Monday, September 26, 2011

The Nominative Function and Inscriptions in Indus Script

Imprint from a Near Eastern cylinder seal, without an inscription, from Period IV,
2000 - 1500 BCE (after Collon 2005: 53, no. 206).  Note the man and various animals.

I previously discussed reasons for thinking that the Indus script was a proto-writing system, including the frequencies of signs (large numbers of rare signs versus a small core of frequent signs), the brevity of inscriptions (averaging 4 to 5 signs per inscription), and low rate of apparently random repetition of signs within inscriptions, a fact that suggests phonetic information is lacking.  I noted also the hypothesis that Harappan stamp seals contained the same sort of information as Near Eastern cylinder seals.  This would include names of humans and deities, occupational terms (or symbols for them), relational elements (such as “son of so-and-so”), perhaps a few toponyms (names of locations), and possibly even an occasional qualifying phrase (such as “the great”).

Korvink discusses this hypothesis that the Harappan seals contain names and titles, terming it the nominative function (2007: 71-74).  He rules out this hypothesis for two reasons.  First, although most seals (and other inscribed objects) contain very short inscriptions, there are some that contain multiple units of information.  Second, “inscriptions vary in length on functionally similar objects” (2007: 71). 
Indus seal M-314 with inscription containing 17 symbols,
the longest continuousinscription to date on any object.

For his first reason, he cites the seal with the longest inscription found in the first two volumes of the Corpus, M-314.  It contains 17 signs in three lines the content of which is not easily interpreted as a name with ancillary information such as titles, honorifics, place names, or occupations (2007: 72).  The first line begins with CARTWHEEL plus DOUBLE QUOTES, which Korvink analyzes as a prefix.  This same prefix is found on many other seals.  Three “fish” symbols follow in a common medial sequence (FISH UNDER CHEVRON / WHISKERED FISH / FISH).  At the end of this line is a common terminal sign, SPEAR.  Since the line begins with a common prefix and ends with a common terminal, it is a self-contained unit which we can symbolize as PMT (prefix, medial, and terminal). 

The second line begins with a sign resembling the letter “A” but with one short leg, caged by eight short strokes (DOUBLY CAGED AY).  This sign is a singleton (or hapax), so it cannot be analyzed further with statistics.  It is followed by CUPPED SPOON, a symbol frequently found in medial segments where it is commonly paired with THREE POSTS.  Here, though, there are no “posts.”  These two signs – DOUBLY CAGED AY / CUPPED SPOON – form the medial segment of the second line that contains no prefix.  The line ends with TRI-FORK TOPPED POT / POT, a relatively common terminal.  The final symbol in this sequence is the most frequent sign and the most common terminal in Harappan script (POT).  The sign immediately preceding it appears often in this position.  The presence of these last two signs indicates that the second line is a self-contained unit, just as the first line is, its content symbolized as MT. 

Korvink does not attempt an analysis of the final line due to the low frequency of the signs in it.  Presumably a third self-contained unit, it includes seven signs: THREE POSTS / CIRCLED DOT / PANTS / MAN HOLDING DEE-SLASH / TRI-FORK / VEE IN DIAMOND / PRAWN in Korvink’s study, which derives from Mahadevan’s concordance.  In my own database, based on my interpretation of the photographs in the Corpus, it is slightly different (names in bold differ from Korvink’s version):




Changing the identification of the four signs in bold does not alter the basic analysis as (1) PMT, (2) MT, (3) M. 

Korvink concludes his analysis of these three lines as indicating that the seal is not a title with ancillary information.  If a more typical PMT inscription contains a title, then this long inscription would seemingly have to contain three different titles, one per line.  Korvink asks, “Would it not be impractical to have more than one title on a seal or sealing?”  The question implies that an affirmative answer, that it would indeed be impractical to encode two or three different titles on one seal.  Being impractical, it must not have been the case. 
Lapis lazuli bead from Mari, Syria, with proto-cuneiform inscription: "For Mesanepada,
king of Ur, the son of Meskalamdug, king of Kish, has consecrated" (Aruz 2003: 143).

My own conclusion differs, although I agree with the basic analysis of three units of information.  For my reasoning, I must refer back to some of the early inscriptions on Sumerian, Babylonian, and Elamite cylinder seals (see my earlier post).  In Collon’s Period II, dating to 3000-2334 BCE, one often finds a single name whether of a human or of a deity, e.g., Ishma-ilum, the name of a ruler of Kisik (2005: 31, no. 84) or Anzu-Sud (2005: 28, no. 81).  On other seals, there is a name plus a profession, sometimes an additional name in a patronymic, and occasionally a place name (Shara-bar-a, scribe of the Lady (2005: 28, no. 83), HE-kung-sig, priestess of Pabilsag (2005: 31, no. 93).  This pattern is also found in the following Period III, 2334-2000 BCE: Ninmelila, wife of Ur-Dada (2005: 35, no. 110), E-gissu, son of Bibbi, the carpenter (2005: 36, no. 116).  Longer inscriptions more often include a second name, with qualifying elements associated with each name: AN-zabazuna, strong king: Tishatal, the scribe, is your servant (2005: 36, no. 121, in Hurrian), (Gudea, governor of Lagash: Abban, the scribe, is your servant (2005: 36, no. 114).  By Period IV (2000-1500 BCE), some inscriptions are longer still, although most continue to include only a name or two and an occupation or relationship:  AN-LUGAL-X, city ruler of the god Assur, son of Ikunum, city ruler of the god Assur (2005: 43, no. 136), alongside Ipiq-Ishtar, servant of Adad-ranka (2005: 43, no. 140), Shamash, Aya, names of the sun god and his divine wife (2005: 45, no. 153), and Harashta (2005: 43, no. 152, in Hittite).

Thus, in a single inscription, three units of information occur often enough, sometimes more more.  If we term the elements PN (for a person’s name), DN (for a deity’s name), OCC (for an occupational term), PAT (for patronymics and other relationship expressions), and LOC for place names, we can analyze the previous inscriptions simply.  A name alone is PN or DN, as found in the first two examples cited above and the last one (all of these being PN).  The inscription of HE-kung-sig contains the three elements PN / OCC / DN.  Ninmelila’s inscription also contains three elements: PN / PAT (for “wife of) / PN2.  E-gissu’s inscription includes four elements by this analysis: PN1 / PAT (“son of”) / PN2 / OCC.  Tishatal’s inscriptions includes five: PN1 / OCC (“strong king”) / PN2 / OCC / PAT (“your servant”).  Occasional inscriptions are even longer: Ilum-muttabbil, purification-priest of Inanna of Zabalam, son of Shu-ili (2005: 45, no. 157).  This can be analyzed as containing six elements: PN1 / OCC / DN / LOC / PAT / PN2. 
Stele from Mesopotamia of Early Dynastic I period, c. 2900-2650 BCE,
with text on fringed skirt: "Ushumgal the pab-shesh priest" (after Aruz 2003: 53).

The seals of those of highest social rank may be the longest for this period:  Shamshi-Adad, beloved of the god Assur, viceroy of the god Assur, son of Ila-kadkabu (2005: 47, no. 173).  The seven elements here include: PN1 / PAT (“beloved of”) / DN / OCC / DN / PAT (“son of”) / PN2.  Another example is Ishar-Lim, king of the land of Hana, son of Iddin-Kakka, beloved of the gods Ilaba and Dagan (2005: 51, no. 199), including eight elements PN1 / OCC (“king”) / LOC / PAT / PN2 / PAT (“beloved of”) / DN1 / DN2.  Thus, where a cylinder seal contains only PN or DN, this may be comparable to Indus seals with only M.  Where the cylinder seal inscriptions include two elements – PN / OCC – this may be comparable to Indus seals with MT.  The addition of relational expressions to the previous two might be comparable to the Indus sequence PMT.  If these suppositions are provisionally assumed correct, then we do find cylinder seals with more than one sequence, comparable to the longer Indus inscriptions.  The three-line inscription contains PMT + MT + M, which could be considered comparable to PAT / PN1 / OCC1 + PN2 / OCC2 + DN.  I have no evidence to show that this is exactly what M-314 indicates, of course.  But such content is indeed conceivable, given the Near Eastern parallel from cylinder seals.  Thus, I consider that the nominative function of Indus seals remains a plausible hypothesis.
From Queen Nefertari's tomb, a portion of the painted hieroglyphic inscription,
including her cartouche with its inscription, "Nefertari, beloved of the goddess Mut"
(photo by author of replica painted on papyrus).  Her titles are in the column on the
right, including "great queen, lady of the two lands."

To quote another example, this time from Egypt, during the 5th Dynasty and the reign of Userkaf, a priest of Hathor describes himself thus: “steward of the palace, governor of the New Towns, superior prophet of Hathor, mistress of Royenet, king’s-confidant, Nekonekh” (Breasted 1906: 100).  This is not a cylinder seal inscription, but the name and titles found in a will.  The queen Mertityotes, from the reign of king Khafre, includes these phrases on a false door: “king’s-wife, his beloved, devoted to Horus, Mertityotes” (across the top); “king’s-wife, his beloved, Mertityotes, beloved of the Favorite of the Two Goddesses, she who says anything whatsoever and it is done for her” (down the right side); and “great in the favor of Snefru, great in the favor of Khufu, devoted to Horus, honored under Khafre, Mertityotes” (down the left side) (Breasted 1906: 88).  Each one of these descriptive sequences includes more than three titles for a single individual.  Of course, wills and funeral monuments provide a good deal more room for such descriptions than a small seal.  But these examples demonstrate the fact that a single individual might indeed have several titles.

To continue with Korvink’s analysis, taking Mahadevan’s analysis of the terminals as indicators of social class does not improve the situation.  The first line of the inscription M-304 ends with the SPEAR, which Mahadevan interprets as indicating the warrior class, while the second line ends with POT, which Mahadevan interprets as indicating the priestly class.  Presumably, an individual could not be a member of two social classes at once, since in historical India warriors were of the Kshatriya caste and priests of the Brahmin caste.  In the Bronze Age, though, India was not yet bound by a rigid caste system as caste is thought to have evolved between about 800-550 BCE (Ions 1983: 8).  In the earlier society, warriors probably were distinguished from priests as individuals, but they may not yet have constituted distinct social classes.  We may compare early Rome at a period considerably later than the Indus civilization, where the average man expected to play many roles in life, growing the family’s food as a farmer, but taking up sword or spear to act as a soldier when the need arose, in time becoming the pater familias, or head of a household (typically leading an extended family that might comprise three generations).  As the pater familias, he would also act as a sort of priest in the family cult, leading the worship of the ancestral and household gods, the lares and penates.  If such a social structure obtained in the Indus Valley, a single individual might then play the role of farmer (or other profession), warrior, and something akin to priest in a single lifetime.  In the urban society of the Indus Valley, there may well have been more specialization of professions than this brief discussion suggests, as well as some incipient social classes.  But Harappan archeology does not suggest a highly stratified society such as was found in contemporary Mesopotamia and in Egypt (Possehl 2002: 57).  In conclusion, then, I think the nominative function has not been ruled out for Indus seal inscriptions.  In any case, Mahadevan’s interpretation of the terminal signs as indicators of social class may well be incorrect.
Detail from Indus seal M-809 with inscription: BACK CEE & CEE / STRIPED MALLET / BI-QUOTES //
STRIPED TRIANGLE / TRIPLE TRIANGLES / SPEAR (over unicorn and cult stand).

Korvink continues his discussion of this function with a second objection, namely, that inscriptions of varying lengths appear to serve the same function.  He gives examples of seal inscriptions that contain some of the same signs, but which vary in length (2007: 74).  I do not see these particular inscriptions in my database, but there are others that are almost the same.  So I will cite the ones I can find in my database, as this will not affect Korvink’s analysis.

104                         STRIPED TRIANGLE / TRIPLE TRIANGLES (not found in my database)


H-187                    STRIPED TRIANGLE / TRIPLE TRIANGLES // POT / COMB (MT).




Korvink states, “Surely the last inscription could be considered a title under the nominative hypothesis, whereas the first of these is far too simple to be considered a title (there is no terminal or prefix on the seal)” (2007: 74).  While I do not find the medial pair STRIPED TRIANGLE / TRIPLE TRIANGLES as a complete inscription, it may appear on one of the tablets that I could not make out or on an abraded seal that I read differently.  I assume it is indeed there somewhere.  If we assume that the medial segment of an inscription (M) contains a name, then the argument is that the simplest inscriptions – of M only – could not serve the same function as those with additional elements.  However, as noted above, cylinder seals from the Near East serve essentially the same function(s) regardless of whether they contain a single name in the inscription or several elements.  Specifically, the seal inscribed Anzu-Sud (PN only) served the same functions as that inscribed Gudea, governor of Lagash: Abban, the scribe, is your servant (PN1/ OCC1 / LOC / PN2 / OCC2 / PAT). 
Again, then, on the basis of comparisons with the Near East, the nominative function remains a viable hypothesis for the Indus seals.  The inscriptions may represent individuals, reflecting occupations and social relationships, perhaps also including connections with deities and possibly with places.  None of this can as yet be demonstrated, but it remains a possibility.


Aruz, J., ed. 2003. Art of the First Cities: The Third Millennium BC from the Meditarranean to the Indus. New York: Metropolitan Museum of Art and Yale University Press.

Collon, D. 2005 and 1987. First Impressions: Cylinder Seals in the Ancient Near East. London: British Museum.

Ions, V. 1983 and 1967. Indian Mythology. London: Chancellor.

Korvink, M.P. 2007.  The Indus Script: A Positional-Statistical Approach. Gilund Press (

Possehl, G.L. 2002. The Indus Civilization: A Contemporary Perspective. Lanham, MD: AltaMira.

No comments:

Post a Comment