The inscriptions from Almosi Gorge

The photographs provided to me by the Archaeology Department of the National Museum of Tajikistan showed that the inscriptions were still on large rocky outcrops, standing at the corner of a rectangular space surrounded by a stone wall that measured 115 by 51 metres (visible on Google Earth) (Figure 2), and the third discovery (Figure 3) was a piece that had detached and fallen off the larger rocky outcrop (Figure 4). The inscriptions, still on the outcrops, were written in a curious script that has not yet been deciphered—the so-called ‘unknown script’ (Figures 5–7). The inscription on the detached piece of rock was written in Greek script, but not in the Greek language. It was immediately apparent to me that this inscription was written in the Middle Iranian language of Bactrian because I could read on it the name and title of a king already known from two other Bactrian inscriptions: ϷΑΟΝΑΝΟ ϷΑΕ ΟΟΗΜΟ ΤΑΚΤΟΕ (shaonano shae ooēmo taktoe) ‘of Wima Takto King of Kings’. The initial letters of the first two words are forms of the Greek rho introduced into the script to represent the Bactrian consonant sh. The Ν and the Ο at the end of ϷΑΟΝΑΝΟ are merged into each other as a ligature. For some reason, there are gaps in the letters of the king's name: line 2 ΟΟΗ, line 3 ΜΟ ΤΑ–Κ, line 4 Τ–ΟΕ. The surface of the rock is very uneven, so the scribe seems to have been avoiding places that were unsuitable for cutting the inscription, unless a closer examination of the stone might reveal a different reason. A curious hooked line extends from the lower right corner of the second Ο of ΟΟΗΜΟ and Lurje has suggested (private correspondence) that this makes the letter resemble the Greek numeral 6 (written with the obsolete Greek letter digamma Ϛ) but the context of the other letters spelling out the king's name suggests that the hooked line could just be a scratch or flaw in the stone.

Figure 2.

The site of the enclosure (bottom-centre) to which the larger rocky outcrop with inscriptions was adjoined on a promontory above a confluence of the Khanaka River and one of its tributaries. Source: image taken from Google Earth Pro, 3 February 2023, facing south-east.

Figure 3.

(a) Bactrian inscription naming Kushan king of kings, Wima Takto; (b) drawing of inscription. Source: image courtesy of the National Museum of Tajikistan; drawing by author.

Figure 4.

The first and largest rocky outcrop with the first inscription in the unknown script and from which the Bactrian Kushan inscription (as indicated by the arrow) had broken and fallen. Source: image courtesy of the National Museum of Tajikistan.

Figure 5.

The inscription in an unknown script on the first rocky outcrop. Source: image courtesy of the National Museum of Tajikistan.

Figure 6.

The National Museum of Tajikistan team documenting the inscriptions in an unknown script on the second rocky outcrop. Source: image courtesy of the National Museum of Tajikistan.

Figure 7.

The first part of the second inscription in an unknown script. Source: image courtesy of the National Museum of Tajikistan.

The inscription revealed itself to be in the name of the second Kushan king, Wima Takto (circa 90–113 CE). The Kushan Dynasty ruled a state that occupied a central position on the ancient Silk Road, spanning southern Uzbekistan, Tajikistan, Afghanistan, Pakistan, Kashmir, and northern India, during the first to fourth centuries CE. According to the Chinese historical text Hou Han Shu, the Late Han Chronicle (Hou Hanshu 88.2921),Footnote ² the first Kushan king, Kujula Kadphises, was originally a subruler of the Da Yuezhi kingdom that controlled southern Uzbekistan, Tajikistan, and northern Afghanistan from the late second century BCE to the early first century CE. He came to power in circa 50 CE by overcoming the other subrulers of the Da Yuezhi. Kujula Kadphises extended his domain to include the Begram-Kabul region of southern Afghanistan, the Taxila, and Sind regions of Pakistan and Kashmir. Wima Takto was Kujula Kadphises's son, who, according to the Hou Hanshu, expanded Kushan rule into north-western India (Hou Hanshu 88.2921).Footnote ³

I immediately responded to Abdulrahman Sharifzoda and told him all this information so that he and his colleagues would know the importance of what they had discovered. The part of the inscription before the name and title was a mystery to me, as I am only familiar with Bactrian from its numismatic use and by reading articles about it, so I sent the images to the leading expert on the Bactrian language, Nicholas Sims-Williams at the Ancient India and Iran Trust, Cambridge. He responded, confirming my reading of the king's name and title, and that I was correct to understand it as representing a possessive form, as the nouns were in the oblique case used in Bactrian to indicate possession. He recognised the first word in the inscription as ΕΙΔΙ meaning ‘this’, but not what followed: ΗΛΟΥΔΟ…, the last three letters disappearing off the broken edge of the stone. He also commented that the same word, in the form ΕΙΔΟ, appeared at the beginning of another Kushan royal inscription, found at the Surkh Kotal monument in northern Afghanistan. This is the second time that an inscription in Bactrian naming this king, paralleled by an inscription in the ‘unknown script’, has been found but, like the new inscriptions, the first one, found at Dašt-i-Nawar in southern Afghanistan,Footnote ⁴ is damaged. There are many similarities between the curious scripts found in both places.

First notices of the new inscriptions

In the first formal announcement of the discovery,Footnote ⁵ Sims-Williams's reading of the Bactrian inscription was mentioned and the connection of the inscriptions with the trilingual inscription found at Dašt-i-Nawar in southern Afghanistan was made.

The National Museum had also shared the images with a Russian scholar, Pavel Lurie (State Hermitage Museum). He proposed a different reading of the Bactrian, but Sims-Williams informed him that the royal name and title had been read in the Bactrian. Since then, he has been working on the unknown script (personal communication) and gave a presentation in Moscow on 25 January 2023 describing the discovery of the inscriptions and focussing on his analysis of the unknown script, showing his own reading of the Bactrian along with that suggested by myself and Sims-Williams. He suggested that the script was a descendant of Aramaic and that the letters of the script were syllabic, with vowel values marked.

In October, the National Museum also shared the images with the Afghan scholar, Davary, and he has since published a booklet on the inscriptions.Footnote ⁶ He repeated the Bactrian reading provided earlier to the Museum, but misreading some of the letters and suggesting that the Bactrian text contained scribal mistakes or represented a variant dialect.

On 11 November 2022, a Tajikistan news agency, Avesta Information Agency, posted a blog about the discovery, Tadzhikskiye uchenyye rasshifrovyvayut drevniye nadpisi, obnaruzhennyye v ushchel'ye Almosi, with a video of the archaeology team at the site; it was by Bobomulloev's cousin, a member of the Institute of History, Archaeology and Ethnography of the National Academy of Sciences of Tajikistan, and quoted Sims-Williams's reading of the Bactrian inscription.Footnote ⁷ This was subsequently reblogged by other sites and was also published online with an English translation by Central Asian Archaeological Landscapes.Footnote ⁸

Following the announcements of the discovery, a team of linguists comprising Svenja Bonmann, Jakob Halfmann, and Natalie Korobzow, who were based at Cologne University, began to work on the new inscription in the ‘unknown script’ and, at a conference on the new inscriptions held in Tajikistan on 1 March 2023, they announced their findings, characterising the script in the same way as Lurje but offering a key to the script that enabled them to recognise parallels in the script on the large rocky outcrop (Figure 5) with the royal name and title in the Bactrian inscription. The Cologne team had been interrogating the ‘unknown script’ since May 2021, when they started work on the Dašt-i-Nawar inscription and its cognates. This previous engagement enabled them to rapidly identify the language of the inscription as a previously unknown Middle Iranian language, related to but different from Bactrian.Footnote ⁹ They also showed that the same name and title together with the word Kushan could also be read in the ‘unknown script’ at Dašt-i-Nawar and recognised further titles of the king in the previously unrecorded Middle Iranian language paralleling those in the Bactrian text of the Dašt-i-Nawar inscription. The relationship between Bactrian and the language of the ‘unknown script’ proposed by Bonmann, Halfmann, and Korobzow is also evident in the solution proposed by Harry Falk,Footnote ¹⁰ based on his reading of Davary's article, that the ‘unknown script’ is a transcription of the Bactrian.

The discovery of further elements and examples of the script will hopefully confirm which solution is correct. Both the Cologne team and Falk offered explanations for the first line of the Bactrian inscription. The former explained the start of the line as the Bactrian ɛιδιηλο,Footnote ¹¹ meaning ‘this’, an alternative form of the word ɛιδο cited by Sims-Williams, followed by an illegible word beginning υ…. Falk's reading was ɛιδι ‘HΛΟΥAΟ or, less likely, ΗΛΟΥA[Θ](Ο)’,Footnote ¹² which he interpreted as ‘a hypothetical elūka-stha(na)’, i.e. place for relics, indicating a burial ground.Footnote ¹³ Davary explained the first line as reading ‘ɛιδι ηαυ(ν)ο’, which he interpreted as meaning ‘dies ist de Eiwān (Terrasse)’.Footnote ¹⁴ All the readings propose meanings that are so far unattested in Bactrian but, given the limited number of texts so far found, this is not unlikely. Bonmann et al. have commented on Falk's reading and shown the problems inherent in reading the inscription as a transcription of the Bactrian.Footnote ¹⁵ The validity of their understanding is further enhanced by their application of their method to the Dašt-i-Nawar inscription and the fragmentary inscription from the Hoq cave on Socotra.Footnote ¹⁶

The identification by Bonmann et al. of the previously ‘unknown script’ as representing an Iranian language close to but distinct from Bactrian recalls Sims-Williams's analysis of the names of Kushan kings.Footnote ¹⁷ He cautiously suggested that the names represented forms from a previously unknown Middle Iranian language—a form of Saka/Scythian, relating to the Kushans or the Da Yuezhi.Footnote ¹⁸ Bonmann et al. also cautiously speculated that the newly discovered language could connect with the Da Yuezhi or Kushan.Footnote ¹⁹ It is suggestive that the Kushan kings used Iranian names from an unknown Iranian language and that, early in their period, they employed an unknown Iranian language, using a script distinct from Bactrian and Kharoṣṭhī, for writing royal inscriptions.

What the new inscription tells us

The importance of the Bactrian inscription and its ‘unknown script’ companion lies in the clarity of the king's name, in its location, and in the further confirmation of the Bactrian reading of the Dašt-i-Nawar inscription as proposed by Sims-Williams.Footnote ²⁰ Until 1993, the name of this king, the second ruler of the Kushan Dynasty, was not recognised where he was mentioned in the Hou Hanshu, the Late Han Chronicle, or in the two other inscriptions in which it appeared at Dašt-i-Nawar Footnote ²¹ and in a Brahmi inscription at Mathura in northern India.Footnote ²² His name was correctly read for the first time on an inscription first discovered in northern Afghanistan at a location known today as Rabatak, published by Sims-Williams and Cribb from photographs sent from Afghanistan by workers for the Halo Trust who were clearing landmines nearby.Footnote ²³ Sims-Williams later had the opportunity to improve on his reading by conducting a direct examination of the inscription, now in the Kabul Museum.Footnote ²⁴ This inscription, also written in Bactrian using Greek script, recorded the erection of a royal shrine on behalf of the best-known Kushan king, Kanishka I (circa 127–51 CE), and included a list of his royal ancestors back to Kujula Kadphises, naming Kujula Kadphises as his great-grandfather, Wima Takto as his grandfather, and Wima Kadphises as his father.

Another inscription could be an inscription of Wima Takto. It is inscribed on a large linga found in northern India, housed in a Shaivite temple in Rey (Reh) village, Fatehpur District, Uttar Pradesh (Figure 8).Footnote ²⁵ Although initially attributed to an Indo-Greek king, it was soon recognised as being of the early Kushan period by Gupta.Footnote ²⁶ The inscription is written in Brahmi in a form that suggests a date in the first to second century CE. It lists the titles of a king using the terminology used for both Wima Takto in the Dašt-i-Nawar inscription and Kanishka I in the Rabatak inscription. The inscription reads mahārājasa rājarājasa mahāmtasa trātārasa dhāmmīkasa jayamtasa ca apra[tihatasa] … (great king, king of kings, the great, the saviour, the just, the victorious and the righteous), matching the royal titles of these two kings in Bactrian: rājarājasa = ϷΑΟΝΑΝΟ ϷΑΟ, mahāmtasa = ΣΤΟΡΓΟ (great), trātārasa = ΒΩΓΟ (saviour), dhāmmīkasa = ΛΑΔΕΙΓΟ (just), jayamtasa and apra[tihatasa] = ΟΑΝΙΝΔΟ (victorious and undefeated). Unfortunately, the last line is partially missing. The tops of the letters would fit the name of Wima Takto, but not any other early Kushan king.

Figure 8.

Map of territory included in the Kushan empire showing locations of Kushan-related sites (●) and Kushan royal inscriptions (). Source: drawn by Robert Bracey.

The repetition of the first part of Wima Takto's name as the first part of his son Wima Kadphises's name was the reason why his name had previously not been recognised in inscriptions. The Hou Hanshu refers to Wima Takto, but only transcribed the first part of his name, so that it was also considered by many as a reference to Wima KadphisesFootnote ²⁷ and some continued to assert so after the Rabatak inscription was published.Footnote ²⁸

Wima Takto's name also appears on three types of coinsFootnote ²⁹ but, before the Rabatak inscription was discovered, two of these coins were thought to refer to Wima Kadphises: a so-called Soter Megas (from its Greek inscription) type issued in Gandhāra and a bull-camel type from Kashmir;Footnote ³⁰ the third (Figure 9), with the same tamga royal symbol as the Gandhāra coin, was only read after the discovery of the Rabatak inscription.Footnote ³¹ This last coin depicts a Kushan king enthroned on a seat with lion legs, mirroring the statue of a Kushan king found at MathuraFootnote ³² that has an inscription on its base naming Wima Takto (Figure 10). The coin and Mathura enthroned representations of Wima Takto also recall two other images of enthroned early Kushan kings at Surkh KotalFootnote ³³ and Khalchayan (Pugachenkova 1971, fig. 54),Footnote ³⁴ suggesting that they may also be images of this king.

Figure 9.

(a) Coin of Wima Takto showing enthroned king inscribed with his name in Bactrian: ΟΟΗΜΟ ΤΑΚΤΟΟ [ϷΑΟ], Soter Megas tamga in left field, Pankaj Tandon collection no. 682.90; (b) drawing of obverse. Source: photograph by Pankaj Tandon, with permission; drawing by author, based on this example and on the example in the British Museum, 1922,0423.27.

Figure 10.

Inscribed image of Wima Takto enthroned, Government Museum, Mathura (still labelled Wima Kadphises). Source: photograph by Robert Bracey, with permission.

Wima Takto, the second Kushan king, no longer to be doubted

The first person to suggest that there was a Kushan king between Kujula Kadphises and Wima Kadphises was MacDowall,Footnote ³⁵ but his name was not known until the Rabatak inscription. The worn state of Rabatak, as first published from photographs, left room for doubt and three leading scholars working on the Kushans rejected Sims-Williams's reading of the name of Wima Takto.Footnote ³⁶ MukherjeeFootnote ³⁷ contradicted Sims-Williams and asserted that the inscription read Sadashkano, naming a son of Kujula Kadphises who was mentioned in a Buddhist donative inscription.Footnote ³⁸ He acknowledged the occurrence of the name of Wima Takto on some of the coins mentioned above and in the inscription at Mathura, but identified them as references to Wima Kadphises, and attributed the Soter Megas coinage to him too.Footnote ³⁹ Fussman also rejected Sims-Williams's readingFootnote ⁴⁰ and doubted the reading of Wima Takto's name in the inscriptions at Mathura and at Dašt-i-Nawar.Footnote ⁴¹ He discounted including the Soter Megas coinage as relating to Wima Takto, attributing it instead to a usurper who interrupted Kushan rule between Kujula Kadphises and Wima Kadphises.Footnote ⁴² MacDowall used Mukherjee and Fussman's analyses to cast doubt on the reading of Wima Takto's name in the Rabatak inscription and followed Fussman in casting doubt on the presence of his name in the inscriptions at Dašt-i-Nawar and Mathura.Footnote ⁴³ He also abandoned his earlier suggestion that there was a Kushan ruler in between Kujula Kadphises and his grandson, Wima Kadphises, suggesting that the Soter Megas coinage should be attributed to the former and the bull-camel coins naming Wima Takto to the latter.Footnote ⁴⁴ Fussman's idea of a usurper issuing the Soter Megas coinage was later taken up by Bopearachchi when publishing a gold coin of Wima Kadphises naming his father, Wima Takto.Footnote ⁴⁵ He accepted the existence of Wima Takto as a son of Kujula Kadphises, but asserted that, between Wima Takto's father and son, a usurper had denied him rule and taken over most of the Kushan territory. Although many scholars are now persuaded by Sims-Williams's reading and recent analyses of the appropriate coins,Footnote ⁴⁶ the views of Mukherjee, Fussman, and MacDowall have a continuing appeal. The recent multivolume history of India, for example, cites their work as the reason for rejecting the analyses by Sims-Williams and CribbFootnote ⁴⁷ on the place of Wima Takto in the Kushan dynastic list as ruler of the whole Kushan realm.Footnote ⁴⁸

The new inscription spells the name of Wima Takto clearly, clarifying the readings of his name in the other inscription. It also adds further to the implausibility of the suggestion that a usurper disrupted Kushan rule and issued the Soter Megas coin across the territory previously ruled by Kujula Kadphises and beyond into India. The inscriptions mentioning Wima Takto have been found in places right across Kushan territory, at Mathura and possibly Rey in northern India, at Dašt-i-Nawar in southern Afghanistan, at Rabatak in northern Afghanistan, and now at Almosi in Tajikistan (Figure 8). Apart from the four coin types mentioned above that include the name Wima Takto, two examples of the Soter Megas coin type from Gandhāra,Footnote ⁴⁹ the bull-camel coins from Kashmir,Footnote ⁵⁰ the coin type showing an enthroned king with the Soter Megas tamga of unknown issue place,Footnote ⁵¹ and the gold coin of his son,Footnote ⁵² there are also Soter Megas coin types with the initial of this king written as vi in Kharoṣṭhī script from Gandhāra and Bactria,Footnote ⁵³ two of which (Figure 11) have the first letters of his name in Greek ΟΟ or Ο.Footnote ⁵⁴ Soter Megas coins have been found across the Kushan-controlled territory from southern Uzbekistan and Tajikistan in the north to Mathura in the south, so their association with Wima Takto and the evidence of the inscriptions show his rule right across the territory held by his father and most of that held by his successors.

Figure 11.

Coin of Wima Takto, Soter Megas type for Bactria, inscribed with his initial vi in Kharoṣṭhī on obverse and Greek O at end of reverse inscription ΒΑΣΙΛΕΥ ΒΑΣΙΛ[ΕΩΝ ΣΩΤΗΡ ΜΕ]ΓΑΣ Ο. Source: Tandon collection no. 581.71, photograph by Pankaj Tandon, with permission.

The evidence for the reign of Wima Takto that can now be assembled from the recognition of his name on inscriptions and on coins from Tajikistan to northern India coincides with all parts of the territory that can be associated with his reign by deduction from the mention of him in the Hou Hanshu. The Chinese chronicle identified Wima Takto as the extender of Kushan rule into India and implied that he was the successor to the territories of the former Yuezhi domain in Tajikistan, Uzbekistan, and northern Afghanistan, which it tells us that his father seized control of those he conquered in central Afghanistan, Pakistan, and Kashmir. The distribution of his inscriptions and of his coins presents a tangible dimension to his rule that was previously unavailable. His achievements can no longer be attributed to his son.

The discovery by Muhsin Bobomulloev and his colleagues at the National Museum Tajikistan adds another piece of vital evidence for the progression and extent of Kushan rule at the beginning of the second century CE, and confirms the information in the Hou Hanshu that Wima Takto was the son of the first Kushan king, Kujula Kadphises. It also confirms the identification of the statue found at Mathura as an image of Wima Takto. The Museum team plans to return to the site to further search for evidence of the nature of the structure to which the rock bearing the inscriptions was attached and further discoveries are awaited. I understand that the Bactrian inscription has now been removed from the site and is safely installed at the National Museum of Tajikistan, Dushanbe.Footnote ⁵⁵