Study guides · Ancient Civilizations

Writing, Decipherment and the Limits of Evidence: Egypt, the Maya and the Indus Valley

How hieroglyphs and Maya glyphs were deciphered, why the Indus script is still unread, and what this teaches about reading ancient civilizations.

Whether a civilization can speak for itself depends on whether we can read its writing. Three famous cases show how different the situation can be. Egyptian hieroglyphs were deciphered in the 1820s, Maya glyphs were largely read only in the late twentieth century, and the script of the Indus civilization remains unread. Looking at why each case turned out as it did is a short course in how scholars know what they claim to know.

Egyptian hieroglyphs and the Rosetta Stone

For centuries after the last hieroglyphic inscription was carved in the late fourth century CE, no one could read the script. Medieval and Renaissance scholars assumed that the signs were purely symbolic, conveying ideas by pictures, and that assumption blocked progress for a long time. The turning point was the discovery in 1799, by soldiers in Napoleon's expedition at Rosetta in the Nile delta, of a stone carved with the same decree of 196 BCE in three scripts: hieroglyphic, Demotic and Greek. The Greek could be read, which gave a fixed point to work from.

Thomas Young, the English polymath, made early progress by recognising that the oval rings called cartouches contained royal names written in phonetic signs. Jean-François Champollion, who knew Coptic, the last stage of the Egyptian language, then showed in 1822 that hieroglyphs combine signs for sounds with signs for words and meanings. His insight that the script is a mix, rather than purely symbolic or purely phonetic, is the basis of every reading since. A well-documented account of the stages of this decipherment is Richard Parkinson's Cracking Codes, published by the British Museum, which holds the Stone.

The Maya script

The Maya of Mesoamerica wrote with a system of glyph blocks on stone monuments, painted pottery and bark-paper books. When Spanish missionaries arrived, they destroyed most of the books, and only four are known to survive. For much of the twentieth century, scholars believed the inscriptions were mainly calendrical and astronomical, the work of peaceful priest-astronomers, and that the script was not a true writing system. Two developments overturned this. In 1960 Tatiana Proskouriakoff, working at the Carnegie Institution, showed that the carved monuments at the site of Piedras Negras recorded the births, accessions and deaths of rulers, which meant they were historical. Meanwhile the Soviet linguist Yuri Knorozov argued from the 1950s that the Maya script was in large part phonetic, using syllabic signs, and he used the sixteenth-century account of the Spanish bishop Diego de Landa, which gave a list of signs, as a key. His ideas were resisted for years, notably by the leading Mayanist J. Eric S. Thompson, but later work confirmed them. In recent decades scholars such as David Stuart and Linda Schele, along with many colleagues, have taken the decipherment far enough that most of the surviving texts can now be read. Michael Coe's Breaking the Maya Code tells the story, including its personal rivalries, and is accessible to beginners.

The Indus script: an unsolved problem

The Indus or Harappan civilization flourished in what is now Pakistan and north-western India from about 2600 to 1900 BCE, with cities such as Mohenjo-daro and Harappa. It left a script of about four hundred distinct signs, found chiefly on stamp seals, but the texts are extremely short. Several thousand inscribed objects are known, most carrying only a few signs, and the longest inscriptions run to about twenty-six. No bilingual text has been found, and the language of the Indus people is unknown, so there is no equivalent to the Rosetta Stone. Many claims of decipherment have been published and none is accepted by the field. Some scholars argue that the signs may not encode a language at all, though most think they do. Because the evidence is so limited, specialists point out that a decipherment could be tested only if a longer text or a bilingual inscription were discovered. Gregory Possehl's The Indus Civilization and Jonathan Mark Kenoyer's work on its material culture give the fullest picture of what can be said from archaeology alone, and Asko Parpola's Deciphering the Indus Script presents the best-known case for a Dravidian reading, with the caveat that it is a hypothesis.

Other cases worth knowing

Linear B, the script used for Greek in Bronze Age Crete and mainland Greece, was deciphered in 1952 by Michael Ventris, an architect and amateur, who showed, to the surprise of many, that the language was early Greek. The earlier Linear A script remains unread. The Inca left no script of the usual kind, but kept records on knotted cords called khipu, which are still being studied. The Harvard Khipu Database Project has collected hundreds of examples to look for patterns. These cases show that a script can be deciphered only under certain conditions: the language must be known or recoverable, there must be enough text, and some fixed point such as a bilingual inscription, a known place name or a known royal name must give a way in.

How to judge a claim of decipherment

Claims to have deciphered an ancient script appear regularly in the press, and a few questions help to judge them. Does the proposal make sense of many different texts, or only of a few signs chosen to fit? Can it predict the content of inscriptions not used in developing it? Is the proposed language one that is known to have existed in the right place and time? Have specialists in the script and the language tested it and found it sound? Real decipherments such as that of Linear B were convincing because they met these tests. Scripts such as the Phaistos Disc from Crete, found in 1908, remain unread because there is too little text to test any proposal.

What this means for reading ancient civilizations

The comparison teaches caution. A civilization whose texts we can read, such as Egypt, can be studied through its own account of itself, though that account is shaped by the scribal elite who wrote it. A civilization like that of the Indus must be reconstructed from objects, buildings, burials and trade goods, and every interpretation is more tentative. Popular books and websites often present speculation about undeciphered scripts or mysterious lost cultures as established fact. A useful test is to ask what the evidence is, who has examined it and whether other specialists have accepted it.

The Ancient Civilizations pages of this site collect archaeological journals, museum collections, university programmes and funding calls for researchers, so that a student can move from these puzzles to the institutions that investigate them.

Further reading

  • Richard Parkinson, Cracking Codes: The Rosetta Stone and Decipherment (British Museum Press, 1999).
  • Michael D. Coe, Breaking the Maya Code, 3rd edition (Thames & Hudson, 2012).
  • Gregory L. Possehl, The Indus Civilization: A Contemporary Perspective (AltaMira, 2002).
  • Asko Parpola, Deciphering the Indus Script (Cambridge University Press, 1994).
  • Andrew Robinson, Lost Languages: The Enigma of the World's Undeciphered Scripts (McGraw-Hill, 2002).

Related on this site

This guide is an introduction written by a research scholar of Buddhist and ancient history from standard scholarly literature. Dates and figures for the ancient world are often debated, and where scholars disagree we have said so. If you find an error, please write to contact@theancientdharma.com.

↑