A few Safaitic inscriptions were even found in Pompeii, on the walls outside a small theatre, probably scribbled by Arabian members of the Roman army.

Inscriptions, Al-Jallad explained, tend to cluster on higher ground, where nomadic herders could keep an easier watch for predators. In a landscape with no other traces of human civilization, the rocks preserved the nomads’ names and genealogies, along with descriptions of their animals, their wars, their journeys, and their rituals. There were prayers to deities, worries about the lack of rain, and complaints about the cruelty of Romans.

Al-Jallad began pulling up every inscription that mentioned migrating in search of rain, and soon he had a long list of terms that had resisted translation. Comparing them with the Greek, Aramaic, and Babylonian zodiacs, he started making connections. Dhakar matched up nicely with dikra, the Aramaic word for Aries, and Amet was derived from an Arabic verb meaning “to measure or compute quantity”—a good bet for the scales of Libra. Hunting for Capricorn, the goat-fish constellation, Al-Jallad found the word ya’mur in Edward Lane’s “Arabic-English Lexicon,” whose translation read, “A certain beast of the sea, or . . . a kind of mountain-goat.” He stayed up all night, sifting the database and checking words against dictionaries of ancient Semitic languages. By morning, he had deciphered a complete, previously unknown Arabian zodiac. “We’d thought that they were place names, and, in a way, they were,” he told me. “They were places in the sky.”

Michael Macdonald amassed a vast collection of photographs of these texts and launched a digital Safaitic database, with the help of Laïla Nehmé, a French archeologist and one of the world’s leading experts on early Arabic inscriptions. “When we started working, Michael’s corpus was all on index cards,” Nehmé recalled. “With the database, you could search for sequences of words across the whole collection, and you could study them statistically. It worked beautifully.”

Enno Littmann, an Orientalist who visited Syria in 1899, with a contingent from Princeton University, and completed the decipherment, labored over what he found on the rocks.

In 1877, an Orientalist from Ottoman Edirne deciphered most of the alphabet, bringing the language of the inscriptions into blurry focus.

he effort to decode the Safaitic texts began in the spring of 1857, when a young Scotsman named Cyril Graham set off from Jerusalem on a tour of Syria.

The history of Arabia just before the birth of Islam is a profound mystery, with few written sources describing the milieu in which Muhammad lived. Historians had long believed that the Bedouin nomads who lived in the area composed exquisite poetry to record the feats of their tribes but had no system for writing it down. In recent years, though, scholars have made profound advances in explaining how ancient speakers of early Arabic used the letters of other alphabets to transcribe their speech. These alphabets included Greek and Aramaic, and also Safaitic; Macdonald’s rock was one of more than fifty thousand such texts found in the deserts of the southern Levant. Safaitic glyphs look nothing like the cursive, legato flow of Arabic script. But when read aloud they are recognizable as a form of Arabic—archaic but largely intelligible to the modern speaker.

It was Safaitic, an alphabet that flourished in northern Arabia two millennia ago, and Al-Jallad and Macdonald are among a very small number of people who can read it.

The effort to decode the Safaitic texts began in the spring of 1857, when a young Scotsman named Cyril Graham set off from Jerusalem on a tour of Syria