🏔️ Behistun: one cliff, three cuneiform languages, and the sealed-envelope test of 1857

Darius I had his account of his rise carved on a cliff in western Iran in Old Persian, Elamite and Babylonian. Rawlinson copied it between 1835 and 1847, but the reading of Akkadian was proved somewhere else: in 1857 four scholars translated an Assyrian text independently and sent their versions in sealed envelopes.

Tags: cuneiform, decipherment, old-persian, akkadian, elamite

Darius I had his own account of how he took the Persian throne carved into a limestone cliff at Bisotun (Behistun), about 100 metres above the ground. The cliff stands on the old road from Babylon to Ecbatana, in what is now Kermanshah province in western Iran. The relief shows Darius with his foot on Gaumāta, the man he says seized power in 522 BCE, facing a line of rebel kings bound at the neck. Around the relief the same story is written three times, in three languages, all of them in wedge-shaped signs: Old Persian, an Indo-European language; Elamite, an isolate; and Babylonian, a dialect of Semitic Akkadian. The site was added to the UNESCO World Heritage List in 2006.

ElamiteAkkadian
🌍 Fly to Bisotun Open the interactive map
⚖️ Compare "name" in the three languages of the inscription
LanguageFamilySpeakersNameWaterOne
Old PersianIndo-European (Iranian, Old, Southwestern)Extinct (~6th-4th c. BCE; royal Achaemenid)𐎴𐎠𐎶/naːma/𐎠𐎱𐎡/aːpi/𐎠𐎡𐎺/aiva/
ElamiteLanguage isolate (Elamite)Extinct𒄭𒅖/hiʃ/—/—/𒆠𒅕/kir/
AkkadianAfro-Asiatic (Semitic, East)Extinct𒈬/ʃumu/𒀀/muː/𒁹/iʃteːn/
🔍 Open Akkadian — Akkadian

Three versions, carved in stages

The Elamite text was carved first. Elamite was the main written language of the Achaemenid administration: the Persepolis Fortification tablets of 509–493 BCE are mostly in Elamite. The Babylonian and Old Persian versions were added later. After Darius's campaign of 520–519 BCE a figure of the Saka chief Skunkha was added to the end of the line of captives. It was cut straight into the first Elamite text, so the Elamite was copied out again to the left of the Old Persian. In a badly damaged passage of the fourth Old Persian column (§70), Darius says he made a form of writing in ariyā, "Aryan", that had not existed before. This is read as a claim that the Old Persian script was created on his orders. The surviving lines are hard to make out on the rock, published readings differ, and two new studies of the passage appeared in 2024, one by Marco Fattori and one by Saber Amiri Parian and Keyvan Mahmoudi.

Copying the rock

Henry Creswicke Rawlinson, an East India Company officer posted to Kermanshah, worked on the monument from 1835 to 1847. He began copying the Old Persian text in the 1830s, and in 1844 he went back, recopied the Old Persian and copied the Elamite. Old Persian was the way in, because its script has only about 36 sound signs plus a word divider. Georg Friedrich Grotefend had already picked out the names Darius, Xerxes and Hystaspes in inscriptions from Persepolis in 1802. Rawlinson published the Old Persian text with a translation in the Journal of the Royal Asiatic Society in 1846–47, and Edwin Norris published the Elamite version in 1855. The three languages have almost nothing in common: the word for "name" is nāma in Old Persian, hiš in Elamite and šumu in Akkadian.

The Babylonian text is on a separate block to the left of the relief, and it cannot be reached from the ledge. Rawlinson's own report, printed in Archaeologia in 1851, says the local climbers declared that block unapproachable, until in the autumn of 1847 "a wild Kurdish boy, who had come from a distance, volunteered to make the attempt". The boy squeezed his way up a cleft in the rock, fixed ropes, and took paper squeezes of the signs while hanging in front of the cliff face.

Grotefend's three names, and "name" in all three languages
CuneiformSign by signReading
𐎭𐎠𐎼𐎹𐎺𐎢𐏁da-a-ra-ya-va-u-šaDārayavauš Darius
𐎧𐏁𐎹𐎠𐎼𐏁𐎠xa-ša-ya-a-ra-ša-aXšayāršā Xerxes
𐎻𐎡𐏁𐎫𐎠𐎿𐎱vi-i-ša-ta-a-sa-paVištāspa Hystaspes
𐎴𐎠𐎶na-a-manāma "name" (Old Persian)
𒄭𒅖hi-išhiš "name" (Elamite)
𒈬MUšumu "name" (Akkadian, one logogram)

Not the key to Akkadian

Behistun is called the Rosetta Stone of cuneiform, but for Akkadian the label overstates its part. Rawlinson's copy of the main Babylonian text was not published until the end of 1851. By then Edward Hincks, rector of Killyleagh in County Down, Ireland, had already worked from other Assyrian and Babylonian inscriptions. In 1846 he showed that the Babylonian script spelled out syllables and that its language had much in common with the Semitic languages. In 1847 he published a list of 76 sign values, and in 1850 he explained logograms and the use of several signs for the same sound. Kevin Cathcart concluded in 2011 that the Behistun Babylonian played no part in the first stage of the decipherment of Akkadian.

Four sealed envelopes, 1857

The test that settled the question was made on a different text. In March 1857 William Henry Fox Talbot, one of the inventors of photography, sent the Royal Asiatic Society his translation of a newly lithographed inscription in a sealed packet. The inscription was the clay-prism text of the Assyrian king Tiglath-Pileser I (reigned about 1114–1076 BCE), from Ashur. Rawlinson was already at work on his own translation. At his suggestion the Society's secretary, Edwin Norris, asked Hincks to make one as well, and the Paris-based Julius Oppert offered to try too; each worked without seeing the others' versions. On 20 April 1857 Norris wrote to Hincks that the translations would stay sealed until they were opened at the May meeting. In late May a committee that included the historian George Grote, William Whewell of Trinity College, Cambridge, and the Egyptologist Sir Gardner Wilkinson compared the four. Hincks and Oppert had had less time and sent shorter versions, but where the translations covered the same passages they largely agreed. The Society took this as proof that Akkadian could be read, and the four translations were printed side by side that year.

The Babylonian version on the rock is badly damaged, and the records of it have not lasted either. Rawlinson's squeezes from 1844 and 1847 were already decaying when Leonard King and Reginald Campbell Thompson climbed back up to the inscription in 1904, and they have since disintegrated. Elizabeth von Voigtlander's 1978 edition was based mainly on latex squeezes that George Cameron made in 1957, and those have deteriorated too. Johannes Hackl went back over the evidence in 2025. He concludes that her additional readings cannot be verified and that about two-thirds of the Babylonian text should be considered lost.

Sources

More on this

See also: Trivia · Word Map — 168 core words · Han Map — Han character readings

Open the interactive map: Word Map · Han Map · Name Map · Time Map · Word Order · Tree

Part of LangMap — a linguistic visualization project. This is a static, crawlable summary; the interactive maps offer pronunciation audio, filters, and a globe view.

Made by Heuron Inc. · GitHub · cho@heuron.com · LangMap home