the indus seals and the twenty-seven-sign outlier

The Indus script has 5,000+ known inscriptions. Most are on stamp seals, steatite squares roughly 2 to 5 cm on a side. The seals were pressed into wet clay to mark goods. Objects bearing these marks have been found across 250,000 square miles, from Shortugai in Afghanistan to Lothal on the Gujarat coast. Everyone who used them agreed on what they meant. Nobody now does.

The corpus is short. Most inscriptions run five or six signs. 400+ distinct signs have been catalogued, depending on who is counting. Parpola's sign list gives one count. Wells gives another. They disagree at the margins.

The longest known inscription is 34 signs, on a fragmentary object recovered at Mohenjo-daro. The second longest is 26. Between them sits DK-6847, a copper tablet from the same site, catalogued in 1931 and rarely reproduced. DK-6847 has 27 signs.

I found it in Mahadevan's 1977 concordance. Entry 2913. The photograph is poor. The line drawing is clearer.

What is unusual about DK-6847 is not the length. It is the arrangement. The signs are not in a single line. They are laid out in three rows of nine, reading right to left in the top row, left to right in the middle, right to left again in the bottom. Boustrophedon, on a tablet the size of a matchbook. No other Indus inscription does this. Mahadevan notes the arrangement in a footnote and does not return to it.

The middle row contains the "fish" sign (Parpola 59) three times. The fish sign is one of the most common in the corpus and has been proposed as a logogram for "star" on the argument that the Dravidian mīn means both. This is speculation. There is no bilingual. There has never been a bilingual.

The tablet is copper, not steatite. Copper Indus inscriptions are a smaller subset, around 240 known. Most are tools or ingots with brief marks. DK-6847 is neither. It is a flat rectangle, 4.1 by 2.7 cm, with no evidence of use as a seal. It was not pressed into clay. The signs are incised, not raised in mirror. Whatever it was for, it was not for stamping.

I counted the signs. Twenty-seven. I counted them again. Twenty-seven.

Three rows of nine. Three cubed. The corpus contains the sign for what Parpola numbers 176, which he calls the "endless knot" and Wells calls "the loop," a closed figure crossed by a horizontal bar. It appears once in DK-6847, at the center of the middle row. Position 14 of 27. The exact middle.

The photograph plate in Mahadevan's concordance was taken from the Archaeological Survey of India's holdings. The current location of DK-6847 is listed as the National Museum, New Delhi. I emailed to ask whether the tablet could be photographed at higher resolution. The response was that the object is in storage and the accession record does not currently match a physical item on the shelf. They are looking.

Parpola's 1994 book reproduces DK-6847 as figure 7.14. The caption gives the sign count as 26. Mahadevan gives 27. I have compared the drawings. Parpola omits the leftmost sign of the bottom row. It is present in Mahadevan and in the 1931 field photograph. I do not know why it was dropped.

The signs on the seals are the same across 250,000 square miles. The people who used them agreed. There are 27 signs on DK-6847 or there are 26. The catalogue does not agree with itself.

I added DK-6847 to the list.