Professional Documents
Culture Documents
A Partial Decipherment of The Indus Vall
A Partial Decipherment of The Indus Vall
Proposed Phonetic and Logographic Values for Selected Indus Signs and
Readings of Indus Texts
Steven Bonta
In Memory of Iravatham Mahadevan
(1930-2018)
Foreword
The arduous task of decipherment requires both the collection and comparative analysis of a
deductive reasoning and educated guesswork using all available evidence, both inscriptional
and artifactual. Unlike the code-cracking carried out by cryptographers, the decipherment of
decipherer must enter and attempt to render intelligible a completely alien universe of
discourse, remote in time and cultural assumptions. Although epigraphers have developed
typologies of writing systems as linguists have for languages, no two writing systems work in
precisely the same way. This is especially the case for writing systems, like the Indus, that
approaches for recording human language were developed, from China to Mesopotamia to
seeking comparisons and parallels with other writing systems, but must also be alert to the
Regardless of the state of archaeological and epigraphic data, the task of successful
decipherment always involves two crucial steps, namely, the narrowing down of possibilities
by discerning evidentiary patterns in the written material under consideration, and the
formulating and testing of hypotheses based on both internal (patterns of sign distribution and
out as much information as possible based on internal distributional evidence gleaned from
exhaustive comparison of the many brief inscriptions constituting the available Indus corpus.
We were able to conclude by the mid-90s that certain very predictably patterned sign fields in
standardized weights, an insight that, in the years since, proved to be the initial entering
wedge for the decipherment of many texts in the present work. Those initial results,
quarters. We then undertook, time permitting among competing obligations (like carrying out
ethnolinguistic PhD research among a minority Tamil-speaking fisher caste in Sri Lanka), to
extract whatever information we could about the script as a whole from these
“M[etrological]” sign fields, and also to analyze the remaining sign field types for clues about
the typology of the underlying language. By 2010, we had found a considerable amount of
evidence that, after all, the main language underlying the script was in fact not likely to be
Dravidian but instead Indo-Aryan, all of the long-held and fiercely-defended assumptions in
favor of the “Dravidian hypothesis” notwithstanding. In our first monograph on this subject,
published by The Epigraphic Society in 2014, we detailed this initial evidence, concerning
which some of our early claims have withstood the test of further research, while other claims
have fallen by the wayside. Yet a method for actually achieving a more thorough
decipherment continued to elude for another decade, despite sustained work on the problem
through 2015.
Discouraged by one failed attempt, described elsewhere in this monograph, that consumed a
couple of years of work, we concluded that further progress on decipherment was likely
unattainable. However, after a hiatus of more than four years, we decided to undertake a new
approach amid the stifling ennui of the Covid pandemic and the stress of losing a parent to
that disease. The new method, described in detail herein, would have been impossible with
the resources available in 1996 or even 2006, but it yielded rapid and spectacular results over
the course of roughly eight months of very arduous work. After the first three months of
renewed labor, while trying to maintain a non-aspirational perspective, there could no longer
be any doubt: the initial “anchor values” posited for roughly a half-dozen signs led to other
values, which began yielding results, like names and titles, that were not only contextually
highly plausible, they in many cases were new to this investigator. This is most especially the
case with the several king names, compound personal names, and repeated occupational and
honorific titles that emerged from the decipherment process, and were confirmed only by
verification that such unknown names and titles are actually in Vedic or Puranic sources. All
During the first stage of this process, we deliberately confined our initial work to a single
concordance, the Mahadevan concordance, with a view to subsequently using the Interactive
Concordance of Indus Texts (ICIT), with its photographic component and additional material,
such as Dholavira inscriptions, as a check to either confirm or negate our conclusions. Once
we transitioned to the ICIT, we not only found all of our sign values (save one!) confirmed,
we also found additional useful data, from Dholavira and elsewhere, that furnished a few
The plan of this work is sixfold: 1) summarize the results of previous work, including ours; 2)
detail the methodology used to arrive at the proposed sign values; 3) present the sign values
and a thorough (though not repetitively exhaustive) list of wholly or partially readable
inscriptions for each sign; 4) set forth for the reader’s convenience a list of what we regard as
the most significant wholly or partially deciphered inscriptions from the corpus; 5)
summarize what we know about the script and about Indus culture, based on the newly-
“patterned inscriptions”; and 6) detail the considerable work yet to be done, including the
figuring out aspects of the inner workings of the script, including rules of compound
formation.
Owing to the uncertain nature of decipherment, it is impossible for this work to be free from
error, but we are confident that the proposed sign values are in the main correct, having
checked and rechecked all of them many times over many additional months of work, setting
aside any for which we have concluded the evidence to be too scant or too uncertain, and
attempting multiple times to falsify all proposed values, especially those arrived at earliest,
that in many instances proved essential as anchor values for ascertaining other signs. But we
are confident enough in the overall results, having experienced unlooked-for confirmation so
many times for proposed sign values arrived at only after exhaustive trial and error and the
process of elimination. We believe that these results will, in the main, withstand the test of
time; as such, we hope they will constitute a genuine beginning for what must be a long and
collaborative process of decipherment, a work that, like all previous decipherments, will
The existence of a major Bronze Age civilization in South Asia was first brought to light in
the 1920s with the excavation of two major sites in Pakistan—Harappa by Vats and
sophisticated urban culture, dating from the fourth through the second millennium BC, whose
vast extent was gradually uncovered thanks to later discoveries of outposts as far-flung as
Shortugai on the Amu River. The core territory of the so-called Indus Valley or Harappan
Civilization stretched from Afghanistan and Pakistan to central India at least as far as the
Delhi area, including at least five large urban sites (Harappa, Mohenjo-Daro, and the still
and numerous smaller sites. Indus artifacts have also been found at several sites in the Middle
In addition to its enormous extent, the Indus civilization is remarkable for its uniformity and
for the exceptional quality of its urban engineering. Its cities and towns are laid out on
essentially the same pattern throughout, while Indus sewer and drainage systems were not
matched in sophistication until the height of Roman civilization more than almost two
millennia later. The very bricks used in construction were of identical dimensions
everywhere. Indus weights, one of the commonest type of artifact, were also noteworthy for
On the other hand, Indus cities are distinctive for their lack of readily-identifiable palaces or
religious structures. A number of large buildings, such as the “Great Bath” at Mohenjo-Daro
and the “granary” at Harappa, have been found, but their exact uses remain unclear,
especially regarding the “granary.” Few statues or other prominent iconography like that
found in Mesopotamian, Egyptian, and Mediterranean sites, let alone in historical India, have
Writing, however, has been found, most conspicuously in the form of brief, highly stylized
inscriptions on steatite seals, but also on numerous potsherds, small tablets, seal impressions
or tags, implement heads, and even a single sign board unearthed at Dholavira. The seals in
particular often display interesting iconography in miniature, usually animal field figures, but
also occasionally peculiar scenes of obvious mythologicial or religious portent. The most
common field figure is the so-called “unicorn bull,” a bovine animal seemingly bearing a
single forward-curving horn (although it may be simply a conventionalized side view hiding
a second horn), with an odd lamp-like object on the ground in front of it. Other field animals
include oxen and zebu cattle, as well as animals no longer found in desertified Pakistan and
northwest India: gharials, elephants, and rhinoceroses. Other seals depict tantalizing scenes
with horned gods, hunting heroes, tree worship, and fantastic multi-headed and hybrid-bodied
No sooner were Indus seals and other artifacts with writing discovered than scholars began
trying to read the inscriptions. Although little progress has been made towards decipherment
in the century since the Indus seals and tablets were first discovered and cataloged, it has not
been for lack of effort. The first noteworthy attempt to launch a decipherment was G. R.
Hunter’s 1933 analysis of the 800-odd inscriptions then available from Sir John Marshall’s
three-volume excavation report on Mohenjo-Daro, the famous Pakistani site that remains the
single most fruitful source of Harappan texts. Hunter’s modest pioneering work remains one
of the soundest decipherment attempts, although some of his conclusions have been negated
by more data.
Not long after Hunter, Henry Heras became the first to propose what is known as the
“Dravidian hypothesis,” the notion that the Harappans were Dravidian speakers, and that the
Dravidian languages represent the primordial language stock in South Asia. Heras’ method,
which consisted of simply equating signs to Dravidian sound values based on supposed
graphology, and then trying to make the resulting readings intelligible, did not yield
convincing results, but both his methodology and the Dravidian hypothesis have proven
enduringly popular outside of the Subcontinent. With rare exceptions, Western scholars ever
since have tried to make the Indus inscriptions fit a Dravidian solution, but so far, no such
Beginning in the 1950s and 1960s, scholars in the Soviet Union and Finland renewed efforts
to decipher the script. Yuri Knorozov, the famous pioneering Mayan decipherer, wrote
several thoughtful papers on the script, while in Finland, Asko Parpola, who favors the
Dravidian hypothesis, began a decades-long study of the script that resulted first in a
1994, and most recently a comprehensive multi-volume photographic corpus of the Indus
inscriptions, all of which have been extremely significant milestones in Indus script research.
The late American archaeologist Walter Fairservis, another proponent of the Dravidian
hypothesis, published a book in 1992 that sought—as others had done before him—to
decipher the script by ascribing Dravidian sound values to signs based solely on their
assumed graphology. Quite aside from its aspirational nature, this methodology has
consistently yielded very different results even insofar as the assumed graphology is
concerned. Parpola, for example, in common with most investigators of the script, assumed
In 1977 the late Iravatham Mahadevan, the celebrated decipherer of the Tamil-Brāhmi
inscriptions, produced a compendious and still-useful (if somewhat dated) concordance of the
Indus corpus then available. While Mahadevan’s work has been superseded in many ways by
Parpola’s photographic corpus and by Wells and Fuls’ online interactive concordance (see
following), his concordance is very well-organized and contains a number of inscriptions not
found elsewhere. His material is used extensively in this work, in conjunction with that of
Wells’ and Fuls’ ICIT, and all inscriptions cited from the Mahadevan’s concordance are so
indicated by MH.
Even more recently, Bryan Wells and Andreas Fuls have created the Interactive Concordance
of Indus Texts (ICIT), an online resource containing many thousands of inscriptions complete
with photos, an invaluable feature for artefactual context. The ICIT is searchable by many
parameters, and has the added benefit of a large number of inscriptions from Dholavira and
elsewhere that are not to be found in any other source. Wherever possible in the main body of
this work, I have tried to cross-reference inscriptions as to their site-specific catalog number,
Several other decipherment attempts are worth mentioning. In 1978 Mitchener produced a
little-known book that attempted what few investigators apart from Knorozov have attempted
in any systematic way, namely, attempt to discern patterns of affixation by comparing similar
inscriptions. This method yielded very significant results for Kobers and Ventris in their
case system reminiscent of Indo-Aryan. Although his work did not garner wide attention, it is
noteworthy for being the only work of Western scholarship (aside from this one) to propose
an Indo-Aryan solution, and for its modest aim of trying to elucidate certain prominent
In 1982, S. R. Rao published an expansive claim of decipherment, also believing the Indus
language to have been Indo-Aryan. His method consisted partly in trying to plug in Indo-
Aryan word values based on graphology, and partly in trying to show that the Brāhmi script
of Mauryan India was based on Indus writing, and that Indus sound values could be deduced
plausible than the various Dravidian readings inspired by the same methodology.
Finally, Kinnier-Wilson’s 1974 book that tried to find common ground between Indus and
Sumerian writing is worth noting, if for no other reason than for his interesting treatment of
Indus numerals. Like Mitchener, Kinnier-Wilson did not attempt a grand, all-encompassing
decipherment, but his modest work is informative and thought-provoking for its brevity.
Besides these, there have been dozens of other claims of decipherment which far exceed the
aims of this work to survey exhaustively. There have also been many claims that the script is
undecipherable, and one, prominent a few years back, courtesy of Witzel, Farmer, and
Sproat, that the Indus inscriptions do not constitute writing at all. We will consider this claim
That the writing of a major civilization remains unread generations after the decipherment of
the Egyptian hieroglyphs and many of the languages written in cuneiform is a source of
frustration and contention. Yet the reasons for failure to decipher the Indus script cannot be
ignored: unlike the Egyptian hieroglyphs and Sumerian and Akkadian cuneiform, the Indus
writings present no lengthy texts, the entire corpus consisting of brief and often repetitive
inscriptions on small items like seals, potsherds, and miniature tablets. There are no lengthy
writings on stelae, papyrus, or temple walls. There are also no bilinguals like the Rosetta
Stone, and no known toponyms or other information from historical sources that could serve
as plausible “entering wedges.” The underlying language has not been proven, but a large
number of investigators in many countries have come to believe that the underlying language
is probably Dravidian. The so-called “Dravidian hypothesis” has enjoyed near-orthodoxy for
many decades, and, as earlier mentioned, was first suggested in the 1930s, not long after the
Indus inscriptions were first brought to scholarly attention, by Heras. It is not our purpose
competent scholars with thorough grounding in Indology—that have appeared over the last
century. But the fact that none of them have yielded plausible readings or any generally-
accepted results has led a number of investigators to claim that—barring some dramatic new
We tentatively suggest that 1) failure to decipher the script is not necessarily because it is
impossible, but because no one has yet deployed the proper methods and 2) the repeated
“Dravidian hypothesis” is so compelling, why have so many different investigators and teams
that have adopted it—including the longstanding Finnish effort, a sustained Russian effort
several decades ago, and the American archaeologist Fairservis, to mention but a few—not
only failed to produce compelling readings, but also had very little overlap in results?
III. Methodology.
In this study we advocate an approach designed to reduce as far as possible any reliance upon
with a view to clarifying not only the patterns of sign distribution, but also sign frequency
and randomness of occurrence. All three of these qualifying features rely in the first instance
on an ability to segment the texts; otherwise, statistical surveys of whether signs occur in
they are likely to be phonetic or logographic, have little utility. It is only once we can
segment the texts with some reliability that we are in a position to assert whether a sign is
simple syllable, a full word, an affix, or some other such, regardless of its overall position
within an inscription possibly representing several words. Thus it is a significant fact that
many of the most commonly-occurring signs in the Indus signary, like and , have very
rigid patterns of distribution that do not agree with an assumption that they should therefore
represent the commonest sounds. , for example, almost always occurs either in final or
penultimate position, and when it does not, it can usually be shown by comparative text
segmentation to occur in final or penultimate position with respect to a preceding sign series
which, together with , can be separated from a following sign series. I do not intend here
to rehearse exhaustively the results of many years of comparative text segmentation, which
have been published elsewhere (and which, in many cases, have also been discovered by
other investigators), but only to summarize the results. Moreover, it bears mentioning that,
with distributional evidence; while several of the first sign values are arrived at by
majority of sign values were arrived at purely by exhaustive comparison and elimination of
possibilities, as described further on, and nearly all such signs have no obvious graphology
On one feature of Indus writing all investigators agree, and have been shown to be correct:
the direction of Indus writing is ordinarily right to left, with occasional exceptional instances
A certain number of inscriptions present more or less random aggregations of signs, in that
they cannot be resolved into any obvious sign “fields” of apparently different meanings.
These inscriptions we have termed “complex” inscriptions (after Wells 2011), not to suggest
that they are unusually complex or opaque, but that they do not resolve themselves into the
highly regularized sign fields found in the majority of Indus inscriptions, to be discussed
(MH3105)
(MH2114)
(MH2422)
(MH1018)
(MH1038)
(MH1056)
(MH4024)
(MH5078)
(MH1406)
(MH4003)
On the other hand, a large majority of Indus inscriptions do present very strong, predictable
patterning, which can be resolved into several sign field types, whereof inscriptions may
display one, some, or all of the fields, nearly always in a canonical order. These “patterned”
inscriptions constitute the bulk of the Indus corpus, with many of them occurring multiple
times.
left/end, as follows: P-field (P), M-field (M), C-field (C), and T-field (T). Additionally,
many canonical inscriptions contain a second line (usually, although not always, found on a
separate face of the object) consisting of a series of vertical strokes and a U-shaped sign .
Table 2 illustrates a few examples of patterned inscriptions, with fields separated by spacing
T C M P
(MH2335)
T C M P
(MH1013, etc.)
T C
(MH4650, etc.)
T C
(MH3120)
T C M
(MH2046)
T C M
T C P
T C P
(MH4441, etc.)
(MH2183)
T M
(MH3074)
T M
(MH4056)
T M P
(MH6206)
T M P
(MH1400)
T M P
T M
(MH4015)
T M
(MH4028)
T M P
(MH2426)
T C MC
(MH1456)
T C M P
(MH2446)
T C M P
(MH2541)
T C M
(MH2654)
T C M P
Note from the above that P-fields nearly always (counterexamples are not shown on this
table, but do exist) incorporate one of three signs in leftmost position, , , and (less
commonly, and not shown on this table) ; these three signs we have called the “juncture
signs.” Two of them, and , are never found anywhere except in P-fields. These signs
likely signify some kind of noun case or noun case-like relationship, and as such are likely
Most of our previous work having been done to establish the structural patterning of M fields
and their implications for the likely content of the other three field types in patterned
resolved into three subfields, the oval (O) subfield, the fish (F) subfield, and the very
common sign pairing , which I designate MS (for “measure”), for reasons that will
become apparent. These three subfields typically occur in the order MS-F-O in Indus texts,
meaning that we would read them in the order O-F-MS. There are occasional additional
entries in canonical M fields, such as the 12-stroke numeral , but we will neglect them for
now. Suffice it to say at this stage that the typology O-F-MS is by far the most common M
field configuration. The O subfield consists of at most two signs from the inventory / ,
, and ; other signs with similar ovate graphology, like , , , and , never occur
in M-fields. The F subfield consists of at most three signs from the inventory , , ,
never occur in M-fields. These two subfields appear to designate two series of weights, while
the MS pair has the aspect of what I have elsewhere termed a “pleonastic number-
measurer,” i.e., a term meaning something like “in the amount of,” “equivalent to,” or
“exactly.” The described internal structure of M-fields may be observed in Table 2 preceding.
, , , , , , and . All of these signs are clearly compounds built up from the
fish grapheme , and all clearly have a shared function within the context of the F subfield.
2) The fish grapheme has a broader distribution than the six compound fish signs. The
most conspicuous context for the fish grapheme other than the F subfield is left-adjacent to
(MH1041)
(MH1365)
(MH1376, MH2374)
(MH2594)
(MH3246)
(MH3351)
(MH4073)
(MH4141)
(MH9822)
:
(MH2524)
(MH1019)
(MH2229)
(MH7243)
(MH4171, MH4873)
(MH1314)
(MH4356)
(MH4377)
(MH2128)
(MH2233)
(MH4116)
(MH4009)
This implies that has some association with numerals or counting. By extension, the
compound fish signs in F subfields, and the F subfields themselves, must have similar
connotations.
3) The F subfield (and the M field as a whole) is always found either directly right-adjacent
established as an “autonomous” sign field (i.e., a sign field also found by itself, implying its
value as a full word or word sequence independent of an M field). These two patterns of
distribution are shown in tables 4 and 5; with this and in other tables in this work, the relevant
(MH7247)
(MH8017)
(MH7220)
(MH3074)
(MH1088)
(MH1155)
(MH2193)
(MH2675)
(MH1277)
(MH1324)
(MH1380)
(MH1534)
(MH4285)
(MH4702)
(MH4056)
(MH4673)
(MH2523)
(MH2574)
(MH3213)
(MH4143)
(MH1531)
(MH4467)
(MH2469)
(MH1109)
(MH2818)
(MH1053)
(MH5052)
Table 5: M fields right-adjacent to “autonomous” C fields (+T field):
(MH1369)
(MH1628)
(MH4263)
(MH1629, MH2863)
(MH5084)
(MH4237)
(MH9022)
___
(MH2087)
(MH2360)
(MH5237)
(MH1373)
(MH1279, MH1487, MH2181, MH2258)
(MH2291)
(MH5237)
(MH3121)
(MH2906)
(MH2821)
(MH7101)
(MH5277)
(MH1005, MH4018)
(MH1120)
(MH2643)
(MH2326)
(MH2221)
(MH1351)
(MH4101)
(MH1208)
(MH1087)
(MH1549)
(MH2537)
(MH3120)
(MH1021)
(MH3307)
(MH1370)
(MH4076)
(MH9071)
(MH4661)
(MH4371)
(MH4650, MH2600, MH5404, MH6128)
(MH2325)
(MH2046)
(MH1040)
(MH2375)
(MH2606)
___ (MH1323)
(MH2489)
(MH4670)
(MH7056)
(MH2222)
4) Based on all of the preceding data, it is clear that M fields cannot be transcriptions of
names or other words or word groups as morphosyntactic linguistic phrases. Instead, they
5) The only types of notations that appear to conform to all of the data are metrological
such denoting month, another the day, and another the year, which would be present in all M
fields; but what we find instead is that all three subfields, while always occurring in the same
order when all three are present, do not necessarily always co-occur—F subfields frequently
occurring without O or MS, for example). Hence the interpretation of the three subfields of
the M field is that the O and F subfields are notational representations of two series of
measures (probably weights), while the MS appears to be, as already noted, a term
denoting measurement, in general, that may also denote some conventionalized basic amount.
Based on the likely metrological content of M fields, it follows that they represent some kind
reason that the content of at least P and C fields, as well as of complex inscriptions lacking
M-fields, would be expected to consist of family and personal names, titles, and possibly
commodities and assets, and even toponyms. All of these would most likely be nouns, both
common and proper, and therefore constitute in the aggregate the best subdomain of the
fields would presumably be morphosyntactic transcriptions of words and word groups. What
follows in this study, therefore, is primarily a partial decipherment of the content of complex
between notational writing, such as we have observed in M fields, and transcriptive writing,
$5.99/lb. (notational); five dollars and ninety-nine cents per pound (transcriptive)
6’4”/255 lb. (notational); six feet four inches, two hundred fifty-five pounds
(transcriptive)
6.02214 x 1023 mol -1 (notational); six point zero two two one four times ten to the
entirely or nearly so. Note also that notational and transcriptive writing may be mixed within
Mr. (notational) John Doe: (transcriptive) $325.27 (notational) past due (transcriptive)
The preceding example mixes transcriptive and notational texts, with the latter consisting of
abbreviations, numerals, and non-alphabetic symbols ($). In general, it will be observed that
notational texts are often characterized by abbreviation and non-sequentiality (i.e., the
sequence of symbols as written does not necessarily reflect the sequencing in utterance; we
do not say “dollars three hundred twenty-five,” but instead “three hundred twenty-five
dollars,” although the symbols representing “dollars” is written before the numerical
amount). As the modern English examples illustrate in part, information often conveyed via
notational writing, across languages and cultures, tends to include numbers, measures,
transactional records, personal titles, calendrics, locations, etc. From the evidence, M fields
are clearly mostly notational and not transcriptive in character. As a result, their internal
syntax may be presumed not to be reflective of language syntax, and many entries in M fields
may well be abbreviated and otherwise conventionalized forms. Later we will examine
several other variants of M fields found in the Indus inscriptions that are similarly notational
in character. The primary focus of this study, however, will be decipherment of the
transcriptive portions of Indus texts, for these alone may possibly yield actual sound values
Having established the notational and exceptional status of M fields, we posit the
fields and C fields within patterned inscriptions. The first task, therefore, is to establish, as
completely as possible and based on information from these transcriptive texts, what can be
known with certainty or near certainty about the typology of the language that underlies the
writing, the nature of the writing system itself, and the content of the inscriptions.
As to the typology of the language, it is a near-certainty that more than one language is
represented in the Indus inscriptions, but it seems a fair supposition that the bulk of the
inscriptions not of West Asian provenience probably represent the same underlying language.
Moreover, in speaking of the underlying language, we are not necessarily referring to the
vernacular language or languages of the Harappans, although they are generally assumed by
other investigators to be one and the same. But there are numerous instances of non-
vernacular languages being used in writing and record-keeping, and of names, etc., being of
foreign origin (as with the Mittani, e.g.). So we are in nowise warranted in assuming
dogmatically that the language of the Indus inscriptions coincides with whatever vernacular
or demotic languages or languages were used by Harappans. Indeed, there is some evidence
that this may not be the case; many investigators have noted the remarkable uniformity of the
inscriptions, regardless of their time depth. Although many of the inscription-bearing objects
were never seriated with respect to time depth by the early excavators of Indus sites, it is
nonetheless apparent that the inscriptions exhibit little variety, despite representing a
sampling of writings spanning many centuries. This fact would seem to suggest some highly
Before proceeding, we ought to address the well-publicized concern, posed in recent years by
Farmer, Witzel, and Sproat (2004), that the “Indus script” may not be writing at all, but is
instead some other kind of symbology. Farmer et al claim provocatively that the patterns of
distribution, including sign frequency and variety of context, do not coincide with any
plausible writing system. This point of view has been contested by Rao et al (2009), but in
this investigator’s opinion, Farmer et al have hit upon a very significant point that previous
investigators have brushed aside, but that any honest researcher must sooner or later come to
grips with. Even the casual student will soon notice the extreme regularity and repetitiveness
of a large portion of the Indus inscriptions, coupled with the extreme abundance of certain
signs and sign pairings (such as and ) that do not seem to square with the
randomness and frequencies associated with natural language; this is true even of the
transcriptional portions of the Indus corpus, i.e., P and C fields and “complex” inscriptions.
Yet the overall number of identified signs in the Indus signary is suggestive of some kind of
logosyllabic script, implying that a significant number of the more common Indus signs
ought to coincide with common CV-type syllables that tend to predominate in all languages,
with their values by exploiting the search features of online dictionaries of classical South
Asian languages. Our resources were the online version of the Monier-Williams Sanskrit
dictionary and the Cologne Online Tamil Lexicon, both maintained by the University of
Cologne. These dictionaries are searchable by initial, medial, and final substring, meaning
that it was possible to ascertain the approximate lexical frequency in initial, medial, and final
position of any CV syllable in either language. We first recorded the relative frequency in
initial, medial, and final position of a large number of expected CV syllables in both Sanskrit
and Tamil, and then sought to match them with “distributional dossiers” drawn up of every
Indus sign with more than 20 or so occurrences. Dictionary entries are not corpora, and are
expected to yield different patternings of syllable distribution, to some extent, from more
natural, randomized texts. At the same time, it was evident that the Indus texts do not
represent anything like a natural sampling, owing to their brevity and repetitive character. In
preparing the “distributional dossiers,” multiple occurrences of the same sign sequences were
conflated, so as to have a list of inscriptions that more closely resembled the non-repetitive
inventory expected from a lexicon, and relative frequency of signs was also taken into
account in comparing Indus signs with candidate CV syllable values. Yet in spite of this, no
signs were found that closely resembled the pattern of occurrences of any of the common CV
syllables in either Tamil or Sanskrit, and only a very few that somewhat approximated the
expected distributions of such syllables. One problem was that the majority of common signs
that did not appear to be logographic or to coincide with some kind of grammatical function,
like case marker or determiner—i.e., signs that were assumed to be purely phonetic—still
appeared much more frequently in initial position than normal distributions would suggest. In
the end, the methodology of trying to match CV-syllables with candidate Indus signs was
largely unsuccessful for both languages, with the exception of a handful of signs that did
As a result of this unsuccessful effort, we were forced to conclude that the Indus signary
2) The Indus script represents a language with a radically different typology from the
3) The inscriptions are highly stylized and limited to certain types of information that
Because of our earlier conclusions regarding the nature of the script, and in particular the
notational character of M fields, possibility 3 was deemed the most likely. As mentioned
previously, earlier work had shown that many of the inscriptions contained numerical and
metrological annotations that likely indicated assets; clearly fields containing such material
(like F subfields or “fish sign clusters”) would be expected to be highly stylized and non-
random, a factor generally not taken into account by statistical studies of the Indus corpus as
a whole. In addition, if a large proportion of the inscriptions contained names and titles, other
constraints might be operative, since names and titles do not promiscuously incorporate
random words, but instead tend to favor certain terms and morphemes (patronymic
designations and names of deities, for example). Thus an inventory containing a long list of
names and titles would also be expected to skew towards repetitiveness, and not pattern like
syllables, common or conventionalized elements of names and titles, allows for the
possibility of common signs representing more complex syllabic values than simple CV, such
as, e.g., CVC, CVCV, etc. In general, we are forced to suppose that any sign that occurs as a
sign (especially ), or as a standalone P field minus the left-adjacent juncture sign — must
represent an entire word or word stem, and hence is likely to represent a minimal form CVC
or CVCV, since C fields and P fields are most likely to represent nouns, and most nouns tend
across language to have a minimal form CVC or CVCV (although some word classes, like
pronouns and prepositions — and even some nouns — can have the sound shape CV). As one
example, consider the rather common sign . This sign occurs very frequently as , with
no other signs in combination with it. The general reading here must be that represents an
entire word-morpheme, with terminal sign constituting either some kind of affixal
morphology or something else — a predicate, a determiner, etc. — that at once completes the
meaning and is extrinsic to . As it turns out, there is not enough data to say much more
about , except that it most likely denotes a full word root and has a likely minimum sound
shape CVC, VCV, etc. — i.e., more complex than CV. On the other hand, signs like and
never occur singly with left-adjacent, but instead often form sign pairings or clusters
that have the aspect of irreducible words or word-roots, such as and . Such signs
are thus much more likely to have simple CV-type syllabic value and, as we shall see with
these and certain other signs, this turns out indeed to be the case.
Thus, we adopt as a working hypothesis that the Indus signary, as long suspected, does
indeed contain both simple CV-type syllables as well as many signs representing full word-
morph value with more complex sound shapes, such as CVC and CVCV. Many of these latter
types of signs are more common than might be expected because they likely represent entries
Given the brevity of most of the inscriptions, it is unlikely that the bulk of them represent
sentence-length texts with much information regarding typological characteristics like word
order. Many statistical studies of the Indus texts have made various claims about evidence for
word order, etc., but the problem with such claims is that, as we have already shown, a large
proportion of the texts are resolved into different sign fields, whose constituent signs behave
very differently from one another and clearly have very different functions. Treating the
distributions of signs like “fish clusters” within M fields, which seem to be notations of assets
of some kind, as equivalent to and interchangeable with those of signs in P fields or C fields,
which apparently contain names, titles, and possibly other nouns, will obviously not yield
useful results. Our approach, as stated previously, will be to treat M fields separately from P
fields, C fields, and complex inscriptions that, by definition, do not include M or T fields.
With regard to these latter three contexts (P and C fields and complex inscriptions), there is
one distributional feature that stands out even to the casual investigator, and which appears to
convey a very significant piece of typological data — data that, because of long-standing
biases among many investigators of the script, has been ignored or misinterpreted. Consider
Table 6:
Set 1:
MH1013, etc.
MH2600, etc.
MH2144, etc.
MH4335, etc.
Set 2:
MH3122, etc.
MH2380, etc.
MH1010
Set 3:
MH3122, etc.
MH2517
MH3078
Note that “etc.” indicates multiple occurrences of the inscription in question besides the index
number given. In the case of set 1, all four inscriptions given, including the longest, have
multiple occurrences. Each inscription also includes terminal sign leftmost, i.e., in final
position. From this we may infer, at least as far as the first three inscriptions are concerned,
that, whatever the purport of , the three common sign sequences occurring right-adjacent
— , , and — all are likely to be words, and most probably nouns of some
sort, whether denoting names, titles, commodities, locations, or some such. But now consider
MH4335, one of the longer known Indus inscriptions, and also an inscription that occurs at
least 11 different times, appears to combine , , and (in that order, left to
right) with left-adjacent (i.e., in absolute final position), and with an M cluster
— occupying the C field, preceded, as is so often the case, by a typical M cluster and
followed, also per a very familiar pattern, by terminal sign . This inscription is absolutely
canonical as to the patterning of M field, C field, and T field, except that the C field is of
some kind of compound sign sequence that can readily be resolved into three more basic
constituents, each of which is presumably a full word or word root in its own right,
presumably equivalent to each of the other two as far as its lexical class is concerned.
We see something similar with the examples in Set 2, with the two very frequent and
repetitive C clusters (a common sign that presumably has the force of an entire lexical
entry) and both occurring right-adjacent to terminal sign , and also being combined
MH3078 with a different autonomous sign pair also found elsewhere several times as a
discrete word, .
These examples, and others like them evident throughout the Indus corpus, are strongly
suggestive of compound words, most likely compound nouns. The examples shown above
have been chosen for clarity, but many others might be adduced, although some are more
difficult to pick out for those unfamiliar with the script and its patterning, because of
complicating presence of M-fields, P-fields, and T-fields other than . Suffice it to say that
examples of such apparent compounds may be found both in C fields and P fields.
This is potentially a very significant typological datum that, as far as this investigator is
aware, has not been previously pointed out. Its potential significance resides in the fact that,
of the two major South Asian language groups, only one characteristically makes widespread
much so, that the rules for the formation of such compounds are traditionally a major source
of vexation for students. The language is Sanskrit, for which a very large number of names
and titles (both divine and human), in particular, are in reality compounds, a fact of South
Asian nomenclature from prehistory down to the present day. For example, devarāja- is a
compound of the nouns deva- and rāja-, candrasekhara- is a compound of candra- and
sekhara-, and so forth. In general, the luxuriance and complexity of compounds increases
from the Vedic to the classical language, but such compounding is a significant typological
course, in words borrowed from Indo-Aryan). There are, to be sure, a few words of honorific
purport, like thiru-, that may be prefixed to a name, but the wholesale compounding of
multiple nouns in the formation of names is not a significant typological feature of Dravidian.
Words in Tamil, etc., may be long and complex, but that is a consequence of the luxuriance
is possible, of course, that the Dravidian languages have changed their typology with the
passage of millennia, and that a proto-Dravidian or early Dravidian language from the
Harappan period may have made more general use of compounding than historical Dravidian
languages do. But the evidence available ascribes most likely to Indo-Aryan the extensive
noun compounding of the sort strongly suggested by internal patterning in some C and P
This is admittedly circumstantial evidence; after all, Indo-Aryan is not the only language
family prone to noun compounding. But we find it compelling enough to warrant asking
whether there are any other such pieces of circumstantial evidence, drawn from the corpus
itself, and not from, say, claims about the absence of horse remains, that might point in a
We regard the next most important piece of evidence to be the “long stroke” numeral sign
often found in clearly numerical contexts (such as the very frequent pairing ), but also
element of a two sign pair; examples of this distributional trait include the very common pair
already mentioned, as well as and ; see Table 7 for some examples of such
occurrences.
Table 7: in word-final position
MH4482, MG4486
MG2015, etc.
MH2426
MH3160
MH2446
Note that the extremely frequent pairing frequently occurs either in combination with a
single common word, most likely a noun. From the other inscriptions in Table 7, we
conclude, because they all occupy single lines of text and are comparatively brief,
and are also likely single words; and to these examples might be added several more
from the corpus involving less-common signs that will not be considered further; see, e.g.,
MH4500 and MH1431. From the appearance of the sign , it seems reasonable to assume
that, in many contexts, it is, straightforwardly enough, the numeral 3. Its appearance in what
appear to be non-numerical lexical contexts suggests the hypothesis that, perhaps, it
represents a common syllable or affix, found often in word-final position, that sounds like the
word for “3” in the Indus language. Again, comparing Dravidian and Indo-Aryan, we find
that the Dravidian root for “three” has the form *mu, *mun(u), *munru, or something similar
(for which see the Dravidian Etymological Dictionary), which forms are not suggestive of
any kind of significant Dravidian word or affix familiar to this investigator. On the other
hand, the Indo-Aryan “three” has the forms tri-/tra-, which are highly suggestive to anyone
familiar with Sanskrit. Sanskrit has two very productive related affixes, the agentive affix
–tṛ (as in pi-tṛ, ‘father,’ and mā-tṛ, ‘mother’) and the “implementive” affix –tra (found in
many extremely common words like mā-tra-, ‘measure; exactly; Brahmin;’ etc., pā-tra-, ‘pot,
vessel, container’, go-tra-, ‘clan, sodality,’ pū-tra-, ‘prince, son,’ kṣe-tra-, ‘field, unit of
property,’ and hundreds of others). A third related affix, -trī, is often found in feminine forms
of words in –tra, but also may form words in its own right, such as ne-trī, ‘eye; leader.’
Leaving unresolved for the moment which of these three values is likely to represent (and
it may represent more than one of them), we conclude that Indo-Aryan has several strong
candidates for a productive affix-type sign that also sounds like the word for the numeral
Another suggestive piece of evidence involves the + stroke numeral field found so
frequently as a second line on inscribed objects, typically on the side opposite the other line.
numeral larger than 4 is ever found in such contexts, even though many higher stroke
numerals are found frequently in other contexts. Because of the pot-like appearance of
coupled with stroke numerals, this sign field has been assumed to represent some kind of
numeral or measurement of some type, perhaps indicative of the carrying capacity or number
numeral or with in lieu of a stroke numeral. These occurrences are shown on Table 8:
Table 8:
MH5484
MH4448, MH5461,
MH5460
MH4363, MH5250
MH5473
MH5283
MH4487, MH5481
The use of in such contexts would seem to be clear enough, given the value of and
various “fish sign compounds” found in M fields as denoting some common unit of
weight/value. The occasional presence of in + stroke numeral-type fields only serves to
reinforce the likelihood that these very common secondary sign sequences are also somehow
The occasional occurrence of in such fields is at once more problematic and suggestive.
appears to be a ligature of the “anthropomorph” sign with the “bow” sign , both
frequent signs in their own right. Anthropomorph signs ligatured with other signs are a
common and significant feature of the Indus signary, and evidence I have set forth elsewhere
suggests (Bonta 2014: 108) that, whatever the significance of the anthropomorph grapheme,
the primary phonetic value is carried its companion ligatured sign. In other words, the bow
grapheme would seem to carry the phonetic value of . For evidence supporting the
i. and :
MH2466
MH2027
MH1087
MH2228 ___
___
___
ICIT 2646
ii. and :
MH2638
MH2413
MH2519
MH2390
MH4260
On the assumption that the sound value of might coincide with or resemble the sound of
the word for “bow” in the Indus language, we may ask whether either Dravidian or Indo-
Aryan possesses any word for “bow” whose sound shape might make sense in a context
this case, both Dravidian and Indo-Aryan yield plausible candidates. Dravidian vil, ‘bow’ is a
near or exact homonym for vil, ‘price, cost,’ while Sanskrit dhan-us, ‘bow,’ is a near-
homonym for both dhan-a, ‘money; wealth; property’ and dhānya-, ‘grain’; any of these
would be contextually apt, but it is worth noting that this is the only instance in which we
were able to find a plausible Dravidian as well as Indo-Aryan value for a sign whose
Dravidian.
Another possible hint as to the identity of the Indus language is the distribution and
graphology of what I have previously styled the “staff sign” / , which we regard as two
allographs of a single sign, both of which resemble staffs of grain, as many previous
investigators have noticed. Both the Mahadevan concordance and the ICIT treat these as
separate signs, but inasmuch as their patterns of distribution are virtually identical and there
are no instances of these two signs co-occurring, there seems to be no basis for other than
assuming them to be allographs. We have shown them as separate signs in what follows only
to facilitate correlation with material in the Mahadevan concordance and the ICIT.
/ has by far the greatest propensity of any Indus sign other than to co-occur with
stroke numerals. Additionally, this sign occasionally occurs in repetitive groupings of three
or more. This data, shown on Table 10, strongly suggests that / , like , probably has
some kind of metrological value, probably another weight. This is not a novel claim;
Fairservis, for example, believed / to be some kind of dry measure (Fairservis 1992: 60-
61).
Table 10: and adjacent to stroke numerals and in multiples of three or more:
MH1220, etc.
MH2370
MH1091, etc.
MH2572, etc.
MH1361
MH2858
MH1143
MH1076
MH1301
MH1131
MH1243, etc.
MH2198
MH4238
MH2298
MH2127
MH1411, etc.
MH7063
MH2008
MH1207
MH1246, etc.
MH2387
MH4843, etc.
MH1422, etc.
MH4669
MH7072
MH1309
MH1439
MH1025
MH2025
MH5057
MH4047
MH2949
MH2950
MH2322
MH7201
___
/ is not always found in such contexts, however, and we will have much more to say
about this important sign in other contexts further on. For our purposes at this stage, however,
it is very difficult to look at inscriptions of the kind found in Table 10 and not conclude that
and, more specifically, some unit of weight — is far more likely, given the similarities
between the distribution of this sign and the fish sign , which we have already established
to be a fundamental unit of weight, and, moreover, the source of an entire series of weights.
/ does not participate in “clusters,” and does not appear to form metrological compounds
of ligatured signs along the lines of the various “fish compounds” (although, as we shall see
further on, this sign does appear to form ligatures in other contexts). It is likely to denote a
relatively heavy unit, for which the kinds of notations seen on Table 10 represent significant
assets or transactional amounts. The question we now ask is: Is there any unit of weight,
likely to participate in such metrological notations, whose name is in any way suggestive of a
staff or blade of grain, or anything similar to it, in either Dravidian or Indo-Aryan? We were
unable to find any plausible Dravidian candidates, but there is, as we have already noted in
previous work (Bonta 2014: 95, 113), a very plausible candidate in Sanskrit/Indo-Aryan,
namely, the pala, a very common, relatively heavy weight of gold and silver, whose name
also means “straw.” While there is some uncertainty, due to the variability of weights and
standards over time, the pala was equivalent to 4 karshas or 64 māshas, or somewhere
between 704 and 1120 grains Troy (i.e., roughly between two and three Troy ounces). Thus a
tentative value / = pala is a very neat solution under an assumption that the Indus
connection with Table 7 furnishes a powerful clue to the identity of our language. The
following is also discussed in our previous work (see, e.g., Bonta 2014; 104), and represents
one of our earliest and most significant insights, because it yielded values for two very
important signs that we were ultimately able to use as “anchors” to ascertain other sign
almost always found as the leftmost/final element in such clusters, typically right-adjacent
either to a T field or a C field. A few examples of this very common distributional trait may
be seen on Table 7. This data has been shown previously, in Table 2, and, at risk of seeming
Table 11: in M fields (M) relative to T fields (T), C fields (C) and P fields (P):
MH4015
T M
MH4028
T M P
MH2426
T C M
MH1456
T C M P
Mh2446
T C M P
MH2541
T C M
MH2654
T C M P
Sanskrit in the masculine, feminine, and neuter forms, roughly as follows, according to
Monier-Williams:
mā-trā (feminine): “measure (of any kind), quantity, size, duration, number, degree &c.
mā-tra-m (neuter): “an element, elementary matter … measure, quantity, sum, size, duration,
measure of any kind (whether of height, depth, breadth, length, distance, time or number
e.g. …artha-mātram, a certain sum of money; krośa mātre, at the distance of a Kos; māsa-
anything, the whole or totality, the one thing and no more, often = nothing but, entirely,
notation would mean something like “exactly X,” “precisely the amount X,” etc. Note that
either the feminine (mā-trā-) or neuter (mā-tra-m) form might fill our requirement, and note
also the possibility that the masculine mā-tra-s, denotative of a broad class of Brahmins,
might be an admissible reading in some contexts. Most importantly, note that the proposed
productive, related Sanskrit affixes, -tṛ or -tra. Our reading of suggests the two values
= tra/trā and = ma/mā. This supposition is lent further support by the distributional
behavior of , which does not occur in contexts suggesting a complex, full word value, but
instead is always paired or clustered with other signs in C fields and P fields, with fairly
random distribution. This, in combination with its high overall frequency, suggests that is
a strong candidate for a common CV-type syllable, and ma is typologically among the most
To summarize, then, we have identified a body of evidence, arising entirely from an analysis
of the internal distributional workings and graphology of the script, all of which points in the
1) Strong evidence for the formation of noun compounds, a typological trait very
2) Evidence for a productive affix suggestive of the number “three,” as evidenced by the
sign , which likely coincides with the Sanskrit value –tra-. This evidence is
suggesting a possible value of dhana-, ‘money, property,’ this common word being a
4) The very plausible reading of pala for / , another metrological sign or unit of
weight analogous to , based not only on the importance of the pala as a unit of
weight, but also because its other meaning, ‘straw,’ is suggested by the graphology of
this sign.
Taken individually, any of these might seem circumstantial, but taken as a whole, they
present a very significant body of evidence — all taken from the script itself and not from
any pre-existing assumptions or cultural biases — that, after all, the principal underlying
language of the Indus writing system is an early form of Sanskrit, presumably somewhat
Much of the above we have already expressed in an earlier and very imperfect iteration of
this project. In that work, we set forth the following values, which, we are happy to say, have
withstood many tests and which ultimately became among my “anchor” signs which we were
able to use as an entering wedge to decipher a significant number of other Indus signs:
= min(a)
/ = pala
= dhan(a)
= ma/mā
= tra/trā
In addition to these, we have elsewhere suggested that the so-called “jar sign” and “arrow
sign” represent common predicates having to do with ownership, namely, ‘be’ and ‘be
worth,’ respectively. Because these signs have often been assumed to represent case endings
or some other kind of affixal morphology, the notion that they in fact represent entire
predicates might seem counter-intuitive. It is worth restating the reasons for rejecting their
frequently occurs to the left of sign sequences, like fish/oval clusters, that
(MH6129)
(MH3074)
Why, if and represent grammatical affixes like gender markers or case endings,
would they occur only singly in such contexts (i.e., instead of occurring separately
and discretely with each lexeme to be marked with a given grammatical feature)?
2. has several other forms, apparently akin in graphology but differing strikingly
in distribution from , although not from one another: , , and . These three
signs, as shall be discussed in detail later on, are, unlike , frequently found in
rightmost position, i.e., in the prefix environment, rather than in T fields. Assuming,
as seems warranted, that , , and are derived from while being distinct in
grammatical affix can have modified versions occurring in the initial or prefix
environment.
3. Besides and , there are a number of other signs and graphemes typical of the
or be associated with them. Certain of these — the “bearer” sign and the enclosing
marker, then , , and would have to be markers for different cases (or perhaps
different genders, numbers, or declensional realizations of the same case), with other
additional affixal information. If were regarded as some other type of affix, then
, , and would then be other markers of the same property. If, for example,
is the Dravidian masculine gender marker –an, then , less frequent and marked with
respect to , would most likely be its feminine counterpart -al. The other terminal
signs, however, would likely denote entirely different kinds of affixes, a conclusion
that militates against the distributional evidence for their constituting a set of signs
with analogous functions (i.e., their occurring in contrastive distribution with and
).
impoverished logosyllabary; as [we and] others (Farmer et al 2004, e.g.) have noted,
the Harappan signary is lacking in signs that occur both frequently and randomly,
representing case affixes, given that ancient writing like proto-Cuneiform and early
Maya only developed grammatical affixes well after the initial invention of writing
especially in view of the success such an approach has had in prior decipherments,
such as the work of Kobers and Ventris on Linear B (see Chadwick 1958), it does not
appear to be defensible with the Indus Valley script. Such an interpretation would
leave us with an unwarranted multiplicity of cases or other affixes that would make
no sense contextually (Why, e.g., would the same lexeme be marked with several
different interpretation of and other terminal signs is in order. (Bonta 2014: 76-78)
We persist in the opinion that represents the fundamental Indo-Aryan predicate root AS,
‘be,’ as both a copula and as a marker of possession, the latter being a characteristic way to
mark ownership in South Asian languages by saying, in effect, ‘such and such is/are of/to so-
[T]here is an explanation for that fits very nicely with South Asian language
typology, namely, that this very common sign in fact represents the predicate ‘be’ in
either Indo-Aryan (as- or bhū-) or Dravidian (uḷ-, ā-, or iru-). In both Indo-Aryan and
Dravidian, there is no separate word or root meaning ‘to have, own, possess.’ Instead
of saying, e.g., “I have such and such,” speakers of Indo-Aryan and Dravidian say,
anciently and in modern times, “Such and such is to me” (similar to English
complement, like ‘this book is mine’ and ‘the money is Jim’s’). For example, in the
Rig Veda, we find expressions like yā vah śarma śaśamānāya santi, ‘The shelters
[śarma] which you [vah] have [santi] for the zealous man [śaśamānāya]’ [literally,
‘which shelters are to you for the zealous man’] (RV I, 85:12).
From the medieval Pāli commentary the Purana Vinaya Sannaya, we find phrases
like hiraññam me atthi, ‘I have gold’ [lit. ‘gold is to me’]. And from the Buddhaghosa
the Vinaya Piṭaka, a part of the Pāli Tipiṭaka) a contract for debt is given as
asukasmin nāma thāne mama hirañña suvaṇṇam atthi, tam tuyham hotu, ‘In such a
place, I have a suvarna of gold, you may have it’ (literally, ‘In such a place, a suvarna
of gold [hirañña suvaṇṇam] is to me [mama atthi], to you [tuyham] let it be [tam
hotu]).’
In modern South Asian languages, similar usages are found. In Colloquial Sinhala, for
example, the verb tiyenәwa, ‘be,’ is used with a dative case subject to indicate
rendered enakku oru puttakam irukkiratu (enakku = ‘to me,’ oru puttakam = ‘a book,’
irukkiratu = ‘is’).
Nor are such uses unique to South Asia. Although most western European languages
have a word corresponding to ‘to have, possess,’ the word corresponding to ‘be’ in
various forms is frequently used to signify ownership. For example, as we have noted
above, in English we have constructions such as ‘this book is Harry’s’ and ‘the red
car is hers’. Spanish has similar constructions, using objects of the preposition de
Este libro es de Juan (‘this book is Juan’s; [literally] this book is of Juan’)
frequent across languages (at least in languages with an explicit copula). (Bonta 2014:
78-79)
Since we have established the likelihood that the Indus language is Indo-Aryan, the most
plausible value for is AS, which may be held to represent various inflectional values of
this root (asti, santi, e.g.). In other words, as far as this predicate is concerned, we see no
basis for supposing that Indus writing attempts to reproduce phonetically verb forms that vary
inflectionally according to number, person, etc. We suggest that, owing to the extreme
compression of form required by the seals and other Indus source materials, this predicate is
indicated only by a “root sign,” as it were, which may not convey all phonetic information,
This hypothesis is lent additional credibility by the nearly identical behavior of which, for
arguments similar to those given above regarding , is also very unlikely to represent any
kind of case marker or other inflectional affix. We also continue to regard the meaning and
probable reading suggested in previous work (Bonta 2014: 105-106) to be viable, namely,
= ISH, where ISH/iṣ- may carry the primary meaning ‘be accepted or regarded as; be worth,’
as given in Monier-Williams. This supposition is motivated in part by the meaning of iṣ- ‘be
accepted as, be worth,’ an expected predicate in the context of the inscriptions (i.e., ‘Worth
such-and-such an amount,’ ‘Good for such-and-such and amount,’ etc.), and in part by the
near-homonym status of the most common Vedic Sanskrit word for ‘arrow,’ iṣu. Another
possible value in some contexts is the near-homonym ĪŚ/īś-, ‘own, possess,’ where =
and :
Table 12: , , and Distributional Contexts
i. Left-adjacent to :
MH2402
MH1024
MH1041
MH2298
MH1140
MH7045
MH2041
MH1145
MH2537
MH1021
MH4104
MH8017
MH4241
MH1155
MH4263
MH2614
MH2193
MH1388, etc.
MH1420
MH4014
MH2643
MH1419 ___
MH1017 ___
MH1404
MH4113
MH4263
MC
MH6132
PC
MH2447
PC
MH4238
PC
MH1419 ___
MC
MH2253
MC (?)
MH2916
PC
MH4801
MC (?)
At first blush, it might seem that , , and are either variants of when occurring
after the very common P cluster ; but this supposition is greatly weakened by the facts
five other examples. Another possibility that suggests itself, given these three signs’ frequent
adjacency to M clusters, is that they are themselves metrological signs, like the fish and oval
series indicative of some unit of weight or other form of reckoning. But this is also highly
unlikely, given the propensity of , , and — in stark contrast to fish and oval signs
— to occur in P clusters as the rightmost or absolute initial entry. We suggest that these three
signs, like , have a strong association with M clusters, but are not themselves
metrological per se. The best hypothesis that we have so far been able to set forth for the
identity of these three signs is that they are nominalizations of having the meaning, in
effect, of ‘property’ or ‘goods.’ Such words are found in both Indo-Aryan and Dravidian;
with Tamil, the term iruppu, a nominalization of the root iru-, ‘be,’ means ‘balance on hand,
surplus, stores, merchandise, wares’ (Burrow and Emeneau 1984, entry 480), while Tamil
uṭai, ‘wealth, property’ (ibid., entry 593) appears to be a nominalization derived from the
other Tamil/Dravidian root for ‘be’, uḷ-. In Sanskrit, meanwhile, we find the very suggestive
word astimat, ‘possessed of property, opulent’ (Monier-Williams 1990: 122), whereof the
very common Sanskrit affix –mat means ‘having, possessing,’ and the stem asti-,
transparently derived from the verb root as-, ‘be,’ must mean, or have meant, ‘property’ at
some time in the past. Indeed, the word has been conveyed down to modern languages of the
Subcontinent, despite being absent from Monier-Williams. The most common word for
‘property’ in both Hindi and Tamil, for example, is āsti, which the latter language
presumably borrowed from Sanskrit, and the former inherited directly. Because we have
assembled significant evidence pointing towards an Indo-Aryan rather than a Dravidian Indus
the meaning ‘property.’ The three main variants of this sign may possibly correspond to the
singular, dual, and plural numbers, either of the property itself or, perhaps, of the owner(s).
In sum, the following values, arrived at in earlier work (Bonta 2014: 113), we still regard as
valid:
= mīn(a)
/ = pala
, = dhan(a)
= ma/mā
= tra/trā
= AS, ‘be’
/ / = ASTI, ‘property’
= ISH, ‘be accepted as, be worth’ (or possibly in some cases, ĪŚ, ‘own, possess’)
vexed, inasmuch as these signs occasionally seem to correspond to literal phonetic values as,
asti, etc., as we shall see further on. But in general, when they are deployed in their canonical
contexts, they have every appearance of either logograms or of greatly abbreviated notations
The other sign values proposed in Bonta 2014 we no longer regard as valid, because they
arose from an important error in that earlier work. At that time, we proposed that the P field
likely represented actual goods, like gems, grain, etc., whose values were expressed by left-
adjacent M clusters. The basis for this argument was the fact that / , ‘pala,’ also occurs in
suggested that the rather common P cluster sequence / probably meant something
like ‘palas of grain (dhanya-) or money (dhana-). We then provisionally assigned values
= ratna (‘gem’) and = mani (‘gem, jewel’) based on the assumption that all P clusters
must represent such expressions, and that, overall, the four basic fields could be construed as
follows:
Unfortunately, this neat set of assumptions did not bear up under further scrutiny. For
example, if C fields contained (presumably) names and titles, then why does / also turn
overall, the very great diversity of sign arrays to be found in the P field seemed far too great
than could be accounted for by assuming this field to be denotative of what would surely be a
fairly limited set of possible commodities — gems, precious metals, grains, and so forth.
Moreover, as noted previously, P fields resemble complex inscriptions found on some seals,
and it seems most improbable that complex inscriptions would simply denote some
commodity and nothing more; names and titles would be far more likely in such contexts.
The solution to the problem of / = pala suggested itself several years after the
publication of Bonta 2014, and ultimately led — after many more years of fits and starts —
to the partial decipherment of the Indus script which will be presented in the following
sections. In brief, we finally realized that / may stand not only for pala (‘straw; unit of
weight’) but also for pāla, a common word meaning ‘guard, protector, keeper, etc.’ and a
very common final element in names and titles (more than 300 different such are recorded,
e.g., in Monier-Williams). The implications for this broader reading have turned out to be
very significant. For one thing, it is clear that Indus signs may indeed represent sound shapes
of the form CVCV, as we have already suggested. For another, vowel length is apparently not
distinguished in Indus writing, opening up the possibility that many signs may have at least
two phonetic values corresponding to consonant configurations (like p-l-) containing both
long and short vowels. And finally, we see that signs may have multiple meanings, allowing
for signs associated in some contexts with some specific function like mensuration to be
deployed in other contexts with completely different meanings. Thus / , when deployed
left-adjacent to stroke numerals, is to be read pala and stands for the unit of weight. But in
general, when deployed in P and C clusters (where it is usually in final position, except for a
pāla and means ‘guard, protector,’ usually in a compound with some preceding/right-adjacent
sign or cluster.
The first such word ending in –pāla that we were able to identify arose naturally from the
very first list of anchor signs given above: sequence / = dhana-pāla, i.e.,
dhanapāla, ‘guardian of treasure, treasurer (AV); king; name of a grammarian and of various
other personages.’ Some examples of this sequence are shown in Table 13, with relevant
Table 13: /
i. Subsequence /
MH4624 ___
MH2449
MH1092
MH4131
ii. Other:
MH1400 [line 3 of 3]
MH1087
ICIT 2646
MH9091
MH2271
Note that the subsequence / appears to read dhanapāla ASTI (āsti), i.e.,
‘property of the treasurer,’ or some such. For the other sequences, note that / in
MH1087, MH2018, MH9091, and MH2271 occurs as the leftmost element, along with a left-
that / is a final element or title in a longer name or title compound, which agrees
with our reading of / as the title dhanapāla, titles often (though not always)
occurring after names in early Indo-Aryan, not preceding them as is the case in English. Thus
Table 14: /
MH4650
MH4161
MH2176
MH8011
MH2600
MH1081
MH7223 ___
MH4075 .
Although its most frequent context is right-adjacent to terminal sign in the C field, /
may also occur in the P field (MH1081) and in non-patterned contexts (MH2176,
MH8011). The preliminary value assigned to this sign (which subsequently was modified
slightly, as will be clarified further on) was = bhū, such that / = bhū-pāla, i.e.,
bhūpāla, ‘king, prince; name of various attested individuals’ (lit., ‘earth-protector’). This
value was arrived and verified after many weeks of work and comparison, and ultimately fit
perfectly with its various contexts, as will be seen later. For now, we note in Table 15 the
following occurrences of in conjunction with known signs (shown with relevant segments
underscored).
names and titles; a complete reading of this inscription will be given further
on]
From the data in Table 15, we see that occurrences of with known signs / = dhana
square with the expected incidence of -bhū as a final element in compound names and titles;
as mentioned in Table 15, we will be able to give complete readings for both MH2114 and
contexts, such as and , to be a stroke numeral [check with visuals in Wells]. These
Table 16: / +
2109
2348
2564
The reading of would be bhū-tra-, but this corresponds to no word or syllable sequence
in attested Indo-Aryan. On the other hand, the very common sequence pu-tra-, ‘son; child
represent putra- would require making one of two assumptions, either a) the word putra-
descends from an unattested earlier form *bhū-tra- or b) Indus signs may represent more than
one set of sounds — in this case, = bhū and pu. Regarding the first, the origin of putra-,
unlike that of most other common words in –tra, is something of a mystery; according to
Monier-Williams, the etymology of this word is doubtful, being perhaps connected with the
root puṣ-, ‘nourish,’ and has also been given the fanciful derivation from put-tra, ‘preserving
from the hell called Put.’ On the other hand, a hypothetical form *bhū-tra- would mean
something like ‘produced by/born from/fashioned of the earth,’ a very logical term for a son
or offspring.
Regarding the second possibility, we have already suggested that vowel length is not
distinguished for the polyvalent sign / , so it is not unreasonable to consider that might
also be polyvalent, representing not only both u and ū but also different bilabial stops. Since
the evidence for the origin of putra- is purely conjectural, we consider the polyvalent
interpretation to be much more likely; further on we will see that a number of Indus signs
appear to be polyvalent in that they conflate similar but phonemically distinct consonant and
vowel sounds apart from vowel length, and also vary as to the inclusion or exclusion of
To capture conflated vowel length, we will hereafter use upper case vowels, and to express
conflated consonant values, we will use upper case consonants, with clarifying notations in
= min(a)
= dhan(a)
= mA [ma, mā]
= trA [tra/trā]
= PU [bhū, pu].
= AS, ‘be’
/ / = ASTI, ‘property’
= ISH, ‘be accepted as, be worth’ (or perhaps sometimes ĪŚ, ‘own, possess; belong to’)
Another sign worth considering at this stage is the very common but highly idiosyncratic sign
, which usually (though not exclusively) occurs in P fields right-adjacent to juncture sign
, often simply as the pairing (the most frequent sign pairing in the Indus corpus), but
also frequently with some other sign or signs right-adjacent — meaning that must be a
sign common in its own right as a standalone name or title, or that can occur commonly as
the final element of a name or title. This last property is akin to pāla, but, unlike , / /
A significant clue to the identity of is evident from cross-checking this sign against the
MH1305
MH7107
MH1400 [line 3 of 3]
MH8104
MH8006
= dhana-X
= pāla-X
= dhanapāla-X
= PU-X
Again, our one additional requirement is that be a common standalone sign denoting an
important name or title (having established with the sequence = dhanapāla that P fields
The value for that best fits all of these conditions, ascertained by the method previously
mentioned of cross-checking each of the known values shown in a searchable online lexicon,
is pati, with meanings ‘master; owner, possessor; lord; ruler, sovereign; husband.’ This word
frequently stands alone with one or more of the senses indicated (and in fact is still in use
today in modern South Asian languages like Hindi), but also may, like -pāla, be suffixed to
other words, names, and titles. The above relevant sequences would then be read,
preliminarily, as follows:
= dhana-pati
= pAla-pati
= dhanapāla-pati
= PU-pati
Of these, dhana-pati and PU-pati are straightforward; dhanapati means ‘lord of wealth’ and
is also an attested name, while bhūpati means ‘monarch, prince’ (lit., ‘earth-lord,’ a near
synonym for bhūpāla) and is also an attested name. Neither palapati nor pālapati is attested
Moreover, balapati is a common and well-attested word, with the meaning ‘general,
commander’ (lit., ‘lord of strength’), and since we have already suggested that Indus signs
may signify more than one similar consonant value, the supposition that may signify bala
Finally, /pati itself, with its particular range of meanings, would be expected, in the
It may seem to the reader that such evidences are too scant or too selective to be conclusive.
There are many other contexts in which , and the other signs so far considered, may be
seen to occur, and for many of these, we will also offer very plausible readings. The purpose
of these early illustrations is to show how the method of cross-checking led to the assignment
of what at first were placeholder values for certain common and idiosyncratic signs, which
values eventually found sufficient and repeated confirmation as the process of decipherment
advanced to warrant their adoption as deciphered signs. Finding even workable placeholder
values for most signs was a tedious, lengthy process of trial and error, especially in the early
going, when dozens of plausible values often presented themselves even after repeated cross-
checking. We do not intend to burden the reader with the details of this process and the many
false leads and early misidentifications entailed thereby, although they are preserved, in all
It will be noted that several of the first signs deciphered — , / , , and , for
that the ascribing of sound values based purely on graphology and presumed rebus values is
claims for the Indus script have consisted almost entirely on the method of positing sound
values for most or all of the signs in either Dravidian or Indo-Aryan based on supposed
graphology alone. This method has invariably produced grotesque and indefensible readings.
important part of decipherment, it cannot be the default or sole method. In general, the
distributional properties of signs will yield far more information than trying to fit individual
signs to words and sounds based on apparent graphology. Accordingly, in producing the
following list of values for deciphered signs, I have relied almost exclusively on the method
of repeated trial and error, in conjunction with cross-checking against known sign values, as
exemplified above for the first few signs deciphered. As a result, sign values will seldom
In general, we have found the Indus signs to consist of both CV-type syllables and “root”
signs, the latter class generally corresponding to a minimal form CVC(V). Some CV-type
syllables may also occasionally correspond to roots, but for the most part, such signs will
coincide with affixes (CV-type affixes like su- and -ka being very common in Sanskrit).
As we have seen already, polyvalence as to both vowel and consonant quality is an expected
characteristic of this writing system. Consonants and vowels that may have more than one
phonemic value are represented in uppercase italic; PU, e.g., which, as we have already seen,
may well correspond both to bhū and pu. For each sign given, all attested values are also
listed. In general, many Indus signs appear to conflate stop consonants according to place of
articulation; thus P can potentially represent [p], [b], [ph], or [bh]. Vowel length also appears
to be irrelevant, although not all signs necessarily connote both a short-vowel and long-vowel
form, as we have found / to do. Moreover, [i]/[ī]/[e] and [u]/[ū]/[o] are ordinarily
conflated, certain signs representing Ci and Cu being found also representing Ce and Co,
they only seem to have one value; one such sign, which we have already seen, is , which
Many geminate signs seem to represent the feminine (usually -ī) form of a masculine root
sign, gemination in non-numerical contexts often (though perhaps not always) signifying the
feminine gender. This is most likely the case with signs/geminates / , / , and /
In this lengthy section, we will furnish the values we have obtained by the method previously
described for 56 Indus signs that we regard as fully deciphered, i.e., for which either a
phonetic value or values or a logographic reading have satisfactorily been established. Table
logographic values have been assigned, but which will not be given individual treatment
further on; they will be illustrated in context under the discussion, in section 7, of the full
meanings of patterned inscriptions). Moreover, certain signs for which the general sense is
understood and has been discussed previously (fish compounds like , etc., and oval
compounds like , etc., which are found in M clusters), but to which no specific values have
i. CV:
= jA [ja, jā]
= sA [sa, sā]
= rA [ra, rā]
= vI [vi, vī]
= hu
ii. VCV:
iii. CVC(V):
/ = var(a)
iv. CCV:
= pra
v. Logograms:
/ = RASHTRA
= LAKSH(A)
= SENA
=ARYA
= DHAR(A)
= SAM
= RAKSH(a)
= RAJA(n)
= DHANYA
= DEVA
= PATI
= ASHVA
/ = GO [go]
= ISH, ‘be accepted as, be worth’ (and perhaps also ĪŚ, ‘own, possess; belong to’)
= ŚRIPATI/SURIPATI
= ŚRIPALA
= ARYAKA
Following is a detailed inscription list for each sign, with full or partial readings. Since many
Indus signs remain undeciphered, most readings are partial, but the majority of readings for C
fields and P fields are complete as far as the fields themselves are concerned. The primary
source for inscriptions is the Mahadevan concordance (MH). Wherever possible, we have
checked MH entries against entries in the Interactive Concordance of Indus Texts (ICIT), and
have tried to give index numbers for both corpora, as well as the catalog number for the
artifact and the artifact type, wherever we were able to find them. Some inscriptions given in
MH are not in ICIT, and subsequent to our work with MH, we found a number of important
inscriptions in ICIT (such as all inscriptions from Dholavira) that are not in MH; in every
case, new inscriptions in ICIT confirmed values already ascertained from MH. Moreover, in
a few critical cases, ICIT, owing to the availability of visual images of almost every
several cases, revealed important new information that not only confirmed values ascertained
from MH, but furnished important new readings as well. Finally, the ICIT lists vastly more
signs than MH, some of which were unknown to MH, and many more of which were treated
by MH as allographs. In some cases, we have listed allographs given in the ICIT (and also in
MH), and tried to show them separately, but in other cases, we have conflated them without
comment, their allographic status seeming to us not to warrant in-depth discussion. For
example, the ICIT treats and as two distinct signs, despite the fact that their respective
distributions are identical, the basis for this claim being several inscriptions in which the two
signs appear to co-occur; but we are unable to see any basis for assuming that these are other
than scribal variations, or perhaps the result of millennia of uneven wear. Again, the ICIT
lists a large number of allographs for , which vary according to the number and placement
of the internal hatch marks; all of these were conflated by MH, and after very careful
evaluation of every one of these signs given in ICIT, we have concluded that MH was
correct. On the other hand, one sign with several “allographs” given in MH as turned out to
be at least two, and possibly three separate signs, based on data in ICIT; at very minimum,
the sign , regarded by MH as allographic, is almost certainly a separate sign from , based
not only on its consistently distinct graphology but also on its completely different patterns of
occurrence. Accordingly, we removed and its assigned value from our list of deciphered
signs, and have concluded that the value once assigned to this sign is probably erroneous.
Finally, although the creators of the ICIT have frequently justified, via personal
communication, their well-reasoned intention to list every conceivable allograph, lest critical
information be lost in some arbitrary conflation (as has demonstrably happened in several
instances with MH), their standard is somewhat unevenly applied; on the one hand, for
example, they are painstaking in showing every allograph of , as noted previously, but
they conflate all allographs of , which exhibit varying numbers of internal hatches, clearly
the ICIT; rather, it is to underscore the admittedly subjective nature, in dealing with texts of
such brevity, of differentiating between separate signs and mere allographs. In the end,
investigators must make the most judicious choices they can, bearing in mind the statement
made years ago by the late Gregory Possehl in the presence of this investigator, that proper
In the following pages of data, the inscriptions are treated according to the following format:
By ‘primary index #’ is meant the first source from which the inscription as transcribed was
obtained. In most cases, this simply means the first source where the inscription was
encountered and studied, but in a few cases, it means that the primary source is more
Portions of the inscription that are transcribed/read are underscored, unless the entire
inscription is readable. The readings of individual signs in the transcription are separated by
dashes; hence, a three-character sign sequence will be transcribed with three discrete sound
Certain sources are frequently referred to in the descriptive material related to inscriptions,
such as the Monier-Williams Sanskrit Dictionary (MW), the Rigveda (RV), and the
Atharvaveda (AV). Also referenced occasionally is the useful online source Wisdom
Library/wisdomlib.org (WL). Moreover, the catalog numbers, from Joshi and Parpola 1987,
are indexed according to site, with a letter abbreviation followed by a dash and associated
M (Mohenjo-Daro)
H (Harappa)
L (Lothal)
K (Kalibangan)
Sktd (Surkotada)
B (Banawali)
Blk (Balakot)
Ad (Allahdino)
Ns (Nausharo)
Krs (Khirsara)
C (Chanhujo-Daro)
Inscriptions from Dholavira and certain other recently-excavated sites are recorded nowhere
else and hence have no standardized catalog number nor any other index number.
: [DEVA]
a. Comments:
This sign may be regarded as a logogram, since it never appears to have any reading other
than deva. The rather common geminate form is more likely to be read devi, the
feminine form of deva, although the geminate devadeva, ‘lord of gods; epithet of Brahma,
b. Inscriptions:
ICIT 1594 [ATi-DEVA/atideva, ‘supreme god;’ note that the ICIT has these two
is well above, but only very slightly to the left of, , which in turn appears
to be turned on its side; in other words, this is not a line of writing, and the rest
of the surface is completely unworked] [H-688; prismatic seal w/o field figure;
surpassing all gods’; devādhideva- also attested, but in a Jain context; sudeva
means not only ‘good or real god/lord,’ but is also a well-attested name, MW;
likely reading ‘god of gods, the real god’] [M-634/ICIT 3084; seal (broken)
attested, but very plausible name, cf. devaka, ‘deity; divine,’ and well-attested
note that devavarman is attested as a name; lit. meaning ‘armor of the gods,
image and a hole through the middle; ICIT represents this inscription as
, even though there is no second visible on the object.
bovine animal possibly roped to a sacrificial stake, a human figure, and some
kind of structure.]
unicorn bull.]
bull]
6/ICIT 1881; seal w/ unicorn bull; note that in this inscription, is facing
opposite of the usual direction, but since this is the only unambiguous case of
this, and because the source of this seal is Kalibangan, I assume, contra the
ICIT, that this is merely a local graphological variant of the very common
sign .]
[second line]
the gods’)] [M-1309/ICIT 3614; small rectangular seal with no field figure]
figure]
MH1623
a. Comments:
This sign appears most often to represent the frequent prefix su-, ‘good,’ but may also
represent, among other variants, the syllable so- in the important name Soma. Inasmuch as
the script does not ordinarily distinguish sibilants from shibilants, may also stand for śu,
śo, etc.
b. Inscriptions:
also an attested name (WL); dhana-śara could also mean ‘bow and
śoka, ‘sorrow; attested mythical name (“1. a son of Droṇa and a Vasu; 2. A
54/ICIT 5506; square seal with no field figure; very doubtful representations
and readings, since the seal is extremely worn and the characters very crudely-
rendered.]
___
unicorn bull]
MH4222 [SU-PAla-/supāla- (unattested, but supālaka is an attested
supati, ‘good lord, good husband’] [H-458/ICIT 1448; seal w/ unicorn bull]
etc.’] [possibly M-1233/ICIT 3565; seal in very poor condition, with no field
___
___
MH7080 [ka-SU-/kaśu-, RV man’s name] [L-4/ICIT 1986; seal w/
unicorn bull]
bull]
: [Su/su, ṣu]
a. Comments:
The value for this sign must be reckoned as provisional, since it is very unclear whether tṛṣu-
b. Inscriptions:
Index)]
Index)]
surpassing all gods’; devādhideva- also attested, but in a Jain context; sudeva
means not only ‘good or real god/lord,’ but is also a well-attested name, MW;
likely reading ‘god of gods, the real god’] [M-634/ICIT 3084; seal (broken)
a. Comments:
This sign most often represents bhū, whose original meaning is ‘be(ing), become(ing), being
produced, etc.,’ but also has come to mean ‘earth’ (i.e., ‘that which exists’), in which sense it
appears in many titular words like bhūdhana, ‘king, prince’ and bhūpāla, ‘king, prince.’ As
discussed elsewhere, its status in the sequence /PU-tra is somewhat uncertain, since the
origin of the word putra-, ‘son; child; offspring’ is not clear. If it arises from an earlier
*bhūtra- (lit., ‘earth born’ or perhaps ‘means of being/becoming’), then the IV form could be
a faithful rendering of an earlier form, and bhū would then become the only value for this
sign. We have opted for an assumption of broader phonetic applicability based on the absence
of any direct evidence for a form *bhūtra and on the fact that other signs do seem to display
b. Inscriptions:
MH4260 [PU-DAna-/bhūdhana-, ‘king, prince’] [H-611/ICIT 1551;
very unclear; ICIT renders it as - -, but this seems wrong, since no lower
but with the three lines much closer together, the same width as , and
having every appearance of being a single sign composed of two long vertical
lines flanking a single short one. However, this sign is not inventoried in the
differentiate these two signs, and given the clarity of the writing in these two
examples and their shared provenience, it seems likely that they are in fact two
different signs.]
preceding.]
unicorn bull.]
bull]
rhinoceros; MH has a second line , but this is in fact a ritual object similar
to the feeding troughs and chattra-like figures found in front of other animal
field figures.]
ruler’) Note that here we are assuming that the geminate has the effect
Bhupala;’ also 5404, etc.] [M-1452/ICIT 3730; copper tablet with horse-like
is the earth’)]
[line 2]
and MH has , while the ICIT has ] [M-25/ICIT 2553; seal w/ unicorn
bull]
no field figure]
ruler, etc.’; note that no image of this object appears to be available, so it is not
clear whether in fact = , these two signs having very similar and
unicorn bull]
that in the attested form of this name, vibhū- > shortened vibhu-] [M-
the leader, the queen’] [K-10/ICIT 1885; seal w/ unicorn bull; note that
ICIT has for the first/rightmost , but this is clearly not warranted from
(saha=’great, mighty’ or ‘with, having’); note that the sign is almost totally
figure]
ox]
given name and divine epithet.] [M-220/ICIT 2744; seal w/ unicorn bull]
MH3105 [ATi-RAJA-SU-mA-PU/atirāja somabhū, ‘high king Somabhu’
bull]
where X-bāhu- denotes any of a wide range of attested (and perh. unattested)
sthūla-bāhu-, etc.), with the common affix –ma(t), ‘having, possessing’ (as
a. Comments:
This sign does not occur in contexts typical of word/root signs, but always occurs in C fields
and P fields in conjunction with at least one, and frequently more than one, other non-
juncture sign. It is one of the more common and randomly-distributed signs in the Harappan
signary, suggesting that it is likely to coincide with a CV that does not commonly have full
word value. The value PA was arrived at after considerable trial and error.
b. Inscriptions:
MH1013 [also many others] [PA-hu-jA AS/bāhuja asti, bāhuja-, ‘Kshatriya
857/ICIT 3269; seal w/ unicorn bull; other examples of this inscription both
etc.); with this reading, is most likely ĪŚ, ‘belong to’; OR bhājya-, ‘to be
(saha = ‘great, mighty’ or ‘with, having’); note that the sign is almost
field figure]
two characters are partly effaced and inscribed in an unusual style that casts
some doubt on whether MH and the ICIT have transcribed them correctly]
ox]
1035/ICIT 3920; seal w/ unicorn bull; partly effaced, and is not included
in ICIT transcription]
very plausible; Indian given name)] [M-141/ICIT 2668; seal w/ unicorn bull;
line 2 of 2]
MH4335 [-ARYA-RAJA-PA-hu-jA-PU-PAla/-āryarāja-
field figure; many tablets with this inscription from Harappa, many of them
partially effaced]
name (new spelling)] [M-487/ICIT 2970; very worn three-sided seal with
animal and human figures on 2 sides; the ICIT does not have initial , which
may be correct, given the very worn and barely legible condition of this
object]
to Pkt. Mava; Kannada has it as a deity name; this sequence occurs several
grass,’ and also a given name (son of Angiras)] [M-1676/ICIT 2224; seal w/
unicorn bull]
notation on reverse]
ICIT 601 [see preceding] [H-2021; rectangular tablet with tree image on
reverse]
MH7062 ___ [ibid.] [L-92/ICIT 2046; small rectangular seal w/o field
image]
where X-bāhu- denotes any of a wide range of attested (and perh. unattested)
sthūla-bāhu-, etc.), with the common affix –ma(t), ‘having, possessing’ (as
potsherd; note that the is separate (above and to the right) of the other two
ICIT 643 [vI-PA-/vipa-, ‘learned man’ OR (much more likely) vibhā-, ‘light,
luster; king, prince;’ attested name] [H-2198; circular tablet w/o field figure;
: [hu/hu]
a. Comments:
type. Determining the value of this sign, in conjunction with , was one of the lengthiest
exercises in cross-checking and trial and error of this stage of the investigation. The assigned
b. Inscriptions:
857/ICIT 3269; seal w/ unicorn bull; other examples of this inscription both
where X-bāhu- denotes any of a wide range of attested (and perh. unattested)
sthūla-bāhu-, etc.), with the common affix –ma(t), ‘having, possessing’ (as
name (new spelling)] [M-487/ICIT 2970; very worn three-sided seal with
animal and human figures on 2 sides; the ICIT does not have initial , which
may be correct, given the very worn and barely legible condition of this
object]
MH4335 [-ARYA-RAJA-PA-hu-jA-PU-PAla/-āryarāja-
partially effaced]
1035/ICIT 3920; seal w/ unicorn bull; partly effaced, and is not included
in ICIT transcription]
ox]
: [jA/ja, jā]
a. Comments:
One of the few CV signs that also has common full word value in the context of the seal
inscriptions (ja = ‘race; tribe; clan, etc.’). Frequently found in word-final position
representing the common Skt. element –ja, ‘born,’ which is often the final element in
b. Inscriptions:
such; 5 total exx.] [M-501/ICIT 501; small rectangular seal w/o field figure]
tablets]
‘husband’; attested Puranic name] [H-321/ICIT 1343; tablet w/o field figure;
bull]
a. Comments:
We originally obtained the value for this sign thanks to the very common pair , mā-tra-,
which has been discussed in considerable detail. The additional value maha/mahā for initial
position is more conjectural, but arises from the following considerations: 1) maha/mahā is
an extremely common name/title element in Sanskrit, and therefore among the top expected
words in our corpus, but we have failed to find it anywhere in the inscriptions (as, e.g., +
another sign identifiable as ha, or as a distinct sign that equates to that value); 2)
everywhere fits the data, yielding exceptionally clear values — except in certain initial
position occurrences, where we are met with semi-hypothetical values like *mapala and
*mapa. as maha/mahā in such contexts resolves this issue very neatly; 3) the phonetic
difference between ma/mā and maha/mahā, particularly in speech, would not be particularly
great.
b. Inscriptions:
related to Pkt. Mava; Kannada has it as a deity name; this sequence occurs
several times, but its meaning is uncertain)] [M-286/ICIT 2807; seal w/
elephant]
figure]
bull]
MH does not give final , but the identity of this partly-effaced sign appears
MH1018 [mA/mahā]
(marya=mortal, man)]
w/ unicorn bull]
w/ ox]
a. Comments:
Usually represents prefix vi-, but also found in word-final position. Occasionally may signify
avi-, ‘kind, favorable;’ see MH 4584 below. The graphology of this sign is possibly
motivated by phonetic similarity with dvi-, ‘two.’ Whether this sign should be considered
b. Inscriptions:
potsherd]
unicorn bull]
ICIT 643 [vI-PA-/vipa-, ‘learned man’ OR (much more likely) vibhā-, ‘light,
luster; king, prince;’ attested name] [H-2198; circular tablet w/o field figure;
śarmarāja; = ?; note that there are at least three possible readings for vi-?-
MH1087 [vI-para-DAna-PAla-/vipradhanapāla-;
vipra-, ‘inspired, wise, learned; sage, seer; priest, domestic priest’ (very
literally ‘bow and arrow,’ warrior’s name?; also dhanasāra- an attested name;
ruler, etc.’; note that no image of this object appears to be available, so it is not
clear whether in fact = , these two signs having very similar and
unicorn bull]
w/ unicorn bull]
(MW)]
possessor; etc.’; nāthapāla = attested king name and name of a royal family
‘lover, husband; given name’] [C-29/ICIT 112; square seal w/o field figure]
śripati (see preceding exxs.) (MW) ] [H-2246/ICIT 484; tablet w/o field
figure; there are a total of five such tablets in ICIT, although the other four
‘inspired, wise, learned; sage, seer; priest, domestic priest’ (very common
attested name and name-element; devavī, ‘gratifying the gods (RV)’] [H-799;
that in the attested form of this name, vibhū- > shortened vibhu-] [M-
the gods’)] [M-1309/ICIT 3614; small rectangular seal with no field figure]
figure]
iconography; other exx. of this inscription appear to show a figure with a tiger,
etc.]
: [kA/ka, kā]
a. Comments:
seems to coincide mostly with the very common and productive suffix –ka, which can
have a diminutive sense and also an attributive sense; in the contexts of the seals, it also
seems to have a possessive sense, as in satyadevaka, ‘Satyadeva’s.’ This sign can also signify
the syllable –kha-, as in sukha-, ‘pleasant, happy, agreeable, etc.,’ found as an element in
b. Inscriptions:
figure]
MH1018 [mA/mahā]
‘following sound’ and pāraka, ‘saving, delivering; etc.’ are also admissible
forms, if less likely); also found in a number of compounds, usu. as the final
entry]
etc.’] [possibly M-1233/ICIT 3565; seal in very poor condition, with no field
___
___
attested, but very plausible name, cf. devaka, ‘deity; divine,’ and well-attested
note that devavarman is attested as a name; lit. meaning ‘armor of the gods,
: [PATI]
a. Comments:
This sign represents the very important value pati, ‘lord, sovereign; husband’ (a word that
persists even in modern Indo-Aryan languages). It is a very frequent, usually final, element,
in names and titles. By itself, it can refer to a deity, an earthly authority, or a husband.
b. Inscriptions:
unclear] [H-922/MH4474; tablet w/o field figure; there are many other tablets
from Harappa with on one side; MH erroneously represents this and
in the corpus]
man; king;’ attested name] [M-1189/ICIT 3540; broken square seal w/o field
figure; line 2 of 2]
ruler, king’] [L-60/ICIT 2027; very damaged square seal w/o field figure;
possibly incomplete]
MH1305 [DAna-PATI-/dhanapati-; dhanapati = ‘lord of wealth; treasurer’
very plausible; Indian given name)] [M-141/ICIT 2668; seal w/ unicorn bull;
line 2 of 2]
obscured and not identifiable with certainty, and the ICIT very plausibly
location]
[4-caru (4 oblations)]
6/ICIT 1881; seal w/ unicorn bull; note that in this inscription, is facing
opposite of the usual direction, but since this is the only unambiguous case of
this, and because the source of this seal is Kalibangan, I assume, contra Wells
and Fuls, that this is merely a local graphological variant of the very common
sign .]
Agni, Krishna, etc.; vasupatnī, ‘mistress of wealth,’ attested as a name for the
cow in RV (MW)] [M-1357/ICIT 3656; small square seal w/o field figure]
supati, ‘good lord, good husband’] [H-458/ICIT 1448; seal w/ unicorn bull]
the leader, the queen’] [K-10/ICIT 1885; seal w/ unicorn bull; note that
ICIT has for the first/rightmost , but this is clearly not warranted from
(usually shown with unicorn bull) but no field animal figure; first line of text
is ]
: [BALAPATI]
a. Comments:
This sign occurs only once, but it is quite transparently a ligature of and , i.e., an
alternate for the sequence - - documented elsewhere in the P field (cf. MH7107). We are
MH5091
: [SENA]
a. Comments:
The value of this uncommon sign was arrived at rather straightforwardly by cross-checking
against signs with which it co-occurs, and arriving at the only possible syllabic value. Its
meaning, ‘army, host,’ is less important, perhaps, than the fact that this word occurs often in
b. Inscriptions:
no field figure]
people (MW)] [M-396/ICIT 2892; small rectangular seal w/o field figure]
MH2125 [RAJA-SENA-kA/rāja senaka or *rājasenaka (senaka and
: [RAKSHA]
a. Comments:
is an infrequent sign that happens to occur in environments where its value can be
b. Inscriptions:
figure]
ICIT 513 [vI-RAKSHA-S(a)va-/virakṣa-aśva-, ‘horse guard’?
MH1018 [mA/mahā]
a. Comments:
This is one of two signs that can represent the common sound pra, but unlike the other ( ),
can also represent the common disyllable para (and possibly pāra) which – unlike pra –
can have full word value. It has a number of allographs varying in the number and
positioning of the internal hatch marks; a careful comparison of these signs shows broad
overlap in patterns of occurrence among one another, suggesting that Mahadevan’s conflation
of this sign series is correct. Note that ICIT does conflate the various allographs of that
b. Inscriptions:
1035/ICIT 3920; seal w/ unicorn bull; partly effaced, and is not included
in ICIT transcription]
broken square seal, apparently w/o field figure; 3rd line of text unreadable]
MH2806 [+5 other exx.] [para/para, pāra, well-attested names (MW)] [M-
w/ unicorn bull]
MH3510 [para-RASHTRA/pararāṣṭra, ‘from a foreign country;
very plausible; Indian given name)] [M-141/ICIT 2668; seal w/ unicorn bull;
line 2 of 2]
‘husband’; attested Puranic name] [H-321/ICIT 1343; tablet w/o field figure;
people (MW)] [M-396/ICIT 2892; small rectangular seal w/o field figure]
MH1087 [vI-para-DAna-PAla-/vipradhanapāla-;
vipra-, ‘inspired, wise, learned; sage, seer; priest, domestic priest’ (very
literally ‘bow and arrow,’ warrior’s name?; also dhanasāra- an attested name;
‘lover, husband; given name’] [C-29/ICIT 112; square seal w/o field figure]
‘inspired, wise, learned; sage, seer; priest, domestic priest’ (very common
: [j(a)ya/jaya, jya]
a. Comments:
This is a highly repetitive sign found often as a self-contained word/name root (as ) and
as . The value jaya turned out to be the only one consistent with available data.
b. Inscriptions:
grass,’ and also a given name (son of Angiras)] [M-1676/ICIT 2224; seal w/
unicorn bull]
3228; seal w/ unicorn bull; also a number of other inscriptions with /jaya]
possessor; etc.’; nāthapāla = attested king name and name of a royal family
a. Comments:
This sign coincides with the common prefix sam-, and appears to have no other variant
forms.
b. Inscriptions:
bull]
unicorn bull]
a. Comments:
This sign appears to consist of a bow grapheme, which represents the phonetic value(s), and
b. Inscriptions:
MH4624 ___ [DAna-PAla-/dhanapāla-, ‘guardian of treasury, treasurer’; well-
unicorn bull]
treasurer;’ note that the ICIT has instead of , which is possibly a correct
distinction, although the former sign only occurs twice] [M-735/ICIT 3159;
unicorn bull]
figure]
MH4487
[H-894/ICIT 1776; tablet w/o field figure; note use of /dhana- in
ICIT 658
MH 5283
also an attested name (WL); dhana-śara could also mean ‘bow and
literally ‘bow and arrow,’ warrior’s name?; also dhanasāra- an attested name;
vipra- ‘inspired, wise, learned; sage, seer; priest, domestic priest’ (very
‘inspired, wise, learned; sage, seer; priest, domestic priest’ (very common
MH5473 [ICIT 1320/H-298; tablet w/o field figure; ICIT has facing
in the opposite direction, and regard as a separate character, although there are
only two possible such occurrences; it is not clear what the orientation of
‘protectress,’ the feminine of pāla, and found in some attested female epithets
Index)]
name (MW)] [Ns-60 (Nausharo); small rectangular seal w/o field figure]
MH9091 [DEVA-DAna-PAla-; -4-PAla/deva
: [DAna2/dhana/dana]
MH5490
name]
‘beloved offering;’ note that all attested names (there are several dozen) are in
cāru-, not caru-] [H-93/ICIT 1149; seal w/ elephant; another seal, C-80 from
name (MW)] [K-59/ICIT 1920; small rectangular seal w/o field figure]
attested name’] [M-1189/ICIT 3540; broken square seal w/o field figure; line
2 of 2]
illegible]
: DHANYA
a. Comments:
/dhani or dhāni+ the anthropomorph , which appears to represent –a- in some contexts
= dhanya-, ‘wealth’ or dhānya-, ‘grain.’ The physical contexts in which this sign occurs –
b. Inscriptions:
Index)]
[DHANYA-PATI/dhānyapati, an attested profession/type of officer,
‘wealth’ (alternate form for dhana; both dhānyam and dhanyam are neuter);
: LAKSH[A]
a. Comments:
Another sign of relatively few occurrences, but whose contexts are extremely consistent with
one root-syllable. The meaning of this term is ‘mark, sign’ and also a “lakh,” the Indian
b. Inscriptions:
bull]
and MH has , while the ICIT has ] [M-25/ICIT 2553; seal w/ unicorn
bull]
bull]
a. Comments:
This sign appears to denote the common “implementive” suffix –tra when appearing in word-
final position, but likely represents tri-, ‘three,’ when appearing word initially. When
agentive suffix –tr/-tri is required, then we suggest that the geminate may be used, on
b. Inscriptions:
ruler’] [Dholavira; small broken rectangular seal w/o field figure (reverse not
shown)]
bull]
(saha=’great, mighty’ or ‘with, having’); note that the sign is almost totally
effaced, and might be another sign] [H-2099/ICIT 786; tablet w/ no field
figure]
sovereign, ruler’) Note that here we are assuming that the geminate has
the effect of transforming /tra into tri, on analogy with /DEVA and
Index)]
a. Comments:
This sign most often represents the common prefix ati-, ‘very, exceedingly,’ although it also
sometimes seems to stand for prefix adhi-, ‘highest, above,’ as well. Alternate reading āt is
ICIT 1594 [ATi-DEVA/atideva, ‘supreme god;’ note that ICIT have these two
is well above, but only very slightly to the left of, , which in turn appears
to be turned on its side; in other words, this is not a line of writing, and the rest
of the surface is completely unworked] [H-688; prismatic seal w/o field figure;
MH1348 [see preceding; in this variant, only the –min of somin is enclosed by
the quadripartite circumgraph, but it seems clear that the meaning is the same
(only back and part of the nape of the animal are preserved)]
bull]
MH7032 [see prev.] [L-90/ICIT 2044; small rectangular seal w/o field
figure]
ICIT 591 [mA-ATi/mati, ‘devotion, prayer, hymn, sacred utterance; wish; opinion;
ICIT 2211 [see preceding] [Mohenjo-Daro (no catalog number); very small broken
seal with no field figure and practically no field at all; see also M-1067/ICIT
ruler, king’] [L-60/ICIT 2027; very damaged square seal w/o field figure;
possibly incomplete]
‘supreme king’(or adhirāja-, ‘emperor,’ but this seems less likely); -śarman,
bull]
unicorn bull]
figure]
MH does not give final , but the identity of this partly-effaced sign appears
name; type of divine being); also a final element in many royal titles
see Indian Epigraphical Glossary, pp. 8-9)] [see also 3 other inscriptions] [H-
MH1018 [mA/mahā]
surpassing all gods’; devādhideva- also attested, but in a Jain context; sudeva
means not only ‘good or real god/lord,’ but is also a well-attested name, MW;
likely reading ‘god of gods, the real god’] [M-634/ICIT 3084; seal (broken)
no field figure]
: [RAJA(N)]
a. Comments:
The fact that this sign so often occurs in P-clusters as - (and occasionally as ) tells us
that , like , , and , among others, must denote a word/root with the force of some
kind of title or other important honorific. After a very long process of trial and error, using
the usual method of cross-checking potential morphemes across multiple contexts, RAJA(N)
emerged as the only plausible candidate. Note that while rājan is an -n-stem noun, the –n
appearing in certain of the oblique cases, it is usually rāja- in compounds and other combined
forms. Note also that , unlike , , and , never occurs right-adjacent to juncture sign
, showing instead an overwhelming affinity for the juncture sign that may be a marker of
plurality. This may be because of the age-old and cross-cultural penchant for referring to
b. Inscriptions:
unicorn bull]
‘supreme king’(or adhirāja-, ‘emperor,’ but this seems less likely); -śarman,
bull]
unicorn bull]
figure]
śarmarāja; = ?; note that there are at least three possible readings for vi-?-
and name of royal family; also ‘governor of the state’)] [M-1838/ICIT 2355;
MH4335 [-ARYA-RAJA-PA-hu-jA-PU-PAla/-āryarāja-
field figure; many tablets with this inscription from Harappa, many of them
partially effaced]
(usually shown with unicorn bull) but no field animal figure; first line of text
is ]
a. Comments:
This sign, which has two graphological variants, mostly occurs in the common pair /
names.
b. Inscriptions:
protection’ (not attested, but presumably a name)] [H-1832; tablet w/o field
w/ ox]
Index]
MH2415 [vI-PU-vAra-mAna/vibhuvarman (attested man’s name); note
that in the attested form of this name, vibhū- > shortened vibhu-] [M-
attested, but very plausible name, cf. devaka, ‘deity; divine,’ and well-attested
note that devavarman is attested as a name; lit. meaning ‘armor of the gods,
figure]
talisman with jewels;’ attested name (of a merchant, MW); full reading is
a. Comments:
This sign often represents the important Sanskrit agentive suffix –man, especially in the
unicorn bull]
‘supreme king’(or adhirāja-, ‘emperor,’ but this seems less likely); -śarman,
w/ ox]
Index)]
that in the attested form of this name, vibhū- > shortened vibhu-] [M-
attested, but very plausible name, cf. devaka, ‘deity; divine,’ and well-attested
name (MW)] [Ns-60 (Nausharo); small rectangular seal w/o field figure]
figure]
talisman with jewels;’ attested name (of a merchant, MW); full reading is
illegible]
a. Comments:
This sign has a little bit of overlap with , but in general, connotes the affix -māna/-
mana (as well as –mani as a geminate ), and is often frequently in initial position,
whereas most often denotes the affix –man. There is, however, some alternation (or
scribal error), such as H-1148 shown below, in which is used in a context where is
normally preferred.
b. Inscriptions:
[reverse]
of somewhat plausible readings here, but this one is uncertain)] [ICIT 99/C-
and MH has , while the ICIT has ] [M-25/ICIT 2553; seal w/ unicorn
bull]
bull]
unicorn bull]
fixed standard of measure;’ in this last sense, this term is found in inscriptions
a. Comments:
This sign has many occurrences but comparatively little variation of contexts; most
occurrences are as , and the only value that we were able to find consistent with this and
the other occurrences is ARYA, ‘noble; Aryan.’ Still, we regard this identification as quite
b. Inscriptions:
such] [5 total] [M-501/ICIT 501; small rectangular seal w/o field figure]
and name of royal family; also ‘governor of the state’)] [M-1838/ICIT 2355;
MH4335 [-ARYA-RAJA-PA-hu-jA-PU-PAla/-āryarāja-
field figure; many tablets with this inscription from Harappa, many of them
partially effaced]
ICIT 3557 [-ARYA-t(a)ya-/-ārya-taya, ‘noble protector’
is also a name and title of various men and deities] [M-1276/ICIT 3583; small
bhāryā-, ‘wife’] [M-416/ICIT 2912; circular seal w/ ox; these two characters
are partly effaced and inscribed in an unusual style that casts some doubt on
w/ unicorn bull]
a. Comments:
This sign is found most often in the combination / śarman-, ‘protection; blessing,’ a
b. Inscriptions:
treasurer;’ note that the ICIT has instead of , which is possibly a correct
distinction, although the former sign only occurs twice] [M-735/ICIT 3159;
unicorn bull]
‘supreme king’(or adhirāja-, ‘emperor,’ but this seems less likely); -śarman,
also an attested name (WL); dhana-śara could also mean ‘bow and
literally ‘bow and arrow,’ warrior’s name?; also dhanasāra- an attested name;
śarmarāja; = ?; note that there are at least three possible readings for vi-?-
a. Comments:
b. Inscriptions:
woman’s name] [M-953/ICIT 3348; seal w/ unicorn bull; see also 2 other
notation on reverse]
ICIT 601 [see preceding] [H-2021; rectangular tablet with tree image on
reverse]
grass,’ and also a given name (son of Angiras)] [M-1676/ICIT 2224; seal w/
unicorn bull]
MH7062 ___ [ibid.] [L-92/ICIT 2046; small rectangular seal w/o field
image]
: [dhu/dhu]
a. Comments:
This infrequent sign has provisional value dhu, a value that fits observed data nicely, but
which may be regarded as very provisional owing to the paucity of available data.
b. Inscriptions:
śripati (see preceding exxs.) (MW) ] [H-2246/ICIT 484; tablet w/o field
figure; there are a total of three such tablets in ICIT, although the other two
a. Comments:
Provisional value, motivated above all by MH4345 in combination with certain singular facts
about the artifact type. Of the 22 occurrences given in ICIT, only 1 (MH4355/ICIT 1237) is
found on an artifact without a field figure (small rectangular seal). Of the 21 remaining
objects, ten feature field figures and scenes other than the standard unicorn bull motif. These
include not only rhinoceroses (2) and elephants (3), but also elaborate and unique scenes that
seem to reflect important mythic themes. For example, ICIT 3538 (M-1186) depicts the
famous “seven rishis” scene, with a line of seven pigtailed figures below a votive scene
involving a kneeling worshipper, a hybrid man-beast, and some kind of deity in a stylized
peepul. ICIT 2829 (M-310) illustrates a figure seated in a tree above a tiger. ICIT 2825 (M-
306) shows a heroic figure holding two tigers at bay, one with each arm. ICIT 2816 (M-296)
depicts two stylized unicorn bull heads sprouting from a peepul. ICIT 2815 (M-295) shows
an amalgam of three tigers. Together these constitute a very significant portion of the total of
unusual, non-standard seals with detailed iconography other than the canonical unicorn bull,
and all in what appears to be statistically significant association with the relatively
uncommon sign . These circumstances suggest that many of these seals might pertain
either to officials especially important to the state or have a function particularly significant
thereto.
standalone status not closely allied morphosyntactically with the rest of the inscription. We
have proposed the value in such contexts to be kāra, ‘royal revenue, toll, tax, tribute, duty’
(MW). In this connection, it is perhaps not coincidental that the elephant — normally a very
unusual field figure, accounts for three of the 21 total field figures associated with this sign
— and kara happens to be the standard word for the elephant’s trunk (as well as the human
hand). Hence it is at least possible that this animal appears on objects associated with the
collection of revenues.
Finally, kāra also has several other meanings, which could apply in cases like MH2014
below, including ‘husband, lord, master’ and ‘hymn of praise, act of worship.’
Taking into account all of these factors, I have tentatively assigned the value
kara/kāra/kar, this being a set that seems to accord best with the limited data available for
this sign.
b. Inscriptions:
and to the right. Additionally, the field figure (rhinoceros) is facing opposite
the canonical direction (head towards the end of the inscription rather than the
in fact occur alone below longer lines of texts on several other objects] [M-
: [S(u)ri/śrī, suri]
a. Comments:
Another very common and important sign found in P-fields, most often singly in combination
checking its rather limited number of occurrences in conjunction with known signs like
and (see below), in conjunction with the observations b) that this sign, like and ,
e.g., should represent some kind of common name/title that can stand alone as a single word
+ ). The only values that matched these criteria — and matched them extremely well —
were the possibly related terms śrī and suri, whereof, in this researcher’s opinions, the former
b. Inscriptions:
MH2064 [S(u)ri-/śrī, suri- ] [M-1165/ICIT 3525; seal w/ tiger]
unicorn bull]
field figure]
this small object, and is quite worn, but recognizable nonetheless] [H-878;
attested name’] [M-1189/ICIT 3540; broken square seal w/o field figure; line
2 of 2]
MH3221 ___ * [ma(hA)-S(u)ri-/mahāśrī- (attested as an epithet of Lakshmi)]
: [S(u)ri-PAla/śripāla]
a. Comments:
This sign is a straightforward ligature of and , and although it only occurs once, the sole
substituted for . There are a number of other ligatures involving , perhaps because of
how frequently this sign appears as a conventionalized element in names and titles.
b. Inscriptions:
2340 [note the similarities between this inscription and 1048 above (
/ : [RASHTRA/rāṣṭra]
a. Comments:
This sign is a quasi-ligature, with the two graphemes involved in its composition -- /rA
and /tra accounting for the initial and final sounds of the word represented. The
form the important and expected title rāṣtrapāla (‘kingdom-protector, sovereign’; attested
b. Inscriptions:
protector,’ attested name ] [ICIT 2481; tablet w/ unclear animal figure; plus 5
more occ., all scratched on the reverse side of copper tablets with ox OR
MH5096 [-RASHTRA-vI-PAla/-rāṣṭra-avipāla??;
avipāla = ‘shepherd’]
: [NATHA/nātha]
a. Comments:
Provisional value, since this sign, while quite common, has very few completely distinct
environments. While -pāla has a large range of potential preceding words and prefixes,
almost none are likely candidates for a single word occurring with with significant
frequency (MH3064 and 3 other occurrences). Besides nātha, ‘protector, lord,’ kumāra,
‘prince,’ is also possible, but we regard it as unlikely given its complex sound shape (3
syllables; we have found no other simple sign in the Indus signary that corresponds to a
three-syllable word) in conjunction with the graphological simplicity of this sign that gives
no hint of compound structure. We regard nātha as by far the likeliest potential value.
b. Inscriptions:
bull]
MH3064 [NATHA-/nātha-] [2 other inscriptions][H-69/ICIT 1128; seal w/
unicorn bull]
possessor; etc.’; nāthapāla = attested king name and name of a royal family
a. Comments:
This sign seems to have two chief values. It is often found as svā in conjunction with /-min
as -svāmin, ‘lord’ (very common ending for names, honorifics, and titles), and it also
sometimes occurs as the word/root sava, ‘commander.’ It may also represent sva-, ‘one’s
b. Inscriptions:
very unclear; ICIT renders it as - -, but this seems wrong, since no lower
but with the three lines much closer together, the same width as , and
having every appearance of being a single sign composed of two long vertical
lines flanking a single short one. However, this sign is not inventoried in
common ending for names and titles, and note that this sequence always
supreme God or supreme lord (being Shiva in Shaivite contexts, e.g.). This is
an interesting but not unique instance of being repurposed from its usual
righteous or holy person, specif., one “fit to receive a dāna” (Cologne Digital
bull]
ICIT 388 ___ [-s(a)va-mIna-/-svāmin-; see preceding] [H-1874; tablet w/o field
figure]
unicorn bull]
image and a hole through the middle; ICIT represents this inscription as
a. Comments:
The uses of this sign are somewhat limited, considering the expected frequency of a syllable
like –sa-. However, this syllable is also contained within other signs of the form CVCV, as
with, e.g., /sAha and /s(a)vA, which could help account for its relative scarcity as a
standalone sign.
b. Inscriptions:
[Karanpura; seal w/ unicorn bull; “votive object” shaped like peepul leaf]
attested name, and devaka is a variant or attributive form of deva. Affix /-ka
śripati (see preceding exxs.) (MW) ][H-2246/ICIT 484; tablet w/o field
figure; there are a total of five such tablets in ICIT, although the other four
[sahA/saha, sahā]
a. Comments:
The similar graphology of this sign with /SA is probably not coincidental. Note that this
sign, in contrast to , can serve as a full word/root (see, e.g., 2154 below).
b. Inscriptions:
(saha=’great, mighty’ or ‘with, having’); note that the sign is almost totally
figure]
reverse (shown below); several other tablets with same text from same
location]
[4-caru (4 oblations)]
and to the right. Additionally, the field figure (rhinoceros) is facing opposite
the canonical direction (head towards the end of the inscription rather than the
in fact occur alone below longer lines of texts on several other objects] [M-
[t(a)ya/taya, tya]
a. Comments:
This sign occurs overwhelmingly as /sa-tya, although the sequence / ādi-tya is not
b. Inscriptions:
name; type of divine being); also a final element in many royal titles
see Indian Epigraphical Glossary, pp. 8-9)] [see also 3 other inscriptions][H-
attested name, and devaka is a variant or attributive form of deva. Affix /-ka
a. Comments:
This sign has some overlap with , but the latter usually appears in word-final position, as
an affix, and apparently cannot appear word-initially. is one of several “long stroke”
b. Inscriptions:
MH7080 [ ka-SU-/kaśu-, RV man’s name] [L-4/ICIT 1986; seal w/
unicorn bull]
‘barter, exchange’]
names] [C-35/ICIT 117; tablet w/o field figure; the ICIT has these three signs
reverse featuring horned animal with human face behind kneeling figure
seal with no field figure; very doubtful representations and readings, since the
___
very plausible; Indian given name)] [M-141/ICIT 2668; seal w/ unicorn bull]
[cAru/caru, cāru]
a. Comments:
This is another sign like / that is found both in transactional contexts and as an element
in names, with two corresponding values that differ only in vowel length. caru is by far the
more common value, and is normally found left-adjacent to long-stroke numerals as, e.g.,
numerals higher than listed in the ICIT (1390 and 375) are very doubtful and appear to be
misreadings. The meaning of caru is “a kind of vessel (in which a particular oblation is
prepared), saucepan, pot RV, AV, [etc.]; …. an oblation (of rice, barley and pulse) boiled
with butter and milk for presentation to the gods or manes” (MW). Aside from the evident
pot-like graphology of this sign, the facts that (unlike / ) it only occurs on a very
particular type of object and only in association with a very limited set of possible stroke
devotional, i.e., a notation of oblations performed. The fact that the important word caru
corresponds not only to a key oblation but also to a vessel or pot is in itself highly suggestive.
However, in a small number of instances, occurs in P-fields and in complex inscriptions,
always found on seals, usually with a unicorn bull field figure, but at least once with an
elephant (H-93/ICIT 1149). It therefore must be the case that has a second value allowing
it to be used as an element in names/titles, usually as the initial element. For this, we posit the
important value cāru, ‘dear, beloved; beautiful, lovely,’ well-attested name [cf. Latin carus],
which is frequently found in names and titles, usually as the initial element (cārudeva, e.g.).
We have not been able to offer a reading for any of the few occurrences where is not the
initial element (M-192/ICIT 2717, M-1714/ICIT 2346, and M-279/ICIT 2800, e.g.), but a
number of names ending in -cāru are listed in MW, though far fewer than those listed with
cāru- as the initial element. Note also two occurrences in MH (not shown in ICIT) of
generally representing feminine forms in –i, we read as the well-attested form cār(u)vi-,
cāru).
We regard the readings caru/cāru for , in conjunction with this sign’s graphology, its
association with a limited set of stroke numerals, and the clear separation of its two types of
inscriptions/seals), to be a nice set of additional evidence squaring with the proposition that
the language of most of the Indus inscriptions is Indo-Aryan. In the following list of
donative contexts.
b. Inscriptions:
ICIT: C-80/ICIT 60; seal w/ unicorn bull (very worn) and H-93/ICIT 1149;
seal w/ elephant]
MH4132 [see prev.] [H-454/ICIT 1444; seal w/ unicorn bull; the ICIT
a. Comments:
This sign is quite infrequent, but its few occurrences do suggest a reasonable provisional
value.
b. Inscriptions:
broken square seal, apparently w/o field figure; 3rd line of text unreadable]
ICIT 130
[pra-ANa-/prāṇa-, ‘breath of life,’ attested name (of a Vasu,
unicorn bull]
[mIna/min, mīna]
a. Comments:
has two contexts. One is the basic “fish sign” that occurs frequently in M-fields, either in
clusters with compound fish signs built on this same grapheme, or left adjacent to various
stroke numerals. In these contexts, it is probably to be read as mīna, ‘fish,’ a word that may
have been related to mā-/mī-, ‘measure,’ and which, in various forms, was a standard
international unit of weight/money across the ancient world. The other context is in field-
final position in P-fields or complex inscriptions, and is to be read –min, particularly in the of
b. Inscriptions:
common ending for names and titles, and note that this sequence always
sign; Īśa is one of the most important divine epithets, and usually denotes the
supreme God or supreme lord (being Shiva in Shaivite contexts, e.g.). This is
an interesting but not unique instance of being repurposed from its usual
righteous or holy person, specif., one “fit to receive a dāna” (Cologne Digital
bull]
image and a hole through the middle; ICIT represents this inscription as
unicorn bull]
a. Comments:
This sign is an exceptional example of a sign representing a CV-type syllable that also
equates to a full word/root (nī-, ‘leader, guide’) which, although it often occurs in compounds
as the final element - nī, may also occur alone (MW). This sign occurs most often as - /
b. Inscriptions:
the leader, the queen’] [K-10/ICIT 1885; seal w/ unicorn bull; note that
ICIT has for the first/rightmost , but this is clearly not warranted from
sovereign, ruler’) Note that here we are assuming that the geminate has
the effect of transforming /tra into tri, on analogy with /DEVA and
in the corpus]
wife’]
supati, ‘good lord, good husband’] [H-458/ICIT 1448; seal w/ unicorn bull]
MH4267 [-PATI-NI-/-patnī- (rarely patni-), ‘female owner; mistress; lady;
6/ICIT 1881; seal w/ unicorn bull; note that in this inscription, is facing
opposite of the usual direction, but since this is the only unambiguous case of
this, and because the source of this seal is Kalibangan, I assume, contra the
ICIT, that this is merely a local graphological variant of the very common
sign .]
[second line]
unicorn bull]
‘barter, exchange’]
unicorn bull]
ICIT 3656 [vAsu-PATI-nI/vasupatni-, fem. of vasupati, ‘lord of wealth,’
man; peepul tree;’ nimāna-, ‘measure, price’; not an attested compound, and
MH does not give final , but the identity of this partly-effaced sign appears
: [p(a)ra/pra, para]
a. Comments:
This sign has considerable overlap with ; it is not clear what rules, if any, govern the
ICIT 130
name (new spelling)] [M-487/ICIT 2970; very worn three-sided seal with
animal and human figures on 2 sides; the ICIT does not have initial , which
may be correct, given the very worn and barely legible condition of this
object]
the former] [K-56/ICIT 1919; very worn and crudely-drawn seal w/ unicorn
given name and divine epithet] [M-220/ICIT 2744; seal w/ unicorn bull]
[rA/ra, rā]
a. Comments:
b. Inscriptions:
no field figure]
iconography; other exx. of this inscription appear to show a figure with a tiger,
etc.]
MH1605 [vI-rA-/vīra-, ‘hero;’ general honorific prefixed to many
unicorn bull]
‘lover, husband; given name’] [C-29/ICIT 112; square seal w/o field figure]
[ŚRIPATI/śrīpati, sūripati]
a. Comments:
This sign appears to be a ligature of /śrī and /pati, with śrīpati a very common
word/title, meaning ‘king, prince’ (lit., ‘lord of fortune’), as well as a very well-attested name
and divine epithet. The possible alternate reading sūripati is plausible, though not attested,
and far less likely; but I have included as a consequence of the possible ambiquity for
itself, for which sūri is a possible alternate value. Moreover, it is even possible that, in the
older language, sūri and śrī arose from a common source word (like, e.g., sure and secure in
English). Whatever the case, is frequently found as a standalone sign in a P-field (like ,
analogy with other such signs—be a common word representing an important name, title, or
honorific. For these considerations, along with the very suggestive graphology, we regard
b. Inscriptions:
the gods’)] [M-1309/ICIT 3614; small rectangular seal with no field figure]
śripati (see preceding exxs.) (MW) ] [H-2246/ICIT 484; tablet w/o field
figure; there are a total of four such tablets in ICIT, although the other three
OR ŚRIDEVAPATI]
a. Comments:
This sign, which is evidently a ligature of and , is most likely to be read śrīpati-deva,
not only because of MH2422 and MH2098 below, but also because we also observe those
two signs in that sequence in MH2579 (see above under ). This sign, while much
less frequent than or the other signs common as single-entry P clusters, nevertheless
a. Comments:
This sign occurs only once, but its graphology (in conjunction with certain other apparently
compound signs with an “attached stroke,” like below) is suggestive enough to warrant a
reading. Because of the paucity of evidence, this reading is of course provisional, but does, in
b. Inscription:
MH2500 [DArA-PAka/dharbaka-, attested Puranic name, MW OR darpaka, n.
[ARYAKA/āryaka]
a. Comments:
This proposed value is admittedly provisional, based on the provisional value for in
combination with the proposal that the “stem” ligature may represent common affix –ka.
b. Inscriptions:
respectable man; attested name;’ *āryakapāla- not attested but very plausible]
3006; copper tablet w/ hare on reverse; plus 9 other inscriptions, all scratched
a. Comments:
This common and important sign has a number of allographs, which are reduced to two in
this accounting. It was the first sign whose two primary values (pāla and pala) furnished the
first major key to deciphering a large number of inscriptions, and is probably the most
important sign to the decipherment so far. As such it is found very frequently in both asset
notations (usually left-adjacent to a stroke numeral) and as the final element in many names
and titles.
b. Inscriptions:
doubtful, since the two rightmost characters appear to have an additional stem-
zebu]
obscured and not identifiable with certainty, and the ICIT very plausibly
bull]
treasurer;’ note that the ICIT has instead of , which is possibly a correct
distinction, although the former sign only occurs twice] [M-735/ICIT 3159;
MH1087 [vI-para-DAna-PAla-/vipradhanapāla-;
vipra-, ‘inspired, wise, learned; sage, seer; priest, domestic priest’ (very
broken square seal, apparently w/o field figure; 3rd line of text unreadable]
‘inspired, wise, learned; sage, seer; priest, domestic priest’ (very common
MH1220 [‘3 palas’] [H-585/ICIT 464; seal w/ zebu (also 2 other inscriptions,
including tablets)]
MH2858 [H-335/ICIT 1356; tablet shaped like a hare (?) (also 6 other
inscriptions)]
ICIT 1913 [K-50; seal w/ fantastic hybrid figure with many-branched headpiece]
‘wealth’ (alternate form for dhana; both dhānyam and dhanyam are neuter);
ICIT 232 [5-PAla] [H-1355; rectangular tablet with three circles with center dots
on reverse]
MH1346
MH1131
MH4238
MH7005
unicorn bull]
MH2176 [PU-PAla-DEVA/bhūpāla deva, bhūpāla = ‘king, lord, etc.’;
unicorn bull.]
bull]
has a second line , but this is in fact a ritual object similar to the feeding
troughs and chattra-like figures found in front of other animal field figures.]
ruler’) Note that here we are assuming that the geminate has the effect
and name of royal family; also ‘governor of the state’)] [M-1838/ICIT 2355;
bull]
MH7201 [line 1 of 2]
MH7032 [see prev.] [L-90/ICIT 2044; small rectangular seal w/o field
figure]
MH5310 [note that the asset notation (3 palas; short strokes) is different
MH1246 [M-1063/ICIT 3437; broken seal that probably had a field figure
inscription)]
MH1422 [H-1996/ICIT 379; tablet w/o field figure (also 5 other inscriptions)]
MH7072
perhaps bhāryā-, ‘wife’] [M-416/ICIT 2912; circular seal w/ ox; the last two
characters are partly effaced and inscribed in an unusual style that casts some
possessor; etc.’; nāthapāla = attested king name and name of a royal family
protector,’ attested name ] [ICIT 2481; tablet w/ unclear animal figure; plus 5
more occ., all scratched on the reverse side of copper tablets with ox OR
3006; copper tablet w/ hare on reverse; plus 9 other inscriptions, all scratched
MH2600 [PU-PAla AS/bhūpāla asti, ‘belongs to the king;’ also 5404, etc.] [M-
1452/ICIT 3730; copper tablet with horse-like field figure (looking back over
shoulder)]
MH4335 [-ARYA-RAJA-PA-hu-jA-PU-PAla/-āryarāja-
field figure; many tablets with this inscription from Harappa, many of them
partially effaced]
pāla]
3006; copper tablet w/ hare on reverse; plus 9 other inscriptions, all scratched
[vAsu/vasu, vāsu]
a. Comments:
This sign seems to have one main sound value (with a possible variant in -ā-; see 8020
below), but with two very different meanings. In P-fields and within names and titles, it has
reference either to a class of gods (the “Vasus”), or to the meaning ‘good, beneficent,’ which
is a very frequent element in attested names. Right-adjacent to the sign , it appears to have
something to do with assets, inasmuch as – like the fish grapheme and its compounds in
M-fields -- it has a number of different compound forms found in the same context ( ,
, etc.), which suggests a series, like the fish series, with a shared function. Note also that
appears to be an element of the sign , which is part of the “oval series” of signs that
also participate in M-clusters alongside fish signs, and which seem to form a metric series
apart from the fish signs. As it happens, the word vasu also means ‘wealth, goods, riches,
property’ (MW) in the Vedic language. The meaning of the very idiosyncratic sign has yet
to be ascertained.
b. Inscriptions:
MH1170 [see preceding for this and many other examples with
‘wealth, goods, property, etc.’, the M-cluster would likely signify the exact
MH2244 [see preceding] [H-650/ICIT 1565; small broken seal w/o field figure
Agni, Krishna, etc.; vasupatnī, ‘mistress of wealth,’ attested as a name for the
cow in RV (MW)] [M-1357/ICIT 3656; small square seal w/o field figure]
6/ICIT 1881; seal w/ unicorn bull; note that in this inscription, is facing
opposite of the usual direction, but since this is the only unambiguous case of
this, and because the source of this seal is Kalibangan, I assume, contra the
ICIT, that this is merely a local graphological variant of the very common
sign .]
[second line]
wealth; king, ruler’; this section probably means ‘property (ASTI) of the
king’]
ICIT 4198 [-vAsu-PAla/-vasupāla, ‘guardian of wealth; king, ruler’]
attested Puranic name, and Vasumani is a name used in India to the present
day] [M-401/ICIT 2897; small rectangular seal w/o field figure (several other
a well- attested Puranic name, and Vasumani is a name used in India to the
name of at least two rishis, namely “1. Name of a Rishi with the patronymic
Aindra (author of Rigveda X, 27, 29, and part of 28); 2. [Name] of another
Rishi with the patronymic Vāsiṣṭha (author of Rigveda IX, 97, 28-30)” [MW];
also the name of a grammarian. This inscription is from the famous “Lord of
a three-faced deity seated in the lotus position, with a horned headpiece. Both
writing, but we do not concur, since one figure, in the upper right, is much
larger than any of the signs in the inscription (which signs are written linearly,
and are ll approximately the same height), while the second anthropomorph is
found well below the inscription, lower than the horned headpiece and
sandwiched between an elephant and a tiger. Thus the horned deity appears to
be depicted as the lord of beasts and men, which latter category is represented
by the two anthropomorphs. We propose that this seal be known as the “great
seal of Vasukra.”] [M-304/ICIT 2823; large broken seal with beasts and
headpiece].
a. Comments:
This sign has few occurrences, and is recognized as a separate sign only by the ICIT, MH
having conflated left-facing and right-facing “quadruped” signs. However, there is a clear
distributional contrast between this sign, on the one hand, and the various allographs of the
right-facing quadruped signs. The value assigned to this sign must be reckoned as
provisional, but it will be noted that, in addition to its suggestive graphology, seven of the
eight occurrences of this sign noted in the ICIT have very plausible readings (the eighth
occurs adjacent to an as-yet undeciphered sign), which lend, in this researcher’s opinion,
b. Inscriptions:
ICIT 2139
[ASHVA-PA/aśvapa, ‘groom;’ note that the ICIT has this as a one-line
inscription, but is clearly incised well above and apart from the other two
/ [GO/go]
a. Comments:
This sign has several allographs, and its provisional assigned value is much more tentative
than that of .
b. Inscriptions:
with bangled arms and a horned headdress; at least seven total occurrences of
this inscription, including five items with the abovementioned horned deity
depicted]
bull]
unicorn bull]
a. Comments:
This character is the first whose decipherment was only made possible by data in ICIT, for
which 18 total occurrences are given (plus three occurrences for ). MH, by contrast, only
has four occurrences of , none of which are in environments that allow for cross-
checking, versus five for . Several instances of given in MH show clear graphology
as and have a strong overlap with a number of inscriptions for found in ICIT, so this
conflation of two allographs appears to be well-motivated. Note that this sign appears to be at
least a partial homonym with , and the distinction between the two is unclear, although
b. Inscriptions:
ICIT 82 ___ [-kara-mAna-/-karman, ‘deed, action, etc.’; common
gatherer;’ overall meaning ‘property of chief tax collector’] [H-773; tablet w/o
field figure]
karasvāmin, attested name (of a holy personage)] [K-7; seal w/ unicorn bull]
act or rite; pious act; offering;’ not attested as a name or title] [M-623; seal w/
unicorn bull]
righteous or holy person, specif., one “fit to receive a dāna” (Cologne Digital
bull]
name of at least two rishis, namely “1. Name of a Rishi with the patronymic
Aindra (author of Rigveda X, 27, 29, and part of 28); 2. [Name] of another
Rishi with the patronymic Vāsiṣṭha (author of Rigveda IX, 97, 28-30)” [MW];
also the name of a grammarian. This inscription is from the famous “Lord of
a three-faced deity seated in the lotus position, with a horned headpiece. Both
writing, but we do not concur, since one figure, in the upper right, is much
larger than any of the signs in the inscription (which signs are written linearly,
and are ll approximately the same height), while the second anthropomorph is
found well below the inscription, lower than the horned headpiece and
sandwiched between an elephant and a tiger. Thus the horned deity appears to
be depicted as the lord of beasts and men, which latter category is represented
by the two anthropomorphs. We propose that this seal be known as the “great
seal of Vasukra.”] [M-304/ICIT 2823; large broken seal with beasts and
headpiece].
From the preceding material, we regard the following inscriptions as yielding the clearest and
most likely readings of complex inscriptions, P clusters, and C clusters arising from the
/devātideva-, ‘god surpassing all gods’; sudeva means not only ‘good or real god/lord,’ but is
also a well-attested name, MW; likely reading devātideva-sudeva, ‘god of gods, the real god’
dhanapāla-, ‘lord treasurer;’ Full reading ‘To/for the lord treasurer 4 palas’ (seal with
unicorn bull)]
attested name, and devaka is a variant or attributive form of deva (tablet w/o field figure;
bhūpāla AS, ‘belongs to the king’ OR ‘belongs to Bhupala’ (potsherd; appears to belong to a
lid)]
MH2176/ICIT 3390/M-998: [PU-PAla-DEVA/bhūpāla deva, bhūpāla = ‘king, lord,
‘king, lord, etc.’; bhūpāladeva = ‘earth-lord god,’ or some such (seal w/ unicorn bull)]
Vasudeva’ or ‘lady Vasudeva;’ Vasudeva is a very common and important name (seal w/
‘who gratifies the gods,’ dhanarakshaka = attested name (of Kubera; MW); 2. atirāja
field figure)].
king Somabhu’ (somabhū, ‘Soma-born,’ is an attested name in MW, epithet of Mercury and
also the name of the founder of the lunar dynasty; seal w/ unicorn bull)]
of the Soma sacrifice’ (seal impression; no field figure; MH includes in its representation of
‘high king Somin’ (somin, ‘performer of the Soma sacrifice, Soma-priest’); the function of
the quadripartite circumgraph surrounding somin is unclear, but the name/title somin has
shown up (without circumgraph) in another inscription, also in the P-field (ICIT 2135 above);
Somin’ (somin, ‘performer of the Soma sacrifice, Soma-priest;’ see preceding); in this
variant, only the –min of somin is enclosed by the quadripartite circumgraph, but it seems
clear that the meaning is the same as with MH1373 preceding (broken seal w/ unicorn bull;
only back and part of the nape of the animal are preserved)]
likeness; idol, statue’ OR para-ātma(n), ‘great soul’, epithet of Brahma, etc. (seal w/ unicorn
bull)]
MH1018/ICIT 3553/M-1203: [mA/mahā]
[ATi-mA-RAKSHA-kA/ātmarakshaka, ‘bodyguard,
king’s name in MW (small partly broken rectangular seal w/no field figure)]
atirāja(n), ‘supreme king’(or adhirāja-, ‘emperor,’ but this seems less likely); -śarman,
‘protection; blessing,’ very common final element in Brahminical names (seal w/ unicorn
bull)]
creator (epithet of Brahma or Shiva)’; attested personal name (seal w/ unicorn bull)]
creator (epithet of Brahma or Shiva)’; attested personal name (seal w/ unicorn bull)]
attested name, and devaka is a variant or attributive form of deva (very worn tablet w/ no
field figure)]
man’s name); note that in the attested form of this name, vibhū- > shortened vibhu- (seal w/
unicorn bull)]
mAna/devakavarman (not attested, but very plausible name, cf. devaka, ‘deity; divine,’ and
well-attested name, and –varman, ‘protection, defense; armor; common surname element;
note that devavarman is attested as a name; lit. meaning ‘armor of the gods, having divine
appellation for Ksatriyas; attested Puranic name (a son of Uśīnara) (seal w/ ox)]
epithet of Soma; attested name (square seal w/ chattra-like ritual object usually shown with
wealth; n. of Agni, Krishna, etc.; vasupatnī, ‘mistress of wealth,’ attested as a name for the
man; king’; attested name of Kubera and of several authors (seal w/ unicorn bull)]
‘treasurer; rich man; king;’ attested name (broken square seal w/o field figure; line 2 of 2)]
MH4624/H-467/ICIT 1455 ___ [DAna-PAla-/dhanapāla-, ‘guardian of treasury,
dhanapāla- ‘lord treasurer’; 4-pala, ‘4 palas’; overall meaning of this inscription, which
exhibits both meanings of , is approximately ‘to/of the lord treasurer 4 palas’ (seal w/
unicorn bull)]
/vipradhanapāla-; vipra-, ‘inspired, wise, learned; sage, seer; priest, domestic priest’ (very
bull)]
(seal w/ unicorn bull; other examples of this inscription both on seals and on tablets)]
righteous or holy person, specif., one “fit to receive a dāna” (Cologne Digital Sanskrit
of at least two rishis, namely “1. Name of a Rishi with the patronymic Aindra (author of
Rigveda X, 27, 29, and part of 28); 2. [Name] of another Rishi with the patronymic Vāsiṣṭha
(author of Rigveda IX, 97, 28-30)” [MW]; also the name of a grammarian. This inscription is
from the famous “Lord of the Beasts” or “Paśupati” seal, which depicts a number of animals
surrounding a three-faced deity seated in the lotus position, with a horned headpiece. Both
MH and ICIT regard two anthropomorphs standing among the animals as writing, but we do
not concur, since one figure, in the upper right, is much larger than any of the signs in the
inscription (which signs are written linearly, and are ll approximately the same height), while
the second anthropomorph is found well below the inscription, lower than the horned
headpiece and sandwiched between an elephant and a tiger. Thus the horned deity appears to
be depicted as the lord of beasts and men, which latter category is represented by the two
anthropomorphs. We propose that this seal be known as the “great seal of Vasukra.”]
Glossary, p. 148); ‘performer of an action’ (MW); full reading is ‘property of the official, 7
palas’]
The values proposed for these five signs, as well as the rationale behind their respective
decipherments, deserves more in-depth treatment, given the prominence of these signs in the
corpus, especially in P clusters. First of all, given the fact that we initially identified the
suggests that these fields contain names and titles. Secondly, given that each of the five “P
signs” given above occurs frequently by itself in a P field accompanied only by a juncture
sign / / means that each of these signs must have some full word or word-root value,
and moreover a value that is common and expected in contexts involving names and titles.
1. : This sign has several interesting distributional and graphological traits that led
ultimately to its decipherment (the first among the P signs to be deciphered). First, it may be
found right-adjacent to all three juncture signs. Second, frequently occurs as geminate
, which we have posited to denote the feminine in many instances (although it may also
be a reduplicated form). Third, often occurs as the final element in what appear to be
appellative sign sequences ( , e.g.), suggesting that this sign is a common final element in
been established as PU thanks to its association with as the very common - , etc. We
thus had the relationship = PU-X, where PU was presumably bhū-. The hypothesis that
eventually suggested itself was = DEVA, a very common name and title element (usually
the final element in compounds), with the geminate = feminine DEVI, also a very
common name and title element. It is worth noting that the reduplicated form devadeva is
The hypothesis = DEVA was given powerful confirmation when we later hypothesized
to have the value ati, or something similar, and eventually noticed that the rather odd
sequence - in MH2069 could very plausibly be read as devātideva, ‘the highest god,
god above gods,’ common divine epithet, and a term we had not previously been familiar
with. Other readings that subsequently emerged, such as -, sukhadeva (attested name),
only further strengthened this reading. As to the meaning of this sign, it should be noted that
deva- may connote not only a divine being, but also any human of very high or lordly status.
any other “juncture sign.” In fact, the pairing is the most frequent sign pairing in the
entire Indus corpus. Unlike , it does not occur as a geminate, but it does have an interesting
suggesting that, like (but unlike ) this sign is both a very common standalone name/title
and a common final element in names/titles. After a very lengthy and exhaustive exercise in
cross-checking and comparison, we concluded that the only value consistent with all of these
data was = PATI, with the frequent P field combination = PATI-nI (i.e., patnī, the
feminine of pati). The meanings of this sign include ‘lord, master,’ ‘owner,’ ‘ruler,’ and
‘husband;’ in this last sense in particular, the word persists in modern Indo-Aryan languages
like Hindi. The feminine patnī means ‘wife’ as well as the feminine equivalents of all of the
other previous definitions. It is a very common final element in names and titles, some of
them overlapping with forms in , e.g., , bhūpati, ‘king, ruler’ (lit., ‘earth-lord; cf.
, bhūpāla).
3. : Unlike and , this sign generally does not occur as the final element in compound
names/titles. Instead, it either occurs singly in P clusters right-adjacent to any of the three
complex inscriptions. The only common name/title element which we found to coincide with
this set of parameters was śrī/suri, whereof we believe there is ample reason to believe that
the former may be primordially derived from the latter (in much the same way, perhaps, that
monsieur is a conflated form derived from mon seigneur, or ma’am from madame). In more
contemporary language, śrī is primarily an honorific used in conjunction with other words
(like Mr. or Mrs.), but we suggest that just as many such titles in various languages (like
monsieur, señor, etc.) may serve either as standalone words or as titles appended to names, so
too śrī/suri in an earlier form of Sanskrit was probably both a term meaning ‘lord’ or
something similar, as well as an honorific roughly equivalent to “the honorable,” e.g. Noting
also the graphologies of this and the preceding sign, we concluded that another common “P
title, śrīpati, and that the less common “P sign” is likely a conflation of and , with
4. : This sign occasionally occurs as a geminate and even once as a triplicate, and is almost
always right-adjacent to juncture sign when in P fields. The single triplicate occurrence
(ICIT 1191, seal without field figure/Harappa) is best explained, in our view, as [Masculine
+ Feminine ], for which the reading rāja-rāj(a)ni, ‘king and queen,’ seems most likely,
this being an attested “dvandva” compound. However, the value = RAJA(N) was first
reached by the usual cross-checking, in conjunction with the fact that this sign, unlike and
, frequently occurs as the semi-initial element in P clusters and complex inscriptions (as -
), and almost never as the final element. In other words, it had the appearance of a title
only, one that always is associated with a following name, and which is not normally an
element in names as –deva and –pati are. These characteristics all matched the value
RAJA(N). It should be noted that the sequence - may well be read as adhirājarāja-,
masculine) and not - (-patnī-, feminine), whereas the latter would be expected if the
initial title - were feminine. It is also interesting to note that of the “P signs,” only
and reduplicate, and, correspondingly, of the major titles/honorifics only rāja and deva
noted the similarity of the sign to a stylized crescent moon, and the association of India’s
first dynasty (the Lunar Dynasty) with that heavenly body. It is perhaps also worth noting
that the name Somabhū, which appears on one very worn seal, is a name associated with that
dynasty.
5. : This sign was by far the most difficult to interpret of the common P signs. It usually
occurs in P fields paired with , and is never found as the final element in any P cluster, C
cluster, or complex inscription. However, like , it also has an alternative set of contexts
with the appearance of some sort of metrology or asset notation, they being , +
that fit cross-checked inscriptions, was a plausible name-title for P fields, and might be
expected in asset notations was vasu, which is a very common name, an important class of
deity, and bears the meanings ‘wealth, riches; gold; jewel, gem.’
From all of the foregoing, we have been forced to conclude that the vexing “patterned
inscriptions,” consisting of various combinations of P field, M field, C field, and T field, may
be resolved into both notational (M field, T field) and transcriptive (P field, C field) textual
domains, with the former two denoting, respectively, metrological notations and abbreviated
predicates, and the latter two—as laid out in the previous sections—consisting of a range of
names and titles, including both personal names and theonyms, as well as honorifics and a
few occupational titles. But how are such inscriptions to be read in their entirety, or,
otherwise put, how can we make sense of this seemingly odd juxtaposition of entries, in the
First of all, it should be evident that that the function of Indus seals is not necessarily
identical to the attested functions of Mesopotamian and other seals. Compared to, e.g.,
Middle Eastern cylinder seals, Harappan seals have different shapes and far briefer
inscriptions. We have seen with deciphered texts given in this study that some Indus seals
simply bear a name, or a name plus a term indicating “property of,” “belongs to,” or some
such, which we might style appellative texts. What we have not seen is evidence of lengthy
votive formulae or lavish biographical information about the bearer. Also, the Indus seals are
remarkably uniform in design, with the overwhelming majority being either the “unicorn
bull” and chattra-like structure as a field figure, along with a raised boss with a hole, or are
small and rectangular, with a single line of text and a hole drilled through the middle. Only a
small minority of the larger “raised boss”-style seals have field figures other than the
“unicorn bull,” and relatively few of the smaller rectangular-style seals have field figures.
Moreover, as even a cursory perusal of any Indus corpus will reveal, many of the inscriptions
are highly repetitive, a fact that, by itself, militates against any notion that such artifacts are
information specific to the bearer, Indus seals are not found in funerary contexts (Parpola
1997).
Finally, it is very clear, as we have shown in this and preceding research, that the Indus seals
often bear numerical and metrological inscriptions, which have helped to serve as an entering
wedge for decipherment. But now we must turn our attention to what these
fields, whether or not they co-occur with T, C, or P fields. What follows is a brief typology of
M field subtypes. In Table 19, we list a small sampling of inscriptions that exemplify these M
field subtypes. The list of examples given is very far from exhaustive, but should suffice to
convey a sense of the abundance as well as the consistent patterning of these types of
inscriptions. In this table, only MH index numbers are given; suffice it to say that these
inscriptions are found across object type, as will be made clear further on.
1. + stroke numeral:
(MH4171, MH4873)
(MH4009)
2. + + stroke numeral:
(MH1365)
(MH2229)
(MH4377)
3. + “fish cluster”:
(MH3074)
(MH1277)
(MH1380)
(MH4285)
(MH4673)
(MH2523)
(MH7247)
(MH2324)
(MH6206)
(MH4112)
(MH1088)
(MH4702)
(MH4056)
5. + “fish cluster”:
(MH1053)
(MH4467)
(MH1109)
6. / + stroke numeral:
(MH2127)
(MH2572, MH3099)
(MH1143, MH1229)
(MH1076)
(MH1131)
(MH2198)
(MH2298)
(MH2387)
(MH4843, MH5100)
8. + :
(MH1068)
(MH1326)
(MH1170)
(MH1345)
(MH3160)
(MH6207)
(MH4028)
(MH2123)
(MH5237)
(MH2821)
(MH7101)
(MH1208)
(MH4101)
(MH3120)
(MH3307)
(MH1370)
(MH1040)
(MH2654)
(MH2426)
(MH2446)
(MH2375)
(MH2325)
(MH4016)
15. + or compound:
(MH1238)
(MH1269)
(MH1096)
(MH5469)
(MH1447)
(MH4022)
(MH1456)
(MH4672)
19. + or compound + + “fish cluster” + “oval cluster”:
(MH2541)
The most basic types of M clusters are shown under 1, 2, 6 and 7, namely, the fish grapheme
straightforward:
= ‘6 /minas’
= ‘3 /palas’
= ‘4 palas’
The context of such inscriptions cannot be overlooked. Many inscriptions of this sort are
found on large seals with bosses and animal field figures, signifying that, beyond any
reasonable doubt, these “seals” do sometimes have metrological information which has every
appearance of representing notations of assets. The following data on Table 20 gives artifact
Table 20: Numbers and metrology on seals and tablets (catalog numbers from Joshi and
Parpola 1987)
From this data, we see that, without question, many of the seals contain nothing more than
numerical/metrological data. In the case of the above examples, we are confronted with seals
whose inscriptions read nothing more than “X palas” or “X minas.” Such seals must either be
The next most complex form of inscription with an M field is the type shown on Table 19,
Having ascertained the values of and , the readings of these two inscriptions, of a type
also typical of canonical seal inscriptions, appear to be along the lines of:
pala-4-CASE-pati; ‘lord-CASE, 4 palas’
In this analysis, we are assuming that “juncture sign” represents either a noun case or
case-like relationship, most likely akin to some genitive or dative sense (although it might
conceivably also denote, or include, some sort of honorific sense). The meaning of the above
two inscriptions, then, becomes something like ‘To/for/of the lord(s), 4 palas’ and ‘To/for/of
These seemingly incongruous readings make perfect sense if we bear in mind that anciently,
temples functioned as banks, both for lending and depositing. This was the case in the ancient
Middle East and Mediterranean; the very words money and mint (as well as French monnaie
and Spanish moneda, both meaning ‘currency’ or ‘coin’) are derived from the famous Roman
temple of Juno Moneta, which was used as a repository for coinage (moneta itself apparently
being derived originally from monere, ‘to warn’). In India, temples have been centers of
banking for millennia, and in many cases, still have enormous holdings of wealth
they recorded deposits of assets at a temple or temple treasury, for which the seal was issued
in exchange as a claim on those assets, which could presumably be used to redeem them, or
as an earnest in other transactions, establishing creditworthiness, for example. As such they
Also plausible is that these, or at least some such inscriptions, were donative in character,
recording some significant donations or oblations, which then became “badges of honor,” as
it were, for the donors. Donative inscriptions of this kind, which sometimes seem to have
been accorded talismanic portent, were often inscribed not only on temple and cave walls, but
also on pottery and other objects, and were a significant feature of protohistorical and
ritual or poetic services, or simply just for religious/spiritual merit. Dana as a socio-
South Asian religious history, having antecedents in the Vedas and then in later post-
Vedic literatures. The concept of dana is not monolithic and, throughout its lengthy
history, different groups and thinkers have contemplated its meaning and place in
texts dealing with the concept of dana) like the Danakhanda (Book on Gifting) which
additions to the concept and regularly recorded these additions in written form
directly upon material and cultural artefacts, such as pottery, reliquaries, plaques,
We now have the beginning of a general hypothesis as to the overall purport of so-called
“patterned inscriptions,” namely, that they may represent a record of personal assets, either
on deposit with, or donated to, some authority — a temple, a sovereign, a treasurer, etc. —
and that the M field and P field correspond to a notation of such assets and the custodial
The next most complex context for M fields involves a final/leftmost , with various
permutations of the “fish” subfield, the “oval” subfield, and in the right-adjacent M
field. In many instances, such configurations also include a P field; several examples of this
accepted/regarded as,’ the latter seems somewhat more plausible, without, however, entirely
excluding the former. Thus, e.g., more likely reads ‘worth ,’ but might
or
reading is either simply ‘is worth (one) measure [mātra-]’ or ‘owns (one) measure [mātra-],’
commodity like gold or silver. We remind the reader that mātra- following a fixed number or
amount has the sense of ‘exactly; in the amount of,’ whereas by itself simply means ‘[a]
measure,’ i.e., some very common standard transactional amount whose precise value has not
been determined.
owns/is worth [weight sequence] CASE-pati/ ‘to/for/of the lord(s); owns/is worth
,’ where /pati would refer either to a deity or some kind of civil authority pertaining
owns/is worth ,’ where /deva most likely refers to a god(s), but might also refer to
owns/is worth the amount of/exactly ,’ where, again, /pati would refer either to a
contextual meanings of these two common terms, bāhuja is an ancient term for a Kshatriya or
member of the warrior caste (the caste said to predominate in India’s mythical Lunar
Dynasty). Bhūpāla, meanwhile, is one of a number of terms for ‘ruler’ that appear on the
seals, and may have referred to a local or regional sovereign instead of the head of state (
/ati- or adhirāja may have conveyed this latter sense). This would account for the relative
character of inscriptions with . Note also that, inasmuch as C fields appear to denote the
owner, either of the artifact itself or of assets denoted thereby, P fields appear to denote the
deity or authority figure in whose name or under whose care the assets were deposited,
donated, or pledged. This is consistent with the notion of denoting a local sovereign,
where repetitive occurrences of this sequence in the C field would then denote assets of the
local sovereign. On the other hand, occurs overwhelmingly in P fields and in complex
inscriptions, but only once in a C field, as , which is consistent with a figure of high,
quasi-divine status in whose name assets could be pledged or deposited, as if for a god.
owner, the M cluster a notation of specific assets, and the P cluster the deity or other
authority in whose name or to whose account the assets are deposited, recorded, or pledged.
The P cluster may simply refer to ‘lord(s)’ [ ] or ‘god(s)’ [ ], as already noted, but may
also frequently denote the ‘Vasu(s)’ [ ; a general term for an important group of deities
denoting important natural phenomena like earth, wind, fire, dawn, light, the moon, etc.] or
convention (in much the same way as epithets like Mahesh and Mahā are always understood
to denote Śiva in modern Hinduism). Additionally, there are many P clusters ending in -
that correspond to names and titles ending in –pati; as there are hundreds of possible
corresponding names, epithets, and titles, most of these remain undeciphered. Also occurring
multiple times in P clusters as the authority to or for whom assets are being pledged,
lord of the council or assembly [ ] also appears. Following on Table 21 are shown a
(MH2654)
exactly/the amount of , to/for/of the high king (adhirājarāja) OR high queen (atirājani)
(MH2446)
(MH2375)
(MH2325)
AS-para- [ ]- CASE-pAla-DAna-DEVA
‘Belongs to Para [very common name, including of kings] , for/to/of the lord treasurer
(MH4020)
Vishnu]’
(MH1087)
interpret, in a general way, all such inscriptions found in this monograph, as well as the many
more inscriptions conforming to these patterns, which constitute a majority of the corpus of
known inscriptions.
Of the varieties of patterned inscriptions shown on Table 19, there remains only to treat items
15-19, which involve various M clusters with and its compounds. , as we have shown
previously, has the value VASU, which in M fields probably equates to vasu-, ‘wealth,
riches,’ but might also refer to ‘jewel(s)’ or ‘gold,’ two other asset-related meanings for this
word. Unfortunately, because the meaning of , the sign always found left-adjacent to
and its compounds in M fields (and seldom found anywhere else) is unclear, and because the
relationship between and its compounds is similarly unclear, it is difficult to draw any
conclusions about the relation of this sign to other elements in M clusters, other than that its
general meaning is most likely ‘wealth,’ perhaps in some standardized sense for which
various multiples (the compounds) have also been conventionalized. It also seems probable
that , on analogy with and , probably denotes some kind of predicate, but one that is
Summarizing what we have been able to learn from the mass of brief, repetitive, and mostly
notational inscriptions that constitute the Indus corpus, we have found a large number of
names and titles, including a few king names (Somabhū, Somin, Śarman, and [probably]
Ratisena, e.g.), as well as a large number of common names (Para, Ravi, Raviprabhu, e.g.),
caste-related names and epithets (Varman, Śarman, Vipra, and Bāhuja, e.g.), and royal titles
(rāja, ati-/adhirāja, pati, patni, deva, devātideva, śri, śrideva, ātmarakṣaka, dhanapāla,
bhūpāla, bhūpati, bhūdeva, etc.). These and many other names, titles, etc., that all emerge
from the values obtained for the signs discussed in this work, together with the other
evidences detailed at the beginning of this study, constitute evidence far beyond any
reasonable doubt that the chief language underlying the Indus inscriptions is an early form of
Sanskrit, and that the inscriptions themselves are largely resolvable into two types, namely,
appellative inscriptions that simply name the bearer of the seal or whomever it represents (the
“complex” inscriptions) and notational inscriptions which, while often containing names or
titles in P and/or C fields, have as their chief purpose the notation of property or assets, either
as transactional or depositary records (the “patterned” inscriptions). All of the uses of these
latter types are still far from clear, but the purport of many of the inscriptions strongly
suggests a relationship with temple donations or deposits. That some inscriptions may in fact
the pot in which it is prepared, one of the five ‘great sacrifices’) + stroke numerals.
Indus society, like historical India, was characterized by caste and presided over by various
princes and monarchs. The very old and worn seal bearing the inscription atirāja Somabhū,
‘supreme king Somabhu,’ is particularly suggestive, inasmuch as Somabhū was the name of
the legendary founder of the “Lunar Dynasty,” the supposed offspring of the moon (Soma)
and regent of Budha or Mercury. It is also perhaps significant that the Lunar Dynasty has
always been associated with the Kshatriya caste, and the term bāhuja, ‘Kshatriya,’ is
extremely prominent in the Indus inscriptions. None of this is to say that the Indus
civilization was the Lunar Dynasty, but that it seems possible that some of the elements of the
Lunar Dynasty mythology may have been inspired by features of this civilization.
Unsurprisingly, the seals, tablets, and potsherds of the Indus civilization are not historical
documents, and it is still unclear whether the Indus writing was ever used consistently outside
such brief and often abbreviated contexts (the isolated example of the “Dholavira signboard”
being perhaps the exception that proves the rule). After all, it seems possible that pre-
memorize, and transmit its vast body of Vedic and Epic Sanskrit literature. It is possible, of
course, that the brief inscriptions on Indus seals and other tiny artifacts represent the sole
surviving remnants of a sophisticated writing system whose literary and votive effusions, set
down on some perishable material, have been entirely lost to time. But it is equally possible
that for the Harappans, as for the protohistorical Indians, literary language may have been
regarded as something too sacred to be profaned by writing, and was only suitable for oral
The work of decipherment is very far from complete, and includes a number of tasks that,
given the current state of evidence, may be unattainable unless many more inscriptions come
undecipherable given the available evidence, and the same may be said for dozens of
infrequently-occurring signs. On the other hand, there are a number of remaining common
and distinctive signs whose meaning may yield to future iterations of decipherment,
especially if a searchable database with more sophisticated search and comparison algorithms
can be created. These include: , , (and perhaps some of its “allographs,” which are
(in non-numerical contexts), , , , and , among others. Also needing clarification are
the “juncture” signs , , and , as to what case(s) or case-like function they actually
perform and the function of the quadripartite circumgraph used with the “enclosed” fish
Additionally, while we have furnished likely readings for several ligatured signs, it seems
very likely that many more Indus signs are in fact compounds or ligatures, whereof the rules
for formation are still far from clear. In some cases, ligatures seem to represent simple
conflations of two signs that may optionally be written as two discrete graphemes, whose
readings are simply the combined full values of the signs; this is the case with compounds in
incorporate sound elements of constituent signs, but their readings are greater than the sum of
compounds seem to have broad overlap with non-compounded related signs, such as /
and perhaps certain other compounds in ( / , for example). The precise function of
in many compounds is still far from clear, but does not appear to be simply phonetic.
In sum, although the task of Indus decipherment can now be accurately characterized as well
in part, it is very far from complete and, indeed, may never be anywhere near as complete as
the various successful decipherments carried out in the Middle East, Egypt, Mesoamerica,
and the Mediterranean over the past two centuries. This state of affairs would change
drastically were large amounts of new material to be added to the corpus from some hitherto
largely unexcavated site such as Ganweriwala, and in particular were some trove of longer
writing samples to be discovered. Barring such an event, however, the task of decipherment
must continue cautiously and without undue expectations, with the modest goal of teasing as
much nomenclatural, metrological, and economic data from the available material as
possible, and contribute whatever additional kernels as we can to our limited stock of
Bonta, Steven
1996. Topics in the Study of the Indus Valley Script. Master of Arts, Brigham Young
University.
2014. The Indus Valley Script: A New Interpretation. Epigraphic Society Special
Chadwick, John
Online:https://www.sanskrit-lexicon.uni-
koeln.de/scans/PUIScan/2020/web/index.php
Online:https://www.sanskrit-lexicon.uni-
koeln.de/scans/MWScan/2020/web/webtc2/index.php
Cooper, Jerrold S.
2004. “The Origin of the Cuneiform Writing System” (in Houston [ed.], pp. 71-99).
Fairservis, Walter.
2004. “The Collapse of the Indus Script Thesis: The Myth of a Literate Harappan
Civilization,” Electronic Journal of Vedic Studies, Vol. 11, no. 2, pp. 19-57.
Hunter, G. R.
1934. The Script of Harappa and Mohenjo-Daro and its Connection with other Scripts.
London.
Online: https://www.indus.epigraphica.de/
Mahadevan, Iravatham
1977. The Indus Script: Texts, Concordances, and Tables. New Delhi.
Marshall, John
https://www.sahapedia.org/donative-epigraphy-ancient-south-asia]
Mitchiner, John E.
Monier-Williams, Monier
Parpola, Asko
1997. “Seals of the Greater Indus Valley,” (in Collon [ed.]: 1997, pp. 47-53)
2010. Corpus of Indus Seals and Inscriptions (vol. 3: New material, untraced objects,
Rao, S. R.
1982. The Decipherment of the Indus Valley Script. Asia Publishing House, Bombay.
Sircar, D. C.
1966. Indian Epigraphical Glossary. Motilal Banarsidass, Delhi/Varanasi/Patna.
Online:
https://archive.org/details/indianepigraphic00sircuoft/page/n5/mode/2up?view=theater
Wells, Bryan
2011. Epigraphic Approaches to Indus Writing. Oxbow Books, Oxford and Oakville.
Wilson, J. V. Kinnier
Wisdom Library
Online: wisdomlib.org