Sometime around 3300 BCE, officials in the Eanna temple at Uruk began pressing signs into small pillows of clay. These earliest tablets, written in what scholars call proto-cuneiform, number about five thousand from Uruk alone. Within a few centuries the signs had become a full writing system able to record speech, and that system lasted longer than any other script in the ancient world, from before 3000 BCE until shortly after the time of Christ.

From tokens to tablets

Where did the first signs come from? The archaeologist Denise Schmandt-Besserat noticed that small clay objects in simple shapes, such as cones, spheres and disks, turn up at sites across the Near East from about 8000 BCE. Schmandt-Besserat proposed that writing grew out of small clay counting tokens. In her account, people first used tokens to stand for measures of grain or heads of livestock, then sealed them inside clay envelopes, then pressed the tokens on the outside of the envelope to show what was inside, and finally realized they could skip the tokens and just draw the marks. Her theory is influential, but critics point out that many tokens may have been game pieces or other objects and that the match between token shapes and later signs is imperfect. Most scholars accept that accounting played a large role in the birth of writing, even if the token story is not the whole explanation.

Wedges in clay

At first scribes drew signs with a pointed tool, but drawing curves in wet clay is slow and messy. Over time they switched to pressing the corner of a cut reed into the clay, which left marks with a triangular head. The word cuneiform means "shaped like wedges", from the Latin cuneus, a wedge. A tablet could be dried in the sun for a letter or receipt, or baked hard for a text meant to last.

One script, many languages

The first language written in cuneiform was Sumerian. Linguists classify Sumerian as an isolate with no known relatives: it is not related to Akkadian, Hebrew, Persian or any other known tongue. Akkadian, which came to dominate Mesopotamia, belongs to the Semitic family, along with Hebrew, Aramaic and Arabic. Scribes adapted cuneiform to write Akkadian and later many other languages, including Elamite in Iran, Hittite in Anatolia and Hurrian in the north. For centuries the script was the shared medium of diplomacy across the Near East.

Spoken Sumerian probably died out around 2000 to 1800 BCE, but it did not vanish. Scribes kept Sumerian as a learned language, much as medieval Europe kept Latin. Prayers, hymns and literary classics were copied and studied in Sumerian for nearly two thousand years after anyone spoke it at home, often with Akkadian translations written between the lines.

Schools and numbers

Becoming a scribe took years. The edubba, or "tablet house," was the scribal school where boys, and a small number of girls, learned to write. Students copied long lists of signs, words and names, then proverbs, model contracts and finally works of literature. Only a small share of the population could read and write, though some groups, such as merchants, learned enough to handle their own letters.

Scribes also learned mathematics. Mesopotamian scholars counted in a system based on sixty, combined with groupings of ten. Sixty divides evenly by two, three, four, five, six, ten, twelve, fifteen, twenty and thirty, which made fractions easy.

Breaking the code

After cuneiform died out, its meaning was lost for about seventeen centuries. The key came from a cliff in western Iran. The Behistun inscription, carved in three languages, was the key to deciphering cuneiform. In it, the Persian king Darius I described his rise to power in Old Persian, Elamite and Babylonian Akkadian. The British officer Henry Rawlinson copied it at great personal risk between 1835 and 1847. Old Persian was the simplest of the three and was cracked first, building on earlier work by Georg Grotefend. With it as a guide, Rawlinson, the Irish clergyman Edward Hincks and others worked out Akkadian.

Many people doubted their results, since Akkadian signs could have several readings. So in 1857 the Royal Asiatic Society in London sent a newly found inscription of the Assyrian king Tiglath-Pileser I to four scholars: Rawlinson, Hincks, William Henry Fox Talbot and Jules Oppert. Each translated it separately and sealed his work. The 1857 test showed that four scholars' separate translations largely agreed, which persuaded the world that cuneiform could truly be read.

Worked example: A schoolboy's tablet

Imagine a round clay tablet from a house in the city of Nippur, around 1750 BCE. On one side, in neat signs, the teacher has written a short proverb. On the other, in larger and shakier signs, a student has copied it, and you can see where he smoothed the clay to fix a mistake. What can they tell us? They show how literacy was passed on: by copying, repetition and memory, not by reading for pleasure. And they show that the great literature of Mesopotamia survives partly because generations of tired children copied it out, line by line, under a teacher's eye.