A CLUE TO THE DECIPHERMENT OF THE SO-CALLED
SHELL SCRIPT
By B. N. Mukherjee, Calcutta
One ofthe undeciphered scripts of ancient India is known to epigraphists
as Shell-script {Sankha-lipi)} The script (lipi) is so called because ofthe
superficial resemblance of some of its characters to a conch-shell (sahkha).
One of the inscriptions in this script is noticeable on the back of a stone
horse which has been dug up about a hundred years ago "near the ancient
fort of Khairigarh in the Kheri district" ofthe then North-Western Province
and Oudh and now of U.P. The stone sculpture was later transferred to the
museum in Lucknow.^ It is at present a part ofthe reserve collection ofthe
State Museum, Lucknow. Its accession no. is H219.
I had a chance of examining the inscription on the back of the stone
horse in March, 1983 when Mr. (now Dr.) R. C. Sharma, the Director ofthe
State Museum in Lucknow, kindly allowed me to handle the sculpture. As I
closely scrutinised the inscription and looked at it from deifferent angles
(by moving around the horse), the following features became clear (figs 1,2
and 3).
(1) The inscription ranges from near the rear end of the back of the
horse to a position near the neck. Its total length is 92 cm and the maxi¬
mum height at one place 33 cm.
(2) The inscription consists most probably of six characters.
(3) The style of writing is ornate and cursive.
(4) There are some signs which are clearly comparable with subscript r
(in cases of the first and the fifth characters) and medial vowel signs (at
least in cases ofthe first and the fomth characters) as used in some forms of
the Brahmi script.
' For detailed discussions on the problems relating to the so-called Shell script
see R. Saloman: Shell Inscriptions, Calcutta 1980, (cited below as SI), pp. If;
" Undeciphered Scripts of South Asia", Aspects of Indian Art and Culture, S.K. Saras¬
wati Commemoration Volume, (edited by D. C. Bhattacharyya and J. Chakra¬
varti), Calcutta 1983, pp. 201 f
Of the few unsuccessful attempts made so far to read inscriptions in the Shell characters the most well-known are those cormected with Ci-Aruton Shell inscrip¬
tion (found in Java). The readings of Brandes, Kern and Pleyte were justifiably
questioned by J. Ph. Vogel. As pointed out by him, Brandes and Kern them¬
selves did not publish their readings. Their suggested readings were quoted by oth¬
ers. Vogel observed that "it is quite possible that the two scholars to whom they are due regarded them as merely provisional, as neither of them published his read¬
ing himseir (J. Ph. Vogel: "Earliest Sanskrit Inscriptions of Java" . Publicaties van den OudAeikundigen Dienst in Nederlandisch-Indel (1925), p. 24). For the defects
in the readings offered by K. P. Jayaswal and H. Sarkar, see SI, pp. 60-61.
^ V. A. Smith: "Observations on the Gupta Coinage". Joumal of the Royal Asiatic
Society of Ireland and Great Britain (cited below as JRAS) (1893, p. 98.
The adding of diacritical marks, indicating sounds of vowel letters to
consonants, is indeed a distinguishing characteristic of Brahmi and Kha¬
roshthi (or Kharoshti) and of other Indian scripts derived from the former.
So the appearance of the feature concemed in the inscription in question in
a Brähmi using zone may suggest that it is written in a script derived ulti¬
mately from some form of Brähmi. Armed with this hypithesis, one may try
to relate the characters with comparable forms of Brähmi letters.
A part ofthe first characters (see figs. 2 and 3) can be recognized as an
ornate form of Brähmi palatal sa with its horizontal stroke, used for con¬
necting the two side arms in the middle of the letter, broken into two
strokes, slanting downwards and touching each other (fig. 3). A stroke
recognisable as a regular sign for subscript r is attached to the basic letter
and is embellished with the addition of a wide curve. There is a horizontal
line attached to the top ofthe basic character {sa). Above the top line and
also above the right side ofthe main figure (sa) one can notice two semiver-
tical lines, which may be taken as indicating the sound of medial i. The
lines concerned are topped by an ellipse, vrith a semi-vertical stroke above
its lower base, which figure by itself stands for medial i. Two slanting lines,
issuing out ofthe upper part ofthe ellipse and running side by side, initiate
a spiral pattern to the left ofthe basic figure (sa). Ultimately one ofthe lines
completes the spiral pattern and the other ends up near the left end of the
ornate subscript r. The whole figure appears to be a very ornate form ofthe
conjunat letter sri (fig. 3).
A line, coming out of the junction of the upper end of the right arm of sa
and its topline is so extended as to touch a (tail-like) portion on the left side
ofthe second character (figs. 2 and 3). Similarly a line issuing out ofthe
right arm ofthe palatal sa (noted above) touches the lower end ofthe right
stroke of the second character. If we interprete these features as indicators
of cursive writing and embellishments, the rest of the second character can
be recognized as representing a somewhat cursive from of one ofthe varie¬
ties of "tailed" ma, known to have been in use in and about the Cupta age'
(fig. 3). The letter concemed is here slightly slanting towards right.
A curved line, again apparently an indicative of cursive writing, issues
out of the top of the right stroke of ma and touches the base of the third
character (figs. 2 and 3). It can be read as ha, with its left upper end touch¬
ing a slightly curved (semi-arched) line ending with a serif on the right (fig.
3). The character appears to be a little slanted towards right. The slightly
curved line may stand for an ornamented version of the sign of a medial
vowel, probably e (fig. 3). A curved line, again an indicative of cursive writ¬
ing, issues out ofthe right hand stroke oiha and touches the loop ofthe next
figure.
^ One may find some similarity between the second character and a variety of tri¬
partite ya. But by accepting the character as ya, we cannot make a sensible reading ofthe inscription. Moreover, such an attempt does not find support from an inscrip¬
tion in regular Brähmi characters of the Gupta age engraved on the neck of the
horse. This inscription has been discussed by us in the present article.
The fourth character (figs. 2 and 3) can be convincingly deciphered as an
ornamented form ofthe conjunct letter ndra, if it is taken to have been writ¬
ten with a heavy slant towards right (and in an anti-clockwise direction)
(fig. 3). A curved stroke is attached to the right side ofthe arm above (here
appearing as if to the left of) the loop of na. This may be the sign for medial
ä (fig. 3). This stroke and also a curved line, issuing out ofthe loop of na,
touch the next character. These features betray again the cursive style of
writing.
The fifth character (figs. 2 and 3) becomes legible if we look at it from the
right side of the inscription. In that case the basic character may be recog¬
nised as an ornate form of da, moved towards right and anti-clockwdse by
an angle of almost 90" from its usual vertical position. It consists of a figure
looking almost like a semi-circle, open to right, with the addition of two
short strokes at the ends. The lower stroke not only extends on both sides
of the lower end of the semi-circle, but also has its upper end curved
towards left. The latter feature makes the line looking like a semi-vertical
one (fig. 3). A long curved line, dravm apparently for the sake of embellish¬
ment, is attached to the lower end of the lower stroke. To the upper stroke
is attached an ornate sign which may be taken to stand for medial i or i. In
this case a comparison of this sign with that ofthe medial vowel f attached
to sra (character no. 1) may tempt us to accept it as indicating medial
vowel i. So the fifth character may be read as di (fig. 3).
If we look at the last character (figs. 2 and 3) from the right side ofthe
inscription, it can be at once recognised as basically cinsisting ofthe letter
ta with subscript y (fig. 3). It is placed very much to the right (anti-clock¬
wise) of its usual position. The conjunct letter is adorned with additional
features. The letter ta is put within an ellipse. A winding line, (issuing out of
the left side ofthe ellipse), touches the lower part ofthe latter (near the end
of the left hand curve of ta) and also contacts, by its lower end, the left side
of the curve of the subscript y. The end of the subscript y is extended to
form an exotic design to the right ofthe letter tya. Ultimately a part ofthe
design moves to left, makes a curved pattern and ends by touching a cres¬
cent-like symbol set between the fifth and the sixth characters (i.e. between
di and tya) (fig. 3).
The above discussion allows us to decipher the inscription in Shell char¬
acters as §ri-Mahendräditya. Here the readings ^ri, ndrä and tya are cer¬
tain. Since the inscription deciphered as ^ri-Mahendräditya makes sense,
the readings of other three letters (second, third and fifth), as suggested
here, are also highly probable. The form of the letter ma is akin to one of
its usual varieties of the Gupta age, though a similar form occasionally
appears in the Kushäna period. The general palaeographie features are
also not against the attribution of the inscription to the Gupta period.''
* For justification of our remarks about the palaeographie features of the basic figures concerned, see A. H. Dani: /w<iian PaZaeo^rap%, Oxford 1963, pls. Xa,Xb,
XII a, XII b, XIII a, Xlllb, XlVa, XIV b and XV b.
Support for the above reading is found from a Brähmi inscription on the
right side ofthe neck ofthe horse. The inscription consists of three lines. It
is heavily mutilated at present (fig. 4). However, the palaeographie featur¬
es of the existing letters are strong enough to attribute the epigraph to the
Gupta period.
V. A. Smith reads . . . dda Guttasa deyadhamma perhaps in this epi¬
graph.' But the extant portion ofthe record does not support such a read¬
ing. We can read asv{o)rasa at the end of line 1, yäjino at the end of line 2
and sya at the end of line 3. The expression yäjino (genitive singular ofthe
Sanskrit word yäjin, "a sacrificer") clearly indicates that the inscription is in Sanskrit.
The evidence of last letter ofthe inscription (. . . sya, meaning "oP) sug¬
gests that the stone horse belonged to a person whose name was engraved
in the space preceeding the letter sya. The person concerned lived, accord¬
ing to the palaeographie data, in the Gupta age.
This inscription, referring to the owner of the horse, should have been
the first one to be engraved after the figure ofthe animal had been sculpted.
On the other hand, the Shell inscription "is so neatly and perfectly integra¬
ted into the form ofthe horse that there can be no doubt that it was planned
and executed together with the statue itself'.' So the stone horse, the
Gupta Brähmi inscription and the Shell inscription were products of one
single process of action and they should be taken as contemporanious of
one another.
The Brähmi inscription on the neck ofthe horse calls its owner as "a sac¬
rificer" (yäjin > yäjinah). Since he is referred to as such in an epigraph
engraved on a stone horse, the sacrifice in question could well have been a
horse - sacrifice (asvamedha). This inference receives support if we can
read at the end of the first line the word asvorasa, meaning "the principal horse" or "the most excellent of horses". The term asvorasa ohvioxisly refers
to the stone horse or to the horse of which the sculpture was a replica. The
word asvorasa of line 1 is also apparently related to the sacrifice alluded in
the term yäjino (yäjinah, "ofthe sacrificer") appearing in the very next line.
In that case the sacrifice concerned had been a horse-sacrifice and the
stone horse was a replica of the animal which had been sacrificed.
The sacrificer, alluded to in the Brähmi inscription, must have been a
ruler ofthe Gupta age. And since the stone horse is known to have been dug
up in a locality which could well have been within the limits of the Gupta
empire, the sacrificer should have been an Imperial Gupta monarch. We
know of two Gupta emperors who performed horse-sacrifice. One of them
was Samudragupta, whose title was ParäkramänJca. The other was his
grandson Kumäragupta I, whose title was Mahendräditya.
It appears that either of these two Gupta emperors was responsible for
erecting a replica of the horse killed in the horse-sacrifice performed by
him. The custom of preparing a replica of an object (particularly a perish-
' JRAS, 1893, p. 98.
' SI, p. 16.
able one) used in a Brahmanical sacrifice is well attested by ancient Indian
sources. For an example, we can refer to the inscription on one ofthe two
stone posts found ar Isapur and now kept in the Archaeological Museum at
Mathura. The inscription, dated in the reign ofthe Kushäna king Väsishka,
records the setting up sacrificial posts {yüpäh) by a person after having per¬
formed a sacrifice.' The stone posts or yüpas "are apparently copies ofthe
wooden posts that were actually used in the sacrifice and were set up to
commemorate the performance of a large sacrifice".*
The short shell epigraph of six characters, engraved together with the
Brähmi inscription at the time of setting up the stone horse, should also
allude (like the Brähmi record) to the sacrifice or/and sacrificer. The occur¬
rence of the first letter, which can be interpreted only as an ornate form of
the letter sri (meaning "prosperous" or "illustrious") indicates that the next
five characters contain the name or title ofthe sacrificer.' The name or title
must be alluding to Samudragupta or Kumäragupta I, each of whom per¬
formed the sacrifice. So the name or title concerned must be either Samud¬
ragupta or Paräkramärdca (title of Samudragupta), or Kumäragupta or
Mahendräditya (title of Kumäragupta I). Ofthe four alternatives, we shall
have to opt for the last one, since the first letter of the name or title (the
second character of the Shell inscription) can be read as ma, but certainly
not as sa, or pa or ku.
The evidence ofthe Brähmi inscription ofthe Gupta age, engraved on the
neck of the horse, cinfirms the reading ofthe Shell epigraph as i^ri-Mahen-
dräditya. This proves the viability of our reading ofthe Shell inscription,'"
which has been offered above without taking the help of the Brähmi inscrip¬
tion.
The stone horse seems to be a replica of the horse killed at the time ofthe
performance of horse-sacrifice by the Gupta emperor Kumäragupta I (c.
A.D. 415/16-455/56). So the inscriptions and the sculpture concemed can
be dated to about the first half of the 5th century A.D.
Our reading of the first inscription confirms the theory of the Shell
script's connection with Brähmi. It appears that the Shell script was basi¬
cally derived from Brähmi. Like the latter, it was written left to right.
The letters of the Shell script, however, did not always exactly conform
to the forms of the corresponding Brähmi vowels or consonants. Some¬
times their shapes were exotic or grotesque.
The characters were written in a cursive style. Superfluous lines often
connected the characters with one another, perhaps to indicate inter alia
' H. Lüders: Mathurä Inscriptions (edited by K. L. Janert). Göttingen 1961,
p. 126. The first word in line 7 is yüpäh (see ihid., p. 295) and not yupah as read on p. 126.
* Ibid.,
' For comparable forms, we can refer to the expressions like &ri-Samudragupta
and Sri- Kumäragupta in inscriptions and coin-legends and Sri-Mahendra and Sri-
Mahendräditya in coin legends.
Another Shell inscription is noticeable on the hip ofthe stone horse. Our read¬
ing of this inscription is being published elsewhere.
the ductihty of the script. As a result of the cursive style of writing, the
characters were often slanted anti-clockwise, and so towards right (and
occasionally clockwise and so towards left).
Sometimes the letters were moved anti-clockwise by an angle of 60-90°
to the right of their usual vertical positions. If all the letters of an inscrip¬
tion of one line were moved anti-clockwise from their usual vertical posi¬
tions by an angle of about 90", the characters would acquire a "horizontal"
orientation. Such an inscription may appear to an uninitiated viewer as
having been written vertically.
The letters were at times put within a circle and/or a decorative pattern.
Post-consonental vowel diacritical marks were generally extended to
develop ornamental flourishes. Extra strokes were now and then added to
different sections of the extended diacritical marks. Additional lines and
designs, having no organic link with the characters, were occasionally
placed near the area inscribed with the actual letters and their diacritical
marks.
All or most of these features, well-executed, made a Shell epigraph a
beautiful record of calligraphic art. But at the same time such type of calli¬
graphy obscured the basic letters and made the reading ofthe record quite
difficult.
Therefore for deciphering an epigraph in the Shell script one has to be
very careful. He will have to look at the inscription from different and all
possible angles (since the letters might have been at times inserted at unu¬
sual positions). In this scrutiny lies the clue to the understanding ofthe
direction and pattern of writing of a Shell inscription.
A close examination of a record in the Shell script will enable an epigra¬
phist to recognise one or more letters, the form or forms of which, bereft of
its or their ornamental and superfluous additions, can be compared with
that or those of a character or characters of the Brähmi script of a particu¬
lar age and ofthe region in which the record has been discovered. Then the
epigraphist should try to compare the forms of all Brähmi letters of the
zone and period with whose ofthe characters ofthe Shell inscription, bear¬
ing always in mind the points mentioned above about the possible shapes
and positions of letters and the plausibility of the presence of decorative
strokes, signs and pattems. These efforts are likely to make the reading of
a Shell epigraph distinctly possible.
The Shell inscriptions are always brief, generally incorporating not more
than nine characters. The inscription on the back ofthe stone horse proves
that Sanskrit was the only language or at least one ofthe languages used in
writing the Shell script. The first inscription shows that a Shell epigraph
could consists of a name and/or a title. Such brief inscriptions could also
perhaps contain short descriptions of objects or (ritualistic) formulae.
There is no doubt that the composers and scribes deliberately made the
Shell epigraphs records of difficult reading. There must have been some
reason for shrounding them with viels of obscurity. Short Shell epigraphs
inscribed on (or by the side of) regular donative records (on pillars, colums,
etc.) and on sacred or commemorative objects, hard surfaces (bare rock,
Fig. 4
cavewalls, etc.) in areas of pilgrimage, seals inscribed with names of deities etc., may suggest a religious affiliation of the script. Inscriptions engraved
in a mystifying style might have been added to sanctify the objects con¬
cemed in a marmer which would compell the viewers to look at the records
with reverence but without understanding them immediately. From this
point of view the Shell script may be looked upon as hieratic version of the
Brähmi script.
The above discussion does not solve all of the problems concerrung the
Shell script. It, however, offers the first definitive reading ofa Shell inscrip¬
tion and a clue to the reading of epigraphs in Shell characters.
List of Illustrations Fig. 1
The Shell inscription on the back of a stone horse in the State Museum, Lucknow (a view from above the back of the horse).
Fig. 2
The main portion of the Shell inscription.
Fig. 3
An eye copy of the Shell inscription.
Fig. 4
A Brähmi inscription on the neck of the horse.
HARIHARIHARIVÄHANA LOKEÖVARA IN NEPAL
Die Entwicklung einer Erscheinungsform Lokeävaras vor dem Hintergrund religiöser Auseinandersetzung zwischen Buddhismus
und Hinduismus in Nepal - dargestellt
anhand literatur- und kunstgeschichtlioher Zeugnisse.
Von Horst Brinkhaus, Hamburg
Die Sädhanamälä ist bekanntlich eine Sammlung buddhistischer Medi¬
tationsanweisungen, welche etwa aus der Zeit vom 6. bis 12. Jahrhundert
nach Christus stammen. Nach den Vorschriften einzelner Sädhanas sollten
bestimmte Erscheinungsformen tantrischer buddhistischer Gottheiten
meditiert werden. Zur Unterstützung der Vorstellungskraft haben Bild¬
hauer und Maler auf der Grundlage solcher Sädhanas künstlerische Dar¬
stellungen der entsprechenden Gottheiten geschaffen. Umgekehrt können
aber natürlich auch einzelne Sädhanas selbst von bereits bestehenden
Skulpturen oder Gemälden angeregt worden sein.
Irmerhalb der Sädhanamälä sind uns zwei Sädhanas überliefert, die
einer Erscheinungsform Lokesvaras bzw. Avalokiteövaras unter dem
Namen Harihariharivähana gewidmet sind; in der Textedition von Benoy¬
tosh Bhattacharyya erscheinen sie als Nr. 33 und Nr. 34'. Welche
Gestalt unter der zunächst rätselhaft anmutenden Bezeichnung Harihari¬
harivähana zu verstehen ist, wird dort ausführlich und in beiden Sädhanas
übereinstimmend erklärt: Lokeövara (ich werde mich im folgenden auf
diese Bezeichnung beschräiüten) sitzt auf Visnus Schultern, Vi^nu sitzt,
wie üblich, auf Garudu und dieser wiedermn auf einem Löwen, dem Reittier
Lokesvaras. Vi§nu, Garudu und der Löwe, die hier alle drei als hari
bezeichnet sind, fungieren gemeinsam als Lokesvaras vahana.
Künstlerische Darstellungen von Harihariharivähana Lokeivara sind,
soweit ich weiß, bislang ausschließlich im Kathmandu-Tal in Nepal
bekannt geworden^. Während eines zweijährigen Aufenthaltes in Nepal
hatte ich Gelegenheit, intensiv nach künstlerischen Ausgestaltungen die¬
ser Erscheinungsform Lokesvaras zu suchen, und ich habe in der Tat im
Kathmandu-Tal eine ganze Reihe älterer und jüngerer Darstellungen in
Form von Skulpturen und Gemälden gefunden.
Bei meiner Suche konnte ich mich zum Teil von Angaben in der einschlä¬
gigen Literatur zur buddhistischen Ikonographie Indiens leiten lassen: So
hat zunächst eiiunal Benoytosh Bhattacharyya' auf zwei Skulpturen
' Sädhanamälä. Ed. by Benoytosh Bhattacharyya. Vol. 1. Baroda 1925.
(Gaekwad's Oriental Series XXVI). S. 77-79.
^ Hierzu paßt übrigens gut, daß die älteste bekannte Handschrift der Sädhana¬
mälä ebenfalls aus Nepal stammt; vgl. Cecil Bendall: Catalogue of the Buddhist
Sanskrit Manuscripts inthe University Library, Cambridge. Cambridge 1883, S. 174.
' Benoytosh Bhattacharyya: The Indian Buddhist Iconography. Nachdruck
der 2. Auflage 1958, Calcutta 1968, S. 136.