Indus script explorer
What is known
Every sentence on this page is followed by the number of the source it rests on. Observed means what the corpus shows; proposed means a reading someone has argued for, named and cited, never adopted here. Nothing on this page translates anything.
- The Indus Valley script is undeciphered. [1]
- Phonetic and semantic interpretations of individual signs vary widely. [2]
- The points of scholarly consensus are few: that most inscriptions read right to left, that certain stroke-like signs are numerals, and that certain terminal signs share a function. [2]
- Mahadevan's 1977 concordance, the standard reference, records 417 distinct signs in 2,906 texts. [3]
- Nearly 4,000 inscribed objects are known: seals, sealings, tablets, ivory rods and pottery. [2]
- Estimates of the sign inventory run from 417 signs to 694, depending on how variants are counted, and nearly 27 percent of Mahadevan's signs occur only once. [1]
- The texts are short: the average inscription is five signs. [4]
- The longest inscription on one side of an object has 17 signs; two identical three-sided tablets carry 26 signs each. [4]
- The evidence for right-to-left writing is that short inscriptions start at the right edge and leave the left blank, and that the left-most signs of long inscriptions spill onto a second line. [5]
- Seal stamps, engraved in mirror image, read left to right, and a few texts run left to right or alternate direction line by line. [4]
- Mahadevan and Parpola, the two scholars behind the standard sign lists, agree that the script is logo-syllabic and that the language shows Dravidian-like features. [5]
- Farmer, Sproat and Witzel argue that the symbols were not writing at all, and were not evolving toward writing after at least 600 years of use. [6]
- Rao and colleagues found that the conditional entropy of the inscriptions matches linguistic systems rather than non-linguistic ones, which they say increases the probability that the script represents language. [7]
- A 2026 preprint re-tested the question against synthetic baselines and found that the corpus does not match either a linguistic or a non-linguistic baseline cleanly. [8]
- The script is not in Unicode; Michael Everson proposed 386 characters for it in 1999. [9]
- The Unicode roadmap for the Supplementary Multilingual Plane still lists the Indus script against that 1999 proposal. [10]
The Dholavira signboard
10 signs in reading order on the stage above; spoked wheel appears 4 times, at positions 1, 4, 8, 9. Tap a sign to read its entry.
The signs
Glyphs are hand-drawn approximations of the published drawings. The signboard's signs carry form-based codes because no peer-reviewed or primary source that could be opened lists their catalogue numbers.
DH-wheel spoked wheel
On The Dholavira signboard at positions 1, 4, 8, 9 of 10 — 4 times.
Observed
- It is one of the ten unusually large signs of the Dholavira signboard, found at the north gate of the citadel and thought to have been inlaid on a wooden board above the door. [11]
- It is the symbol that recurs on the board: one of the ten symbols is repeated four times. [12]
- The signboard was in use in Dholavira's Stage IV, the classical Harappan phase. [11]
Proposed
No proposed reading is recorded on this page for this sign.
DH-tee T-shaped sign, a bar on a stem
On The Dholavira signboard at position 2 of 10.
Observed
- It is one of the ten unusually large signs of the Dholavira signboard, found at the north gate of the citadel and thought to have been inlaid on a wooden board above the door. [11]
- The signboard was in use in Dholavira's Stage IV, the classical Harappan phase. [11]
Proposed
No proposed reading is recorded on this page for this sign.
DH-crossed-stem crossed sign with a stem
On The Dholavira signboard at position 3 of 10.
Observed
- It is one of the ten unusually large signs of the Dholavira signboard, found at the north gate of the citadel and thought to have been inlaid on a wooden board above the door. [11]
- The signboard was in use in Dholavira's Stage IV, the classical Harappan phase. [11]
Proposed
No proposed reading is recorded on this page for this sign.
DH-lozenge lozenge
On The Dholavira signboard at position 5 of 10.
Observed
- It is one of the ten unusually large signs of the Dholavira signboard, found at the north gate of the citadel and thought to have been inlaid on a wooden board above the door. [11]
- The signboard was in use in Dholavira's Stage IV, the classical Harappan phase. [11]
Proposed
No proposed reading is recorded on this page for this sign.
DH-cross two crossed strokes
On The Dholavira signboard at position 6 of 10.
Observed
- It is one of the ten unusually large signs of the Dholavira signboard, found at the north gate of the citadel and thought to have been inlaid on a wooden board above the door. [11]
- The signboard was in use in Dholavira's Stage IV, the classical Harappan phase. [11]
Proposed
No proposed reading is recorded on this page for this sign.
DH-stroke single vertical stroke
On The Dholavira signboard at position 7 of 10.
Observed
- It is one of the ten unusually large signs of the Dholavira signboard, found at the north gate of the citadel and thought to have been inlaid on a wooden board above the door. [11]
- The signboard was in use in Dholavira's Stage IV, the classical Harappan phase. [11]
Proposed
No proposed reading is recorded on this page for this sign.
DH-hook hooked sign
On The Dholavira signboard at position 10 of 10.
Observed
- It is one of the ten unusually large signs of the Dholavira signboard, found at the north gate of the citadel and thought to have been inlaid on a wooden board above the door. [11]
- The signboard was in use in Dholavira's Stage IV, the classical Harappan phase. [11]
Proposed
No proposed reading is recorded on this page for this sign.
M-342 the jar sign
Observed
- Sign 342 is the most frequent sign in the corpus: on its own it accounts for 10% of all sign occurrences in Mahadevan's concordance. [3]
- It is one of the three signs that most often end a text. [3]
- Mahadevan called it the terminal sign, occupying ten to twelve percent of all Harappan writing, as common as the letter e in English. [13]
- Just four of the 417 signs account for 21% of the 13,372 sign occurrences in the concordance. [6]
Proposed
- Mahadevan reads it as a Dravidian pronominal masculine singular suffix, sometimes with the ideographic sense of a sacrificial vessel with food offerings. [2]
- Mahadevan also suggested that within the script it may have served as a grammatical suffix, possibly a nominal suffix used only by the elite. [13]
- Parpola, per Mahadevan, takes the sign to represent the front view of a bull or a cow with curved horns. [13]
- Wells considered it a terminal marker that might have been a verbal ending with the sound value -ay. [2]
- Bonta reads it as a predicate or root meaning be or belong to, possibly a general predicate marker. [2]
STROKES the stroke numerals
Observed
- Parpola identifies the numerals as a set of short stroke signs of up to nine strokes, arranged in one or two tiers. [5]
- The numerals are among the few signs whose function can be deduced with fair certainty, partly because they consist of groups of vertical strokes. [14]
- On many tablets a pot-shaped sign is preceded by zero to four vertical strokes that clearly stand for numbers. [4]
- Mukhopadhyay counts 22 signs built from long and short strokes in horizontal and vertical patterns and classes them as numerical. [2]
Proposed
- Parpola proposes that groups of small inverted semi-circles found beside the strokes are tens. [5]
- Mukhopadhyay suggests that, since only a limited number of strokes are used across the corpus, the stroke signs may have stood for denominations of a standard metric. [2]
- Parpola argues that a numeral precedes the word it counts, as in Proto-Dravidian, though not in every case. [14]
Counting
The number tool on the stage draws 1 to 9 as strokes, up to 4on one row and the rest in two tiers — the split is this page's rendering choice. Anything larger, and zero, is refused; the reasons are the notes below.
- The stroke numerals run to nine strokes, in one or two tiers. [5]
- Only a limited number of strokes are used across the whole corpus. [2]
- The Harappans' stone weights, found at Mohenjo-daro, run in the ratios 1, 2, 4, 8, 16, 32, 64, 160, 200, 320, 640 and 1,600: binary in the small weights, then decimal. [15]
- Hemmy found no inscription or mark of value on any of the weights. [15]
- The most common weight is the 16th ratio, about 13.7 grams; above the 64th ratio the increments become decimal. [16]
What the number tool links to
- Drawn as strokes: Parpola's set of stroke numerals runs to nine strokes, arranged in one or two tiers. [5]
- Zero: The numerals consist of groups of vertical strokes. [14]
- Too large: The attested stroke numerals run to nine; Parpola proposes that groups of small inverted semi-circles beside them are tens, a proposal rather than a reading. [5]
- Not a number: Phonetic and semantic interpretations of individual signs vary widely. [2]
Sources
- [1]Shruti Daggumati, Peter Z. Revesz (2021). A method of identifying allographs in undeciphered scripts and its application to the Indus Valley Script (Humanities and Social Sciences Communications) · peer-reviewed
- [2]Bahata Ansumali Mukhopadhyay (2019). Interrogating Indus inscriptions to unravel their mechanisms of meaning conveyance (Palgrave Communications 5:73) · peer-reviewed
- [3]Nisha Yadav, Hrishikesh Joglekar, Rajesh P. N. Rao, Mayank Vahia, Ronojoy Adhikari, Iravatham Mahadevan (2010). Statistical Analysis of the Indus Script Using n-Grams (PLoS ONE) · peer-reviewed
- [4]Asko Parpola (eds. Jagat Pati Joshi and Asko Parpola) (1987). Corpus of Indus Seals and Inscriptions 1: Collections in India — Introduction · primary
- [5]Iravatham Mahadevan (1997). An Encyclopaedia of the Indus Script (review of Parpola, Deciphering the Indus Script) · peer-reviewed
- [6]Steve Farmer, Richard Sproat, Michael Witzel (2004). The Collapse of the Indus-Script Thesis: The Myth of a Literate Harappan Civilization (Electronic Journal of Vedic Studies 11-2) · peer-reviewed
- [7]Rajesh P. N. Rao, Nisha Yadav, Mayank Vahia, Hrishikesh Joglekar, Ronojoy Adhikari, Iravatham Mahadevan (2009). Entropic Evidence for Linguistic Structure in the Indus Script (Science 324) · peer-reviewed
- [8]Ashish Nair (2026). How Non-Linguistic Is the Indus Sign System? A Synthetic-Baseline Scorecard (arXiv preprint 2604.17828) · primary
- [9]Michael Everson (1999). Proposal for encoding the Indus script in Plane 1 of the UCS (ISO/IEC JTC1/SC2/WG2 N1959) · primary
- [10]Unicode Consortium. Roadmap to the SMP (Unicode Consortium) · primary
- [11]Archaeological Survey of India (2011). Excavations — Dholavira (Archaeological Survey of India) · primary
- [12]Archaeological Survey of India. Dholavira Symbols 1 (Archaeological Survey of India caption, reproduced at harappa.com) · primary
- [13]Iravatham Mahadevan. The Terminal Indus Sign (interview with Iravatham Mahadevan, harappa.com) · primary
- [14]Asko Parpola (2010). A Dravidian solution to the Indus script problem (Classical Tamil Research Endowment Lecture, Coimbatore) · primary
- [15]A. S. Hemmy (1931). System of Weights at Mohenjo-daro, in Mohenjo-daro and the Indus Civilization, vol. II (ed. John Marshall) · primary
- [16]Jonathan Mark Kenoyer (2010). Measuring the Harappan World, in The Archaeology of Measurement (eds. Morley and Renfrew, Cambridge University Press) · peer-reviewed
What it is
An explorer for the Indus Valley script, built around the difference between what can be observed about it and what has been proposed. The page opens on a stage that fills the screen: the observed half. A pager at its top counts the observed sets the page holds and steps between them; today that is the Dholavira signboard, its signs redrawn by hand from the published drawing. A toggle on the same bar switches the stage to a number tool that draws a whole number as rows of vertical strokes. Everything below the stage is the proposed half: for every sign, what the corpus shows and what scholars have argued the sign is, each argument named and cited, with a numbered source after every sentence on the page.
Tap a sign on the stage and its entry below is highlighted and scrolled into view; the entry separates the observed from the proposed under two headings, and a proposed reading is attributed to the paper that made it and never adopted as the page's own. Type a number and it is drawn as strokes, up to four on one row and the rest split into two tiers, a layout the page labels as its own choice. Type a word, a letter, zero, or a number beyond the stroke set, and the tool refuses and offers one link to the note that explains why, with its source. It is deliberately not a translator: the page never renders letters or words as signs, and it says so.
Every factual sentence is stored as data with the verbatim passage of its source beside it, and a script re-fetches each source and records whether the passage is still there. The glyphs are hand-drawn approximations of the published drawings, not a font and not copied images. Sources are listed at the foot of the page, every citation links to its entry, and the address of any sign's entry can be shared as a link.
How to use it
- Tap a sign on the stage; its entry below highlights and scrolls into view.
- Switch the stage to Numbers and type a whole number to draw it as strokes.
- Type anything else and follow the link to the note on why it is refused.
- Every sentence ends in a source number; click it to jump to the source.
- Share a sign by copying the address after clicking it.
More experiments
- Antikythera — Canvas, Mechanism, Gears
- Flytrap — WebGL, Three.js, r3f
- Lathe — webgl, three.js, r3f