Cuneiform

Cuneiform is a writing system of the ancient Near East that developed in the late fourth millennium BCE; the latest texts that can be dated come from the first century CE. Its name comes from the Latin word cuneus, “wedge”, because its signs are made of wedge-shaped impressions. Scribes pressed the end of a cut reed into soft clay; each press left a mark with a triangular head and a thin tail.

Cuneiform is, together with Egyptian hieroglyphs, one of the oldest known writing systems in the world. It arose in southern Mesopotamia, the land between the Tigris and Euphrates rivers in what is now Iraq. The language behind its earliest tablets is uncertain, though later cuneiform clearly recorded Sumerian. Over three thousand years it was adapted for more than a dozen languages, from Anatolia (modern Turkey) to Iran.

Cuneiform is not an alphabet. A single sign could stand for a whole word (a logogram), for a syllable such as ba or ku, or for a usually silent “classifier” (a determinative), placed before or after a word, telling the reader that it is, for example, a god, a place or a wooden object. Reading cuneiform means working out from context which of these roles each sign plays.

In its mature form, every sign is built from just a few basic strokes: horizontal, vertical and diagonal wedges, plus a small hook-like mark made with the corner of the stylus. Simple signs need only one or two wedges, while complicated ones can have twenty or more. Experienced scribes wrote quickly, and their handwriting, like ours, changed from place to place and century to century, which helps scholars date tablets.

Mesopotamia had few trees and little stone, but its rivers supplied endless clay and reeds. A scribe kneaded clean clay, shaped it into a tablet that could fit in the palm of a hand or be as large as a book, and smoothed the surface. With a stylus of reed, cut to a sharp edge, he or she pressed the signs into the clay while it was still damp.

Early signs were drawn with a pointed tool. By the middle of the third millennium BCE scribes used a stylus with a wedge-shaped tip that was pushed rather than dragged, producing the typical wedges. By turning the stylus, a writer could make horizontal, vertical and diagonal wedges and a small corner-shaped mark, and combine them into signs.

Most tablets were simply left to dry in the sun. Dried clay can be softened and reused, and many everyday tablets were thrown away. Paradoxically, disasters helped preservation: when palaces and libraries burned, their tablets were baked hard in the fire and survived for thousands of years. Important documents were sometimes fired deliberately, and letters and contracts could be sealed inside clay envelopes to prevent tampering.

Cuneiform was also carved on stone monuments, such as royal inscriptions and law steles, cast in metal, and written on hinged boards coated with wax, which could be erased and reused but rarely survive.

Cuneiform began with bookkeeping. In the fourth millennium BCE the cities of southern Mesopotamia, above all Uruk, grew into large communities organised around temples and administrative institutions. These institutions collected, stored and distributed huge quantities of barley, animals, textiles and beer, and they needed ways to keep track of who owed and received what.

Long before this, people across the Near East made small clay tokens of different shapes, many of which seem to have served as counters. By the mid-fourth millennium tokens were sometimes sealed inside hollow clay balls (bullae), with impressions on the outside. Flat clay tablets impressed with number signs also appear before the first tablets combining numbers with signs for goods, people and institutions. These finds tie early writing to accounting, but the purpose of many tokens and how directly they relate to later signs are still debated.

The earliest texts belong to the Uruk IV phase, conventionally dated to about 3350–3200 BCE. The German excavations at Uruk between 1928 and 1976 produced nearly 5,000 proto-cuneiform tablets and fragments, assigned to the Uruk IV and the following Uruk III phases; in 2004 the Metropolitan Museum put the total known from sites of the Uruk culture at about 6,000. The great majority are administrative records; most of the rest are lists of signs and words arranged by theme, which scribes used for training. These early texts record numbers and nouns but hardly any grammar, so scholars still debate which language, if any, lay behind them; Sumerian is the most popular answer, but it is not universally accepted.

In its earliest form the script had a large repertoire of signs (counts differ with how variants are treated), some recognisable drawings and many abstract. Over the third millennium the signs became more stylised. At some point the conventional orientation of the script also turned by 90 degrees, so that lines of writing ran horizontally from left to right; when, where and why this happened is still debated. By the second millennium the original pictures were no longer recognisable.

The key step towards writing language, not just lists, was the rebus principle: using a sign for its sound rather than its meaning. The Sumerian word for “arrow”, ti, sounded like the word for “life”, til, so the arrow sign could be used to write “life”, and then for the syllable ti in any word. A few examples of this occur in the earliest tablets of about 3200–3000 BCE, but its consistent use becomes clear only after about 2600 BCE, when scribes increasingly wrote personal names, grammatical elements and eventually full sentences in this way.

At the same time new signs were created by combining old ones. The sign for “mouth” with the sign for “water” inside it meant “to drink”. Scribes also added phonetic complements, extra syllable signs that hinted at the pronunciation, and determinatives such as the star-shaped sign for “god” placed before divine names. The result was a flexible but complex system of word signs and syllable signs, with a repertoire of several hundred signs that varied by period and language, many of which had several possible readings.

Cuneiform outlived Sumerian as a spoken language, although whether the very first proto-cuneiform texts recorded Sumerian remains uncertain. Scribes also used it for Akkadian, a Semitic language completely different in structure from Sumerian: Akkadian names and words already appear in earlier Sumerian texts, and from the twenty-fourth century BCE, under the Akkadian Empire of Sargon, Akkadian became a major written language. Akkadian, in its Babylonian and Assyrian dialects, became the main language of cuneiform and, for several centuries, an international language of diplomacy. In the fourteenth century BCE, rulers of Egypt, Babylonia, Assyria, Mitanni, the Hittites and the cities of the Levant corresponded in Akkadian cuneiform, as the Amarna letters found in Egypt show.

Sumerian stopped being spoken, probably by the early second millennium BCE, but it survived for two thousand years as a language of religion and scholarship, rather like Latin in medieval Europe. Scribes produced bilingual Sumerian–Akkadian texts and word lists that later helped modern scholars understand Sumerian.

Other peoples borrowed the script for their own languages: the Elamites of south-western Iran, the Hittites and Luwians of Anatolia, the Hurrians of northern Mesopotamia and Syria, and the Urartians around Lake Van. At Ugarit on the Syrian coast, scribes invented an alphabet of 30 cuneiform-style signs, and the Achaemenid Persian kings created a separate semi-alphabetic cuneiform for Old Persian.

In the first millennium BCE Aramaic, written in a much simpler alphabet on parchment and papyrus, gradually replaced Akkadian as the everyday language of the Near East. Cuneiform retreated into temples, where priests and scholars used it for rituals, omens and astronomy. The latest datable cuneiform tablets are astronomical almanacs of the first century CE. For a long time an almanac from Babylon dated to 74/75 CE was considered the last; in 2014 Hermann Hunger and Teije de Jong dated a poorly preserved almanac from Uruk (W22340a) to 79/80 CE from its astronomical contents. Their dating has been widely accepted, but some scholars, including Eckart Frahm (2025), doubt it and prefer to keep the Babylon almanac of 74/75 CE as the latest securely datable text.

Most cuneiform texts are ordinary documents: receipts, ration lists, loans, sales, marriage contracts, adoptions, court cases and letters. Private archives, such as those of Assyrian merchants trading with Anatolia in the early second millennium BCE, let historians follow families and businesses in extraordinary detail.

Cuneiform also preserved the world’s earliest literature. It includes Sumerian hymns attributed to the priestess Enheduanna, proverbs and wisdom texts like the Instructions of Shuruppak, laments for destroyed cities, and epics such as the stories of Gilgamesh. Law collections, most famously the Code of Hammurabi of about 1750 BCE, were written on tablets and monumental steles.

Scholars wrote in cuneiform about mathematics, medicine, divination and astronomy. Babylonian mathematicians used a place-value system based on 60, which survives in our 60 minutes and 360 degrees. Late Babylonian astronomers kept observational diaries for centuries and developed mathematical methods for predicting the movements of the moon and planets.

Scribes trained in schools, called in Sumerian the é-dub-ba-a, “tablet house”. Students copied sign lists, vocabulary, model contracts and literary texts. Thousands of these exercise tablets survive, some with a teacher’s model on one side and a student’s wobbly copy on the other.

In late antiquity the knowledge of how to read cuneiform died out, although exactly when is uncertain; some scholars think a few specialists kept it alive for a century or two after the last dated tablet. European travellers began copying inscriptions from the ruins of Persepolis in Iran in the seventeenth century, and in 1778 the Danish traveller Carsten Niebuhr published accurate copies. He recognised that the inscriptions contained three different kinds of cuneiform script.

The simplest of the three was Old Persian. In 1802 the German schoolteacher Georg Friedrich Grotefend correctly guessed that certain repeated groups meant “king” and the names of Persian kings, and several scholars built on his work. The great breakthrough came from the trilingual Behistun inscription of Darius I in western Iran, carved on a cliff in Old Persian, Elamite and Babylonian. The British officer Henry Rawlinson copied it in the 1830s and 1840s at considerable risk, and he and the Irish clergyman Edward Hincks used it to decipher Akkadian.

In 1857 the Royal Asiatic Society in London tested the decipherment. Four scholars, Rawlinson, Hincks, Julius Oppert and William Henry Fox Talbot, translated the same newly lithographed inscription of the Assyrian king Tiglath-Pileser I independently and sent their versions under seal; a committee then compared them. The versions agreed in their main lines, which convinced most observers that Akkadian cuneiform could really be read. Sumerian, a language unrelated to any living language, was recognised and deciphered later in the nineteenth century.

Excavations at Nineveh, Nimrud and other Assyrian cities in the mid-nineteenth century uncovered vast collections of tablets, including the library of the Assyrian king Ashurbanipal. In 1872 George Smith of the British Museum identified among them a flood story resembling the account of Noah, which caused a sensation and made cuneiform famous.

The study of cuneiform is called Assyriology, because the first large finds came from Assyria. Tablets are held in Iraq and in museums around the world, for example in Baghdad, London, Berlin, Paris, Istanbul, Yale and Philadelphia; the British Museum alone holds about 130,000, one of the largest collections. Because the number of trained readers is small compared with the size of the corpus, many tablets have never been fully published.

Digital projects are transforming the field. The Cuneiform Digital Library Initiative (CDLI) catalogues and photographs hundreds of thousands of objects, and the Open Richly Annotated Cuneiform Corpus (Oracc) publishes edited, translated and linguistically annotated texts online. Cuneiform signs have been encoded in the Unicode standard, and researchers are experimenting with 3D scanning and machine learning; a 2020 study, for example, tested neural networks for suggesting missing passages in broken Babylonian texts.

Cuneiform heritage has also suffered from war and looting. Sites in Iraq and Syria have been illegally dug during periods of conflict, and many tablets without known findspots have entered the antiquities market. Scholars debate the ethics of studying such unprovenanced tablets, and some have been returned to Iraq in recent years, such as the Gilgamesh Dream Tablet, handed over by the United States in September 2021.

The link between clay tokens and writing is one of the best-known theories in the field. The archaeologist Denise Schmandt-Besserat, building on earlier work, argued that the shapes of Neolithic tokens were the direct ancestors of many proto-cuneiform signs, so that writing evolved from a long tradition of accounting. Many scholars accept a link between tokens, bullae, numerical tablets and the earliest numerical signs. Critics, however, have questioned the meanings assigned to particular token shapes, whether finds from widely separated places and periods really formed a single system, and how far token shapes correspond to later signs.

Specialists differ on how much weight to give to inherited accounting practices and how much to deliberate innovation. Early writing drew on several existing recording practices, such as numerical notation, seals and other marks, but exactly how these combined to produce proto-cuneiform is uncertain: some describe a fairly rapid invention by administrators in Uruk around 3300 BCE, while others stress the long accounting background. Robert Englund and Hans Nissen, who studied the archaic Uruk texts in detail, emphasised the metrological systems: different counting systems were used for different commodities, such as grain, animals and land, which they took to show that proto-cuneiform served first as an accounting technology and only later as a way of recording language.

A related question is whether writing was invented once and spread, or several times independently. Egyptian hieroglyphs appear at about the same time as or slightly later than proto-cuneiform. Some scholars think the idea of writing passed from Mesopotamia to Egypt, while others stress the very different signs and structure of the two systems and see independent inventions. Chinese writing and Maya script are generally considered separate inventions.

The language of the earliest tablets is also debated. Because proto-cuneiform records little grammar, it is hard to prove that its users spoke Sumerian. Arguments rest on a few possible phonetic spellings and on continuity with later, clearly Sumerian texts. Some scholars have suggested the presence of other languages in early Uruk, but the evidence remains thin.

Because cuneiform signs are polyvalent, editing a tablet involves many decisions. Assyriologists publish texts in transliteration, a sign-by-sign rendering in Latin letters. Conventions show the scholar’s interpretation: syllabic readings in lower-case italics, logograms in capitals (for example KÙ.BABBAR, Sumerian “silver”, read in Akkadian as kaspum), determinatives as superscripts, and index numbers or accents to distinguish signs with the same sound, such as u, ú and ù. A transliteration thus records both what is on the tablet and how the editor has chosen to read it.

Pronunciation is partly reconstructed. Akkadian is relatively well understood thanks to its Semitic relatives, but Sumerian has no known relatives, and its sounds and grammar are reconstructed from bilingual texts and ancient grammatical lists. Readings of Sumerian names have changed as knowledge improved: the king once read as Ur-Engur is now read Ur-Namma.

Chronology is another challenge. Dates in the third and second millennia BCE depend on king lists, year names and a few astronomical observations, such as the Venus observations of the time of King Ammi-saduqa. Scholars have long debated alternative “high”, “middle” and “low” chronologies that differ by decades or more; the Middle Chronology is used by convention, and a 2016 study combining tree rings and radiocarbon dating supported it for the earlier second millennium BCE.

Finally, the surviving record is uneven. Some cities and periods are documented by tens of thousands of tablets, others hardly at all, and archaeological context was often lost in early excavations and looting. Historians must therefore be careful not to mistake the accidents of survival for the realities of ancient life.