|UFDC Home||myUFDC Home | Help ||
|Table of Contents|
|List of abbreviations|
|Descriptive catalogue of the...|
|The direction of writing|
|Connection with other scripts|
|Analysis of the tables of...|
|Tables with sign-list|
|Appendix I: Museum reference-l...|
|Appendix II: Comparative morphographic...|
|Plates I - XXXVII|
CITATION THUMBNAILS PAGE IMAGE ZOOMABLE
STANDARD VIEW MARC VIEW
|Table of Contents|
Title Page 1
Title Page 2
Table of Contents
List of abbreviations
Descriptive catalogue of the texts
The direction of writing
Connection with other scripts
Analysis of the tables of signs
Tables with sign-list
Appendix I: Museum reference-list
Appendix II: Comparative morphographic table of Proto-Indian and other sings
Plates I - XXXVII
Studies in the History of Culture. No. 1.
THE SCRIPT OF
HARAPPA AND MOHENJODARO
AND IT S CON ECT ION
WITH OTHER SCRIPTS
G. R. HUNTER
With an introduction by
Professor S. Langdon
KEGAN PAUL, TRENCH, TRUBNER & CO.LTD.
Broadway House, Carter Lane, E.C.
Made and Printed in Great Britain *y
PERCY LUND HUMbPHRIS CO. LTD.
12 Bedfor Square, London, W.C.x
and at Bradford
THE TEXTS OF
HARAPPA AND MOHENJODARO.
List of Abbreviations 6
Descriptive Catalogue of the Texts 23
Direction of the writing 37
Connection with other Soripts 44
Analysis of the Tables of Signs 51
Tables with Sign-list 129
I. Museum Reference-list 191
II. Comparative Morphographic Table of
Proto-Indian and other signs 201
Plates I to XXXVII
at the end.
AUTHOR 8 PREFACE.
This work was submitted in manuscript to the university
of Oxford in June 1929, when I was supplicating for the
degree of Doctor of Philosophy. Subsequently the manuscript
has reposed in the Bodleian Library. Permission to publish
it was received from the Government of India, Archaeological
Dept., in November 1932.
It is my pleasant duty here to acknowledge my obligation
to the Archaeologigal Department of the Government of 4ndia
for permission to copy the inscriptions which form the subject
matter of this volume. aince thing volume was written I have
by their courtesy been enabled to copy all the inscriptions
subsequently recovered from Mohenjodaro and Harappa up to
April 1931. On this material I am still working. But it
is important that i should here state that the study of this
new material tends only to fortify most of the conclusions
reached in the volume now offered to the public.
i take this opportunity of expressing my gratitude to
Professor Langdon, who most kindly placed at my disposal his
own researgheg on the subject, and to my wife, who did most
of the monotonous copying and re-copying involved in the
production of the Tables, and whose pen is responsible for
all the actual draughting in this voltge.
To Professor Langdon 1 am also indebted for all
arrangements incidental to the publication of this volume,
as also for reading the proofs.
Nagpur, India. bhe 24th of September, 1933.
G. R. Hunter.
I N T 0 D U C T 1 0 N.
Dr. Hunter has continued his investigations on the
early Indus Valley Script, which he began at Oxford, by copy-
ing many more seal inscriptions, which were excavated by Mr.
Mackay at Mohenjodaro since the material, placed at the disposal
of Mr. Sidney Smith, Mr. Gadd and myself, was available. in
Mohenjo-Daro and the indus Civilisation, three large folio
volumes edited by Sir John Marshall, Probsthain London, 1931,
the script was investigated by the writers named above. Vol.
II, chapter XXII, ijgn-List of Early indua orint, by C. J.
Oadd; Meohanioal a ture of the Jarly Indian Writing, by Sidney
Smith; chapter XXIII, The Indus Script by the writer. Dr.
Hunter has made an intensive study of greater material and has
arrived at many valuable results of classification. Since Sir
John Marshall's book was published, M. G. de Hevesy has called
attention to the soript of the Easter island, Bulletin de la
Society Prehistorigue Franoaise, 1933, Nos. 7-8, Sur Une
oriture ucani nne. There can be no doubt concerning the
identity of the indue and Easter island scripts. Whether we
are thus confronted by an astonishing historical accident or
whether this ancient Indian script has mysteriously travelled
to the remote islands of the Pacific none can say. The age of
the Easter Island tablets made of wood is totally unknown, and
all knowledge of their writing has been lost. This same script
has been found on seals precisely similar to the indian seals in
various parts of anoient Sumer, at Susa and the border land east
of the Tigris.
As to progress in the interpretation the way is completely
barred by the lack of any conceivable clue for even a guess at
a means of interpretation. Here is a civillsation of whose
history nothing has survived. it is impossible to suggest even
the name of an historical person or place of that time in india.
No group of signs gan be suggested as having any definite prow
iunciation and identified with any name which can be suggested,
The only possible clue which suggests itself to me is that the
Sumeriang must have known this script in their intercourse with
travellers from Andla who brought the indian meals to Uumer.
'ragments of lists of archaic signs have been preserved; on
these tablets the Uumeriano identify these archaic signs with
signs of the classical Sumerian and Babylonian script. Naturally
most of the archaic signs preserved and explained on these
tablets are peculiar forms of old SuPerian signs, which can be
fitted into their place in the history of Cuneiform epigraphy,
But a few appear to me to belong definitely to the prehistoric
Indus Valley script. I refer to two tablets both in the British
Museum, 81-7-27, 49 -5O, published In Cuneiform Texts, Vol. V,
P1. 7 and three fragments all apparently from the same tablet,
said to have been excavated in the S.E. Palace at Nimnoud,
K.8520 published by Houghton in Transactions of the society of
Biblical Archaeology, VI 454. All these tablets come from
Assyria, but the script used in the explanations of the archaic
signs is that used in Babylonia circa 2000 B.C., a date not too
far below the period in which Indus Valley seals are found at
lAsh, circa 2700 B.C. It is, therefore, entirely possible that
the Babylonian epigraphists knew the Indus script. Vow the
scribe arranges the signs in order of the well known 4umerian
Syllabary A and in CT.V7 Obv. I there is an extraordinary sign
entered as the archaic form of aJ, usual meaning negative "not",
Sumerian value nu. This is totally unlike any archaic form of
2 and may be the Indus sign 75 or 76 of my sign list, Naturally,
if this thesis be true, all the scribe means to say is that the
Indian sign means "not"; the phonetic value n cannot be inferred
unless tha Indian language is sumer1an. Ibid. Rev. 11 2 there
are extraordinary forms of the sign SAG "heart*, restored by
syllabary AII 52. One of these is identical with io. 87 of my
list and two of them seem to be mere variants. if so, then the
common indian sign No. 87 means "heart", pronounced a, iag in
Sumerian. I do not mean to say that there is any certainty about
this suggestion of the survival of Indian signs in the epigraphical
texts of these Babylonian scribes. Sumeriap texts of this kind
or bilingual Sumerian and indian inscriptiong seem to offer the
only possible help to which scholars may have recourse at present;
for the Sunerians were the only literary people who knew this
only possible help to which scholars may have recourse at present:
for the Sumerians were the only literary people who knew this
writing and language when it was still written and spoken.
Dr. Hunter has presented here all the known material. His
knowledge of all the existing variants of the signs is unsurpassed
and I am glad to have the opportunity of commending his book to
scholars as a trustworthy edition of the texts.
S. Langdon, Oxford, uotober 10, 1933.
The Script of Mohenjodaro and Harappa and its
relation to other scripts.
The material for the above work was provided by some
750 inscribed objeQta unearthed at the abeve-mentioned sites
up to February 1987, These objects were mostly seals, oon-
taining on average about 0 signs apiece. A few copper coins
were also found, and some slabs of'olay impressed. There
were also at narappa several inoised slabs of steatite whioh
appear to have served as receipts.
The signs are clearly of ideographic origin, some readily
reoognisable plotures, e.g. of birds, but moot are oonven-
tionalised, in many oaees beyond recognition of their pic-
torial originals. Graphioally the script bears a close re-
semblance to ProtoEslanite, and a less close to Sumerian of
the Jemdet-Naar and Fara periods, except as regards the
anthropomorphous eigne. The latter bear a lose resemblance
to Egyptian of the Old and Middle kingdoms. The resemblance
to these three scripts seems too lose to be aoeidental, but
whether the connection is due to community of descent or borrow-
ing cannot yet be determined.
One of the cardinal features of the script is a system
of modifying basio signs (a) by internal and external strokes
asiilar to the g= modifications in Sumerian. These do not
always alter the sense or pronunciation (b) by the addition
of one or more short strokes. The latter do modify at least
the sound. These strokes are applied on exactly the same
principle as in Brahbi, and with the same effect. Indeed
the entire Brahmi 'alphabet' is shown to be derived from the
script of Mohenjodaro and Harappa. It is also shown that
those scholars were not mistaken who connected Brahmi with
the South Semitic and Phoenician scripts. For there is
much evidence to sgow tiat these also were derived from the
script of uarappa and Mohenjodao (which.1 have called Proto-
I1nlan). It is thus seen that Proto-Indian forms an import-
ant link in the history of the evolution of the alphabet from
pictographic writing. The method adopted in elucidating the
script has been to tabulate every ooourrence of each sign
together With those signs whose morphography suggested the
possibility of their being variants. In this way certain
sign sequences showed themselves to be of common ooourrerme.
ThuW it was possible to reoogrnis variarts and also words.
The languages of Harappa and Mohenjodaro are shown to
have been one and the same. it has not been possible to
determine from the material at hand the identity of this
language. It appears however to be monosyllabic. It does
not appear to be the language of the Proto-El4Imte tablets.
It is possible on the latter to recognize those sign groups
which constitute proper names. similarly on the Proto-
Indian seals the bulk of the legend is always a proper name.
eany signs are ooemon to both scripts, but the sequences are
quite different# if then there are no proper names in common
it is not likely that the languages are closely related.
Many of the signs of the Cypriote syllabery bear a
close resemblance to Proto-Indian signs, but the phonetic
values of the latter, as far as these can be determined from
Brahmi and the semitio scripts, are irreconcilable with the
Uypriote phonetic values. If connection there be it must
have been at a period before Proto-Indian became a phonetic
The script reads normally from right to left, but ooca.
sioally from left to right, and sometimes boustrophedon.
In the latter case the signs ape sometimes reversed, but not
always. it is Certain that the reversal of a sign had no
effect on its signifi cane. The reading is over the backs
of the animal signs, as in aeroitio, but the anthropomorphous
signs face the direction of the writing.
It has been possible to determine the significance of a
few of the signs from the regularity of their oourrence in
particular positions and contexts: In particular (a) the
numeral signs,(b) the ordinal suffix, (0) the word for
'servant' and its determinative, (d) the ablative suffix,
(e) the dative suffix, (f) the word for 'slave' and its
determinative, (g) the word for 'son'. The coins bear the
same names as the seals, votive tablets, and receipts, but
of course without the dedicatory preface often found on the
seals and votive tablets, and without the ablative suffix
common on the reoeipts and not uncommon on the seals and
The work is divided as follows: (1) Introduction,
(2) Descriptive catalogue, (3) Museum catalogue, (4) The
direction of the writing, (5) Oonneotion with other scripts,
(6) Analysis of the Tables of Signs, (7) The Tables of Signs
with a sign list, (8) A Comparative Table of Proto-Indian
and allied signs, (9) An Appendix giving an analysis of
Sumerian ideograms, with a view to elucidating their pioto.
graphic signifioanoe for the purpose of comparison with
THE SCRIPT O0
HARAPPA AND MOHENJODARO.
And its connection with other scripts.
LIST OF ABBREVIATIONS.
Arohaeological survey of Indi4, Annual
Ounningham, Archaeological survey of
Oambridge History. o India.
Delaporte, Musee du Louvre, Ostalogue des
UylInidres Orient lx.
Delegation en Perse, M6moires.
The introduction to the present volume.
Journal of the Royal Asiatic Society.
Certain unpublished photographs of impres-
sions from Proto-Indian seals.
Revuo d'Assyriolog e.
INTR 0 DU TIO
The existence of the script dealt with in this work
has been known to Orientalists for half a century, or more.
But it was not till the Archaeological Department of the
Government of India took in hand the systematic excavation
of the ancient sites now known as Harappa and Vohenjodaro
that any considerable number of texts was forthcoming. Even
now the texts we possess, though numerous, are very short,
being mainly confined to engravings on seals. No stelae
have as yet been found.1 Nevertheless it is felt that the
texts at our disposal are sufficiently numerous to justify
the present attempt to collate them and classify their signs,
and draw certain inferences regading the nature of the script.
Plates I to XXXV00 indicate the extent of the discoveries of
insoribed objects up to th- close of the excavating season -
late February 1927. They are reproductions of autographic
copies made by the writer at the Museums of Mohenjodaro and
Harappa during March and April 1927. They reproduce, then,
all the script that we at present possess with the exception
of the following, which have already appeared elsewhere:-
1. Nor have we a single example of the olay tablet, so common
C. Except no. 11, below, and the few texts which in the Tables
appear with their number preceded by P in col. II. These
are taken from unpublished photographs to which the author
has had aooess.
4. V n
s. AV4V 0 k
9. Vv ,'
R.A. Vol. 22, page 99.
J.R.A.S. 1925, P1. X, p. 698.
C.A.R. Vol. V,
PI. XXXIII, and
pp. 699, 700.
J.R.A.S. 1912, pp. 699, 700.
HI. Vol. I P. X
U.H.I. Vol. I, P1. XI.
n In nI
R.A. Vol. 22, p. 56.
D.C.O.O. Vol. I, P1. 2, No. 8b.
30. /' ,, mD.0.0.0. P1. 25, No. 15.
Del. en Perse, Vol. II, p. 129.
1. Reoopied from the originals.
2. Shading indioates that the text is defaced or broken end
II. U (; 1
le. 1 4a
I4. ri I
17. vn '
1V -t t" 'A
1. Copied from the original in the Louvre Museum. The
original is a seal, circular, of stone dark green in colour.
The signs are written in the upper semieirole parallel to
the oiroulterence. The lower semicircle shows a bull.
2. Nos, 12 to 55 are reproduced here with the signs as they
would read on an impression. The photographs in the Illus-
trated London News reproduce the actual seals. Those in
the A.S.I.A.R. and the 'Times' do the same.
5. The Illustrated London News published other seals besides
those given here. Their texts will be found on Plates
I to CXXXV among the others, being copied direct from the
originals in the Museums of Mohenjodaro and Harappa.
Not yet published.1
A.S.I.A.R. 1P95-1924, P1. XIV.
R I XlX.
"The Times", Feb, 26th .926.
The Illustrated London News,
Oct. 4th, 1924, 3
a a I
The Illustrated London News,
Oot. 4th, 1924.
19. ; O c
20. Vt,::" A l.
21. V "11, "
22. Y, (
26. A' V j
28. I I
The Illustrated London News 4-10-24.
if U U S
I? if U i
if if U i
1. Nos. 15-19, 21, 25, 27, 29, 30, were republished in
Archaeological Survey of India, Annual Report, 1923-1924.
50. n The Illustrated London News,
Oct. 4th, -1924.
The Illustrated London News,
31. 4 1oc'/9X
m6. 4 K1
A cursory examination of the script of Mohenjodaro
and Harappa will reveal that it is distinctive. At is
neither Sumerian, nor any other known script, though it
bears certain resemblances to several. Some of these are
doubtless coincidental, since in the very nature of pigto-
graphio writing it is hardly possible to avoid some similarity
in depicting the same object. A closer examination will
establish that it is precisely the commoner signs of our
tezts that are the most distlitotive e.g. V
At the same time it would be rash, in the present state
of our knowledge on the subject, to rule out of court the
hypothesis of a common descent from some remote ancestor for
the script of Harappa and any other piotographio script.
We krow so little, after all, of the ultimate photographic
ancestry of any script, even Uumerian.
Let us now refer briefly to circumstances and considera-
tions that should be borne in mind when examining this
Race. It is not likely that the originators of the
script were Aryans, since the latter are not believed to
have entered India before 1200 B.C., at the earliest,
whereas the script, as proved by Mr. Maokay's find at Kish,1
existed many centuries before that date. It is probable
that the Indus Valley prior to the arrival of the Aryane was
inhabited by Doavidians, and that the Brahuls of the neigh-
bourhood are a remnant of this stock; but this is not
certain, nor would it exclude the possibility of a riverine
or maritime folk of a different race being responsible for
Mohenjodaro and Harappa.
There is a natural temptation to look for a connecting
link between the agglutinative languages of ancient Sumer
and Elam and the a;glutinative languages of Modern India;
and in this connection not only Brahul is of interest, but
also the ancient tongue so-far represented by a solitary
cun\iform inscription from Herat. It is of course obvious
--------.----------------- 7 ----------------------------
2. See Sayoe. Antiquity. June 1927, p. 208.
that the finding of a linguistic connection between Sumerian
or Atzanite or the language of the Herat seal on the one
hand, and any modern language of India of pre-Aryan origin
on the other, taken in conjunction with the undoubted faot
of intercourse between India and Sumer and Elam, would be
a likely slue to the identity of the language of our inecrip-
tions. But so far this connection has not been found.
Meanwhile, in looking for it the peculiarities of the qunda
languages should not be ignored. That their present speakers
are even more primitive than the Dravidians is historically
not repugnant to the possibility of their ancestors having
evolved an elaborate civilization five thousand years ago.
It Is unfortunate that little information of an ethnolo-
gical order has been yielded by the exoavations:--a few
skeletons, the position cf which leaves it open to doubt
whether their owners were not the viot;ms of a mediaeval
'dcoity'; and a couple of busts of which sir John Marshall
has stated that their heads are unlike those of any modern race
of Indian. But one would like to know whether any anthro-
pometrioal survey of the region has been made, and especially
of the predominantly Brahui tracts of Baluohistan.
However, it Is equally possible that the people of our
script were a seafaring race, foreign to the India into which
they had penetrated up the navigable Iudue and its affluents.
In support of such a contention it might be urged that the
sites so far known of this civilization are confined to the
banks of navigable rivers; that the fish (?) sign is peculiarly
in evidence in their script; that they certainly brought
bitumen overseas (from Mesopotamia ?) for the swimming bath
at Xohenjodaro; and that while an abundance of seals have
been found which were certainly used for stamping the sealings
of merchandise, as is proved by the sealing acquired by
V, Soheil (no. 8 above), which still bears on it the traces
of the fabric to which it was attached, such sealing are
noticeably absent among the finds at Mohenjodaro and Harappa;
suggesting that the seals were principally employed for
stamping merchandise destined for abroad, and that Mohenjodaro
was a great emporium.
It is also to be remarked that the houses are all small
and surprisingly uniform in their dimensions, and that nothing
resembling a king's palace has so far been discovered. This
would also seem to point to a demooratio (or oligarchic)
trading community rather than to a native monarchy. Were
these people the Phoenicians of the East? There are times
when one is almost tempted to credit the legend of a lost
Atlantis, placing it, however, rather in the Pacific and
around Easter island than in the Atlantic, and to wonder
whether there, in early times, did not arise a Neolithic
civilization and neolithic script which, spreading thence
West and East overseas was the ultimate parent alike of
Central American and Indo-Sumerian civilization. One thing
that is certain is that there was much more travel and inter-
course in archaic times than has been generally supposed.
The history of navigation, from the time when the ocean-
going ships of Tyre were succeeded by the coasting galleys
of Athens down to the days of the Northmen, seems to be one
of decay rather than progress. But before the Phoenicians
it would seem to have been otherwise, and what was a daring
voyage of discovery for Nearchus was perhaps a commonplace
of normal trading for the sailors of Mohenjodaro. Indeed,
it is possible that the sailors of Mohenjodaro embarked upon
voyages much longer than that from the Indus to the Euphrates.
I would invite a comparison of the seaj1 published as
1. Provenance Crete, part of the Demargne collection,
D.C.C.O. p. 94. There are several similar 3-faced,
prismatic seals from Crete in the Ashmolean Museum, Cxford.
No. S1,a,b,o,d, (6 88) on plate 59 of M. Delaporte's Musee
dU Louvre, Cylinders et Uaohets Orientaux, with the triangular
prismatio objects of similar size found at Harappa (PI. XX,
Noe. 62-83). The design on the side 16B of this. ret(n seal
may be compared with A lsae Table L=I, col. IV) in Proto.
Date. Seals like the one found by Mr. Mackay have been
found in abundance at various levels at Mohenjodaro and
Harappa. The square seal portraying a bull, with one horn
visible, standing in profile (facing right), with the symbol
in front of his fore-feet, and the text written horizontally
across the upper portion of the face of the seal is the
commonest find at either site. Now this is the only Indue
Valley find in Mesopotamia that can be approximately dated,
unless we accept as of Indian provenance the seal found
recently by Mr. Woolley, and accept also the genuineness of
the guneifbor characters it bears. The latter, which was
recently on temporary exhibit in the Assyrian basement of
the British Museum would appear to belong to the third
millenntum B.C. The Kish seal also is not later than
2000 B.C. Meanwhile in India itself, while there is evidence
of intercourse with Mesopotamia, that evidence is insufficient
to emble us satisfactorily to date any particular stratum of
the ruins. There are a few square seals of black marble,
similar in shape and size to those found in Meaopotamia of
the archaio period. Some of these bear no legend, and have
therefore not been included in these plates. But the ordinary
-------- -------- ----- --- ---~.r,, ,---- - -- - -- - -
1. See Plate I, No. 390.
2. Some of the pottery shows affinities with that of Monesian,
of Susa of the seoond period, and Jemdet Nasr, cirea 3500 B.C.
square seal with inscription, that has been yielded in
hundreds by Mohenjodaro and Harappa, is different as to
material, shape, and the ring attachment on the reverse
from these archaic seals. On Sumerian and Elamite analogy,
then, one would be inclined to ascribe the archaio-looking
sea3 to the fourth millennium B.C.; while on the evidence
of the Kish seal one would ascribe the ordinary seal with
ring attachment to the third millennium and perhaps to part
of the second also. This does not preclude the possibility
of their survival into a later period.
The few circular, flat, olay objects, sometimes bearing
a stamped inscription, and in appearance not unlike Phoenician
Tesserae, which have been yielded by the excavations, may be
of later date, There are objects very similar in appearance
frca Susa, exhibited in the 'balle dite de Mastaba' of the
Louvre. Another object apparently of late date is the
fragment of a silver bar shown on Plate 2XVII (No. 518).
If the signs thereon are cuneiform of the 'nucleiform'
variety, as they appear to be, it would seem that here we
have a Babylonian export of comparatively late times. And
this is about all the material we at present possess that can
assist us in dating our texts.
It is clear then that we have no ascertained upper and
lower limits, except that the lower limit was probably pre-
Buddhist since a Buddhist stupa of the third century B.C.
crowns the acropolis (T) of MQhenjodaro. Again the complete
absence of Aobaemenid remains at Mohenjodaro suggests that
it was evacuated at latest before the establishment of Persian
rule in that area. The upper limit may well be beyond
4000 B.C. The considerable depth of superimposed buildings
all in burnt briok, evidently of suncessive epochs, which
the excavations at Mohenjodaro have revealed suggest that
this civilization had a very extended duration. It is true
that the script seems to have undergone remarkably little
transformation throughout the period. But this need not
surprise us when we remember the history of the monumental
script of Egypt. The comparatively rapid changes in Meso-
potamian cuneiform may be attributed partly to the invention
of the clay tablet, and partly to the influence of foreign
conquerors with no interest, religious or national, in pre-
serving either ounbrous forms or ideographic values.1 But
in the Indus Valley the negative evidence is clear that the
clay tablet failed to establish itself, while there is no
positive evidence of foreign conquest. The various suooes-
cive cities of Mohenjodaro do not appear to have been burned.
Language. If as I think professor Langdon Is right in
deriving the Brahmis script from that of Harappa and Mohen-
Jodaro,3 it follows that some of the latter's signs had
acquired phonetic values by the time they were borrowed by
the Hindus or that which is equally possible by an earlier
race who passed them on to the Hindus. But little else
follows. It certainly does not follow that the 'Indiase'
of Harappa and Mhenjoedaro spoke Sanskrit as Colonel
Waddell appears to have thought any more than that the
Phoenioieas spoke Greek! The possibility that the people
of Uohenjodaro were the ancestors of the Brabui has already
Civilization. The people of the Indus Valley were in
point of general civilization similar to their Uumeriea and
1. Of. the writing of Anzanites in the cuneiform script in the
days of Naram-Sip.
2. Another olphabetto script that may owe something to that of
Harappa and Mohenjodaro is the South Semitic.
5. In an article not yet published.
Babylonian contemporaries. Their briokworkl is excellent;
especially in the aonstruotion of their drains, which remain
watertight to this day. Incidentally the size of the sur-
face drains suggests that the rainfall, if seasonal, was
heavy. Perhaps the monsoon visited Mohenjodaro in those
days. There is no inherent meteorological improbability.
In 1926 Karachi received over 10 inches of rain in two suc-
cessive days, though the normal annual rainfall in modern
times is under 10 inches. The apparent absence of irriga-
tion works at Mohenjodaro would also suggest that in ancient
times the rainfall was adequate. The presence of the elephant
and the rhinoceros, and the absence of the oamel in their
glyptio designs supports the same conclusion. These people
were clever oraftsmen, working in many metals and stones.
They made e*oellent pottery, whioh they deoora'ed with taste.
some of these designs are still in local use today.2
Method of writing. The examples of direct writing that
we possess are confined to objects of copper and stone.5 On
clay we have only stamped impressions. But it is obvious that
the literature of this people was not confined to the 700 odd
seals and amulets ete. unearthed. The absence of lengthier
documents among the finds suggests that for ordinary purposes
perishable materials were used. That clay was not among them
has already been inferred. Perhaps they utilised skins, as
Herodetus tells us the Phoenicians did, perhaps papyrus or
1. It is interesting to note that in point of size and shape
the bricks are similar to modern bricks, and quite different
from the large square Babylonian brick. They resemble
rather the bricks excavated by Professor Langdon at Jemdet
Naer. All the bricks are burnt. The finding of these
perfectly-made, modern-looking bricks even at the lowest
levels is one of the curiosities of Mohenjodaro.
2. See an actiole by the writer in the 'Times of India, Illus-
trated Weekly', May 7th, 1927.
3. Except for two signs scratched on a piece of pottery. See
l. II, Bo. 21.
silk. The signs themselves, on some of our seals, suggest
the influence of painting with a brush, being splayed at the
extremities. It is quite possible that here we have indica-
tions of a change of style due to the introduction of a new
writing material, which, as future specimens come to light,
may be of aid in dating our finds. The signs are traced
vertically from top to bottom, and are arranged horizontally.
The animal, in oases where there is an accompanying animal
design, is usually plooed immediately below the script, and
faces to the right.2 There are, however, some half-dozen
oases in whibc the animal faces left.5 The large number of
signs yielded, after allowing for probable variants, makes it
olear that the script is not alphabetio. It was probably,
like Sumerian, a mixture of the phonetic and the ideographio.
The first point to determine in any attempt to elucidate the
script is the direction in whi h it reads. In accordance
with Egyptian usage one would expect it to begin over the
head of the subjaoent animal and read towards the tail, ie*.,
in our case, from right to left. And this, as we shall
presently show, is what we do find. It is interesting to
note however that in the body of our tests the animal designs
face to the left; that is the script reads 'over their
backs so to speak, as in thi Mooritio inscriptions. The
anthropomorphoug signs however face right.5 Another
1. See PI. I, BOS. 89, 901, 409. There were several other
examples showing an approach to this style of script. But
it was not found feasible to reproduce in the autographs
minute variations in the thickness of the signs.
2. It is of course to be understood that when speal:in of
direction in connection with seals it is always the direction
of the impression taken from the seal that is intended.
-. Nos. 513 to 517.
4. See in particular PI. XIV et seq. Nos. 277,292,365,406,451.
5. See Table 2LX,
observation is that the second line, when the space left by
the subjacent animal permits, is frequently complete on the
left; while, if sufficient signs to fill the line are not
required, it is the space to the right that is left vacant.
This in some instances is due to boustrophedon writing. But
where we find two-lined inscriptions with both lines reading
from the right, and in the second line a blank space left on
the right, we may attribute this to an artistic or epigraphic
tradition which required the end of the last line to contain
the end of the inscription, just as the beginning of the
first line contains the beginning of the inscription. The
Sumerians evidently had the same convention. Reading from
left to right thql9ft the left end .of the last line blank.
Of. Oudea, cylinder A, Gol. I, oases 8, 10, 14 and passim
in Sumerian Inecriptions.
The dominant impression mentally registered after a
survey of the sites and the remains of lhbenjodaro and
Harappa, and especially of the inscribed objects, is that
this civilization was independent: remarkably independent
when its undoubted commercial connection with Mesopotamia
is recalled. Consider the evidence of epigraphy alone.
Among nearly 800 inscribed objects we have, -to date, not a
single inscribed brick tablet,cylinder,1 Qone or mace-head.
This civilization vanished. How, when, and why is at present
a mystery. The evacuation of Mohenjodaro seems to have been
peaceful, and, judging by the comparative paucity of the
finds of intrinsic value, deliberate. Probably a sudden
shift in the course of the Indue it is now four miles
distant was sufficient cause. But for the abandonment
of the whole region a wider explanation must be sought.
---------------- --- --- --- --- --- --- --- --- --- ---
1. The cylinder seal found at Susa is presumably the work of
a Hesopotamian craftsman to the order of an Indian client.
Possibly progressive dooiooation of the neighbourhood was the
cause. Meanwhile, this civilization does not appear to have
vanished without leaving aey influence on its suooessors.
As already stated, Professor S. Langdon detects its influenoe
on the Brahmi script, Sir John Marshall on Hindu religious
symbols. But for Colonel Waddell's supposition that the
people of Mohenjodaro and Harappa were the ancestors of the
Hindu Aryans there is at present no evidence.
In the present fragmentary nature of our knowledge it .s
not possible to arrive at any final conolugion regarding the
Proto-lndian esript and its affinities, The provisional
o0nglusions that I have reached on an examination of the
evidence are these:-
1. The script as we have it is mainly phonetic.
2. It had a pictographic and ideographic origin.
5. That origin was many centuries before 5000 B.C., as is
shown by the highly conventionalised form of the majority
of the signs at that date.
4. There are olear affinities with Sumerian and Proto-
Elamitio, which, in the case of Sumerian, increase as
the difference in date increases, i.e., the resemblance
of the script of Mohenjodaro to that of Jemdet NWar
(3500 B.C.) is suoh greater than its resemblance to the
.umerian of contemporary date (5000-2000 B.C.), showing
that the common ancestry (or mutual borrowing) of the
three scripts dates to before 4000 B.C..
5. That the homomorphous signs (Table XLIX), which are
invariably silhouette, and are thus in marked contrast
to the Sumerian (which used the head, neok and bust,
but never the complete silhouette) suggest borrowing
6. That the superficial (T) regemblances to Cretan, suggest
the possibility of the existence in remote times of a
very widespread raoe uiing a single piotographic
system of writing.
7, That the arebmi, Sabaean, a portion of the Cypriote
and a portion of the Phoenioian eoripts are derived
from Proto-Indian, due in the last three oases to
oonmercial interoourse by sea via the Arabian Sea,
the Red Sea and the Mediterranean. It is possible
that the Indians had the monopoly of seaf.aing as far
as the Gulf of Suez, which would account for Hiram's
eagereess for an alliance with Solomon that would
allow the Phoenicians to establish a base at Eziongeber.
1. 3rd Kings, Ch. IX 20 28.
Descriptive Ostalogue of the Texts,
los. 1 to 80. Stamped impressions.
No. 1. A lump of burnt olay, bearing i the centre the
imprint of a complete seel. This is the only object of its
Vind hitherto found in India. The only other one known was
found in Mesopotamia (see No. 8 of the introqdntion). Beneath
the inscription is n animal in profile, facing to the right,
with only one horn visible. Below his head is a symbol,
probably he majority of the inscribed seals of
ohenjodaro and Harappa portray an animal in profile facing
right with either this symbol, or ===: or a plant,
placed below the head. It is suggested that the animal
represents a divinity, and that the accompanying symbol repre-
sets an offering. With regard to the meaning of the script,
it is probable that the seals were intended to serve mudh the
same purpose as the Iesopotamian cylinder seals, and that their
legends are, therefore, similar in meaning. A reference to
the sign-list will reveal similar sequences in signs on the
seals Noe. M. 70, 232-234, 462-464, 477.
9os. 2, 3. Plat rectangular slabs of clay. There is no
design aooompanying the legend.
4. A piece of clay, shaped like a button. The inscription
on the front hemisphere is aooompanied by a subjacent bull2
with two horns visible, with the 1t= symbol at his feet)
beneath the inscription on the rear hemisphere is a rhinoceros ()
--------------------------CII- CI ---------- -- -Cli~--------- -----~
1. All sizes are approximately as shown in the plates, except
as otherwise stated in these notes.
2. The subjacent animal is always to be understood as facing
to the right, unless otherwise stated.
5. Olay. No accompanying design. Face flat. Reverse
6. Fragment of a small thin slab of clay. A decorative
design is impressed on the reverse.
7. Thin clay slab.
8. Similar in shape to the small, three-face, prismatic
objects that are common at Harappa (see H. 62-85, 87, 88).
On each face the legend is accompanied laterally by an animal
design a bull with two horns showing,and the =-- lay.
9. Thin flat slab of clay.
10. Rectangular stamp on a fragment of pottery. The only
instaQne of stamped pottery on these sites.
11. In shape a dice. Yellow in colour. Impressed on all
six aides; on three sides two sets of parallel lines crossing
each other at right angles; on the fourth side two parallel
lines; on the fifth side a bull with defaced superscription;
on the sixth side the text shown, with a bull subjacent.
No. 12 it identical, while there was also a third dice similar,
but without any legible script.
15. Clay. Face flat; reverse convex.
14. Circular. Face and reverse flat. Olay. Below the
legend is a bull (?). On the reverse is a decorative design.
This object is about 8 mm. thick. In shape and size it is not
unlike a Palayraean tessera.
15. Clay slab. The script and design on faoe and reverse
are identical. The design which aooompanies the script
laterally to the left is apparently a rhinoceros.
16. Similar to 15. Olay.
17. Clay slab.
18. Three-faced prism of clay. The signs extend vertically
over two faces.. The design on the third face is most inter-
esting as tending to establish the sacred nature of the bull
on our seals, and also the orientation of our signs. It
is clear that the men are walking from left to right, holding
(t-rI" wi It)f( y'1?) 7fi ga. W i, X= 4. e, Ia
the standards in front of than. signs ^bear a
strong resemblance to the last man (reading from the right).
19. Clay slab. Reverse, two goats.
90. Clay slab. The space to the right is occupied by a
goat. Beneath its head is Y The reverse is identical.
The paucity of stamped slay at eohenjodaro some 18
articles compared with the large number oa seals about
450 is noteworthy.
No. 21. Inscribed pottery.
This is the unique example of inscribed (as distinct
from stamped) pottery. The V is three inches in height.
The signs are roughly soratohed with a sharp instrument on a
round plate, or dish, about 1 toot in diameter. Probably the
owner's identification mark.2
1. Bee Plate I, No. l8o.
2. fT. Harappa No 863, where the signs are probably a builder's
Nos. 22, 23. Inscribed stone.
Fragments of black marble bracelets, or anklets. The
signs are clearly and cleanly inzised with a sharp instrument.
These are the only examples at Mohenjodaro of direct writing
(as distinct from the preparation of seals) on stone.
Nos. 24-61. inscribed copper.
These pieces of copper, thin rectangular slabs about
*th of an inch thick, of a standard size, would appear to be
pieces of money. As far as is known they are unique, nothing
similar having beep found in archaeological sites in other
countries. Op the reverse they bear animal designs similar
to those on the seals. The writing is now difficult to read
owing to corrosion. The fact that several of the inscriptions
are identical suggests that they give the names and titles
of rulers, of the issuing authority, or of the place of issue*
It is hardly possible that they give the value
or weight of the eoine, since we find entirely different
legends on coins of the same size, weight, and material. Now
it will be found on examination that almost all of the sign
sequences found on these coins qan be paralleled from the
seals: indeed. in two oases the complete legends are identical:
viz. t:e coin M. 42 and the seal M. 481; the coin M. 54 and
the seal I. 1 Similarly the sequences Vy V' V ,V
'l, rA, 1 vIo which are
found at the end (left) of the copper-ooin inscriptions, are
likewise found at the end of the seal inscriptions, as a
glance at the Tables will show; and the sequences FM ,
f , Ili , 'i', ill I ~f ';,' w Ihich are found
at the beginning (right) of the copper-ooin inscriptions are
likewise found at the beginning of the seal inscriptions.
It is olear then that we have on the coins the same kind of
inscription as on the seals, and, from our universal experi-
enoe of seals in all countries and all epochs, this can only
be Proper Names. So then the copper inscriptions set forth
the names, titles, or styles of the persons who issued the
coins, probably the rulers of the state. With this thought
in mind we may re-examine nos. M. 24-51. It will be shown
later1 that V is but a 'spelling-out' of V Nos.
24531 then are identical, and might have all been written as
Nos. 30 and 31 l A .. TIhese signs are to be read from
right to left.2 They probably constitute the ruler's style.
The last sign is so generally last as to be almost certainly
a suffix, The first sign is very like the Hittite sign for
'King', and the second like the Hittite sign for 'land'.
One is tempted to regsa the 1 as the suffix of the genitive
case and read 'King of the land'.
Another conclusion that may be drawn from these copper
plaques is that the signs used in our inscriptions are inde-
pendent of the accompanying animal design. Nearly all these
coins nave an animal design on the reverse, in some oases too
indistinct to determined but No. 30 has clearly an elephant,
while No. 31 has something quite different. But their
legends are identical. No. 43 has an animal looking like a
reindeer, with three plants or trees at his feet; no. 44
shows a hare. Other designs, as far as I was able to discern
them, are the bull4 (in 32, 33 the head is turned to look
backwards towards the tail), a tiger5 and a goat.6 The
----------- ------------------- ------------------------
1. In the analysis of Table A.
2. See pp. 31 et seq.
3. But see page 55 below Note 1.
4. Noe. 25-29, 32, 33, 48, 51, 53, 55, 57.
5. No. 60.
6. No. 61.
seals also witness to the mutual independence of the animal
designs and the legend.
No. 62. Terra-ootta seal.
This is the only example at Mohenjodaro of a terra-ootta
No. 63. Oopper seal.
An inoised piece of oopper, in shape quite unlike los.
24-61, some 4*5 om. long, 1 om. wide and '75 cm. thiok. The
insoribed face is flat, the back rounded. From the reversed
orientation d the writing on the original it was clearly
intended as a seal, and I have autographed it aooordingly as
from an imprint.
Nos. 64 to 123. Stone Reotangular seals.
Mostly of limestone or steatite. The inscribed surface
is flat but the reverse is convex, varying in thickness from
2-3 am. at the edges to 7-12 mm. at the centre. At the
oentre they are perforated breadth-wise by a single hole.
There is no aooompanying design either on the face of the
seal or elsewhere. The rectAngular pieces of stamped olay
(see Aos. 1-20) were probably obtained from seals similar to
these. It will be noted that on the seals, as on the oopper
ooins, the commonest final sign is V and the next com-
monest f (with variants).
Similar to Nos. 64-123; but not perforated.
Stmlar to Nos. 64-125. perforated; but with flat
Instead of convex reverse.
Similar to Nose 127-129; but inscribed on reverse as well
The top and bottom sides are blank.
Nos. 134-141, 145, 145, 147-153.
seals of the same type as Nos. 155-457, except that there
is no visible design acooppanying the script.
No. 139 is interesting, as being the longest inscription
hitherto found, and the only one running into three lines.
Noe. 142, 144, 146
have not got the usual perforated projection on the reverse.
Nos. 144 and 146 are peculiar as to size, and are correspond-
ingly thin, (about 2 am.). They are of the size shown in
Grey limestone. OCirular. Flat. Inscribed on face
and reverse. Unperforated.
Square. Surface flat. ScLes perpendicular. Thickness
from 5 to 10 mm. Reverse flat except for a perforated pro-
jection or attachment. Mostly white, yellowish, or light
grey in appearance, and ogmposed of limestone or steatite.
These seals are remarkably uniform in their proportions, and
appear to be of standard sizes. They are all accompanied
by the bull, standing in profile and facing right (See Plate
I, No. 390). One horn and one ear only are depicted.. The
bull in these seals is invariably of the European and not the
Indian type. The horn Us usually shown plain without the
parallel 'shading' show in Io. 390. Beneath the head
almost invariably appears the symbol the principal
varieties of whioc are giyeY on Plate I. This is the dis-
tinctive seal of both Haappa and Mohenjodaro, outnumbering
all the other seals. It wrll be observed that nearly half
of these seals end with the sign V
The peculiarity of this seal is that the boss on the
reverse side is InsOribed with the sign )
The face and reverse have the ordinary bull design. The
top and bottom sides are blank and perforated by a hole, for
stringing the seal*
0os. 441 to 509.
Square seals, similar in shape aend appearance to Nos. 155.
457, but with different designs aooompanying the legend.
441-456. Design, Indian bull (see Plate I, No. 4491.
457-475. Design, as in Plate I, No. 453.
476. The Indian bull, but in place of l we have the
symbol + apparently a plant or tree.
477. Design as in Plate I, No. 455. Inscribed on the upper
edge as well as on the face.
478-487. Design, an elephant (see Plate I, No. 478).
488-494. Design, an anmal resembling a rhinoceros. Before
the forefeet the symbol 'k .
495. Design, a boar (?) with 1=f .
496. Design, a beetle (?).
497. It seems doubtful whether the sign shown in the plate
Is intended as a legend. Accompanying it .s a three-headed
498. Design, a crocodile.
499. The left side of the square contains a tree; the
lower half a dog (T)
500, 501. Design, a tiger (see Plate I, No. 500).
502. Design, a deer.
505. Design, an animal difficult to identify.
504. The script is at the bottom of the seal, most of the
remaining space being occupied by a tree (See Illustrated
London News, 27-2-1926).
505. Below the script, from right to left appear a horned
lion, a horned man, and a tree. The lion and the man face
006. The middle spaoe is oogupied by a decorative design.
507, 509. Design, a fiotitious animal with two horns and
508. The left side is oooupied by a tree. The lower half
of the seal contains a tiger.
Jos.. 10-512. Oiroular seals.
510. Design, bull. Similar to seal in the Louvre (see
Introduction, page 3, no. 11).
Jos. 511, 512. Fragmentary. The design in each case
seems to have been a central oiroular body, from which pro-
trnded several heads. There would appear to have been seven
on Ho. 512. Of the four heads visible, two possess two horns
apiece, the third possesses one, and the fourth none. If the
remaining three heads (T) possessed 2, 2, and 1 horne respeo-
tively, we have here perhaps the beast with seven heads and
ten horns' familiar to the writer of the Apooalypse.
Square eoals, with animal designs, similar to those
already noted, but with the subjaoent animal facing left.
It is possible that here we hUre examples of engraversl
mistakes, as is not unknown in Mesopotamia, the animal and
legend being engraved as though for direct vision instead of
- D..-- -----. Pa---e I, ---o. ---a (T. 1).
1. Of. D.O.O.O. Planehe I, li. 7a (T. 13).
for viewing on an impression.
515. Design, buil. One horn visible. In plaoe of the
symbol we have an object ; apparently a plant.
Compare this with the signs in Ool. IV of Table XVII.
514. Bull as in PI. I, 390 (but facing left).
515. Design, an animal not identified, with one horn
516. An animal with two horns, spread thus 1
517. Design, the hind-quarters of a bull are visible.
It was noteworthy that at Mohenjodaro the inscriptions
other than seals were practically confined to copper coins.
At Harappa, however, while we have only one copper coin, we
h cai-^l^V'tiM r "'^.A
have a fairly large collection of inscribed That
they are not seals is shown both by the orientation of the
signs and by the nature of the incisions. They are, for the
more part, thin whitish slabs of limestone; very brittle,
and lese than 8th of an inc, thiok. It will be observed
that, while they contain few signs that do not also occur at
Mohenjodaro, there is a marked difference in the frequency
of certain signs and sequences of signs. But if these objects
are receipts it is pot surprising that their legends
should differ from those of the seals. ai particular we
may note the rarity of the final V in these texts; the
fact that nearly all these obIjcts are worked on the reverse2
as well as the faoe; the appearance of new shapes of certain
signs, e.g. the V is frequently written T The objects
are all flat as to both face and reverse.
1. The horns, when two are drawn, are always depicted frontally,
not in profile, but this is the only pair of horns showing
this particular shape.
2. If the reverse is not shown in the plates it is to be under,
stood that it ta blank, unloes the contrary is stated in those
3. Except as otherwise stated in these notes.
1eo. 28, 29. While the face is flat the reverse is
No. 25. On the reverse a crocodile.
No. 35. Square seal. About 8 mB. thick. Perforated attach-
ment on reverse. Beneath the legend are a few indeterminate
scratches. This text belongs to the group Nos. 125.245.
Nos. 40, 41. Cylindrical in shape. The space to the
left on the reverse sides is occupied in the originals by the
Symbol placed horizontally. -OE
No. 42. Cylindrical; a hard dark-coloured stone. The
reverse shows a divinity in a shrine (T).
No. 45. Cylindrical. Reverse: a tree.
No. 55. Reverse, a crocodile, with the sign held
vertically in its jaws, and accompanied by the sign 5
written horizontally > in each corner. This would seem
to establish definitely that the sign I is a fish, and
a differently written variant.
No. 61. The signs are about three inches long on the
original, which is a fragment of a large circular stone that
may have served as a door-socket.
Nos. 02-83. Small three-faced prismatic objects of limestone.
Unperforated. All three faces are worked (except in the case
of No. 80) bearing either inscription or design or both. They
are shown complete with design on Plate XXX.
Their significance is discussed in the analysis of
Nos. 84-86. Copper.
No. 84. A copper coin, similar in shape to those found
No. 86. A broken slab of copper about I inch thick.
No. 86. The signs show in the plate appeared on a copper
dagger about 5 inches long. There were several other copper
daggers in the Museum, but they had not been cleaned and so
NOs. 87-122. Impressions on olay.
No. 87. Three-foed prism, two faces of whioh are covered
by a single pair of signs.
No. 89. Reverse, a plant .
No. 90. Cylindrical. The space to the left of the legend
on the reverse is occupied by a plant.
No. 92. Reverse, a plant.
No. 95. Reverse, the space to the left is occupied by a
bull ith two horns standing over the symbol.
No. 95. ace, the space on the right is occupied by what
appears to be a hare. There are six of these slabs all found
together, identical in all respects, including a pronounced
twist that was given to the slabs before burning. P r this
it is olear that a number of these slabs were prepared, impres-
sed with the same 9sal and then baked together. These stamped
clay slabs, manufactured en masse and bearing the owner's
name on face and reverse can only have served as votive tablets.
Doubtless they were placed before the family god to keep him
in mind of the householder's prayers.
No. 101. Fragment of a ring. The legend is on the conoave
No. 102. cylindrical. On the reverse is a centipede.
No. 105. Two identical specimens. The reverse contains a
design that recalls the VI
No. 107. This gives a clue as to the nature of the
aotif. It is clearly a religion emblea or offering that oan be
carried in procession like a standard. Compare Plate I (M.) 18.
No. 109. Reverse: a plant.
-- Plat slabs unless otherwise stated below.
Is Plat slabs unless otherwisee stated below.
No. 110. Faoe: the spaoe to the left is ooupied by a
human figure with tail, standing extreme left and facing
right. Facing him is a seated figure with raised arms and
long hair. Reverse: the space on the left is occupied by
two felines, standing on their hind legs and facing one
another. The space on the right is oooupied by a man seated
up-side down. Suspended from his legs is a large insect.
No. 112. Reverse: in the space on the extreme right is a
No. 115. Glyindrioal.
Ros. 116, 119. Cylindrical. Reverse: a crocodile.
No. 121. Reverse: gonvex.
No. 122. The lower half of the face shows a bull.
Nos. 125-157. Limestone and steatite. Inscribed on the
face only. The shape is as shown in the plates, except that
los. 126, 130 and 135 are squares. The rectangular shaped
seals have convex backs, as in the similar seals from
Mohenjodaro (M. 64-185), and they are similarly perforated.
No. 129 shows one of these. The 'o' in the middle of the
reverse is not a sign but the hole that perforates the back
of the seal. The. square seals have the usual ring, attachment.
No. 137 is of black marble.
Nos. 158-149; 158-161. Inscribed on the reverse as well
as the foe. They are not perforated and are similar in
appearance to the inscribed objects Noe. 1-00, to which grotu
they belong. They are not seals.
Nos. 141, 144. Four-faced prisms.
No. 145. Faces in the space to the right appear five
swastika signs in a row. Reverse: in the spaoe to the left
appear a man and a tiger.
No. 150. A square seal. The face contains a bull but
no legend. The reverse is blank.
No. 151. A square seal of black marble.
los. 162-227. Square seals. Mostly limestone and
steatite. Perforated boss as back, same as Mohenjodaro.
Design eaotly as on the Iohenjodaro seals nos. 155-457.
(Plate I, no. 590). We also note the same sign sequences as
at Mohenjodaro. Olearly the same language as well as the
same script prevailed at both places.
gos. 228-231, 255, 254. Rectangular seals. Flat. Worked
on face and reverse.
No. 229. Reverse: tortoise (T).
Nos. 228, 250, 231. Reverse: crooodile.
Nos. 252, 255. Stamped olays eyl 4ndral in shape.
Nos. 256-859. Square seals, like os. 162-227, but showing
a bull with two horns.
No. 240. Squae seal. Design: elephant.
No. 241. Fragment of a square seal. The spaoe to the
left contains seven men in a row, each holding the one in
front of him by the hand. The men are looking to the right.
No. 242. To the left of the script (T) is a tree, to the
right an animal.
No. 243. Flat square seal* Reverse also flat, no ring
attachment. It is also without the clear out rectangular
sides of the ordinary seal. It is perforated throughout its
breadth by a hole. It thus resemble the archaic seals of
Mesopotamia. It is doubtful whether the sign on this
seal is anything more than a decorative device,
THE DIRECTION OF WRITING.
The orientation of the proto-Indian script is, in the large
majority of cases, from right to left, i.e. the signs are placed
successively in a horizontal row starting from the right. Evi-
dence of this is afforded by a comparison of the sequence of the
signs in texts containing two or more lines on the same face, with
the sequence in single-line texts. Attention may first be
directed to the single-line texts containing V as their left-
hand sign. Of these there are 177 at Mohenjodaro and 31 at
Harappa (see the Plates passim, but especially V, VIII, IX, X, XI,
XII, XIII.) It is clear that a large proportion of our texts -
nearly one third either begin or end in V Now examine
M. 303, 516, 391, 365. In M. 303 V being the only sign in twh
second line is clearly the last sign. If then we read the script
from left to right we must place V at the extreme right of the
text and read j ( V which gives us the sequence
f V which is found nowhere else;
whereas if we read the script from right to left and place V
at the extreme left we get V 6 etc. a sequence of four
signs which occurs no less than five times elsewhere M. 184; 89:
124; 9; and H. 90. while the three signs V 6 occur in
a dozen other texts (see Table I nos. 49-65). Treating M. 516
the same way we get V I T ) which not only gives
us V in its common position but also the sequence ~
Now it is significant that the only other occurrence of the sign
K4 viz. M. 447, shows precisely this sequence. There can
be little doubt then that both the lines in M. 516 are to be read
from right to left (starting of course with the upper line). It
is not to be inferred that the second line is always to be read
from right to left. Cases of boustrophedon writing, though
apparently rare, undoubtedlyy occur. M. 391 is a case in point.
While the upper llke reads from right to left the lower one reads
from left to right. This reading gives us V CC~I II
No other reading is tenable in face of the
evidence of M. 161; 162; 462; taken in conjunction with the evi-
dence of Table LII, which shows V CC nine times and OCV
net once. No. M. 365 however is clearly not boustrophedon.
That the second line in this text is to be read in the same
direction as the single-line texts is clear from the sequence
V WI which is found eleven times, while V V is
nowhere found. The two lines of U. 368, then, are to be read in
the game direction. That this direction is from right to left is
indicated by the position of VV which in single-line
texts is found almost invariably as a left-hand group. (See Table
VI). We may now examine the other inscriptions containing mere
than one line on the same face. M. 139 is our longest in-
scription containing three full lines of script. Each line is to
be read from right to left. In the case of the first line this
is proved by the sequence A which is one of the commonest
sequences in our texts, occurring twentyone times (see Table
LZvXmF). In the case of the second line it is proved by the
sequence Vf f to which we have already referred; and in
the case of the third line by the sequence ff which occurs
elsewhere five times, while its reverse kf T Is nowhere found.
Regarding M. 141, the position of P as a right hand sign
makes it probable that the first line reads from the right.
Regarding the direction of the second line there i@ no evidence,
as the signs thereof are nowhere else found in association.
M. 151. The first line shows a sequence normal in single
line inscriptions (see Table XXV), and therefore reads from right
to left. Regarding the second line the evidence is scanty.
0 and Y are not elsewhere found together, and the only
instance where P and ( a&e found is H. 44. But as shown in
the analysis of Table LMVI X and X are not variants of the
same sign. However, they probably represent allied sounds, as is
explained later, and it is possible that the X V of H. 44, and
the V X of M. 151 are the same word with a dialectal variation
of pronunciation. There are many such instances of dialectal
variations recorded in the script, as we shall see. Provisionally
then I have assumed that we have in H. 44 and M. 181 the same
word, and have accordingly read the second line of M. 151 from
left to right.i
M. 162. The first line is from right to left. This is*
clear from the four signs on the left, a sequence we have already
examined under No. 391. The second line also reads from right
to left. If we read it otherwise we have Ca final pre-
ceded by V which Is nowhere found, whereas V W is found
in seven other cases. (see Table XC.)
M. 450. The sequence 0 on the right of the first line
is one of the commonest in the script. It occurs in this
position in single line inscriptions thirty times, or if we
treat 0 as a variant of 0 sixty-eight times (see Table
XXIV). It Is clear then that the second line is to be read to
the left, not to the right, of the first line, therefore the
reading of all single line inscriptions with "l or "0
on the extreme right are to be read from right to left. Taking
these inoriptions together with those ending in V we have no
less than 247 inscriptions which demonstrably read from right to
left. This may be accepted as conclusive evidence of the normal
direction of the writing at Mohenjodaro and Harappa, at least as
----------------------------- --- n----------- o-m-m- mewm-- m---- n- -
1. In the tables I have written out the texts with more than
one line as they would have been written had the scribe
placed all the signs in one line. This was essential
for purposes of comparison. The reader can readily
discover whether any given text in the Tables has more
than one line by referring to the list immediately pre-
oeding the Tables.
regards single-line incbriptions and the first line of multiple-
line inscriptions. It remains to consider the direction of the
writing of the second line in the remainder of the inscriptions
with more than one line, i.e. to determine how many of them are,
like Hittite, boustrophedon. In regard to no. M. 450 there is
virtually no evidence. V is never found followed by I
and only once by ( (M, 40) and that in a context where the
latter sign clearly associates with the sign preceding it and not
with the 4 Still in this, as in all cases where no evi-
dence is obtainable from sign sequences In other texts, I have
for purposes of transcription assumed a right to left reading, as
this is the reading on the majority of second lines where the
direction can be determined.
M. 193. Not bgustrophedon, as is never found as a final
sign. (see Table XXI).
M. 230. Not boustrophedon, since (i) 1 is never final,
(ii) Af V medial is found in M. 355, (ill) : : final is
found twice (see TableXCII).
M. 341. Probably boustrophedon, since )) is often final
(Table LVxI) while H only onoe (Table LVI). Neither sign is
found elsewhere following so that the evidence is very
M. 232. Boustrophedon, since + is a common sequence
(see Table XLVI and its analysis).
M. 417. Not boustrophedon, since(4) is found elsewhere
followed by and Y is found similarly preceded by '1
while no element of a sequence i"tY 6) is found anywhere. .
M. 447. Not boustrophedon in view of the sequence 0
Cf. M. 516 discussed above.
M. 453. Not bogstrophedon, since 1111 is never final (see
TableXXXI) whereas :D: is, as already noted.
-. See Tables -- -- -------- -------, X----VI--
1. See Tables ZI, XXVI,
M. 477. The third llne is from right to left in view of
the sequence HIW 00 (see Table VIII).
M. 499. Not boustrophedon since 4 is final in M. 508
M. 506. Read from the right ID V"
M. 514. O is unique. But it is probably a defective
form of The latter is not found elsewhere associated with
W but it does appear following A in M. 11S. The evidence
is thus very slender, but such as it is, it points to a
H. 107. The top line reads from right to left. Cf. M. 20
S* The bottom line is probably bougtrophedon, giving
with the reverse, which is from right to left, the sequence + I
which is fairly common. (see Table XLVI).
H. 241. The evidence is practically all. If 0 is a
variant of the g group then a comparison with M. 366 would
suggest a right to left reading.
I. 24. The sign in the second line is clearly to be read to
the left of the first line, and the signs In the third line to
the left again. This gives us V V % as our final
sequence. We have this sequence in H. 58; while V 1 is a
ooEgon final sequence as already noted.
M. 235, 237, 245, a53, 409, 492, 508, I.19, H. 166, all
contain a single sign in the second line. In every case the
first line is to be read from right to left nd the sign of the
second line as the final sign. It will be noted that the second
line sign in nos. 1. 235, 409, I. 19, is a variety of ,
which in the single line inscriptions also is invariably final
(see Table XXVII nos. 1, 3, 4); in nos. 246, 492, it is l ,
which is also normally final (see TablexCIS, which is further
ponfirmation that the script reads usually tfos right to left.
M. 133 is interesting as containing not only two lines on
the same face, but legends on three other faces. By a comparison
of the sequences with those found elsewhere the reading can be
established as fellows. Begin with top line of face, read right
to left. Then second line of face left to right. Then reverse
right to left. Then right side left to right. Then left side.
It will be seen that the reading is boustrophedon throughout,
Another peculiarity is that in the lines where the direction of
the writing is reversed (i.e. left to right) the form of the non-
symmetrical signs is also reversed, on the Hittie plan. Thus we
have for ) (see Table LSZlU) (AA( for.)AA)
(see H. 227).
The consideration of this inscription brings us to our next
category of multiple line inscriptions viz. those that have only
one line on each face, but have more than one face inscribed.
M. 132. Clearly boustrophedon. The face reads from right
to left, while the reverse is clearly left to right, being the
last three signs of the face reproduced in reverse order. Again
the direction of the writing in\ease is proved by the fact that
elsewhere T is invariably final (see Table LTV).
H. 118. Boustrophedon. The face reads left to right, the
reverse right to left. This is less surprising at Harappa than
it would be at Mohenjodaro in view of the fact that at Harappa
mny of the single line inscriptions read from left to right.
In the remainder of these inscriptions the writing is in the
sage direction on each face and that right to left for the most
papt. Those in which the writing is from left to right
(principally Harappa) are so indicated in the Tables by placing
an asterisk against the inscription. For the most part the in-
soriptlons on the different faces seem to be independent of one
another. This is clearly the oase in no. ). 132, noted above,
where the inscription on one side is an abbreviation of that on
the other. An extreme form of this is M. 439, where the sign on
the reverse seems to stand much as an initial does to a name.
Again in some cases the Inscription on either side is identical,
viz. M. 16, 18, H. 145. A large proportion of the inscribed
objects at Harappa have VIII or VIIll on the reverse. It is
clear that in these cases the reverse has no syntactical relation
with the obverse. Returning now to the inscriptions with two or
more lines on a single face: only in two instances M. S03 and 391
have we reason to suppose from the sequences that the signs in the
second line form part of the word or phrase in the preceding line;
while in some cases, notably 6. a89, 193, 230, 453, it is almost
certain that the sense of the first line is complete in itself,
and that what follows is an additional name or title.
No. 139 indeed looks like a Sumerian 'burgul' seal, a seal
with the names of three different men (perhaps as In Soier,
fashioned for the purpose, combining the names of the parties to
a contract in a single seal). It is significant also that this
seal alone of all the square seals bears no glyptic design, whIoh
again recalls the Sumerian contract seal.
It remains to remark that at Harappa there are several in-
stanees of single-line inscriptions reading from left to right.
At Mohenjodaro there are only two (M. 513, 515).
IHE CONNECTION WITU OTIER SCRIPTS.
The discovery of any new script at once suggests a search
among existing scripts for possible ancestors or descendants. In
pursuing this search one naturally first directs one's attention
to those scripts which are (a) contemporary in date and from which
there may have been borrowing, and vice versa, (b) those which ape
found in the same locality at an earlier date, (c) those which are
found in the same locality at a later date. In the present in-
stance category (b) is entirely wanting. In category (a) we have
3Suerian, Proto-Elamite, Egyptian and Minoan. In category (o)
we have Kharoshthi, Brahmi, and Sabaean. With regard to Kbaroehthi
its descent from Aramaic is proved. Net so, I think, in the case
of Drahpi. It is true that Buhlerts derivation of the Brah
syllabry1 from the Semitic scripts has long held the field.
But it was never universally accepted. Cunningham in particular
believed it to be derived from a lost pictographic source. A
detailed refutation of Bubler's equalisations seems unnecessary in
view of the positive evidence set forth in the Comparative Table
(Appendix I ). It will be seen that I accept certain of uBiler't
equalisations with the Phoenician, but these are precisely the,
cases where it seems that the Phoenialan signs themselves are
probably derived from Proto-Indion. Now it may be argued that
the interval of time between the disappearance of the civilization
of yohenjodaro and the first appearance of Brahmi (c. 300 B.C.) is
too great to make a direct descent probable. But what do we know
concerning the lower limits of the Proto-Inrian civilization?
The bricks of the Buddhist stupa at Mohenjodaro lie immediately
. --r-----r----------r----------~------------------ ----
1. It is Incorrect to speak of the Brahmi characters as
alphabetic. No signs except the vowels stand for single
upon Proto-Indian remains. Nothing has so far come to light to
suggest that the Proto-Indin civilization came to an end before
the Aryan invasion. And it aust be remembered that the script
that we possess is all pojomnental seals, sealings and coins.
It is quite possible that alongside of this there may have been a
demotic approximating more closely to the script of the Bran coin
and Asoka inscriptions.
With regard to Sabaean the time interval is less. And
though the inscriptions may not antedate the sixth century, a much
earlier date is claimed for the beginnings of the Minaean empire,
and presumably for the origin of the script also. If distance is
urged as militating against the probability of Babaean being
derived from Proto-Indian, it should be remembered that the dis-
tance from the mouth of the Indus to the Sabaean coast is less
than 1000 miles, that the monsoon winds are absolutely reliable
and sailing conditions ideal, making it possible during six months
of the year to sail from Karaohi to Aden with the shore almost
continuously in sight without tacking once, and during the other
six months to perform the same feat in the opposite direction.
Again, both areas were known to the ancients as Ethiopian. In
view of the fact that both the form and the names of some of the
Sabaean signs have not yet been satisfactorily amounted for, it
has seemed to me legitimate and desirable to bring out in tabular
form the undoubtedly striking resemblances between Babaean and
With regard to contemporary scripts:
Many of the signs bear a remarkable resemblance to the mona-
mental script of Ancient Egypt. The entire body of anthropo-
morphous signs have Egyptian equivalents which are virtually
exact. And it is interesting to note that not one of these
anthropomorphous signs have the remotest parallel in Sumerian or
Proto-Elamite. On the other hand there are many of our signs
that are exactly paralleled in the Proto-Blapite and Jemdet-Nasr
tablets, such as I that have no conceivable morphographio
equivalent in Egyptian. One is bound to conclude that the pre-
sumption is strong that our script has been borrowed in part from
Egypt, and in part from Mesopetamia.l* Of course there is a
considerable proportion of signs that are common to all three
scripts, auch as the signs for tree, fish, bird. But this is
coincidental, and indeed inevitable in the very nature of picto-
graphy. It is only safe to draw inferences of causal connection
where the less obvious and more conventionalised ideograms,
especially those that are so conventionalised that their picto-
graphio origin is hardly determinable, show a marked correspond-
ence; and in a lesser degree, where easily recognizable pioto-
graphs show the same variations. Now the latter is very marked
as between our script and Proto-Elamite, as will appear from a
study of the Comparative Table.
The resemblance of our script to Proto-Elamite is closer than
its resemblance to Sumerian. This is natural in view of the
geographical proximity of Baluchistan to Slam. The resemblance
to Sumerian is not really apparent till we reach the Jemdet-Nasr
period. Now the script of that period (B.C. 3500) is so closely
related to Proto-Elamite that Professor Langdon affirms a common
ancestry of the two. This would seem to be confirmed by the
evidence of our script, which approaches the Sumerian in
similarity in measure that the latter approaches Proto-Elamjte.
One is led to the conclusion that the element in our script whioh
was borrowed from Mesepotamia was borrowed at a period before the
1. This is just what we should expect, if, as has been
suggested in the Introduction, our people were a race
of overseas traders, like the Phoenicians.
separation of the u8merian and Proto-Elamite scripts. Of course
it is possible that all three had a common ancestry, and that the
Egyptian element in our script alone was borrowed. It is even
possible that all four scripts may have had a common origin.
But this is an enquiry that does not concern us here, and which In
the nature of pictography, would be very hard to solve without the
aid of anthropological evidence as to whether or not there was in
prehistoric times racial affinity between the inhabitants of the
tile, Euphrates and Indue valleys.
The connection between Proto-Indian and Proto-Elamite is so
close that Professor Sayce has suggested that the languages may be
allied.l' This I have endeavoured to teat. There is no doubt
that our texts are entirely proper names (and titles). If the
languages are allied we may expect identity of some at least of the
proper names. Now in the Proto-Elamite tablets it is possible to
detect the proper names with some degree of certainty: see the
analysis of Tablet No. 490 by Father Scheil on page 30 of Vol.
ZVII Mlmolres de la mission archoologique en Perse. Applying his
method I have collated all the proper names occurring on the tab-
lets in this volume and vol. VI containing certain signs that could
be reasonably safely identified with Proto-Indian signs to see
whether in any case the same sequence of signs could be observed.
The method adopted was the same as that adopted in the preparation
of the tables of Proto-Idian texts. The signs selected as
possible equivalents of Proto-Indian signs were ,K'X',X
C: 0 i the various bird signs 0, T
,0,1,A,A l,4,x> 9,3,41 -.P, f,1,WXA, X fIand their variants.
Every occurrence of each of these signs in all contexts that could
1. See Antiquity, June 1927, p. 206.
1. See AntigaitZ, Juno 1927, p. 206.
conceivably be proper names was tabulated. The result was that
out of 355 occurrences the only sequences discernible that tallied
with those of our texts were:
" ) XVII. 34. of. M. 18. J 00 XVII. 73. 5. ) f. H. 53.
SVI. 373. 4.
I XVII. 75. of. H. 137.
4I # cf. Table XLVI. XVII. 17. 1.
of. M. 145.
This is less than might have been anticipated as the result
of mere coincidence, and infinitely less than we should have ex-
pected had there been any causal connection between the scripts.
Indeed the evidence is in the opposite direction, for there are
sequences coitaki.ng signs that are common to both scripts, which,
found frequently in Proto-Elamite, are absent from Proto-Indian,
and vice versa, e.g. M 0 In Proto-Elamite;
St, II f OD in Proto-Indian. It is then fairly
certain that while the scripts are allied the languages are quite
distinct since they have not a proper name, and scarcely the
element of a proper name, in common.
A survey of the possible affinities of Proto-Indian with
Hittite and Minoan is not included here, not for lack of super-
ficial resemblance, but for lack of space and time, and because it
was deemed better to investigate the apparent affinities with
scripts which were already very fully deciphered. An exception
has been made in the case of Proto-Elamite on account of its
proximity both inr time and place to Proto-Indian. The inclusion
of Cypriote in the comparative table was made on the principle
that at this stage of the work of deciphering Ppoto-Indlan it was
desirable to include in our comparative survey all independent and
deciphered scripts, Chinese has not been included because after
a study of the articles of Mr. Hopkins in the J.R.A.S. supplem-
ented by a visit to view his collection, and especially after
receiving the considered opinion of Mr. Hopkins who spent a week
in examining my autograph copies of the Proto-Indian texts, I
concluded that the relationship, if any, was too remote as between
Proto-Indian and the earliest Chinese of the Honan bones, to
warrant a detailed investigation at this stage.
My conclusion is that the Proto-Indian script is connected as
to its origin with Egypt on the one hand, and Sumer-Elam on the
other; that the script is, on the majority if not on all of our
texts, a simplified syllabary of open and closed syllables,
roughly 250 in number, many of them constituting complete words;
that from the open syllables of this script are derived the
Brabmi quasi-alphabetic script, and a large portion of Sabaean;
that it is quite possible that Phoenician and Cypriote are like-
wise modifications of Proto-Indian, which however presupposes a
common meeting ground of their sailors and merchants in the
Isthmus of Suez and the mines of Sinai at least evidence of any
such intercourse at this point would assist in deciding whether
the morphographic resemblances are coincidental or not. This of
course reopens the question of the origin of the Alphabet, and
suggests that Proto-Indian was an all-important link in the chain
of its development from pictographic origins.
Analysis of The Tables of Signs.
Analysis of Tables I and VI.
Lt will be shown on the completion of the analysis
of these tables that we have only 234 distinct signs, apart
from compounds. Now the Brahmi script makes provision for
33 consonantal and 8 vowel sounds (i Ciaherent) i, i, 1, u,
i, 5, o). ono a syllabary consisting of 33 consonants each
articulated with 8 following vowels would give us 264 signs.
The number of syllabic signs required to form a simple
syllabary of open syllables to represent Brahmi sounds 5V
closely approaches the number of signs on our Texts that we
may be moved to assume that our script Is mainly a syllabary
of this kind, as a first working hypothesis; provided of
course we are previously impressed by the evidence of the
Brahmi signs being derived from the lPrto-Indan.1 But this
hypothesis has not been assumed before first investigating
the script to discover whether an ideographic conception was
tenable. It is not. There is clear evidence in Table I
itself of the presence of phonetic elements. We may first
take the sequence V V Of this V "V will be seen to
be a simple variant. V f (variant (, (. ) is clearly
closely allied. For V V is Zollowed by f final or
quasi-final2 in every case save one (T. VI. 31) and U with
1. See evidence of Comparative Korphographic Table.
It is not to be inferred that any relation between the
language of the proto-Indians and the Arya. of the Asoka
edicts is implied. Sanskrit and Pall and the other
Prakrits had by tVis time absorbed the phonetic elements,
notably the cerebral sounds of the Dravidian population.
2. The E in Table VI Nose 18-21 is an independent suffix.
See analysis of Table LVIII.
its variants is astilarly followed in every case save one
(T. VI. 5). Now if we oompare V V, V 7 with V preceded
by other signs, we *hall find that of the 17 signs and sign-
groups found immediately preceding V V, V V no less
than 16 are also found immediately preceding V alone,
and that frequently. Compare T. I. Doe. 6-20o 22-38, with
Tables LV, tXV, X30V1,1 XCII, LIX, XIII, LI, XLVI, XXXI, XXXIV,
VII, XVI. In these 16 combinations the proportion of
occurrences with an intercalated V or U to those without
is as follows:
3:81 37 if'7 1:41;
1:7; 2t; 2:1; 1:4;
Illli 1:20; 3:5; t| 7:2; 1:1.
There seems good reason to conclude that V V 'iV is a
'spelling-out' (as we so frequently have in Sumerian and
Assyrian names) for f It is probable that V whichh
is so often final, is an open syllable-. The principle which
we see in Brghmi of regarding the simple form of every sign
as containing as inherent final S; and the fact that to this
day in the Indian vernaculars words that we should regard as
terminating in a consonant (from their pronunciation) are
always regarded by the Indian grammarians as possessing a
final l and written accordingly; and the fact that in Sumeriak.,
words appear to have been similarly so regarded, since the
Sumerian never taW lugal-na, lugal-ka but always lugala-na,
1. See Analysis of Tablo XXXIV for identification of "!
2. It is possible that this bird sign is a yariant of the
bird sign in No. 21 ea Table I. In that case the propor-
tion will be 2:1, and incidentally all 17 of the signs
preceding f u ,' J will have been found precedingV
3. For identification of 1liU with 1lill see analysis
of Table XXXI.
lugala-ge etc., should make us be prepared to regard signs
which are normally final (as V is) as open syllables:
while a sign which like V U is never final we may
provisionally regard as a closed syllable. If then V i
is a 'spelling-out' it is something in the nature of ak-ka.
Whether this doubling of the consonant in the script had any
counterpart in pronunciation (as in Assyrian) or not (as 1n
Sumerian) is difficult to say. If it had it may well have
been due to the quantity of the preceding vowel. The
combination \ t is peculie.ly interesting. Not only
from its appearance, but from the fact that it is always
followed immediately by V final, we may be sure that it
is a compound and that one of its elements is V The
other element is clearly # (see Table XVI). It is
equally certain that this compound is phonetic and not ideo-
graphic. If.it were ideographic, then by all our knowledge
of ideograpnic writing its meaning must necessarily be
different from How then shall we account for its being.
found invariably in the same circumstances as B ut this
is net all. V and U though closely allied as shown
from their relationship to V are not actually variants.
This is clear from the regularity and difference of their
anteedents. If we take I' as ak we may take as ek .
Probably the selection of one or other of these syllables in
the 'tpel]ing-out' process was influenced by the quality of
the vowel of the preceding syllables, a principle common to
Sumerian and many languages vowel harmony. Now it is surely
most significant that these same alternations of .1 and (
are observable in the compound formed with If then
------'C---C---- ----- -- -------- --I
I. It is of course to be understood that the selection of
any particular consonant or vowel for purposes of illus-
tration Ln the analysis of the Tables is arbitrary. For
the selection of vowels see Analysis of table XXIX.
there is a difference of initial vowel as between V and
V there is clearly a vowel difference between and
SBut if 4 be a full syllable in itself, as
must be presumed, and if that syllable is fully pronounced
in thn compound, then, with this constant syllable intervening,
the carrying of vowel harmony over and in spite of it on to u
would be incomprehensible. But suppose that the syllable W
ia, on combination with V truncated, that it loses its
vowel, that ba-ak becomes bak; ba-ek, bek. Then everything
is explained: the syllable bak'has become b@k under the
Influence of something antecedent. Tn other words the com-
pound represents the contraction or an open and closed syllable
into one 'compound' syllable, and the first element in the
compound has been reduced to a mere consonant, it has lost its
inherent 9, it is what the Sanskrit grammarians all hlant.
This is precisely the principle governing the formation of
compound (Samtukta) signs in Brahmi and Nagari to this day.
If we Hre right the sequence \f 8 (which twice occurs)
is to be read ba-ka, while V 3* is b'ak-ka; a mere
graphic variant as in Sumerian.
With regard to the sign V and its variant-form V ,
the latter is probably original, and may be taken to represent
a pair of arms with hands. This is one of the signs that
shows affinity with Egyptian. See Gardiner, E.G. p. 445.D.28.
The sign V which as we have shown is but IV articulated
with a different vowel, is morphographically so akin to V
that it may well have arisen from it. This would imply the
deliberate differentiation of signs to supply cognate phonetic
symbols. There is abundant evidence of this elsewhere, as
will be noted in the analysis of other Tables. In the present
instance this differentiation will have been made by adding
to \V a horizontal stroke in each half, producing V
The further modification to V ,P is probably of the
gunu order, and without effect on sense or sound.' The
deliberate modification of a given sign to provide a symbol
for a cognate phonetic value would presumably arise first in
the case of syllables which, not forming a complete word in the
language, or forming a word that was difficult or cumbersome
to express ideographically, could not be written otherwise.
It is an intelligent device that the cuneiform users seem
never to have taken, being content to the end to represent
e.g. ah, ti, uh by a single symbol. As far as we know Proto-
Indian would appear to have been the first script to adopt this
device. It is not without interest to observe that Ethlopic
and Brahmi have the same traits. With regard to the shape of
If it probably represents a vase or jar with two handles,
the upper horizontal elements representing the lips of the
vase, the lower its handles. For the variety of its shapes
and its Suaerian and Egyptian affinities see the Comparative
With regard to the meaning of V at any rate of Vf
final, we may say that it is an affix. That it is an affix is
suggested (1) by its normal position at the end of the text,
(2) that it is preceded by well defined sign groups which there
is reason to regard as complete words, either names of gods
used in the formation of proper names, or titles, (3) that
when it is found in the body of the text it is normally
preceded by precisely the same combinations. That it is a
---r-------------- ----------------------------- ---
1. The symmetry of Proto-Indian signs is one of the character-
istics of the script. It is in harmony with the artistic
sense of its users, so abundantly exemplified in their
glyptic designs on these very seals. In the modification
of signs this symmetrical principle was continued, each
equal portion (whether or .) of the sign receiving the
same modifying strokes. See Tables V, XV, XIII, TXXIV,
LXXV, LXXVI, CII, CVII.
1. As often is the case with tumerian gunu signs.
suffix which is not a determinative is probable for the
following reasons: (1) If V be a determinative its
frequency indicates that it is one of a very wide class.
'Man' and 'scribe' ape the only two that see=t possible.
But if it is either of these how do we account for its presence
on the copper coins where we should expect rather the
deterwItnatives of king or ruler? If we reply that the
determinative a man was probably used after all men's
names whether rulers or Dot, then how do we explain the fact
that a large number of typical square seals end in op or
whloh, as is shown in the analysis of Tables XV and LXVI1I,
stand in exactly the same relation to their antecedent words
as Vf does to its antecedent words? So that if fV
is a dterrin.attve then they also are determinatives. ojc
'f V is a determinative after men's name it is only one
of several, and it would be difficult to account for Its
prevalence on the coins, in place of one of the more dis-
tinctive determinatives. While if we are right in decphering
one of these coins 'King of the land', V would have to be
regarded as a determinative either of 'king' or 'land', which
in view of its prevalence on the seals, is impossible. So
nuch for the negative evidence.
(2) That f is a suffixed element in name-formation is
strongly suggested by a comparison of Tables 1 and LXVIII.
It will be seen that 9 like V is normally final.
Like V, if followed by any single sign,it is followed by ,E
Like V it is preceded by well defined sign-groups that
clearly constitute words. But, the three distinguishable
words that precede via., + T( ,
ocCurring as they do 26 times are never once found precedingV ,
while of all the many sign-groupe found regularly preceding V
not one is found preceding Are we to assume that all
the men whose names ended in, say., Enlil, Iannar, -mansum,
were leather workers, and 01 other men whatsoever were scribes?
For that is the position to which we are reduced if we insist
on regarding IV and a as determinatives.
We must now consider the forms 1J etc. Table I, Nos.
348-400, Co'. IV. The first thing we notice is that these
forms are never found at the end of a text. Secondly we note
that they are often found with the same antecedents as V .
Compare Nos. 969-273 with 346-353, 375-6, 387; Nos. 49-65 with
354; Nos. 164-168 with 539. Nos. 243-245 with 360; No.309
with 364; No. 321 with 36~1 No. 43-44 with 367-388; 350-331
with 369; 157-163 with 372, 376; 195-197 with 373, 398;
290 with 382; 138-149 with 392, 399; 215-217 with 400.
The example VY/), 'fIJI/), "Y) '/ ) has this
peculiarity: it is the one combination ogpionly found with V
in which i is not f.nal. In all the other combinations
with V ,tie V is final in the totality or large majority
of occurrences; with Y/) it is not once final, but on the
contrary, in all five occurrences the combination is initial.
But I doubt if this signifies anything more than that this
combination is a name (of a deity?) that lent itself to employ-
ment as an initial element in the formation of proper nams.
When we find i') it ti the same word with a chang of
vowel in the final syllable. In the ease of this word V
would appear to have its normal use as a suffix, and consequently
V~ 4 v also. But there is no reason to suppose that
in their other occurrences the V group are other than the
syllabic element of roots. It is significant that the great
majority of combinations comonly found preceding V' are
not found preceding the V group (i.e. the signs in the
a i f lu, l u. &$0 V k&& 60A" w*i4 *l'
4th column of Table I, pp. 6 & 7). Thus both the form of the
signs, which suggest deliberate differentiation from V ,
and the circumstances of their occurrence combine to show that
they are syllables allied to but not identical with V
Taking this evidence in conjunction with what has been
observed concerning the modification of V we may assume
as a working hypothesis that both in the case of open and in
the case of closed syllables signs were modified by the
addition of short straight lines to represent syllables con-
taining the same consonant but a different vowel.
We may now consider the function of certain signs that
follow V when the latter would otherwise be final.
These are T 1 and A
Now E follows not only V but (which we have seer
is functionally similar to \f ) and a miscellaneous collection
of signs (see Table TVIII). It is probably a suffix.
Allowing for the difference in the number of inscriptions as
between Mohenjodar4> and Harappa this sign is proportionately
seven times as frequent in Harappa, where it appears on 77
texts as against 20 at Mohenjodard. But these are mostly,
business receipts (see analysis Table tVIII). occurs
twice after V and four times after other signs. It is in
every case final. It may be taken as a determinative. (See
Table LXI= ). I /\ in 9 out of its 10 occurrences (aee
Table XI) is final. It follows V 3 times, and of its
1. A further proof that they are not identical Is that V
is found on one and the same seal in conjunction with
other members of the group. it will be observed however
that of the other members no two varieties are found on
the same inscription suggesting that they are mere
variants of each other, or phonetically interchangeable.
This is further borne out by the presence of the same
sequences with different members of the V group, which
are not found with V Cf. Noe. 361 and 381; 355 and
386; 357 and 390; 372 and 378.
other antecedents one is It may be taken as a deter-
minative. ( is final in 6 out of 7 occurrences (see
Table CIV). It follows V 4 times, and 4 twice. It
is probably a determinative.
When final is preceded by V 1l times out of a
total of 15. The proportion is so high as to suggest that it
may stand in functional relationship to the V it
follows. It is perhaps the determinative of the word 'servant'
( V ). That in point of grammar and syntax the
combination of '\f appears to hold the same position as V
simple is suggested by the fact that, like V when final it
Is liable to be followed by the determinative suffixes I and
/, 8 (see Table =LIX, 11, 14).
We may now examine the condition of that which is
neither final nor quasi-final, but truly medial, being followed
by several signs which are clearly words or portions of words,
and not mere determinatives or suffixes. We shall observe a
very interesting phenomenon. V medial is preceded by
signs which form complete words, sometimes complete texts!
The same words which precede \ final. It is then performing
the same function as V final. The sense of the inscription
may then be said to halt at this V medial, just as it does
at V final; i.e. we have reached the end of a word or
phrase, complete with suffix; what follows must be something
new a further name or title. And this inference is confirmed
by an examination of what follows medial V in our texts.
It reveals that medial V is followed by complete names,
sometimes found by themsevlea as complete texts. The best
illustrations of the argument in this paragraph aret-
Table I* 17, compared with I. 14-16 on the one hand, which shows
that V U is a complete textl; and LXX. 2,6,7
on the other hand, which shows that i is a complete
Table I, 41 compared with I, 39 and VII, 1, 49, 45, 48 and
T. I. 512 compared witi T. I, 50 and XI, 2, 19, 27, 38, 37,
39, 78, 97.
T. I, 1;9 compared with T. I, 138 and XI, 28, 46, 47.
T. I, 100 T. I, 103 and I, 206, 209.
T. I, 200 T. I, 122 and I, 200 and Table TLXXVI3,
T. 192 T. I, 191, 195, 194 and Table XII,5, 2,)
T. I, 213 T. I, 212, 211 and T. I. 245-245.
Other eyapples might be given, but these are sufficient to
substantiate our contention.
In eos. 339, 341-347 \V appears.to be used simply as
a syllable forming part of a word; in tbese cases it has
probably no sense-connection with % the suffix.
It remains to consider Nos. 4, 5 and 385 of Table I.
If, as we have reason to believe, Vf V and UV V a"
merely a spelling out of the same word (with a dialectal or
euphonic modification of its pronunciation) which word when
suffixed is usually written V it follows from Nos. 4 and
5 that the full word is a bt-gyllable ak-ka (perhaps ponounced
as though containing a single consonant). Now It has been
urged that this word is a mere suffix. How then do we explain
its appearance alone? A clue to the explanation is afforded
---------------------- ---- -
1. .o. : see analysis of Table XIII, and
on the detachable nature of" and its antecedents see
analysis of Table XXX.
2. ilth regard to the short perpendicular stroke being 4
mere liaison aemi-vowel, virtually equivalent to a point of
punctuation age Table XXIX analysts.
3. Frop which it will apeoer that '/4 is a word in itself.
by No. 585, where V is found alone on each face of the
prism (H. 77). While at Harappa I did not copy the design
accompanying each of these V in the blank portion of the
prism, as I did not at that time appreciate its importance; I
nade a record in my notes however that the design was a figure
like that shown on M. 440, facing right on face (a), left on
face (b), and the figure of a woman (?) facing right on face (e)
In the case of No, 4 (M. 24) the design on the reverse of the
coin was too effaced to be distinguishable, while regarding
No. 5 (M. 503) I observed one horn and a portion of an animal
whose identity I could not determine. Now it has been shown
above that V' and are allied sounds, and that in the
case of tte word \/) V Y ) they are undoubtedly
variant pronunciations of one and the same word. I suggest
then that in V V V V V and in I7 of
Nos. 4, 5, and 380 respectively we have the final element
(suffix) of the word V /) the / ) portion being
represented pictorially by the divine or heroic figure. In
other words / ) is the name of the figure in 1. 77 and
M. 440. If this is so, as what sort of a suffix are we to
regard Vf ? !f the three seals are intended to give the
owner's name, like all other seals, this name can hardly be
gnlil-1-ge or EnlilBra but only warad-ealil; or, to give a
Hindu parallel which will be closer as preserving the order of
the Proto-Indian, not Narayan-ka or parayanrko, but Narayan-
Dass. In other words V final is a suffix not in the
sense of a grammatical suffix but as a suffixed element,
'servant' or the like, used in the formatioD of proper names.
The last 3 signs in Col. IV of Table VI are compounds.
The last is a phonetic compound for V If V and
1. In which case the coins U. 25-51 should be read not 'King
of the 'and' but 'servant (of the) King (of the) land.'
are both closed syllables, as there is reason to believe (see
analysis of Table XXIV) there can be no case of contraction or
elision here. The compound will be either ideographic or
integral (i.e. each syllable being pronounced fully as is the
case with Sumerian compound phonograms) The two preceding
signs are probably phonetic compounds of the integral sort.
The compound is resolved in text No. 5. The reason for
writing integral syllables as a compound Is probably the same
as in Sumerian: vis., that they form one word.
Analysis of Table II.
The similarity of the form of the signs in Col. IV
suggests that they may be variants or represent allied sounds.
That they are not all variants is clear from No. 22, where U
and U appearA on the same text. But that U is closely
allied to J in sound and can take its place, is clear from
a comparison of Nos. 15 and 16. It is interesting to compare
V with If For just as V is clearly a member of
the Vf group (of. I. 401 with I. 391 and I. 51, 139. 531,
357) and probably a graphic variant of J ; so J clearly
belongs to the U group and is probably a graphic variant
for ( which is not found on these texts (perhaps to avoid
confusion with 1 which is ideographically quite distinct).
Again a comparison of Nos. 24 and 25 shows that 0 and r
are variants, which again is parallel in the Vf group.
The sequences in Table II give no direct evidence as to the
value of ) but the analogy of the V\ group suggests
that ) should be regarded as phonetically allied to
and 0 There is nothing repugnant to this in the
sequences, while the mcrphography of the sign strongly supports
such a view. We may cgnelude therefore that U is a
syllable. That the remainder are graphic variants of a sign
which was formed by a deliberate modification of U to
represent an allied syllable
The last two signs in Col. IV are variants of each other.
The sign is a compound of U and T .
Analysas of Table III.
Really no evidence on which to form an opinion. The
similarity of shape suggests that the two signs in Col. IV are
identical. If it is phonetic its rarity is a matter for
surprise, unless it be a, compound. It may possibly be a
compound of U and S, (see Table CIII). It is
seen from and e that U and B are found
elsewhere as compounds. (Sge Tables II and XXVI).
Analysis of Table IV.
That the first two signs in Col. IV are simple variants
is suggested by a comparison of Nos. I & 2. That the 3rd
and 4th signs are also variants is virtually certain from
their shape. That the 6th and 7th are either variants of the
above or at least allied ie laplied by the sequences of
Nos. 5, 6, 7. That the 10th slgn is a variant of the 5th and
6th is suggested by the sequence V The 8th and 9th
are clearly variants of eagh other. The last sign has a
sequence in common with the 6th. Regarding the 7th we can
only note its shape and its initial position in favour of
regarding it as a variant of the group 5-11 (cf. Col.IV). On
the analogy of Tables I, II and VI we may accept this group as
variants. On the same analogy we should be inclined to treat
group 1-4 not as a variant of group 5-11 but as an allied
syllable, or gun variety.
Analysis of Table V.
The principle reason for including the signs in Col. IV
under one Table is their shape. With regard to the 2nd and
3rd we have also the community of the suffixed V The
similarity of shape between these two signs is also most
marked. The additional stroke in the second of them recalls
the addition of strokes to the base form of the sign in Tables
I, II, IV, VI, and suggests that here also we have the
modification of a sign to serve as the symbol of an allied
sound. The 4th sign in Col. IV is sufficiently like the ZSd
and 5rd signs, and sufficiently unlike any other sign in our
texts to warrae. its inclusion in the Table. we may take it
provisionally then as a simple variant of the 2nd slgn. The
Inclusion of tah first sign has less to support it as
regards shape, and the sign would not have 'been. included at all
but for the fact that it is preceded by. A This sign
belongs to a comparatively rare group (Table LXXI) and the fact
that it is twice found preceding T suggested the
possibility that \ (which was otherwise unconnectable
with any sign) might be a later and simplified or cursive form
of W .
Analysis of Table VII.
It is clear from the sequences that the signs in Col. IV
of this Table are simple vaplants, except the last three. With
regard to the sign \ we may compare No. 83 with Nos. 26,
S'; No. 64 with No. 38; No. 66 with No. 58 (there is reason to
think that )) s phonetically allied to Y see analysis
of Table XLITI) No. 66 with No. 55. But the similarity of
-- Or as a-- variant, o --nter--------------------diary for----
1. Or as a 44 variant, from intermediary forms W V
sequences is not very close. In particular it is to be noted
that this sign is not followed by III and does not appear
as initial or quasi-initial, whereas are normally
initial or quasi-initial (i.e. preceded by signs which are
either whole words or prefixes. It will be shown later that all
the members of the fish group, and OC are in the nature of
prefixes). The sign is then related to, but not identical
with, J (of which j is a less complete and probably
later form). Now it will be observed that graphically the
respect in which / differs from W is precisely in
the addition of two short strokes. In view of what has been
said in the analysis of the previous tables we may safely
assume that here also we have a case of a syllabic sign being
modified to represent a phonetically cognate syllable. We
shall also on the same grounds take the penultimate sign in
Col. IV as a simple variant of \ Of course the variation
may be of the gunu order and the syllable still be phonetically
allied. The last sign may be a phonetic compound in view of
its shape and the fact that it is initial, For if W be
the initial part of the compound we should expect to find it
initial in the text, as V is frequently initial or quasi-
Initial. That is the initial part of the compound we
may assume, partly because it appears above the other portion,
and partly on the analogy of Brahmi and its derivative Nageri
which place the second element in a compound either after or
below the first part, (an example of the second part placed
after the first has already been noticed in Proto-Indiar. in the
compound ,). On the other hand if we take our sign
as a compound it is difficult to identify the second element.
Is it ? T (see T. LVIII. Col. IV. last two signs). This
seems the most probable explanation. If we regard it not as a
compound but as a single sign it is to be observed that there
is no sign in Sumerian or Egyptian with which it may be com-
pared. There is of course the sign given as No. T. 24 page
500 of Gardiner's ERyptian Grammar, but this does not contain
the element j which would appear to be an essential part of
the sign. I shall assume therefore provisionally that the sign
is a compound of W and .1
Analysis of Table VITI.
The sign in Col. IV appears to be distinct. Morpho*
graphically its nearest neighbour is But an
examinationn of the sequences in Tables VIII and XV make it
appear most unlikely that this resemblance is other than
Analysis of Table IX.
It is morphographloally improbable that the two signs
in Col. IV are other than simple variants. Again there is
nothing to connect them with any other sign. If their
sequences showed any striking resemblance to the f group
(which like this group seems to represent a plant of sorts) one
might admit the possibility of a causal connective; but they
Analysis of Table X.
The signs in Col. IV are clearly all variants. They
differ only as regards the shape of the enclosed element, and
the varieties of this are precisely the same as the varieties
of that element when it appears.alone (see Table XI) where it
can be shown that they are all variants (see analysis of
Table XI). That the various signs in Col, IV of Table X are
all variants is also evident from the sequences.
004r&'Ai'4k it ,111 14 Ptl d 'A t \y ,4;^A t' Y.a r06.
With regard to the function of this sign, we shall observe
that (a) it is frequently initial, (b) it is never final, (c) it
normally precedes signs that can be shown to be prefixes (like
the fish-group) or sigpngroups that are in themselves whole
words; e.g. in Nos. 9, 11, 84-536 39, 46, 50. It is clearly
then frequently a prefix probably in every case except Nos.
41-45, when it appears to be the second element in the word$E
With regard to the fact that 0 is never found final
while :Q: is so found and the inferences to be drawn there-
from, see analysis of Table 3)CI. It has been noted that this
sign contains two elements 0 and T which elements are
also found independent in our script. (See Tables XI and
XXVI). Are we then to CQasider it as a compound phonogram?
In this case it must be either T 0 or 0- f .
Now if it is Y--0 it ie strange that it is never final.
If it is 0 y it i strange that it is never preceded
by one of the numeral signs which so commonly precede T
I conclude that it Is not a com9ound phonogram but ('in origin)
a compound ideogram as in Sumerian (See Appendix p. b
No. 99). The sign then represents a garden a tree in an
enclosure. It is net likely however that it retained this
sense in our texts. It is difficult to see how a garden or
cattlepen could be utilise9 as a prefixed element in the
formation of proper names, and a very common element withal.
No! In our texts it is doubtless used as a simple phonogram,
homophonous no doubt with the original ideogrammatic value, or
an abbreviation of the latter, but unconnected with it in
1. By prefix is always to be understood prefixedd element in
the system of name-formation" unless otherwise indicated.
2. The motif of the Sumerian parallel is however different.
A closer approximation in motif is the Sumerian sign
No. 20, p. This means 'cattlepen' which may be
the original ideographic meaning of the Proto-Indian sign
meaning. This feature probably holds good of the large
majority of the signs in our texts. They were doubtless all
formerly used ideographically, either in Proto-Indian or in the
scripts where they originated, but have by the period of our
Texts come to be used as mere phonograms. Whether when borrowed
(in the case of those that bear evidence of borrowing) they were
borrowed as idograms or phonograms, must be decided in each
case on the evidence of the comparative T&able. Where a Proto.
Indian sign can be identified both with an Egyptian or Sumerian
sign and with a sign in Cypriote, Brahmi, or Sabaean, and the
phonetic values of the former and latter coincide, we may
infer that Proto-Indian borrowed the sign as a phonogran.
When this is not the case we may infer that the Proto-Indians
borrowed the sign as an ideogram, utilized it to represent a word
in their own tongue of the same meaning, but of course
phonetically different, and passed it on with their own phonetto
value, which would be quite independent of its phonetic value
in the script of origin.
Analysis of Table XI.
It is clear from the sequences that all the signs in Col,
IV except the last two are variants. The characteristics of
this sign are (1) that it is normally final or quasi-final,
(2) that it is normally preceded either by a numeral sign or
by 6 On the significance of the numeral signs see
Analysis of Table XXXI. The sign is presumably a tree. It has
two characteristic forms t wherein the position of the
branches relative to the trunk (or stem, if we consider it a
plant rather than a tree) is symmetrical, and T ', where
it is not. This difference in morphography is marked, and
seems to refer back to the (probable) ProtomElamitic origin
of the sign.1 If we examine Del. au Perse XVII. Pl. III,
So. 17, we shall see three kinds of tree or plant. Two of
them have the upper portion thus and are differentiated
Qnly by the number and position of their lower leaves or
brancnes. They are evidently varieties of the same species2,
sin e in the total they are enumerated together. The third
kind has the upper portion symmetrical thus Y and is
enumerated separately. It is virtually certain that the two
species had separate names in Proto-Elamite. Yet their forms
in Proto-Indian serve clearly to represent one single word:-
are simple graphic variants. The most probable explanation
is that the signs were taken over into Proto-Indian as
ideograms: that in the Indus valley people did not, in the
spoken tongue, differjitate between the two species of plant,
and therefore did not differentiate in their script, but used
the two signs indiscriminately to represent the word which,
for them, covered the two species. From this it follows that
at least one, probably many, and possibly all of the Proto-
Indian signs borrowed or descended from Proto-"amite, or
collaterally descended with Proto-Elamite from a common
ancestor, had at the moment of their borrowing, descent, or
severance (according to the hypothesis we adopt) ideographic
rather than phonetic import, and were on their first appearance
in Proto-Indian ideograms and not phonograms. With regard to
the last signs in Col. IV, they are clearly variants of one
another since they differ only in the curvature of the NV
element. But they are clearly not variants of the remainder
of the group, since (a) the sequences found with the remainder
1. The Proto-Elamitic origin is strongly suggested by the
fact that the varieties of this sign in Col. IV are
precisely the varieties of the sign (signs ?) in Proto-
Elamitic. See the Comparative Table.
2. Op. cit. p. 3.
are conspicuously absent, (b) the sequences found with them
are found nowhere else in the Table. In particular T is
one of the very few signs of common occurrence that are now-
-here found preceded by And this is not surprising
since what follows is always the beginning of a word1,
whereas T is normally (perhaps invariably) the final
element in a word. That it is this rather than a general
suffix is indicated by the fact that it is found after
relatively few sounds, and relatively many times after eagh
of them. We may here anticipate the discussion on the
numerals to remark that Y preceded by a numeral probably
indicates such a word as the Latin secundus, tertius, sextus
etc,, and like the Latin names may be used either alone as in
Nos. 2, 32, 33, 36, 46, 79, 86, 96, 101, or in combination
with another word like Octavius Caesar. It will be noted
that in all but two cases Nos. 28 and 11 it follow the word it
*qualifies, from which we may provisionally assume that in
Proto-Indian the adjective normally follows the noun. It
would seem also that the syllable Y is the ordinal suffix,
or capable of serving as such, like Sumerian -kam. From the
case of No. 11 it would appear that the word for 'eight' in
ProtofIndian was phonetically of such a kind as to coalesce
with this ordinal suffix to form one syllable, e.g. ba-ra
a bra.2 Returning to the last two signs of Col. IV in
Table XI, it is probable that they represent a syllable allied
to The form of modification is however
peculiar. It seems to be different from the form of modifica-
tion by the addition of strokes. It will be discussed later.
1. See analysis of Tables XXX and XXIX.
2. It is of course possible that this and other compound
phonograms may be integral compounds in which both
syllables preserve their full value as in Sumerian.
(See Appendix pp. ). But we have shown that in
one case at least, Nf( there is a strong probability
of elision of the final vowel of the first syllable.
Analyis lof Table XII.
The sequences show that all the signs in Col. IV are
simple variants. The sign is always final except in No. 20.
It is clearly not a general suffix but the second element in a
word, except in Nos. 17-20 vsere it may be an independent word
in itself. The internal strokes are of the Runu order and
do not affect the phonetic value.
Analysis of Table XIII.
That the first two signs in Col. IV are identical with
the third sign is suggested by a comparison of text 1 with 8,
88-92; and 2 with 3.12. They are probably late and simplified
forms of alth regrd to the form itself and
its Sumerian and Egyptian resembling forms, suggest a fish.
With regard to its function we may note first that it appears
to be the second element in certain words, notably T3 0.
Secondly we shall note that in a large number ot contexts it
appears to be a prefix, appearing either as initial, 66, 78,
88, 89, 96, 97, 101, 102, 112, or quasi-initial after other
prefixes or whole words, 23, 33, 34, 41, 90-95, 99, 100; and
usually followed by sign groups which are whole words: 25, 25,
64, 77-79, 88-94, or by the signs which are commonly suffixed;
passim, V 61-63, t 86, 87. Prom this it
is clear that 4 frequently appears as a word complete
in itself which is often used before proper names, see
especially H. 145 and M. 209 where the words which follow
in 88 and 94 respectively are found as complete texts.
unlike does not appear to be intimately
connected with any sign, as is with and II
But apart from this it is surprisingly like Almost
every one of the signs both following and preceding it are also
found with used as a prefix. When to this fact is
added the marked graphic similarity it is difficult to avoid
the conclusion that and are allied syllables
representing dialectal variations of one and the same word.
That the variation is dialectal rather than euphonic may be
inferred from a comparison of 41 with 120, and 34 with 119,
where respectively antecedents and sequents are identical.
Appears from its shape, its position in the texts in
which it occurs, and the sequences in which it is found to be
exactly parallel to They are probably both formed from
by the addition of an Internal stroke. Bkat they are
clearly not identical in view of the fact that both varieties
are found on the same text (see Nos. 126, 127, 130).
Phonetically would appear to be more closely allied
to than to since both appear in sequence with
t This as we shall see is characteristic of every
sign in Col. IV except k and the last two signs (which
are independent of the 'fish' group). Like ,i appears
to be independent of the signs that precede and follow it;
it is a separate word, a prefixed element, not an initial of
final element in words.
Appears from its shape, position and sequences to
be allied in sound and function to and All that
has been said regarding them may be applied to it. That it
is not however identical with either of them may be inferred
(1) from its being found on the same texts (2) from the fact
that it figures as prefix to the group V r 6 to the
exclusion of all other members of the fish group except for a
single instance where 4 is prefixed. (See Table I, 50-66).
It appears to be derived from a by the addition of A
to represent an allied sound which, as a word, is a dialectal
variant of the word which could equally be written with any
member of the fish group.
To may be applied all that has been said regarding
in the matter of function and phonetic value. It
appears to be a modification of by the addition of '"
It figures as prefix to the group V 3 to the exclusion of
every other member of the fish group. Again it constantly
appears on the same texts as other members. Thus its
individuality is clearly established. At the same time its
shape, position, and contexts leave no doubt regarding its
close alliance both in meaning and sound with the other members.
We thus conclude that t ia,),, are all distinct,
yet are all used to write one and the same word. In what
then does the variability of this word consist? Certainly
not in ideography. A scribe might, as in Sumerian, occasion-
ally represent the same word by different ideograms, but he
could not do it on principle; nor will an ideographic explaa-
tibn account for the marked preference for particular forms
in particular contexts. We are driven to admit that the
variation is phonetic. Again this variation is not on grounds
of euphony. The sequences ahow continually different
varieties of the 'fish' sign between identical antecedents and
aequents. The most striking illustration of this is afforded
by Table XII. The variation must then be dialectal varying
from speaker to speaker, or village to village, or period to
period. The next point to consider is the frequency with
which two varieties of the fish sign occur together. In all
these cases the signs and sequences preceding or following the
two fishes can be shown to be independent words. The two
fishes aay be assumed then to constitute a single word in every
1. Excluding the combination pT. and j(| which
have been shown to be separate words.
case. The question is whether all the combinations represent
one and the same word with dialectal phonetic variations, or
whether each variety of combination represents a different
word. We should incline to the latter opinion, were it not
for certain remarkable uniformities, vizs- (1)
S occurs 4 times occurs 3 times
n' 3 n 2
T 3 to tt oI
n 2 "
S5 5 1
A no other varieties of
S zthe combination are found.
It is curious if a l these words are different that they should
occur roughly the same number of tines.
(2) It is curious if they are different words that they
should occur so often in the same positions in the text,
suggesting that their function in name formation is similar.
(3) It is strangest of all that they should be found in the
same sequences. See especially M. 235 with H. 238; M. 318,
317, 485 with M. 139 and M. 507 and M. 453; M. 395 with M.388
and H. 136 and I 26; M. 318 with M. 260, 344 and M. 238;
M. 183 with H. 113 and M. 475; M. 104 with H. 179, M. 501;
M. 54, 317 with H. 179; M. 490 with M. 335 and M. 54.
(4) It is also curious that each variety of the modified fish
should appear in these compounds roughly in same number of times
in proportion to its total appearances. Thus
Appears in these double fish compounds 7 times in a total of
it T t o 17 in a total of
1B in a total of
t n I 19 in a total of
I.e. in each case the proportion is roughly
I think then that the evidence is cumulative and forces us to
the conclusion that in all these varieties of the 'double-fisht
group we have but one word with varieties of pronunciation
that are dialectal or euphonic or both.
We now note another peculiarity. This double-fish word
which like the single-fish word shows wide dialectal variations
is found in the same relative position in the texts and in the
same sequences as the single-fish word. This is beat illus-
trated by Table XII, where we see the word KC preceded
3 times by the double-fish word and 6 times by the single-fish
word. And every time the fish or double-fish is initial (or
quasi-initial). Similarly we find T suffixed to the
single-fish word 16, to the double 6 times; compare also the
occurrences of the two words with II V, V X In fact among
the 33 signs which are found immediately before or after the
double-fish word the on'y ones that are not found in the same
relation with the single-fish word are ) 1, A ,A
and The remainder are not only found, but found
repeatedly. The evidence then is very strong that the
single-fish word and double-fish word are identical. In fact
the latter may be regarded as a spelling out of the former.
We have then the following ways of writing this word, the
phonetic relationship of which I have endeavoured to suggest by
transliteration. The consonant 'b' is of course selected
arbitrarily; the allocation of the given vowels to any
particular variety of 'fish' is believed to be exact, for
reasons that will be discussed later.
BAB 3s BEB IB BOB B-aB0 BI-80
ST-DA SO-DA 31-BI BO-BI 80-BE BI-BE BI-BE BO-BI
It will be observed that the same two varieties of fish
are never found together in our Texts. This would seem to
suggest that whbo the Proto-Indian formed a carative or a
'jangle' by the reduplication of the root, he avoided repeating
the same vowel. The same tendency is observable in many
languages, of. English 'baby', Frenoh bobe, Italian 'bambino'.
It is also clear, if our inferences are correct, that at
least seme of the signs in our script stand for syllables that
are closed at both ends consonant-vowel-oonsonant;- what have
been called 'compound syllables'.
Is it possible from our texts to discover the meaning of
this word which in one or other of its varieties occurs
hundreds of times? We may note first of al that no member of
the fish group it ever found final except ] 1 and that
only in sequences where it may well form the second element in
a word. Nos. 81, 84, 80. Secondly the fish word is often
found initial or quasi-tnitial. Thirdly it sometimes separates
two sign groups which are clearly words, and probably names,
in themselves, e.g., Nos. 73, 12B, 156, 102, 166, 168, 209, 210,
211, 235, 245, 284, 286, 255, 257, 258. To these may be added
all those cases where the fish word is preceded by a sequence
ending in But these sequences although probably com-
plete words are often not names of men but rather in the nature
of a dedicatory formula (see analysis of Table r La). Be that
as it may these three considerations lead me to the conclusion
that the fish-word may very possibly be the Proto-Indian word
for 'sons. IQ this case the word 'son' comes before the name
of the father as in Sumerian and Auzanlte. It is worthy of
note that where a modified form of V precedes a member of
the fish group, the nature of the modification, whether by one,
two, or three strokes, seems to depend on the variety of the
1. And in No. 297: where also it is probably an
element in a word, as 4 is nowhere else followed by
any 'fish' sign.
fish sign or vice versa. See Nos. 128, 156, 909-211; 266,
285, 286. We have aeen that the varieties of the fish sign
are phonetic varieties, and that V ete. are phonetic
modifications of V May we now assume that the number of
strokes by which the V is modified is not immaterial but
indicates different phonetic varieties? If *e it would appear
that the law of vowel harmony was rigorously observed in Protow
Indian speech1 and meticulously recorded in the script. This
has its parallel in Sumerian also. On the whole I think we
cannot reject the evidence of these concomitant variations
and must assume that Vi, 17, represent kA, kl, kul
SThe last sign ip el.IV is possibly 4 graphic variant of
via a lost intermediate form It will be
Observed that the variety of preceding V is Or
it may be an independent sign. It may be connected with
(see Table XIV).
Analysis of Table XIV.
In view of the fact that and are clearly
Jnodifications of Y and respectively we should expect
1. At least in the Case of the liquid vowels. The simple
form V which is probably articulated with a (inherent
in the base form of Brah i and Ethiopic) seems Eore stable,
being found before and 4 and after all
sorts of signs (see Table I). It Is perhaps worth remark-
ing here that if (as I think it is arguable) the Brahmi,
Sabaean and Ethiopic scripts are all derived from Proto-
Indian, and if the Ethiopians were allied in race to the
Ethiopic GedFro.ai s (?) of the Indus valley, then the
extraordinary fluidity in the Ethiopic liquid vowels, may
have its explanation in the similar fluidity of these vowels
in the Proto-Indian parent. By this T do not wish to
suggest linguistic descent, but merely that if there was
racial descent or affinity we may expect the phonetic
peculiarities of the parent (which are determined by the
physical conformation of the organs of speech) to be
manifest in the descendant even when speaking a different
2, See analysis of Tables XXIX and XXXVIII.
to find a form as the base form of the sign in Col. IV.
It is not however found. It almost certainly existed.
Perhaps it dropped out, as many of the Sumerian signs of the
Jemdet Nass period dropped out, its place being taken by
another symbol with the same phonetic value. It is not Im-
probable toat a (see above) we have a modification of
this lost base-form.1 For it is significant that i
preceded by a solitary i and forms part of a word
ending with Neither of these features can be found with
any other modified form of It is probable then that it
is not a modified form of a wh'ch leaves the way clear for
considering it a modification of That is all we can
say at present.
Analysis of Table XV.
All the signs in Col. IV are either variants or allied.
This is indicated (a) by the shape (b) by the position, nearly
always final, (c) by the sequences RI Nos. 3.6, 15, 37, 5$;
and R 0 lea. 2, 26, 29, 43-48. These may be regarded as the
key sequences of this Table. They will help us to decide
whether the various signs in Col. IV are simple variants or
On morphographic grounds we may divide the signs in Col.IV
into two groups; the first eight, ending with Text No. 20t
and the last seven, Texts No. 22 to 52, 21 is of course in-
determinate. Now it will be observed that while the sequence
R1 occurs five times in the first group, it occurs twice only
with the second, and while the sequence ( 0 occurs once only
with the first group it occurs 8 times with the second.
Agaln I- K and R occur twice, Rll thrice with the second
1. For the g caphic nature of the modification
of.C ,r, (Tables LITI, XCVI).
U.9 (Table V).
group and not at all with the first. Conversely R R
appears twice with the first group and not with the second.
We may infer then that the two groups represent two sounds
allied but not identical. It will be noticed that the respect
in which they differ is the addition of.short strokes. In
view of what we have seen in the analysis of the previous
Tables, we may be certain that this indicates a modification of
the vowel of the syllable. The shape of the sign in No. 22 is
peculiar. The vertical foundation or base for the horizontal
strokes has been drawn, but the strokes themselves omitted.
This is probably A error 4n the scribe's part or my own in
copying.1 The additional element may be compared with the
same in the signs C It probably indicates
that the syllable is to be articulated with the vowel U.
Analysis of Table XVI.
Prom the evidence of their shape and sequences there can
be no doubt that the 2nd, 4th and 5th signs in Col. IV are
identical. Again the evidence of the sequence VI ts
so powerful that we must conclude that the third sign T
which has no neighbour in shape among the other signs of the
Frotc-Indian script, is an abbreviated or simplified, probably
later form of r This view is strengthened by the shape
of the last sign la the text (No. 12) which, as we have urged
in the analysis of Table 1, must be regarded as a late form of
the sign V It is interesting to observe that both these
1. It s1 to be regretted that in the case of the majority of
the inscriptions I hove had no opportunity of c'.ecking
my autagralp copies with photographs of seal impressions.
I requested that such pho.ozraphs might be supplied to me
by the Archaeological Department of rbe Government of
InLda, but up to the present they have not been received.
late forms approximate to the shape of the corresponding signs
in Brahml. (See comparative Morphographic Table).
With regard to the first sign in Col. IV the evidence of
the sequea~oes t negative, and this sign is probably independent
As it occur only once it may well be an ideogram rather than
a phonogram, The sign in text 41 may not be a sign at all
but a decorative device. On the other hand it may be the
fuller and Pmre. complete form of tL As it occurs alone
there is no help to be derived from the evidence of sequences.
If it is a sign, it is probably an early form of The
shape oftAs sign r and its variants r Y is
exactly parallelled in Sumerian and Proto-Elamiticj and iU
those scripts also we have no morphographic clue as to its
original ideographic significance. It is hardly likely to be
a man's hand, as we already know the sign for this in Sumerian,
and it was quite distinct from It is possible that
is a compound of + i The fact that the
sign appears In the upper right-hand corner of the seal (which
below contains the design of a many-headed beast) makes it
probable that It is to be regarded as script and not a
The signs in Col. IV that accompany Text 5 Nos. 42-48 are
clearly a doubled form of They are simple variants of
one another. Their significance is argued in the discussion
on Plurals p. 74.
Analysis of Table XVII.
The shape of the first three signs in Col. IV ajp the
evidence -of the sequences makes it reasonably certain that
---------. See sc eo---------s Tabe, -------------------- --------
1. See Mtscoellaneous Table, C 11.
. O" tke wii, ji 4 up 4'id-a" 'J-Laig I M! a n.e 1 & (7. J ) "d
" auR J I* 0 -'t t (' l B *+ rKouk9i Ad sis. %M ti a W4.It
dLCUY IC~L Sn~~ r~ rrl .L)~ ~L Q )YJ- O r.J a$,.
these signs are simple variants of one another. The first
is probably nearest to the original piotogram which doubtless
portrayed a marsh,1 (of. our own conventional way of indicating
a marsh in map-drawing Y ). The key sequence in this Table
is At4 R
Analysis of Table XVIII.
The sign in Col. IV seems to be independent. Its only
near neighbours in sbapb are tf and The
resemblance is not really close in either case, while the
evidence of any connection in the sequences is distinctly
Analysis of Table XTX.
The two signs in Col. IV, opposite texts 23, 24, are
perhaps independent signs; but perhaps allied, since there is
a resemblance in shape though not in sequences. The remaining
signs in Col. IV of the whole Table are undoubtedly simple
variants. The form in text No. 8 should be regarded as
original, showing the tail, back, two ears and hind legs of
an animal. The shape of the ears suggests the jackal. The
ears seem to have undergone progressive conventionalis&tion
and suppression until in text No. 14 they disappear entirely.
We may compare the same phenomenon in Table LIX.
Analysis of Table XX.
The first two signs in Col. IV may be taken as variants
in view of their virtual identity in shape. They do not appear
1. The portion 0 is probably the bulbous root of the
marsh plant indicates the ground line and W the
visible portion. alternatively, the sign may be borrowed
from the sgyptian sign for a papyrus clump.
tq be connected with the signs in any other Table. The
stroke makes one suspect that the base form is ( The
first two signs will then be the base form modified by the
vowel u, the third sign will be the same modified by the vowel o.
Analysis of Table XXI.
The first and third signrin Col. IV may be taken as
identical. The second sign is shown to be a mere graphic
variant by its place in the sequence V Y R which
is the key sequence of the Table. It is 4nferesting as
approximating exactly to the 8abaean form, and may therefore be
regarded as the ultimate form of the sign in Proto-Indian.
The signs in texts 43-54 differ from 6 only in the
number of interior lines, and may therefore be -egarded as
allied. If No, 42 is correctly copied this inference would be
also supported by the sequence A0 But the signs or-
the coins are so faint that it is possible that the sign in
No. 42 may also have contained the interior lines. From the
evidence of the sequences, notably the absence of the key
sequence it is certain that the signs with interior lines are
not mere graphic variants of 6 It is not likely that
this modification by interior lines corresponds to the phonetto
modificatloi that we have observed in the case of signs modified
by the addition of short strokes, firstly because in this case
the strokes are not short, and secondly because in the case of
text 46 their number is too great. The modification appears
to be rather analogous to the modification of Suierika signs to
form gunu signs, In the latter case the number of added strokes
is immaterial. We may infer the same here.
The last sign in Col. IV is probably an indepexnent sign.
Analysis of Table XXTT.
The sign in Col. IV appears to be independent both of
those in Table XXI and those in Table XXIII. It may possibly
however be allied to the last sign in Table XXI.
Analysts of Table XXIII.
The key sequence MA R shows that all the signs in Col.IV
are simple variants. The most complete and probably earliest
form is the last, Text yo. 8.
Analysis of Table XXIV.
Both the shape and position of v and Q in the texts
and the fact that each is normally followed by leads us
to infer that they are graphic variants of one and the same
sign. The forms of this sign appeertng in Nos. 47, 48 are
probably defective. It is not likely that they are other than
variants. Cf. 46 with all the texts containing "1 and
also with No. 1; 47 with 61; in the case of 49 the three
interior strokes were tightly incised on the original and may
have been accidental scratches.
No. 50 would appear to be a modification of r by
prolonging the element V to provide a base for adding
short strokes at right angles. Compare the modification of
the base form in Table XV.
The sign is probably pietographically independent.
It is perhaps an ideogram for 'heaven'; the circle representing
the sky and the interior lines a star. Or it may be a wheel.
Functionally it resembles 0 ) It is not likely however
1. It is certainly not ic rntical with el since it
occurs on the same Text, M. 159, whereas are
never found on the same Text.
that it is phonetically allied. At least 0o such conclusion
could be based on any assumption of euphoric variation, sinee
like 0 ,< it is initial, and like t1em followed by "
It is probably then quite unconnected, like V and
which also seem fun tionally to correspond. We must now
endeavour to ascertain this function since so many of our
tests begin with '", now it will appear
from an analysis of the sequences Nos. 8-43, 46, 51-91, 102,
104-127, that marks a halt in the sense. What follows
is quite independent of what precedes, and constitutes a cgm-
plete word or words in every case; words which are sometimes
found as complete texts in themselves; while no less than
fifty are found as initial in other texts. If we turn to the
analysis of Table XXX we shall find that there also what
follows is invariably a name complete in itself.
"0 etc. is therefore not a prefixed element in certain
proper names but an element unconnected with proper names yet
regularly placed before proper names on seals. What sort of
an element is this? If we may be guided by the Babylonian
analogy we may assume that this element was a dedicatory
formula. "To the god X." Compare also the Herat seal,
geographically so near to the site of the Proto-Indian
civilisation. (Antiquity 1927. p. 206), and m may
then provisionally be assumed to be names of deities and "
the dative suffix. When we have several signs before we
may have as well as divine names some phrase like 'for his
life'. Now it will be observed that ', '0 / appear
in the same position in the texts; that the first occurs 24
ties, the second 10 times, the third 7 times. Furthermore a
comparison of Nos. 104 and 128; 107 and 129; 105 and 156;
106 and 138; etc. shows that the selection of any one of the
three was nct made on grounds of euphonio harmony with the
following word. I conclude that the dative suffix was a word
subject to,phonetic variation. That its normal value was "
and that this value was invariable after syllables whose vwel
was a, such as a but was variable after a syllable
containing a liquid vowel, as, I suggest, was the case with 0
The suffix would still normally be which I will take to
have the value 1, but might be (which we will assume to be
the vowel ) or */ (which we will assume to be the vowel t
pronounced with a labial glide uj or wi). Let 0 a AN.
Let $ a BIL. Then 'To. AN is always AN-l. 'To BIL' is
normally BIL-T, but optionally BIL-I and BIL-UI. The use
of Y as a dative 'suffix does not appear to be confined to
i/ : see analysis of Table XL. The reason for taking
I i i/ to be simple vowel sounds is based on an analysis
of Table XXIX, which show i to be the vowel I or u, and
probably the former, taken in conjunction with the evidence
already noted of t and representing vowel modifications
when inserted in V and elsewhere. If is a vowel there
is strong reason to believe that is also a vowel. And
if and which can both stand as the dative suffix, are
vowels, there is reason to suppose that '/ which is also a
form of the dative suffix, is also a vowel.
We have now to consider Nos. 5-7, 93-97, 146-148. In
these, ases i, 1 are initial and there is no ground
for assuming that their function is other than when followed
by L What then has become of the dative suffix? I
take it tlat in these eases the sign following 0 , 0
Began yith a vowel, and that in consequence the dative suffix
was absorbed or *1id4$ in other words that (I, U P
are closed syllables. In the ease of t this can be
demonstrated (See analysis of Table XXIX).
The last two signs of Col. IV are clearly compounds of
and 0 and variants of one another. The form 6 as
simplification of 0 is not perhaps surprising, but it is
interesting as giving us an exact approximation to the
Analysis of Table XXV.
The resemblance between the two signs is probably
deceptive as there is marked dissimilarity in the sequences
except in the solitary case of the sequence R II (Nos. 6 and
Analysis of Table XXVI.
Although Q is a variant of v 1 0 does not appear
to have any connection with 0 their sequences being entirely
different. This is not necessarily a matter of surprise,
as there is no reason to assume thaT and or 0 and
Were in any way connected as to their pictographic origin
(see pictographic Table). And we have noted above how the
similarity between the designs of the signs in Col. IV of Table
XXV is purely coincidental.
The sign 0 would seem to be coohected with 0 in
view of the occurrence of the sequence I V R which is
found nowhere else. In that case we may probably assume that the
groups in Col. IV Texts 20-24 are modification of 0 by the
addition of strokes corresponding to a modification of the
vowel of the syllable to 1. The group appearing in texts
34-38 may be the compound B 0 The sign in Col. IV against
Nos. 40-41 is probably of independent pictorial origin.
4 may be 0 4 X ; 0 may be0 *'X (cf.
Table LXIV ). is clearly O ; is
0 "+ With regard to the signs in Nos.47-56 ) is almost
certainly equal to(1) which makes us suspect that in this form of
bracket we have really a splitting of the sign ( to make room
for enclosing a sign with which 0 is to be compounded.
In the case of Nos. 54, 55, the compounds would appear to be
'integral' (i.e. each syllable pronounced fUilly without
elision or contraction),
The signs 0 and ( ) are then identical. We
may therefore assume that and ( Y are
also identical. The fact that 0 is like 0 liable
to be compounded with an inserted 'fish' suggests that it is
a phonetically allied syllable, and that in (o ) (Y we
have really one word k 0 differently pronounced. The
fast that is found with the same modifying element 4
as is found with 0 makes us suspect that 0 and 0 are
allied syllables. And from what will be said concerning I
in the analysis of Table XXIX together with what we have
already said about it we may infer that 0 is 0 with the
substitution of I for i as its vowel element.
I would then conclude that the form Q is original ia*d
that 0 is the syllable articulated with i, 0 with e
S is quite independent of 0 w is V artioulated
with u is the same articulated with U.
i a. loui-nI4 t-LoKtl -armo, 4.-a assw vanidt, 7LkS d dcot Ju hr 4 1, wia W'
Ienr 0 aQ 0 e 7Ls -4fuw*eA is aft, jeiId wIA Jt 1 0T rU. iJ nestt1 ) fJoL,-.
**>d hs rn tft < rt4 ra l i
IlE;k S UM h S.ke etll f "uIuL laola4ry .f-lj.sa t1i.t4 4 rt. a aS I 4a. s *,
I ;adr yaI I *# ^ 0"1ih s1ee lWiW LIX "a4d *
Analysis of Table XXVII.
The original pictogram would appear to have been heaven,
plus a covering, plus the shadow of darkness? I Night? The
earliest form would appear to be the second sign in 0ol. IV.
Of this te first sign must be a cursive form, with the
hadow detached (Vnless this is a phonetio modification of
the second by the addition of A ). The third, fourth, and
sixth signs are easily admissible as graphic modifications
of the second. In the fifth the element 0 has almost oom-
pletely disappeared. In the seventh it has been reduced
to "' in the eighth and ninth it has entirely disappeared.
In the tenth and eleventh the shadow is reduced to one stroke
and joined up with the rest of the sign. In the penultimate
sign it is joined to the extremities of the body of the sign to
form loops, In the two preceding (Texts 15, 18) the shadow has
disappeared altogether and of all the original sign only the
covering pall remtnse. Of course in No. 10 the oocurrence
of 6 after I may be a coinoidence. If so the position
of the sign in Noe. 11-14 would suggest that this particular
sign 0 has no connetiot with the rest of the group. Regard-
ing the remainder discussed above the evidence of the sequences
seems inexorable despite the remarkable graphic modifications.
The latter however are susceptible, as we have endeavoured to
show, of a progressive explanation. The key to the identity
of these signs is partly their position as finals, but par-
tioularly as finals after the V There are so few single
signs suffixed to V Again we have the sequences R C]
and. R .
The last sign may be a compound of Y ( f see Table XI).
Analysis of Table XXVIII.
Probably variants. The first sign is obviously the
second reversed. But it is unlikely that that alters its
significance, as throughout these texts reversed and normal
forms of signs seem to be identical. See Tables of ,
(LVIII, LXXIV, LXXXIV).
Analysis of Table XXIX.
It will be noted that the sign sometimes occurs at
the top of the line, sometimes towards the middle. But it
is clear from the sequences that this is immaterial. The
same is true of (Table XXX) and many of the signs in
Table XXXI. It will next be observed that this sign is
often found between sign groups that are whole words and
even whole names, the elements before and after being
found as complete texts. 1f. No. 25 with U. 286 and M. 184;
No. 26 with M. 297, 298 and H. 148, 1. 209.
Now what sort of an element is this which serves to link
together (or separate) words, names, and even texts. Our
first answer would be that it is a mark of punctuation as
in Phoenician, and comparable to the Virama in the later
Indian scripts. But the evidence of Nos. 10, 51 and 19 is
against this explanation. Here is final. If we assume
that here also it is a mark of punctuation used to indicate
the termination of a text, how do we account for the fact
that only three texts out of over 750 are so terminated?
It seems certain that in these texts it has a phonetic value.
But if in these, then also in all the other texts where I+
or '@ are found. Are we then to conclude that had two
distinct values, the one phonetic and the other punctuative?
In view of the ambiguity that this would introduce into the
script, and the fact that elsewhere the script provides so