inspection of the photograph
of this inscription, which was made available to me recently, fails to
31
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

substantiate the existence of an n between the m and ẓ. And given that

this would be the only attestation of the n-variant, I believe we may
return to the reading m ẓ l l. We will leave the discussion of this expres-
sion here and return to it once we have treated in full BES24 1.2,
which will reveal its correct meaning.

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

Safaitic bnyt is no doubt the equivalent of the Classical Arabic bun-
yatun, binyatun, which can refer widely to ‘a thing that one has built,
framed, or constructed’, synonyms of bināʾun (Lane, 262–263). The
latter is used in the Qurʾān to describe the vault of heaven, Q 2:20:
allaḏī ǧaʿala lakumu l-ʾarḍa firāšan wa-s-samāʾa bināʾan ‘the one who
made the earth a bed and the sky a raised construction’. Ugaritic attests
bnwt, which finds a cognate in Akkadian binûtu. The term refers to
‘creatures’, the living beings created by the god ʾilu (DULAT, 233).

w whb-nh ‘and may he grant him’

This prayer for a boon is best understood as referring back to the
‘light’ in the initial clause. The verb takes a pronominal suffix signify-
ing the indirect object for the first time in the Safaitic corpus; usually
the indirect object is introduced by the preposition l-, and this is the
only construction known in the Qurʾān as well. However, prayers such
as hab-nī baytā ‘give me a house’ and hab-nī ʿilmā ‘grant me knowl-
edge’ are attested in Islamic-period Arabic and in Christian hymns as
well, e.g. hab-nī ḥubbā ‘give me love’.50 The same syntax is attested in
the idiom wahabat-hu min ḏāti nafsi-hā ‘she gave herself unreservedly
to him’ (Lahlali and Islam, 2023, 346). Although infrequent, this con-
struction is sufficiently attested in Arabic to support the interpretation
of the pronominal suffix as an indirect object. Perhaps it is significant
that the 3MS object pronoun takes the n augment, which has been
observed in a number of other inscriptions, but its distribution is not
yet understood (Al-Jallad, 2015, 97–98; 2019, 348). Its grammatical
identity is unclear. While the normal imperative of the verb whb is hb,
clear examples of the imperative with a w, perhaps reflecting an inten-
sive D-stem, are attested:
KRS 957: h yṯʿ whb l-sʿd bn ghmn nʿm
‘O Yṯʿ, grant favor to Sʿd son of Ghmn’
Reading the present attestation as an optative suffix conjugation is also
possible. Alternatively, one can understand it as a nominal form whbn,
meaning something like ‘a grant’ or ‘bestowment’.
50 This is the title of the Coptic hymn tarnīmatu ‘hab-nī ḥubbā’. For the full

lyrics of the song see St. Takla Haymanout Coptic Orthodox Church (2025).

32
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

{h}t this
The interpretation of this sequence is challenging. Firstly, there is what
appears to be a misplaced stroke at the top of the letter that is similar,
yet not identical, in patina and stroke weight to the glyph itself that
almost closes the fork. If read as part of the letter, it would produce the

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

word yt, which would be identical to the Aramaic direct object marker;
however, this would make no sense in the present linguistic context.
Moreover, a close inspection of the glyph in comparison to the other
y’s carved in the inscription makes such a reading extremely unlikely.
Unlike the clear y glyphs, which are formed by distinct triangles, this
stroke is curved and does not close the fork. It fails to make contact
with the left arm of the fork and, instead of forming an angle with the
right one, it crosses it roughly at the midpoint. Moreover, the right
arm cuts across the neck of the glyph, a variant feature of the h but not
the y in this hand. It is therefore clear that we are not dealing with an
intentional segment but more likely a chip on the stone or a secondary
addition.51

Image 11: A comparison of the h and y glyphs, with the {h}t centred

I will give a couple of options ranked by likelihood based on parallels

in other texts and Safaitic syntax more generally.

Demonstrative
It is possible to interpret this word as a feminine demonstrative cog-
nate with pre-Classical Arabic hātī. The bare form t is previously
attested in Safaitic, and the addition of the h prefix to it is unremark-
able (Al-Jallad, 2019, 349). If this is correct, then the demonstrative
would refer back to ‘light’, which is feminine.

51 See Al-Jallad (2018) for a discussion on damage altering the shape of glyphs.

33
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

Imperative
One may also consider reading ht as the suppletive imperative verb,
hāt ‘give’. This interpretation, however, should give us pause as we
would expect a conjunction between it and the preceding clause, espe-
cially if we are to assume yet another switch of persons from third to

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

second.52

ḏ ḫlq he who/that which he has created

While relative pronoun ḏ /ḏū/ is regular in Safaitic, the verb or infini-
tive ḫlq following it is met with for the first time here. Nouns derived
from this root exist in Safaitic and appear to denote ‘portion’ or ‘divi-
sion’, often in reference to periods of time (SafDict, 80). The meaning
of the Arabic term ḥlyqh in the Nahal Hever papyri is debated, but
Yardeni suggests ‘custom’ or ‘nature’ (Yardeni, 2014, 308). Cognates in
other West Semitic languages relate to splitting, enumerating, portion-
ing, or dividing. Perhaps relevant is the attestation of the verb ḫlq in
the Northern Middle Sabaic inscription Najrān 1, which Robin et al.
understand as ‘to offer compensation’ (2014, 1073–1077). Classical
Arabic uses ḫalaqa as the unmarked verb of creation, but it still retains
meanings of ‘to measure’ and ‘to proportion’. Kogan reconstructs the
original West Semitic meaning of this verb as ‘to divide’ or ‘to measure’
(2015, 118). As I have suggested earlier in this essay, the Arabic mean-
ing of creation linked to ḫalaqa likely arose metaphorically, originat-
ing from the ancient Near Eastern cosmological myth where creation is
understood as a primordial act of division. If we understand this phrase
as an epithet, its placement at the end of the clause may serve to focus
on the act of granting the author his request. Ḏ-clauses are not infre-
quently right-dislocated in the inscriptions.53

2.4 Discussion
I will now attempt the translation of Section 2. But before approach-
ing this complicated text, a few words on the methodology of inter-
preting Safaitic inscriptions are in order here. Firstly, these texts form
52 A connection with htʾ ‘moment’, ‘portion’ (Hava, 1982, 806a) is difficult as the
glottal stop is stable in Safaitic; the primary sense of this root, however, is ‘to be
crooked’ or ‘to become ragged’. The lexicographers record a similar set of meanings
to the root htt, but this would appear to be a biform of the former root derived from
dialects that have lost the glottal stop.
53 For example, see BES15 161 (Al-Jallad, 2025) where the author displaces the

ḏ clauses introducing his lineage to the end of the inscription following the narrative.

34
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

coherent wholes. In former times, some interpreted individual clauses

as completely independent compositional units, where the prayer had
no relation to the narrative, and the linking of narratives no relation to
each other. I have tried to show recently that this is not the case, and
that a very basic statistical analysis of prayers versus the content of the

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

narrative indicates that indeed there is a positive correlation between
the two units; they are semantically linked (Al-Jallad, 2022, 9–14).54
Moreover, unlike the monumental South Arabian inscriptions which
tend to prefer a concrete understanding of lexemes, the Safaitic inscrip-
tions are imbued with emotional content and religious metaphors, as
the texts cited throughout my 2022 booklet demonstrate. Thus, when
a consonantal skeleton presents us with two interpretations, one meta-
phorical and one concrete, we are not a priori beholden to the latter
meaning. Rather, I would argue, it is the coherence of the entire com-
position that should guide how we arbitrate between competing under-
standings of polysemous words. The first portion of the prayer is clearly
mythological and sets a metaphorical frame for what follows. Its trans-
lation is clear: h lh nr ‘O Allāh, let there be light’. The grammatical
composition of the final clause is open to a number of interpretations
but its meaning is more or less clear, with the exception of option 4,
which I shall discuss further below.
1) w whb-nh ht ḏ ḫlq (taking whb as an optative suffix conjugated
verb and ḏ-ḫlq as its subject)55
‘may he who has created grant him that (fem.sing., referring back
to nr ‘light’)’
2) w whbn-h ht ḏ ḫlq (taking whbn as a nominal form or infinitive of
command, and ḏ-ḫlq as a relative clause with ht as an antecedent)
‘may his bestowment be that (FS, referring back to nr ‘light’) which
he (Allāh) has created’
3) w whb-nh ht ḏ ḫlq (taking whb as an imperative and ḏ-ḫlq as its
subject)
‘and grant him that, O one of creation’
4) w whb-nh ht ḏ ḫlq (taking whb-nh as a suffix conjugation and ḏ ḫlq
as a relative clause)
‘and he granted him that (FS, no antecedent) which he created’

54 See also the interpretation of H1 in Al-Jallad (2017, 120–121).

55 Such constructions are attested, for example, BES 192: …w nẓr h-śnʾ f slm h
lh ‘and he kept watch for the enemy so grant security, O Allāh’ and also, previously
quoted, JSLih 08: ʿbdmnt / ʾṣdq / f rḍ-h / h lh / w sʿd-h ‘ʿAbdmanōt performed the
ṣdq so satisfy him, O Allāh, and help him’.

35
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

The crux of the entire text is the second clause, w ḥmm ġmy bnyt-h,
which lends itself to an abstract, mythological interpretation or a read-
ing in concrete terms. I will offer both paths of interpretation below.
Let us begin with the concrete interpretation. This requires us to sepa-
rate the final two clauses from the prayer introduced by f. The author

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

calls upon Allāh for ‘light’ and then returns to introduce a second
narrative. This could describe a number of things.
A) We could appeal to my reconstructed meaning of ḥmm ‘to
offer’, with its object being ġmy bnyt-h ‘the roof of his construction’.
The referent of the pronominal suffix -h in this case is unclear. No
specific person was mentioned at the beginning of the inscription and,
as we have shown, this text cannot be associated with the building party
of the ṯyt memorial, so this action cannot be connected to an architec-
tural feature of the present mortuary installation. Moreover, the final
clause would not follow logically or grammatically. One could poten-
tially understand it as a paraphrase of what came before, ‘and he granted
him (to the dead man?) that (feminine singular) which he created’.
But what would ht, a feminine singular demonstrative, refer to? Since
‘roof’ is masculine, it would have to refer to the structure itself, bnyt,
but that is not what was offered. In this case, we may understand the
final clauses as a second prayer to be translated along the lines of 1–3
(above).
B) A more agreeable interpretation along concrete lines is to parse
ḥmm differently and understand it as ḥm m- ‘and protect him/those’,
which then renders the following two clauses as extensions of the prayer.
This is formulaically happier. From this, we can produce: w ḥm m ġmy
bnyt-h ‘and protect whoever covered his construction’. In this context,
the third-person pronoun must refer to the generic ‘loved one’ nfs wdd
and bnyt to a mortuary installation associated with the procession of
the dead. One possible structure that would require covering or roof-
ing is the ritual shelter, a type of tent constructed at graves (Al-Jallad,
2022, 37–41). This clause would not refer to the present installation
but more abstractly invokes Allāh to protect anyone who is involved
with the burial and commemoration of the dead, no doubt in reac-
tion to the site. The final clause must then be understood as begin-
ning with a nominal form along the lines of translation 2 above.
C) Finally, we may disconnect the final two clauses entirely from
the preceding text by understanding them as new narratives describing
a type of religious offering. This is anchored in taking ḥmm ġmy bnyt-h
as ‘he offered the roof of his construction’, where the construction
could signify a religious installation, perhaps a mobile tabernacle. The
pronominal suffix would then refer back to Allāh. The final clause can
36
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

then be taken as an indicative statement, ‘and he granted him (Allāh)

that which he created’ or can be construed as a subsequent prayer, ‘so
may he grant him that which he has created’. The latter understanding
is preferable as the feminine pronoun ht is left without a referent in the
former interpretation while in the latter it would connect back to ‘light’

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

and is therefore more coherent. Moreover, while grammatically possi-
ble, the first reading of the final clause makes little sense in context.
The downside of this reading is that the second clause would have noth-
ing at all to do with the initial narrative or the first clause of the prayer,
especially since the asking for ‘light’ is inescapably connected to theme
of death and the darkness of the ancient Near Eastern underworld.
Therefore, this understanding of the second clause, though tempting in
isolation, faces difficulties when considering the entire composition.
While the concrete reading of the texts produces grammatical and
semantically plausible understandings of each section, it creates sig-
nificant inconsistencies when trying to understand the inscription as
a complete whole. In my estimation, interpretations A and C are rather
unlikely, while interpretation B is the most compatible with the narra-
tive and first clause of the prayer, but still requires us to read signifi-
cantly between the lines.
The clearly abstract diction of the narrative component along with
the prayer for nr ‘light’ motivates us to consider seriously a more liter-
ary reading of the final two clauses of the inscription. The theme of
the inscription is clearly death, in an abstract sense rather than that of
a specific individual, triggered by the interaction with a funerary instal-
lation. This phenomenon has been previously attested. The inscription
AMSI 41 records a man stopping by the grave of someone called Wrd,
reflecting on the phenomenon of death itself, and then resigning to the
fact that his own demise is inevitable.56
AMSI 41:
l msk bn ʾsd bn slm w qʿd ʿd wrd f ḏkr h-mt f qṣf f h lt ʿmr ṣdq-k w gnn w
mn m{t} {l}s¹ fṣy
‘By Msk son of ʾsd son of Slm and he sat at (the cairn of) Wrd and remem-
bered the dead and grieved so, O Allāt, grant long life to your righteous
worshipper and protect (him) but from death there is no deliverance’

56 There are other, more mundane examples of this genre. For example, BES15 464
is by a man called Lkf son of ʿlm and he states: w wgm ʿm f ʿm ʿl-ḥbb f ḥbb ‘he grieved
year after year for loved one after loved one’. While the archaeological context of this
inscription was not recorded, I would not be surprised if it is located at a cairn. It is
possible that this man, triggered by a funerary site, etched this expression of grief, which
does not commemorate any single person but is rather a reflection on the pain of loss.

37
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

AMSI 41 clearly deals with abstract concepts rather than concrete actions
and, as such, can serve as a template for understanding the motivation
and content behind the production of what I would call a contempla-
tive text.57 This should motivate us to explore the connection between
the narrative, which expresses a fear of death, and the prayer for nr

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

‘light’. While we know very little, if anything, about the concept of an
afterlife among the Safaitic-writing tribes, it is untenable to suggest
that their mythology existed in complete isolation from the broader
Near Eastern context during antiquity.58 Let us consider the most widely
read Mesopotamian myths of the 1st millennium, the ‘Descent of Ištar
to the Netherworld’ (Lapinkivi, 2010). It begins with a detailed image
of where the life forces (nfs) of humans go after death.
1 To the Netherworld, the Land of No Return,
2 Ištar, the daughter of Sin, set her mind.
3 Indeed, the daughter of Sin set her mind
4 to the dark house, the dwelling of Irkall a,
5 to the house which none leaves who enters,
6 to the road where traffic is one-way,
7 to the house, whose dwellers thirst for light,
8 where dust is their food, clay their bread.
9 They see no light, they dwell in darkness (Trans. Lapinkivi, 2010, 29).

The underworld, the grave, symbolized by darkness is also found in

the Bible and in Ugaritic myth as well and so it seems to be a pan-Near
Eastern concept (Johnston, 2002; Smith & Bloch-Smith, 1988; Stour,
1980). Light, therefore, is the opposite force of death, betokening life
and creation, and I would understand the invocation in the first clause
to be a request for life in the face of death. This parallels AMSI 41,
where the narrative begins with a reflection on death, h-mt, and then
asks Allāt to grant him long life, ʿmr.
Anchored in this understanding, I would take the second clause as:
w ḥm m-ġmy bnyt-h
‘and protect him from the darkness / the covering of his grave’

57 On the edition of this text, see Al-Jallad (2018). The original edition worked

with the text out of context. Upon redocumenting the text during our 2019 cam-
paign, it became clear that wrd in the inscription referred to the burial of a man called
Wrd in a nearby cairn. The author of AMSI 41 was not part of the burial party but
rather reacts to the cairn as he moved through the area. The word ʿd before wrd
should be understood as the cognate of Classical Arabic ʿinda ‘at’, with the assimila-
tion of the n.
58 On the evidence for some manner of afterlife, see Al-Jallad (2022, 78–83).

38
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

The term bnyt, which we have established earlier as referring to a mẓll

‘dark place’, a funerary installation, does not refer to the ṯyt or the burial
of ʿAqrab. Rather, ʾOwais, triggered by his environment, is asking Allāh
for protection from the grave, a euphemism for death. The referent
of the third person pronoun is unclear – does it refer to the unnamed

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

‘loved one(s)’ or to the author himself? I would argue for the latter, as
the narrative does not identify any particular person but rather deals
with the phenomenon of death more generally. Thus, the prayer is
for the preservation of ʾOways’ life. This may be compared to another
Safaitic inscription, which calls upon the god Śayʿ haq-Qawm for deliv-
erance from death.
KRS 68:
l ʾnʿm {b}{n} {w}rl bn dmy ḏ-ʾl ʾs²ll w h śʿhqm ṣmy nqt f {ʾ}{n}-k bġy-h
w qf{y}t-h {w} b-ḫfrt-k fltn m-mt
‘By ʾnʿm {son of} {Wrl} son of Dmy of the lineage of ʾśll and O Śayʿ haq-
Qawm, he sacrificed a camel, for {you are indeed} the one he seeks and
his {path} {and} through your guidance there is deliverance from death’
The term ġmy has roughly the same valence as ẓll, ‘to cover’, ‘to cast
darkness’. Both meanings point to the same conclusion: death. Under-
standing ġmy as ‘darkness’ describes the state of being in the grave and
the underworld, while interpreting it more literally as the verbal noun
‘covering’ describes the act of enclosing the deceased in their burial
cairn (bnyt).
Let us now return to the difficult expression bʾsmẓll. To reiterate, it
occurs exclusively in a memorial context. In most cases it appears sim-
ply above, but it sometimes takes an object introduced by ʿl- and once
even a direct object. I would suggest that the construction be parsed
rather as bʾs m ẓll and be compared to the Quranic expression biʾsa-mā
‘evil is that which’. The construction is always followed by a verb:
Q 2:90: biʾsa-mā štaraw bi-hī ʾanfusa-hum
‘evil is that for which they sold themselves’
Q 2:93 biʾsa-mā yaʾmuru-kum bi-hī ʾīmānu-kum
‘evil is that which your faith enjoins upon you’
Q 7:150 biʾsa-mā ḫalaftumū-nī min baʿdī
‘evil is that with which you have replaced me’
Q 3: 187 biʾsa mā yaštarūna
‘evil is that which they purchase’

Given the context of all that we have now learned, I would posit that
the correct translation of this phrase is: ‘evil is that which casts darkness’.
39
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

The verb ẓll refers to enveloping darkness and its symbolic represen-
tation of death and the underworld. This better explains the object
introduced by ʿl- ‘on’, as darkness falls upon something or someone.
The one case in which the object is introduced by -h may be explained
differently by appealing to the sense of ‘grave’ itself, mẓll, or to the use

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

of the verb to mean literally ‘to cover’ as in the burial itself.
C 1682:
l mʿn bn {k}{s}{l}y bn mʿn bn ṣrmt bn ḥny w ḥll h-dr f h lt slm w wqyt
w ḏkr ḫl-h tm f bʾs m ẓll f wlh w ḫbl-h
‘By Mʿn son of {Ks¹ly} son of Mʿn son of Ṣrmt son of Ḥny and he camped
in this place so, O Allāt, let there be security and protection, and he
remembered his maternal uncle Tm, for evil is that which casts darkness,
and he was distraught and (grief) drove him mad’
KRS 12:
l wny bn ḫlṣ bn tm w wgd ʾṯr ghm f bʾs m ẓll ʿl-ʾḫ-h w wlh l-ʾbd ʿl-ḥbb-h
‘By Wny son of Ḫlṣ son of Tm and he found the trave of Ghm, for evil
is that which casts darkness upon his brother, and he was distraught
forever on account of his beloved’
AMSI 24:
l msk bn slm bn ʾsd bn slm w wgd ʾṯr dd-h f bʾs m ẓll-h / f bʾs mẓll-h
‘By Msk son of Slm son of ʾsd son of Slm and he found the trace of his
paternal uncle, for evil is that which covers him / for evil is his grave’

The final clause would then reiterate the prayer of the first clause, ask-
ing again for light to protect against death, referring back to nr with the
anaphoric feminine demonstrative, ht. Since it is a repetition, I would
suggest that ʾOways appealed to iltifāt, an Arabic rhetorical strategy that
creates emphasis or dramatic effect by shifting persons within a compo-
sitional unit. This is attested frequently in the Qurʾān and also in the
Ancient North Arabian inscriptions:
Q 10:22:
huwa llaḏī yusayyiru-kum fī l-barri wa-l-baḥri hattā ʾiḏā kuntum fī l-fulki
wa-ǧarayna bi-him bi-rīḥin ṭayyibatin wa-fariḥū bi-hā
‘he is the one who sends you forth across land and sea until you are
upon a boat and they carry them along with a favorable wind on account
of which they rejoice’
The same rhetorical strategy is attested in the following Hismaic
inscription:
40
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

KhMa 2:59
sqm l-ʾlh ṣʿb f tḍrʿ w tʿny w tś[d]{d} l-h b-kll m fʿl … w l l-k trḥm ʿl-y
‘he transgressed against the god Ṣʿb and so he supplicated, toiled, and
exerted himself for his (the god’s) sake in all he has done … so who but
you can show mercy upon me’

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

Given the rhetorical function of iltifāt, its presence at this point of the
composition is not unexpected. As such, I would appeal to reading (1)
of the first clause, producing an optative suffix conjugated verb with a
right dislocated subject:
w whb-nh ht ḏ ḫlq
‘and may he who has created grant him that (i.e. light)’
If we do not wish to appeal to iltifāt, then it is possible to appeal to
reading (3), which continues the series of imperatives:
w whb-nh ht ḏ ḫlq
‘and grant him that, O one of creation’
Since the two final readings differ only in the grammatical identifica-
tion of their constituent components, but not their meaning, I will
not consider them as separate interpretations.
So then, we have now in hand a number of concrete interpretations
and an abstract understanding as well. How do we arbitrate between
these? I would propose setting up criteria for ranking the plausibility of
a given interpretation based on what we have discussed above:
1. Genre compatibility: The composition finds a precedent in terms
of genre;
2. General cohesiveness: The entire composition is internally cohe-
sive; the various compositional components (i.e. narratives and
prayers) make sense as a single unit;
3. Environmental context: The interpretation is compatible with the
landscape and its archaeological context;
4. Grammatical: The interpretation does not require ad-hoc gram-
matical explanations, graphic emendations, or leave certain con-
structions unaccounted for;
5. Lexically sound: The meanings of the words are secure (not based
on hypothetical definitions given to hapax legomena);
6. Compositionally regular: The text follows established syntax and
formulaic constructions.
59 See Al-Jallad (2020) for the present interpretation of the text, which differs

in the details from the editio princeps.

41
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

The five interpretations of the final section of BES24 1.2 are as fol-
lows, based on the main division between concrete and abstract under-
standings:

Concrete category:

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

A: ‘So, O Allāh, let there be light, and he offered the roof of his (no
clear referent) (funerary?) construction and bestowed him with that
(no clear referent) which he (ʾws) created/portioned’
B: ‘So, O Allāh, let there be light, and protect whoever covered his/
their construction and may his bestowment be that which he has
created (= light)’
C: ‘So, O Allāh, let there be light, and he offered the roof of his (reli-
gious) construction and he granted him (Allāh) that which he (ʾws)
created/portioned’

Abstract category:
A: ‘So, O Allāh, let there be light, and protect against the darkness of
his grave, and grant him that (light), O one of creation / and may
he who has created grant him that (=light)’
B: ‘So, O Allāh, let there be light, and protect against the covering of
his grave, and grant him that (light), O one of creation / and may
he who has created grant him that (=light)

Concrete Abstract
Criterion A B C A B
1 X X X ✔ ✔
2 X X X ✔ ✔
3 X X X ✔ ✔
4 X ✔ X ✔ ✔
5 X ✔ X ✔ ✔
6 X ✔ ✔ ✔ ✔
Total 0 3 1 6 6

Table 3: Ranking the interpretations of the third section of BES24 1.2

It is no surprise that the interpretations which have to do with offering

roofs of temples and structures fail the threshold of plausibility. Con-
crete A and C require hypothetical and reconstructed lexical semantics,
and leave the anaphoric references of both the pronoun -h on bnyt and
the feminine demonstrative unaccounted for. While Abstract B does
42
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

not succumb to those issues, the narrative mentions an undefined

plurality of the dead while the prayer would seem to refer to someone’s
specific grave, thus the prayer for light directed at anonymous persons
who engage in the burial of an unnamed man disrupts the cohesiveness
of the text. The blessing of individuals who have taken part in a burial

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

does not match the narrative describing the fear of future death. Finally,
it contradicts the contemplation genre, where the author reacts to death
by asking for the preservation of his own life.
The abstract readings score higher because they satisfy criteria 1 and
2. While I have marked both as meeting criterion 3, this assessment is
contingent upon the acceptance of iltifāt within the current context. It
is a rare phenomenon to be sure, but it is securely attested. If one
wishes to object to iltifāt, then we may reduce both scores to 5, but
the score 6 can easily be restored by appealing to the interpretation
of whb-nh as an imperative with a right dislocated vocative, a securely
attested syntactic construction. The abstract readings are maximally
coherent and all pronouns have their referents accounted for; they,
moreover, find a precedent in terms of genre. The criteria are unable to
distinguish between readings A and B, however. One could prefer B
slightly to A, if one wishes to appeal to the fact that the precise sense
of ġmy as ‘to cover’ is better reflected in the Arabic lexica, while its mean-
ing ‘darkness’ is more rarely encountered in poetry (DohaDict, s.v.).
However, this objection loses force once we recognize that the present
composition is drawing on artistic language. Moreover, the metaphorical
sense of ‘darkness’, rather than the concrete ‘covering’, better matches
the prayer for light, and so I ultimately prefer that understanding on
the grounds of balance. I hereby offer my final reading and interpreta-
tion of the text:
BES24 1.2
l ʾws bn ʾdm bn ṣʿd w tʾmr rdf nfs wdd ʿṯrt f h lh nr w ḥm m-ġmy bnyt-h w
whb-nh ht ḏ ḫlq
‘By ʾws son of ʾdm son of Ṣʿd and he feared the procession of fallen loved
ones so, O Allāh, let there be light! And protect against the darkness of
his grave and grant him that (=light), O one of creation / and may he who
has created grant him that (=light)’

3. Analysis: The Mythological Profile of the Ancient Allāh

The role of Allāh as a light-giving creator is compatible with the

mythology of ʾilu in Ugarit, Marduk in the Babylonian tradition, and
indeed Elohim of the Hebrew Bible. The first line of the invocation,
43
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

h lh nr ‘O Allāh, let there be light’, is paralleled by Genesis 1:3 ‘And

Elohim said: Let there be light’. A Nabataean burial inscription from
Hegra (JSNab 2) hints towards the existence of a superbeing in ances-
tral Arabian mythology responsible for this cosmic phenomenon. The
text invokes a deity with a sentential epithet: p ylʿn prš lylyʾ mn ymmʾ

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

mn dy ynpq yt-hm l-ʿlm ‘And may he who separates night from day
curse whoever removes them forever’.60 While Healey very cautiously
suggested that this epithet might be attached to Dusares (Healey,
2001, 93), there is in fact nothing in the inscription itself that would
suggest so. Taking the present inscription into consideration, it is more
likely that the epithet belongs to Allāh, and that JSNab 2 is a unique
Nabataean instance of invoking the creator deity, who is otherwise
only known through personal names.
Section three of the inscription is clearly a litany against death,
which is symbolized through the imagery of falling darkness (ẓll, ġmy),
addressed to the living, creator god (ḥy, ḫlq). It is impossible to ignore
the thematic and structural parallels with Qurʾān 113:
‫ب ا ْلفَ َل ِق‬ ُ ‫يم ُق ْل َأ ُع‬
ِّ ‫وذ ِب َر‬ ِ ‫ِب ْس ِم ۝ الر َّْح َ ٰمنِ الر َِّح‬
In the name of Allāh, the merciful, the compassionate; say: I seek refuge
in the lord of dawn
‫ِم ْن َش ِّر َما َخ َل َق‬
From whatever evil one has created

ِ ‫َو ِم ْن َش ِّر َغ‬

َ ‫اس ٍق إ َِذا َو َق‬
‫ب‬
From the evil of night when it falls
‫ات ِفي ا ْل ُع َق ِد‬
ِ ‫َو ِم ْن َش ِّر الن ََّّفا َث‬
From the evil of women who blow into knots
َ‫اس ٍد إ َِذا َح َسد‬
ِ ‫َو ِم ْن َش ِّر َح‬
From the evil of the envier when he envies

This incantatory surah aims to shield against death, embodied by the

falling of night (= darkness), and then specified by the actions of con-
cealed enemies. Allāh’s light, represented by the epithet ‘lord of dawn’,
counteracts the effectiveness of their malignant magic. Later Ḥadīth
literature attests an invocation to Allāh to shine light directly into the
grave. The following prayers are attributed to Mohammed:

60 For the latest edition of this text, see DiCoNAB 2025, inscription 255.

44
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

ʾallāhumma ǧʿal lī nūran fī qabrī 61

‘O Allāh, grant me light in my grave’
wa-ġfir la-nā wa-la-hū yā rabba l-ʿālamīna wa-fsaḥ la-hū fī qabri-hī wa
nawwir fī-h62
‘and forgive us and him, O Lord of Eternity, and grant him space in his

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

grave and shine light into it’
Darkness as a symbol of the underworld, and conversely light as life,
perhaps survives into the late pre-Islamic period in Jāhilī odes. The
muʿallaqah (ode) of the Jāhilī poet Labīd b. Rabīʿah (d. 661 CE) recounts
a mythic event that describes a wild cow whose calf was lost. She then
found him devoured by grey wolves, who represent the arrows of Fate.63
In a scene that describes her grieving, the stars are blocked out by heavy
clouds and incessant rain, thus producing darkness. She disappears
into the roots of a mighty tree where sand pours upon her. It is clear
that the events recounted here are supernatural—this is no ordinary
animal—and they may have originally conveyed an image of a descent
into the grave, into the underworld seeking after her deceased calf. What
comes next is illuminating:64
wa tuḍīʾu fī waǧhi ẓ-ẓalāmi ka-ǧumānati l-baḥriyyi sulla
munīratan niẓāmu-hā
‘And she shines brightly in the face like the pearl of a seaman, whose
of darkness string had been withdrawn’

The juxtaposition of light and darkness symbolizes life opposing death,

providing context for ʾOways’ plea to Allah for illumination at the bur-
ial site, and, indeed, protection from his own grave and the underworld.
We may now return to the verb ḫlq, found in the epithet of the
third clause. If we wish to understand it in light of JSNab 2 (and other
Near Eastern creation myths, like Gen. 1), then it would seem to sig-
nify the primordial act of creation entailing the separation of light and
61 This belongs to a longer prayer meant to be recited on the way to the mosque.

This particular line is attested in Al-Tirmiḏī (1996, no. 3419), and is reported by
ibni Ḥajar as well (Al-Qaḥṭānī 1988, 23). It is, however, lacking in the recensions
of Al-Bukhārī and Muslim and thus appears to be a secondary addition.
62 This verse terminates a prayer Mohammed is said to have recited over Abu

Salamah, one of his early followers, on his death bed (Muslim 1955–6, Book 11,
Hadith 8).
63 On the connections of this image to Safaitic concepts of death, see Al-Jallad

(2022, 73–77).
64 There are many editions of this text, but I am partial to the interlinear trans-

lation of Johnson (1893).

45
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

darkness or day and night. However, the cosmogeny of Mohammed’s

opponents in the Qurʾān may suggest that the ‘division’ signified by this
verb was in fact the separation of the heavens and the earth, Q 21:30:
ʾa-wa-lam yara lladīna kafarū ʾanna s-samāwāti wa-l-ʾarḍi kānatā ratqan
fa-fataqnā-humā wa-ǧaʿalnā mina l-māʾi kulla šayʾin ḥayyin ʾa-fa-lā yuʾminūn

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

‘don’t those who deny (the Arab ‘pagans’) regard that the heavens and
earth were once joined but then we split them apart and forged from
water all living things; so why won’t they believe (this)’?
My translation is a bit unconventional. The verse makes it clear that this
was not a point that Mohammed’s opponents rejected, but it is rather
to be understood as common ground making their rejection of his
message all the more puzzling. Because of this, and the evidence prof-
fered by BES24 1.2, we are not compelled to understand the account
given in Q 21:30 as the result of biblical syncretism or the reworking
of late antique Christian cosmogonies.65 Rather it is equally possible
that it reflects an inheritance from an ancestral Arabian religion, which
was ultimately related to other Near Eastern creation myths, such as
the Enuma Elish.66
The phrase ḏ ḫlq finds a parallel in the epithet of ʾilu, qn ʾrṣ ‘crea-
tor of the earth’, which is attested in a number of Northwest Semitic
inscriptions, spanning from the 8th century BCE to the 2nd century CE
(Miller, 1980). The verb qny ‘to create’, which is the semantic equiva-
lent of ḫlq, is also applied to ʾilu in the Ugaritic texts (Pope, 1955, 49–
54). BES24 1.2, therefore, demonstrates that the ancient Arabs shared
in Near Eastern ʾilu mythology. This particular myth could perhaps be
sourced to Nabataea given the etymological origins of the name allāh
itself. As such, the conclusion that Allāh is simply ancient ʾilu under
a different name appears unavoidable. But how did the name mutate
in such a way? Robin’s idea of it being a translation is imprecise. After-
all, ʾilu is a proper name and so it would not lend itself to translation
as such, especially with the definite article. I would suggest instead that
the epithet al-ʾilāh ‘the god’ par excellence, was applied to the creator ʾilu
in West Arabia and the Nabataean realm. In time, the epithet simply
replaced the name, resulting in ʾilu being forgotten completely. The
evolution of the divine name Allāt offers an important parallel. Scholars
have previously suggested that the name was the epithet of ʿAṯtar/Ishtar,
specifically her North Arabian manifestation ʿtrsm /ʿAttar-Samē/. This

65 See Anthony (forthcoming) for the latest treatment of Quranic cosmogeny.

66 On the Babylonian creation myths, see Lambert (2013). On evidence for
religious exchanges between North Arabia and Babylon, see Al-Jallad (2020).

46
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

has recently been confirmed by an unpublished Thamudic B inscription,

likely dating to the mid-first millennium BCE, which invokes ʿAttar-
Samē as the daughter of Ruḍaw, h ʿtrsm bnt rḍw. The same mythological
complex continues into Safaitic, but there is no trace of the name ʿtrsm.
The title ‘the goddess’ seems to have fully superseded the original name,

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

resulting in the invocation h ʾlt bnt rḍw, meaning ‘O Allāt, daughter of
Roḍaw’ (Al-Jallad, 2020).
The present inscription likely dates to the turn of the era, some
600 years before Mohammed’s ministry. It, moreover, comes from a
region more than 1,000 kilometres away from the drama of Paleo-Islam
in West Arabia. Nevertheless, the Safaitic inscriptions present a religious
world that is largely compatible with the Quranic opponents. Together
they share in the erection of naṣab stones (nṣb, m(n)ṣb(t), animal
sacrifice (nḥr, ḏbḥ), pilgrimages to shrines (ḥgg), swearing oaths (ʾqsm
b-ʾlh), and the lack of a belief in a tangible afterlife or a bodily resur-
rection.67 One of the main issues that sets the Quranic interlocutors
apart, and motivated some to understand it as the impact of Judaizing
monotheism, was the belief in a single creator deity, and the description
of creative acts in a way that was compatible with the Hebrew Bible.
Yet it is now clear that their similarity to Biblical narratives reflects
a common origin, a cognate tradition, and is not the result of bor-
rowing in late antiquity from neighbouring Jews or Christians. This
fact has significant implications for our understanding of the syncretism
between Ancestral Arabian religion and biblically rooted monotheisms
in the late pre-Islamic period. Robin argued that the etymological simi-
larity between ‘pagan’ Allāh and Christian al-ʾilāh motivated the equa-
tion between the two. I would suggest that it was not necessarily the
etymological connection alone but rather the shared mythological frame-
work that was responsible for this. Ancient Allāh was the creator and
performed acts that made him immediately comparable to and inter-
changeable with the biblical god. The ancient Arabs could have easily
found cosmological similarities with Christians and Jews, without fully
adopting their broader mythologies, particularly the narratives of patri-
archs like Abraham and Moses, or Jesus for Christians
The present discovery, configured with other arguments about epi-
graphic Allāh, I think tip the scales towards understanding him as a
well-defined superbeing with a specific role in the cosmos and mytho-
logical profile. Yet, one of Al-Azmeh’s points remains valid: the epi-
graphic evidence does still suggest that Allāh was a remote and rarely
called upon deity in the most ancient of times, comparable to ʾilu in
67 On all of these practices and beliefs, see Al-Jallad (2022).

47
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

Ugaritic and other Northwest Semitic texts, and indeed Greek Kronos.
Even though ʾOways composed a rather elaborate text invoking Allāh,
he remains a minor deity in Safaitic, in terms of invocational frequency.
One might interpret this evidence in support Al-Azmeh’s idea that
Mohammed elevated Allah to the primary object of worship. This the-

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

ory, however, was posited before the discovery of the bulk of the Paleo-
Arabic inscriptions from Arabia. At the present time, the only deity
invoked in the inscriptions of West Arabia, and indeed all of Arabia,
is Allāh / al-ʾilāh.68 This pattern represents a clear departure from pre-
Islamic practice, where he was rarely invoked and inscribers typically
appealed to other deities. So Al-Azmeh appears to be right–Allāh was
promoted in terms of invocational frequency (but perhaps not in terms
of cosmic power). Yet that promotion predated Mohammed’s enter-
prise. A sequel to this essay, ‘Late Antique Allah’, will attempt to under-
stand this process.

Address for correspondence: [email protected]

68 See Robin (2020) and more recently, Al-Jallad and Sidky (2024).

48
 Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025
Map 1: Location of the site Marabb Al-Hilāl 09 (global)
ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

49
 Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025
Map 2: Location of the inscriptions produced by ʾws bn ʾdm and his sons
ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

50
 Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025
Map 3: Location of the site MHL 09 (detail)
ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

51
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

DICTIONARY AND DATABASE SIGLA

AAWHA Inscriptions recorded by Ali Al-Manaser in a tributary of Wādī

al-Ḫuḍarī in September 2002 and published on OCIANA.
AAWAB Inscriptions recorded by Ali Al-Manaser in Al-Wādī Al-Abyaḍ
in June 2021 and published on OCIANA.

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

AbWH Safaitic inscriptions published in Abbadi 1996.
Al-Mafraq Museum Safaitic inscriptions kept at the al-Mafraq Museum collected
by ʿAbd al-Qadir al-Ḥuṣān and published on OCIANA.
AMSI Safaitic inscriptions recorded by Ali al-Manaser in 2004 at
Wādī al-Ḥašād and published on OCIANA.
ASWS Safaitic inscriptions published in Awad 1999.
AWS Safaitic inscriptions published in Alulu 1996.
C Safaitic inscriptions published in Ryckmans 1950–1951.
CAL Comprehensive Aramaic Lexicon Project: https://cal.huc.edu/.
CDG Leslau 1987.
DohaDict The Doha Historical Dictionary of Arabic: https://www.
dohadictionary.org.
DULAT Del Olmo Lete and Sanmartín 2015.
HHM Hismaic inscriptions published in Hayajneh 2018.
Jacobson Previously unpublished Hismaic inscriptions from Jordan on
OCIANA.
JSLih: Dadanitic inscriptions in Jaussen & Savignac 1909–1920.
KJC Hismaic inscriptions from site C, Wadi Ǧudayyid, Jordan,
published in King 1990.
Lane Lane 1863–93.
OCIANA Online Corpus of the Inscriptions of Ancient North Arabia:
ociana.osu.edu.
RWQ Safaitic inscriptions published in Al-Rousan 2004.
SafDict Al-Jallad and Jaworska 2019.
SIAM Safaitic inscriptions published in Macdonald 1979.
SIJ Safaitic inscriptions published in Winnett 1957.
WH Safaitic inscriptions published in Winnett and Harding 1978.
WTI Thamudic inscriptions published in Winnett and Reed 1970.

BIBLIOGRAPHY

Abbadi, S. 1996) Nuqūš ṣafawiyyah ǧadīdah fī l-ʾurdunn / wādī al-ḥašād. Dirāsāt

al-ʿulūm al-insāniyyah wa-ʾl-iǧtimāʿiyyah [al-ǧāmiʿah al-ʾurdunniyyah], 23, 242–
252.
Al-Azmeh, A. 2014. The emergence of Islam in late antiquity: Allah and his people.
Cambridge: Cambridge University Press.
——. 2021. ‘Further to the Pre-Muḥammadan Allāh’, Arabica, 68, 419–435.
Al-Jallad, A. 2015. An outline of the grammar of the Safaitic inscriptions. Leiden:
Brill.
——. 2017a. ‘Pre-Islamic ‘Ḥamāsah’ verses from Northeastern Jordan: A new Safaitic
poetic text from Marabb Al-Shurafāʾ, with further remarks on the ʿĒn ʿAvdat
inscription and KRS 2453’, Proceedings of the Seminar for Arabian Studies, 47,
117–128.

52
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

——. 2017b. ‘Graeco-Arabica I: The Southern Levant’, in Al-Jallad, A. (ed.) Arabic

in Context: Celebrating 400 years of Arabic at Leiden (Semitic Languages and Lin-
guistics of Arabic at Leiden 89). Leiden: Brill, 99–186.
——. 2018. ‘The earliest attestation of laysa and the implications for its etymology’,
in Macdonald, M.C.A. (ed.) Languages, scripts and their uses in ancient North Arabia
(Supplement to volume 48 (2018) of the Proceedings of the Seminar for Arabian

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

Studies). Oxford: Archaeopress, 111–119.
——. 2019. ‘Safaitic’, in Pat-El, N. & Huehnergard, J. (eds) The Semitic languages.
2nd edn. New York: Routledge, 342–366.
——. 2020. ‘Notes on the language of the Hismaic inscriptions and a re-reading of
Line 1 of the Madaba Hismaic inscription’, Journal of the Royal Asiatic Society,
30(3), 561–569.
——. 2022. The religion and rituals of the nomads of Pre-Islamic Arabia: A reconstruction
based on the Safaitic inscriptions. Leiden: Brill.
——. 2024. ‘Hasaitic’, Thematic dictionary of ancient Arabia [Online]. Available at:
https://ancientarabia.huma-num.fr/dictionary/definition/hasaitic (Accessed: 20 Janu-
ary 2025).
——. 2025. ‘KRS 294’, in Al-Jallad, A. & Macdonald, M.C.A. (eds) OCIANA [Online].
Available at: https://ociana.osu.edu/inscriptions/36747 (Accessed: 20 March 2025).
Al-Jallad, A. and Sidky, H. 2024. ‘A Paleo-Arabic inscription of a companion of
Muhammad?’, Journal of Near Eastern Studies, 83(1), 1–14.
Alulu, Ġ.M.Y. 1996. Dirāsat nuqūš ṣafawiyyah ǧadīdah min wādī al-sūʿ ǧanūb sūriyyah.
Unpublished M.A. thesis, Yarmouk University.
Al-Qahtani, S.B. 1988. Ḥiṣn al-muslim min aḏkār al-kitāb wa-ʾl-sunnah. Riyadh: Safir
Press.
Al-Rousan, M.M. 2004. Nuqūš ṣafawiyyah min wādī qaṣṣāb bi-l-ʾurdunn. Dirāsah
maydāniyyah taḥlīliyyah muqāranah. Unpublished doctoral thesis, King Saʿūd Uni-
versity, Riyadh.
Al-Tirmidhī, A. ʿĪ. 1996. al-Jāmiʿ al-kabīr. Edited by Bashshār ʿAwwād Maʿrūf. 6 vols.
Beirut: Dār al-Gharb al-Islāmī.
Anthony, S.W.A. forthcoming. ‘The creation’, in Sinai, N., Klar, M., Reynolds, G.S.
& Zellentin, H.M. (eds) Biblical traditions in the Qurʾan. Princeton: Princeton
University Press.
Atbuosh, M.A. 2025. ‘South Arabian etymology for the Meccan Kaʿbah’, Bulletin of
the School of Oriental and African Studies, 1–11.
Awad, M., 1999. Dirāsat nuqūš ṣafawiyyah ǧadīdah min ǧanūb wādī sārah/al-bādiyah
al-ʾurdunniyyah aš-šamāliyyah. Unpublished MA thesis, Yarmouk University.
Baasten, M. 2009. ‘Scheppen of scheiden? Over de interpretatie van Genesis 1’, Met
Andere Woorden, 28(4), 3–14.
Beeston, A.F.L. 1979. ‘Nemara and Faw’, Bulletin of the School of Oriental and African
Studies, University of London, 42(1), 1–6.
Brockelmann, C. 1908. Kurzgefasste vergleichende Grammatik der semitischen Sprachen:
Elemente der Laut- und Formenlehre. Berlin: Reuther & Reichard.
Clark, V.A. 1979. A study of new Safaitic inscriptions from Jordan. Doctoral disserta-
tion. Department of Middle Eastern Studies, University of Melbourne.
Crone, P. 2010. ‘The religion of the Quranic pagans: God and the lesser deities’,
Arabica, 57, 151–200.
——. 2016. ‘Pagan Arabs as God-fearers’, in Bakhos, C. & Cook, M. (eds) Islam and
its past: Jāhiliyya and late antiquity in early Muslim sources. Oxford: Oxford Uni-
versity Press.

53
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

Del Olmo Lete, G., and Sanmartín, J. 2015. A Dictionary of the Ugaritic Language in
the Alphabetic Tradition (2 vols), Leiden: Brill.
Della Puppa, C. 2022. The Safaitic scripts: Palaeography of an ancient nomadic writ-
ing culture. Doctoral dissertation. Universiteit Leiden.
Digital Corpus of the Nabataean and Developing Arabic Inscriptions [DiCoNab].
2025. [Database]. Available at: https://diconab.huma-num.fr/ (Accessed: 5 March

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

2025).
Fiema, Z.T., Al-Jallad, A., Macdonald, M.C.A. & Nehmé, L. 2015. ‘Provincia Arabia:
Nabataea, the emergence of Arabic as a written language, and Graeco-Arabica’, in
Fisher, G. (ed.) Arabs and empires before Islam. Oxford: Oxford University Press,
373–433.
Fischer, W. 2012. A grammar of Classical Arabic. 3rd edn. New Haven: Yale University
Press.
Graf, D.F. & Zwettler, M.J. 2004. ‘The North Arabian “Thamudic E” inscription
from Uraynibah West’, Bulletin of the American Schools of Oriental Research, 335,
53–89.
Hämeen-Anttila, J. & Rollinger, R. 2001. ‘Herodot und die Arabische Göttin “Alilat”’,
Journal of Ancient Near Eastern Religions, 1(1), 84–99.
Hava, J.G. 1982. Arabic-English dictionary for the use of students. Beirut: Dār al-Mašriq.
Hawting, G. 1999. The idea of idolatry and the emergence of Islam. Cambridge:
Cambridge University Press.
Hayajneh, H. 2018. ‘A preliminary investigation of an Ancient North Arabian invoca-
tion from the Madaba region of central Jordan’, in Macdonald, M.C.A. (ed.) Lan-
guages, scripts and their uses in ancient North Arabia (Supplement to volume 48 of
the Proceedings of the Seminar for Arabian Studies). Oxford: Archaeopress, 95–100.
Healey, J. 2001. The religion of the Nabataeans: A conspectus. Leiden: Brill.
Huehnergard, J. 2005. ‘Features of Central Semitic’, in Gianto, A. (ed.) Biblical and
Oriental essays in memory of William L. Moran. Rome: Pontificio Istituto Biblico,
155–203.
Jastrow, M. 1903. A Dictionary of the Targumim, the Talmud Babli and Yerushalmi,
and the Midrashic Literature, 2 vols. London, W.C.: Luzac & Co. ; New York:
G. P. Putnam’s Sons.
Jaussen, A. & Savignac, M.R. 1909–1920. Mission archéologique en Arabie. 5 vols.
(Publications de la Société Française des Fouilles Archéologiques, 2). Paris: Leroux/
Geuthner. [Reprinted Cairo: Institut Français d’Archéologie Orientale, 1997].
Jarrar, N. 2024. ‘Developing digital Islamic heritage sites in Jordan: The case of al-
Mafraq’, Digital Applications in Archaeology and Cultural Heritage, 32.
Johnson, F. E. 1893. The seven poems, suspended in the temple at Mecca. Translated
from the Arabic by Capt. F. E. Johnson, Royal Artillery, with an introduction, by
Shaikh Faizullabhai, B. A. Bombay: Education Society’s Steam Press, Byculla.
Johnston, P. 2002. Shades of Sheol: Death and afterlife in the Old Testament. Downers
Grove: InterVarsity Press.
Kennedy, D.L. 2012. ‘The Cairn of Hānī: Significance, present condition and con-
text’, Annual of the Department of Antiquities of Jordan, 56, 483–505.
Kiltz, D. 2012. ‘The relationship between Arabic Allāh and Syriac Allāhā’, Der Islam,
88, 33–50.
King, G.M.H. 1990. Early North Arabian Thamudic E: A preliminary description
based on a new corpus of inscriptions from the Ḥismā desert of southern Jordan
and published material. Ph.D. thesis. School of Oriental and African Studies,
University of London.

54
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

Koehler, L., Baumgartner, W. & Stamm, J.J. 2001. The Hebrew and Aramaic
lexicon of the Old Testament. Translated by M.E.J. Richardson. 2 vols. Leiden:
Brill.
Kogan, L. 2015. Genealogical classification of Semitic: The lexical isoglosses. Berlin: De
Gruyter.
Kootstra, F. 2022. The writing culture of ancient Dadān. Leiden: Brill.

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

Kropp, M. 2017. ‘The ʿAyn ʿAbada inscription thirty years later: A reassessment’, in
Al-Jallad, A. (ed.) Arabic in context. Leiden: Brill, 53–74.
Lahlali, E. & Islam, T. 2023. A dictionary of Arabic idioms and expressions: Arabic-
English translation. Edinburgh: Edinburgh University Press.
Lambert, W.G. 2013. Babylonian Creation Myths. Winona Lake, Indiana: Eisenbrauns.
Lane, E.W. 1863–93. An Arabic-English lexicon. London: Williams and Norgate.
Lapinkivi, P. 2010. The Neo-Assyrian myth of Ištar’s descent and resurrection. Helsinki:
The Neo-Assyrian Text Corpus Project.
Leslau, W. 1987. Comparative dictionary of Geʿez (Classical Ethiopic): Geʿez–English,
English–Geʿez, with an index of the Semitic roots. Wiesbaden: Harrassowitz.
Lindstedt, I. 2023. Muḥammad and his followers in context. Leiden: Brill.
Macdonald, M.C.A. 1979. ‘Safaitic Inscriptions in the Amman Museum and Other
Collections I’, Annual of the Department of Antiquities of Jordan, 23, 101–119,
pls. 35–44.
——. 2000. ‘Reflections on the linguistic map of Pre-Islamic Arabia’, Arabian Archae-
ology and Epigraphy, 11, 28–79.
——. 2014. ‘”Romans go home?” Rome and other ‘outsiders’ as viewed from the
Syro-Arabian Desert’, in Dijkstra, J.H.F. & Fisher, G. (eds) Inside and out: Inter-
actions between Rome and the peoples on the Arabian and Egyptian frontiers in late
antiquity. Leuven: Peeters, 145–163.
——. 2024. ‘Book review: Pre-Islamic Arabia: Societies, politics, cults and identities
during late antiquity by Valentina A. Grasso’, Journal of Near Eastern Studies, 83(2),
343–348.
Mettinger, T. 2004. ‘Amarna Letter No. 84: Damu, Adonis, and “The Living God” at
Byblos’, in Cohen, C., Hurvitz, A.M. & Paul, S.M. (eds) Sefer Moshe: The Moshe
Weinfeld jubilee volume: Studies in the Bible and the ancient Near East, Qumran, and
post-Biblical Judaism. University Park, USA: Penn State University Press, 361–
372.
Miller, P.D. Jr. 1980. ‘El, the creator of Earth’, Bulletin of the American Schools of
Oriental Research, 239, 43–46.
Muslim. 1955–6. Al-Jāmiʿ al-Ṣaḥīḥ. Muḥammad Fuʾād ʿAbd al-Bāqī (eds). 5 vols.
Cairo: ʿĪsà al-Bābī al-Ḥalabī wa-Shurakāʾuh.
Negev, A., Naveh, J. & Shaked, S. 1986. ‘Obodas the god’, Israel Exploration Journal,
36, 56–60.
Norris, J. 2018. ‘A survey of the Ancient North Arabian inscriptions from the Dūmat
al-Jandal area (Saudi Arabia)’, in Macdonald, M.C.A. (ed.) Languages, scripts and
their uses in ancient North Arabia (Supplement to volume 48 of the Proceedings
of the Seminar for Arabian Studies). Oxford: Archaeopress, 71–93.
Norris, J. & Al-Manaser, A. 2018. ‘The Nabataeans against the Ḥwlt—once again:
An edition of new Safaitic inscriptions from the Jordanian Ḥarrah desert’, Arabian
Epigraphic Notes, 4, 1–24.
Online Corpus of the Inscriptions of Ancient North Arabia [OCIANA]. 2025. [Data-
base]. Available at: https://ociana.osu.edu/ (Accessed: 5 March 2025).
Pope, M.H. 1955. El in the Ugaritic texts. Leiden: Brill.

55
 ANCIENT ALLAH: AN EPIGRAPHIC RECONSTRUCTION

Prioletta, A. 2018. ‘New research on the Thamudic graffiti from the region of Ḥimā
(Najrān, Saudi Arabia)’, in Macdonald, M.C.A. (ed.) Languages, scripts and their
uses in ancient North Arabia. Oxford: Archaeopress, 53–70.
Robin, C.J. 2020. ‘Allah avant Muhammad’, Jerusalem Studies in Arabic and Islam, 49,
1–146.
——. 2021. ‘The Judaism of the ancient kingdom of Ḥimyar in Arabia: A discreet

Downloaded from https://academic.oup.com/jss/advance-article/doi/10.1093/jss/fgaf012/8129546 by guest on 16 May 2025

conversion’, in McDowell, G., Naiweld, R. & Stökl Ben Ezra, D. (eds) Diversity
and rabbinization: Jewish texts and societies between 400 and 1000 CE. Cambridge:
Open Book Publishers. 165–270.
Robin, C.J., al-Ghabbān, ʿA.I. & al-Saʿīd, S.F. 2014. ‘Inscriptions antiques de la région
de Najrān (Arabie Séoudite Méridionale): nouveaux jalons pour l’histoire de l’écri-
ture, de la langue et du calendrier arabes’, Comptes Rendus de l’Académie des Inscrip-
tions et Belles Lettres, 2014(3), 1033–1128.
Rohmer, J. 2021. ‘The political history of North-west Arabia from the 6th to the
1st century BCE: New insights from Dadān, Ḥegrā and Taymāʾ’, in Luciani, M.
(ed.) The archaeology of the Arabian Peninsula 2: Connecting the evidence. Vienna:
Austrian Academy of Sciences Press.
Ryckmans, G. 1950–1951. Corpus Inscriptionum Semiticarum: Pars Quinta, Inscrip-
tiones Saracenicae Continens: Tomus I, Fasciculus I, Inscriptiones Safaiticae. Paris:
E Reipublicae Typographeo.
Sinai, N. 2005. The Qur’an: A historical-critical introduction. Edinburgh: Edinburgh
University Press.
——. 2019. Rain-giver, bone-breaker, score-settler: Allāh in Pre-Quranic poetry (Amer-
ican Oriental Series, Essay 15). New Haven, CT: American Oriental Society.
Smith, M.S. & Bloch-Smith, E. 1988. ‘Death and afterlife in Ugarit and Israel’, Journal
of the American Oriental Society, 108(2), 277–284.
St. Takla Haymanout Coptic Orthodox Church. 2025. ‘‫حبا‬ ً ‫[ كلمات ترنيمة هبني‬Lyrics
of the hymn “Give Me Love”]’. St. Takla Haymanout Coptic Orthodox Website.
Accessed at https://st-takla.org/Lyrics-Spiritual-Songs/09-Coptic-Taraneem-Kalemat_
Heh-Waw-Yeh/Habny-Hoban.html, last accessed April 17, 2025.
Stour, M.C. 1980. ‘The Nether World and its denizens at Ugarit’, in Bendt Alster (ed.)
Death in Mesopotamia: papers read at the XXVIe Rencontre assyriologique interna-
tionale. Copenhagen: Akademisk Forlag, 227–238.
Testen, D. 2005. ‘Literary Arabic and early Hijazi: Contrasts in the marking of definite-
ness’, in Benmamoun, E., Eid, M. & Haeri, N. (eds) Perspectives on Arabic linguistics,
Vol. XI. Atlanta: 209–225.
Watt, W.M. 1971. ‘Belief in a ‘High God’ in Pre-Islamic Mecca’, Journal of Semitic
Studies, 16(1), 35–40.
Wilson-Wright, A.M. 2019. ‘Bethel and the persistence of El: Evidence for the
survival of El as an independent deity in the Jacob cycle and 1 Kings 12:25–30’,
Journal of Biblical Literature, 138(4), 705–720.
Winnett, F.V. 1957. Safaitic Inscriptions from Jordan. (Near and Middle East Series, 2).
Toronto: University of Toronto Press.
Winnett, F.V. & Harding, G.L. 1978. Inscriptions from Fifty Safaitic Cairns. (Near and
Middle East Series, 9). Toronto: University of Toronto Press.
Winnett, F.V. & Reed, W.L. 1970. Ancient Records from North Arabia. With contribu-
tions by J.T. Milik and J. Starcky. (Near and Middle East Series, 6). Toronto:
University of Toronto Press.
Yardeni, A. 2014. ‘A list of Arabic words appearing in Nabataean and Aramaic legal
documents from the Judaean Desert’, Scripta Classica Israelica, 33, 301–324.

56