A civilization that left writing nobody can read
The Harappan people used a script, and it has not been deciphered. No decipherment has been accepted by scholars, despite many confident announcements over the past century. That is the main reason this civilization feels silent: we have its cities, tools and jewellery, but not its own account of its kings, gods, laws or history. “Forgotten” here means not only lost to memory but also unreadable, though the artefacts themselves are very well documented.
What survives
Inscriptions are found on several kinds of objects. The best known are small square stamp seals, usually of carved steatite, bearing an animal and a line of signs. Others appear on clay or faience tablets, on pottery, on copper tablets and on ivory or bone rods. Thousands of inscribed objects are known, with a figure of roughly four to five thousand often quoted, though counts vary with what is included.
The animals on seals include a bull, an elephant, a rhinoceros and a creature often called the “unicorn”, which is the most common. A few seals show scenes with human or part-human figures, including the one often called the “Pashupati” seal, whose interpretation is contested. At Dholavira, a group of ten large signs was found near a gate, sometimes described as a signboard, which is one of the few cases of a text meant to be read by many people.
What the signs look like
The script has somewhere between about 400 and 450 distinct signs, according to different catalogues, which is too many for a simple alphabet and too few for a script with a separate sign for every word. Many signs are pictorial, such as fish-like or jar-like forms, while others are strokes or geometric patterns. Most inscriptions are very short, typically only a few signs, and the longest on a single surface are only a few dozen at most.
Evidence from overlapping signs and the way lines were cut suggests that texts were generally read from right to left, although there are exceptions. Some sign sequences show regular patterns, and some strokes are widely thought to be numbers. Beyond that, little is firmly agreed.
Why it remains undeciphered
Several obstacles combine.
- No bilingual text. The Egyptian hieroglyphs were read with help from the Rosetta Stone, which paired them with Greek. No comparable inscription is known for the Indus script.
- Short inscriptions. With so little continuous text, there is not enough context to test proposed readings.
- Unknown language. We do not know which language the script writes, or whether it writes language at all in the sense usually assumed.
- Few clues from neighbours. Mesopotamian texts mention the region, but they do not quote Indus writing.
Other undeciphered scripts show the problem. The Linear A script of Minoan Crete and the rongorongo tablets of Rapa Nui remain unread, while successes such as Maya glyph decipherment depended on longer texts, known languages and related scripts, none of which exist for the Indus signs.
Competing ideas
Scholars have proposed several broad explanations.
A language. Many researchers think the script records a language of the region. Proposals have included an early Dravidian language, supported by specialists such as Iravatham Mahadevan, who compiled a major sign concordance, and Asko Parpola, a Finnish scholar who explored readings based on this idea. Others have suggested an early form of Indo-Aryan or an unknown language. None of these readings are accepted generally.
Signs, not language. Some researchers, including Steve Farmer, Richard Sproat and Michael Witzel, have argued that the signs may be a non-linguistic system, perhaps of religious or political emblems, like heraldry, rather than writing. A statistical study published in 2009 argued that the sign order looked more like language than random patterns, and critics replied that such methods cannot settle the question. The debate is active.
Computer approaches. Researchers have tried to analyse the frequency and order of signs with computers, which can show structure but does not give meaning without outside information.
Why claimed decipherments fail
Announcements of a decipherment appear regularly, including some that read the seals as Sanskrit or as a particular Indian language. A reading is persuasive only if different scholars can apply the same rules and get consistent results, the proposed language fits the dates and setting, and the readings can be tested against new finds. Published proposals so far have not done this. In one well known case from the 1990s, a seal said to show a horse, which would have supported a link with Vedic culture, was shown to have been altered in an image, and reputable scholarship does not rely on it.
What the signs may still tell us
Even without reading, the inscriptions give real information. Their wide distribution across sites hundreds of kilometres apart points to shared practice, and their use on seals fits with marking goods, ownership or identity. Seals with the same signs have also been found at Gulf and Mesopotamian sites, which is relevant to trade. A new find of a longer text, or of a bilingual inscription, would be the most likely route to progress, but it may never come.