MYS0D7QBBHA32C1·Mystery · script

The Dhofari script

Xenia2024 · CC0

Where: Dhofar (Oman), al-Mahra (Yemen), Socotra; related texts in north-eastern Oman⁠[1]
Medium: Mostly painted in caves and rock shelters; some carved on wadi beds and stones⁠[2]
Scripts: Script 1 (vertical columns; variants 1a and 1b) and Script 2 (horizontal lines)⁠[1]
Date: Radiocarbon terminus post quem in the 4th century BC for three inscriptions⁠[1]
Decipherment: Al-Jallad, JEOL 49, published July 2025, from three halḥam abecedaries⁠[5]
Letters: 26 glyphs in Script 1a on Al-Jallad's analysis⁠[2]
Language: Extinct Semitic language related to, not ancestral to, Modern South Arabian⁠[5]
Open data: OCIANA Dhofari records (images CC BY); DASI for South Arabian comparanda (CC BY 4.0 data)⁠[12]
Status: Partly explained
Questions: 0 of 4 answered

People Ahmad Al-JalladEpigrapher (Ohio State); 2025 decipherment; OCIANA director Giuliano CastagnaLinguist (Beijing Normal University); South-east Arabian inscriptions, radiocarbon Geraldine M. H. KingEpigrapher; Dhofar Epigraphic Project M. C. A. MacdonaldEpigrapher (Oxford); classification of Arabian scripts ʿAlī Aḥmad al-ShaḥrīDhofari researcher; Dhofar Epigraphic Project

Catalog keyMYS0D7QBBHA32C1

Across the monsoon hills of Dhofar in southern Oman, and into al-Mahra in Yemen and on Socotra, caves and rock shelters carry short painted texts in an alphabet that matches none of the known scripts of Arabia⁠[1]⁠[2]. Travellers reported them from 1900, Bertram Thomas noted inscribed stones in 1932, and the first systematic record was the Dhofar Epigraphic Project of ʿAlī al-Shaḥrī and Geraldine King in 1991 and 1992, which sorted them into a vertical Script 1 and a horizontal Script 2⁠[1]⁠[3]. Michael Macdonald wrote in 2000 that they had defied decipherment and might not be very informative; a 2001 decipherment by Nayeem found no support⁠[4]⁠[1].

In July 2025 Ahmad Al-Jallad published a decipherment of Script 1⁠[5]. Three inscriptions, too long and too varied to be words, each contained more than 20 distinct letters without repeats; he read them as abecedaries and matched them to halḥam, the letter order of the South Semitic scripts. Comparing letter shapes with the other pre-Islamic Arabian scripts gave candidate values, and the order fixed the rest⁠[5]⁠[6]. On his analysis the letter shapes descend mainly from North Arabian scripts, especially Thamudic B, not from South Arabian, and the language merged interdental and sibilant consonants, which ties it to the eastern side of Arabia and rules out direct ancestry of Mehri and Jibbali, though it is related to them⁠[5]⁠[7].

The claim is recent and its testing has barely begun. Giuliano Castagna, who had surveyed and dated the inscriptions, called it the first reliable study in Science's report, and the Biblical Archaeology Society carried the story in 2025 and 2026⁠[8]⁠[7]⁠[6]⁠[9]. The paper is on Zenodo under CC BY with its files restricted, and its texts are being entered in the OCIANA corpus, edited by Al-Jallad⁠[5]⁠[10]. Of the 4 questions on this site, 0 are answered. The most likely answer to “Has the Dhofari script been deciphered?”, not yet confirmed: the letter order and most values are right, but readings of texts are too uncertain to call the corpus read.

Key questionsset by the editors

Has the Dhofari script been deciphered?Most likely: the letter order and most values are right, but readings of texts are too uncertain to call the corpus read

What language do the Dhofari inscriptions record?Most likely: an extinct Semitic language related to Modern South Arabian but not ancestral to it

All 4 questions

The object

The corpus is a body of mostly painted inscriptions in caves and rock shelters, with some carved on wadi beds and loose stones, from the Dhofar hills (the Qara range and its neighbours) west into al-Mahra and on Socotra, with related texts as far as north-eastern Oman⁠[2]⁠[1]⁠[11]. Al-Shaḥrī and King's survey recorded the bulk of the known texts in 1991 and 1992 in an unpublished report; Castagna's fieldwork found more that they had missed, and no total count has been published⁠[3]⁠[1].

Inscription near Haidi, Socotra, copied by Theodore BentJ. Theodore Bent, 1897. Public domainArtifact

Two script types are recognised. Script 1 is written in vertical columns and is the commoner, with fewer than 40 sign forms as recorded; Script 2 runs in horizontal lines⁠[1]⁠[2]. Al-Jallad distinguishes variants 1a and 1b within Script 1 and counts 26 letters in 1a⁠[5]⁠[2].

The data are scattered. The decipherment paper is a CC BY Zenodo record whose files require a log-in⁠[5]. OCIANA, the Ohio State successor to the Oxford corpus of Ancient North Arabian, has a Dhofari script category with records such as the abecedary KMD 28-31, each with photograph, tracing, DStretch image and transliteration, images marked CC BY⁠[12]⁠[10]. The Ancient South Arabian comparanda are in DASI, whose record data have been CC BY 4.0 since 2026⁠[13]. The script is not in Unicode.

Map of the regionOpen the full map

Region Country

+−

Dhofar Governorate, OmanRegion

A region

Area D, between the Wadi Arzat and Medinat al-Ḥaq roads, DhofarInscription site

A region

Al-Mahra, YemenRegion

A region

A regionNatural Earth

Timeline

Timeline

Squeezed where little happened. Each mark is a published claim, by how it stands:AcceptedMost likelyStill possibleDisputedRuled out

What is known

The inscriptions are alphabetic: the sign count is too small for a syllabary, and three texts list more than 20 distinct letters each without repetition⁠[1]⁠[6]. Their order matches halḥam, the letter order of the South Semitic family, closely though not perfectly⁠[5]⁠[6].

Abecedary KMD 28-31 (Dhofari 1a): photo — Abecedary KMD 28-31 (Dhofari 1a)G. M. H. King and M. C. A. Macdonald, 1991. CC BYArtifact

Abecedary KMD 28-31 (Dhofari 1a): enhanced — Abecedary KMD 28-31 (Dhofari 1a)G. M. H. King and M. C. A. Macdonald, 1991. CC BYArtifact

The date is provisional. Castagna sampled three inscriptions for radiocarbon in 2019, analysed in Vilnius in 2020, which gave a terminus post quem in the 4th century BC; popular accounts place both scripts around the end of the first millennium BC and their disuse before the 7th century AD⁠[1]⁠[6].

Content, on the new readings, is short: prayers that may have been protective, names and graffiti⁠[6]. Script 1a shows a preposition k- and a plural verb ending -m; Script 2 shows mimation on personal names⁠[2].

What has been tested

The decipherment has an internal test built in: the values come from the order of three abecedaries and from shape comparison, and they must then give sensible words when applied to the other texts. Al-Jallad reports first readings that do so, with forms paralleled in Modern South Arabian⁠[5]. Script 2 values have been proposed largely from those of Script 1a⁠[2].

Outside testing has not yet been published. The one recorded outside judgement is Castagna's, who called it the first reliable study⁠[7]. No published reanalysis of the abecedaries or independent reading of a text with the values fixed in advance has been found. Earlier claims were tested and failed: Castagna rejected Nayeem's 2001 readings for double values and unexplained transliterations⁠[1]⁠[14].

The readings

On the script, the decipherment holds that Dhofari is a South Semitic alphabet in the halḥam order whose letter forms derive from a North Arabian archetype close to Thamudic B, not from Ancient South Arabian⁠[5]. The earlier working assumption, reflected in the label "Dhofar inscriptions" and in comparisons with South Arabian, was a local or South Arabian development⁠[1].

On the language, Al-Jallad reads an extinct Semitic language related to the Modern South Arabian languages but not their ancestor, because it had lost the interdental consonants that Mehri and Jibbali keep⁠[5]. The alternative, that the texts are early Modern South Arabian, would require the interdentals to have been restored, which is not expected⁠[5]. Older attributions to the people of ʿĀd mentioned in the Quran are not scholarly readings⁠[7].

Published claims

Published claims and their status

Year	Who	Claimed	Status	Reason
1993	ʿAlī al-Shaḥrī and Geraldine King⁠[3]	The Dhofar inscriptions are written in two related scripts, a vertical Script 1 and a horizontal Script 2.	Accepted	Later surveys and the 2025 decipherment keep the division, adding variants within Script 1⁠[1]⁠[5]
2001	Muhammad Abdul Nayeem⁠[14]	Some of the Dhofar inscriptions can be read, in scripts he named Sa'kalhanic A and B.	Disputed	Castagna rejected the readings because they give some signs two values and leave others unexplained, and no one has taken them up⁠[1]
2024	Giuliano Castagna⁠[1]	Radiocarbon samples from three inscriptions give a terminus post quem in the 4th century BC.	Most likely	No result contradicts it, but it rests on three samples, and the other inscriptions are undated⁠[1]
2025	Ahmad Al-Jallad⁠[5]	Three inscriptions are abecedaries in the South Semitic halḥam order, and they give the sound values of Script 1a.	Most likely	The only decipherment, called the first reliable study by an uninvolved specialist, but no detailed critique or independent reading has been published⁠[7]⁠[8]
2025	Ahmad Al-Jallad⁠[5]	The letter shapes derive mainly from a North Arabian script close to Thamudic B, not from Ancient South Arabian.	Most likely	It follows from his shape comparisons, which no one else has yet checked; the earlier assumption of South Arabian origin was never argued in detail⁠[1]
2025	Ahmad Al-Jallad⁠[5]	The language is an extinct Semitic language related to Mehri and Jibbali but not their ancestor, because it had lost the interdental consonants.	Most likely	It follows from the 26-letter inventory if the decipherment holds; the first readings are short and few⁠[2]⁠[6]

What would settle it

Independent readings by other epigraphers of texts not used in the paper, with the values fixed in advance, would test the decipherment; so would a fourth abecedary found later that fits the same order and values⁠[5]. A published, complete corpus of the inscriptions with images and site data, which OCIANA is beginning to provide, is the precondition for both, and more radiocarbon dates on the pigments would fix the chronology⁠[12]⁠[1].