Pramāṇa · Search · About · API

Indian Epigraphy: A Guide to the Study of Inscriptions in Sanskrit, Prakrit, and the Other Indo-Aryan Languages c11.s1: 6.2 The Era of Decipherment (1835–1860)

Indian Epigraphy: A Guide to the Study of Inscriptions in Sanskrit, Prakrit, and the Other Indo-Aryan Languages c11.s1

Language
English

Contents

6.2 The Era of Decipherment (1835–1860)

c11.s1.1

After the slow but steady progress of the first three decades of the nineteenth century, the study of Indian inscriptions erupted in a blaze of glory in the middle of the 1830s. The "Glanzjahre"²³ of 1834 to 1838 were largely, though by no means exclusively, due to the remarkable efforts and insights of James Prinsep (1799–1840), who came to India in 1819 as assistant to the assay master of the Calcutta Mint and remained until 1838, when he returned to England for reasons of health. During this period Prinsep made a long series of discoveries in the fields of epigraphy and numismatics²⁴ as well as in the natural sciences and technical fields. But he is best known for his breakthroughs in the decipherment of the Brāhmī and Kharoṣṭhī scripts.

c11.s1.2

²³ Thus appropriately called by Windisch, *Geschichte der Sanskrit-Philologie*, 98.

c11.s1.3

²⁴ For a collection of his essays on these subjects, together with extensive supplementary notes by Edward Thomas, see *Essays on Indian Antiquities, Historic, Numismatic, and Palaeographic, of the Late James Prinsep*.

6.2.1 Decipherment of the early Brāhmī script

c11.s1.t1.1

Although many Sanskrit inscriptions had been published since the time of the pioneers Wilkins and Colebrooke, and although some of the later scripts derived from Brāhmī had been successfully read, inscriptions of the Gupta period and earlier remained incomprehensible when Prinsep began his epigraphic researches in the early 1830s; this, despite Wilkins’ having already read, four decades earlier, inscriptions of the sixth century. Renewed progress in the decipherment of early Brāhmī²⁵ began in 1834 with T. S. Burt’s procurement of reliable facsimiles of the inscriptions on the Allahabad pillar,²⁶ including the Aśokan pillar edicts and the “Queen’s Edict” as well as the inscription of Samudragupta. Prinsep’s “Notes on Inscription No. 1 [the Aśokan inscriptions] of the Allahabad Column”²⁷ broke the first ground in the reading of Mauryan Brāhmī script. In this brief paper Prinsep displayed a combination of intuition and methodical thought which would do any modern decipherer proud. He expressed, first of all, doubts as to “whether the language this character … expresses is Sanscrit” (116), noting especially the “rare occurrence of double letters” (i.e., conjunct consonants). Although in subsequent articles between this one and his final breakthrough three years later Prinsep seems to have changed his mind and considered that the unknown language was Sanskrit,²⁸ in the end his first intuition that the inscriptions were not in Sanskrit proved to be correct. Given what little was known at the time of the history of the Indic languages, this was a remarkable insight, since it would have seemed almost automatic to assume that the language of such clearly ancient inscriptions would be Sanskrit.

c11.s1.t1.2

²⁵ Referred to in early works by various names, most commonly “lath [i.e., *lāṭ*] character” and “Indian Pali” (see 2.2.2).

c11.s1.t1.3

²⁶ See JASB 3, 105–13.

c11.s1.t1.4

²⁷ JASB 3, 1834, 114–8.

c11.s1.t1.5

²⁸ See, e.g., JASB 3, 487.

c11.s1.t1.6

Equally impressive was Prinsep’s arrangement, presented in plate V of JASB 3, of the unknown alphabet, wherein he gave each of the consonantal characters, whose phonetic values were still entirely unknown, with its “five principal inflections” (117), that is, the vowel diacritics. Not only is this table almost perfectly correct in its arrangement, but the phonetic value of the vowels is correctly identified in four out of five cases (plus *anusvāra*); only the vowel sign for *i* was incorrectly interpreted as *o*. Moreover, Prinsep also provided statistical counts for each principal consonantal character and its vocalic “inflections” and noted “other forms occurring.” All in all, this table is a model of sophisticated decipherment technique, comparable, mutatis mutandis, to the famous grids developed by Michael Ventris for the decipherment of the Cretan Linear B syllabary²⁹ more than a hundred years later. Prinsep also provided a statistical count of the letters appearing on a page of a Sanskrit text (the *Bhaṭṭikāvya*) for purposes of "aiding the investigation of the powers of the unknown alphabet" (118). Although Prinsep's final decipherment was ultimately to rely on paleographic and contextual rather than statistical methods (cf. Hoernle, *Centenary Review*, 60), it is still no less a tribute to Prinsep's genius that he should have thought to apply such modern techniques to his problem.³⁰

c11.s1.t1.7

²⁹ See John Chadwick, *The Decipherment of Linear B*, 2nd ed. (Cambridge: Cambridge University Press, 1970). On modern decipherment techniques in general, see E. J. W. Barber, *Archaeological Decipherment: A Handbook* (Princeton: Princeton University Press, 1974).

c11.s1.t1.8

³⁰ In his conclusion to this important article, Prinsep voices the opinion, again correct, that the decipherment of the ancient Indian script should be much simpler than that of the cuneiform and Egyptian hieroglyphic writings, and urges on his countrymen to turn "fresh attention to the subject, lest the indefatigable students of Bonn or Berlin should run away with the honor of first making it known to the learned world" (118).

c11.s1.t1.9

Burt's facsimiles of the Allahabad inscriptions were in the meantime also serving as a stimulus to the study of the later Brāhmī script of the Gupta era; in his "Remarks upon the Second Inscription of the Allahabad Pillar,"³¹ A. Troyer published in March 1834 a translation of the inscription of Samudragupta based on a transcription into Devanāgarī by Madhava Rao, head librarian of the Sanscrit College.³² While the transliteration and translation were very incomplete and inaccurate, Troyer did succeed in identifying the names of the kings Candragupta [I] and Samudragupta, though he was inclined to identify them as rulers of the Mauryan, rather than the then virtually unknown Gupta, dynasty.³³ In June of the same year,³⁴ W. H. Mill published an improved, though still far from accurate, version³⁵ of the same inscription.

c11.s1.t1.10

³¹ *JASB* 3, 118–23.

c11.s1.t1.11

³² According to Troyer, "It was principally the alphabet of the Mahámalaipur inscriptions [published by Benjamin Guy Babington in *Transactions of the Royal Asiatic Society* 2, 1828, 258–69] that enabled MADHAVA RAO to transcribe . . . the remains of the inscription" (119); the contributions of Wilkins' decipherment of the "Gya" (i.e., Nāgārjunī) inscriptions is also acknowledged.

c11.s1.t1.12

³³ See also Prinsep, *JASB* 3, 115–6.

c11.s1.t1.13

³⁴ *JASB* 3, 257–70.

c11.s1.t1.14

³⁵ Cf. the comments of Fleet in *CII* 3, 1.

c11.s1.t1.15

Then, in October 1834, B. H. Hodgson presented his discovery of "Ancient Inscriptions in the Characters of the Allahabad Column"³⁶ (i.e., in Mauryan Brāhmī) on columns at Mathiah and Radiah, that is, the Aśokan pillars nowadays known as Lauṛiyā-Nandangaṛh and Lauṛiyā-Ararāj.³⁷ In his "Note on the Mathiah Láth Inscription,"³⁸ Prinsep made several important steps toward the decipherment of Mauryan Brāhmī. First, he noted that "*all three inscriptions* [i.e., the Delhi, Allahabad, and Mathiah/Lauṛiyā-Nandangaṛh Aśokan pillar inscriptions] *are identically the same*" (484), and he identified a recurrent series of fifteen syllables "which may be supposed to be some formula of invocation" (483); this, in fact, eventually turned out to be the introductory formula *devānampiye piyadasi lāja hevaṃ āha*, "The Beloved of the Gods, King Priyadarśin, speaks thus.” Even more importantly, he correctly identified (485) the consonants 𑀬 and 𑀯 as *ya* and *va* on the basis of their resemblance to the corresponding letters in the “Gya alphabet” (see n. 32) and of their frequent occurrence as the lower, that is, latter, members of consonantal conjuncts; and he also seems to have tentatively identified the all-important character for *sa*. These discoveries did, however, lead Prinsep temporarily to the incorrect conclusion that there was “little doubt that . . . the language [is] Sanscrit” (487; see p. 204).

c11.s1.t1.16

³⁶ *JASB* 3, 481–3.

c11.s1.t1.17

³⁷ On the equivalents of the early and modern geographical designations of the Aśokan inscriptions, see Windisch, *Geschichte der Sanskrit-Philologie*, 106 and 108 n. 4.

c11.s1.t1.18

³⁸ *JASB* 3, 483–7.

c11.s1.t1.19

Finally, with regard to the contents of the inscriptions Prinsep commented: “Whether they mark the conquests of some victorious rájá;—whether they are as it were the boundary pillars of his dominions;—or whether they are of a religious nature, bearing some important text from the sacred volumes of the Bauddhists or Brahmins, can only be satisfactorily solved by the discovery of the language, and consequently the import these curious monuments are intended to convey” (487). How typical, once more, of Prinsep’s powers of intuition that each of the three alternatives he proposed turned out to be at least partially correct!

c11.s1.t1.20

The third volume of the JASB also contained one other significant contribution to the decipherment of Brāhmī, this one by J. Stevenson, entitled “Restoration and Translation of Some Inscriptions at the Caves of Carlí” (i.e., Karle).³⁹ In this paper Stevenson was able to correctly identify, by comparison with the text of the Allahabad Gupta inscription published in March of the same year, twelve consonants (*ka*, *ga*, *ja*, *ta*, *tha*, *da*, *pa*, *ba*, *ya*, *ra*, *va*, and *sa*) in at least some of their occurrences in the dedicatory inscriptions at these caves, which date to about the first century a.d.⁴⁰ Stevenson thus succeeded in providing some phonetic values for the oldest form of Brāhmī to date; though how very far he was from a true decipherment can be demonstrated by a comparison of his translation (498) of his inscription A as “To the Triad. I, Arodhana, lord of Jambudwípa (India), the obtainer of victories, of a truly victorious disposition, the commander of the world, the chastiser of the earth, and exalted above paradise, slaughter every foe that rises against me” with the correct translation by Burgess and Indraji:⁴¹ “Sēṭh Bhūtapāla, from Vejayanti, has established a rock-mansion, the most excellent in Jambudvīpa.”⁴²

c11.s1.t1.21

³⁹ JASB 3, Oct. 1834, 495–9.

c11.s1.t1.22

⁴⁰ See 498, “Alphabet as far as decyphered.”

c11.s1.t1.23

⁴¹ *ICTWI*, 28, ins. no. 1.

c11.s1.t1.24

⁴² The text, as read by Burgess and Indraji, is *vejayaṃtitā seṭhiṇā bhūtapālenā selagharaṃ pariniṭhapitaṃ jabudīpamhi uttama*. See also Prinsep’s comments (JASB 3, 487 n. 6) on Stevenson’s readings, which he partly accepts, although his alphabet “does not convert the context into intelligible Sanscrit.”

c11.s1.t1.25

The next major step toward decipherment was made by Christian Lassen in a letter to Prinsep published by the latter in JASB 5.⁴³ Here Lassen correctly read, by reference to Tibetan and “Pali” (i.e., Brāhmī) letter forms, the Brāhmī legend on the reverse of a coin of the Indo-Greek king Agathocles as *agathukla rája*.⁴⁴ With this discovery, the partial knowledge of Brāhmī was pushed back into the pre-Christian era.

c11.s1.t1.26

⁴³ Nov. 1836, 723–4.

c11.s1.t1.27

⁴⁴ On the basis of this important discovery, Lassen could be credited with the first correct reading of an inscription in early Brāhmī, though he cannot be said to have deciphered the script as a whole.

c11.s1.t1.28

After the important contributions of Stevenson and Lassen, the initiative passed back into the capable hands of Prinsep. His next step was presented in “The Legends of the Saurashtra Group of Coins Deciphered” in *JASB* 6.⁴⁵ Here Prinsep offered partially correct readings of the hybrid Sanskrit legends on the coins of Western Kṣatrapa rulers of the third and fourth centuries, though he erred in reading *kṛtrima* ‘elected [king]’ instead of *kṣatrapa* and in reading the name elements *-sena* and *-siṃha* both as *Sáh*. He did, however, correctly identify several conjunct consonants such as *jña*, *mna*, and *śva*, as well as the hybrid Sanskrit genitive ending *-sa*; and this latter discovery provided the clue that set the stage for his greatest breakthrough.

c11.s1.t1.29

⁴⁵ Mar. 1837, 377–92.

c11.s1.t1.30

This came in the modestly titled “Note on the Facsimiles of Inscriptions from Sanchí near Bhilsa, Taken for the Society by Captain Ed. Smith, Engineers,” published in June 1837,⁴⁶ in which Prinsep presented the definitive decipherment of the oldest forms of the Brāhmī script. The final key turned out to be hidden in some two dozen brief inscriptions, dating to the second or first century B.C., which had been copied by Edward Smith from the Buddhist *stūpas* at Sāñcī; concerning these, Prinsep made the following oft-quoted remarks:

c11.s1.t1.31

⁴⁶ JASB 6, 451–77.

c11.s1.t1.32

I was struck at their all terminating with the same two letters, 𑀤𑀸𑀦𑀁. Coupling this circumstance with their extreme brevity and insulated position, which proved that they could not be fragments of a continuous text, it immediately occurred that they must record either obituary notices, or more probably the offerings and presents of votaries, as is known to be the present custom in the Buddhist temples of *Ava*; where numerous *dwajas* or flag-staffs, images, and small *chaityas* are crowded within the enclosure, surrounding the chief cupola, each bearing the name of the donor. The next point noted was the frequent occurrence of the letter 𑀲, already set down incontestably as *s*, before the final word:—now this I had learnt from the *Saurashtra* coins, deciphered only a day or two before, to be one sign of the genitive case singular, being the *ssa* of the Pálí, or *sya* of the Sanscrit. “Of so and so the gift,” must then be the form of each brief sentence; and the vowel *á* and *anuswara* led to the speedy recognition of the word *dánam*, (gift,) teaching me the very two letters, *d* and *n*, most different from known forms, and which had foiled me most in my former attempts. Since 1834 also my acquaintance with ancient alphabets had become so familiar that most of the remaining letters in the present examples could be named at once on re-inspection. In the course of a few minutes I thus became possessed of the whole alphabet, which I tested by applying it to the inscription on the *Delhi* column. (460–1)⁴⁷

c11.s1.t1.33

⁴⁷ A more vivid and personal view of Prinsep’s breakthroughs is provided in the extracts of his letters “in his own spirited language” to Alexander Cunningham quoted by the latter in ASIR 1, ix–xvi. Typical of Prinsep’s “ardent enthusiasm” (xvii) are his cheers ““Chulao bhai, juldee puhonchoge”“ (“Go on brother, we shall soon get there,” in Cunningham’s note, x) and “Hurrah for inscriptions!” (xv).

c11.s1.t1.34

The alphabet presented by Prinsep at the end of this article (475) proves that he did indeed substantially decipher the early Brāhmī script; his identifications of the characters are all correct except for the one that he identified as *ṛ*, which is actually *jha*, and the five others (*gha*, *ṅa*, *jha*, *ña*, and *o*) that he was not able to find. He also correctly concluded that the script is the earliest form and parent of all other Indic scripts (474).

c11.s1.t1.35

Prinsep’s readings and translations of the Sāñcī donative inscriptions themselves are, however, only partially correct; his no. 25, for instance (462), is read as *Vajágato* *dánam* "The gift of VRIJÁGÁN," whereas the correct reading⁴⁸ is *vajigutasa dānaṃ* "The gift of Vajiguta." The inaccuracies in Prinsep's interpretations of the Sāñcī inscriptions (and of the other inscriptions he subsequently published) can be attributed only in part to the imperfect copies of the originals he had available to him; more significant was his inadequate knowledge—which he frequently and candidly confessed⁴⁹—of Sanskrit and other Indian languages. But although none of Prinsep's editions of early inscriptions can be considered anything like standard, this does not in any way diminish his great achievements, which were primarily those of a paleographer and decipherer rather than of a linguist and epigraphist.

c11.s1.t1.36

⁴⁸ By N. G. Majumdar in Marshall and Foucher, *The Monuments of Sanchi*, I.302.

c11.s1.t1.37

⁴⁹ For example, *JASB* 3, 487; 6, 452 and 567.

c11.s1.t1.38

Prinsep next turned his efforts to the interpretation of the Aśokan and other early inscriptions. In the same article (469) he presented a substantially correct reading of the introductory Aśokan formula *devānaṃpiye piyadasi lāja hevaṃ āha* (see p. 205) and some other recurrent passages. He also recognized (confirming his initial intuition; see p. 204) that "the language is ... not Sanskrit, but the vernacular modification of it," that is, Prakrit. But he did not yet realize the significance of the dialectal variations in the inscriptions from different regions, referring to them as "grammatical errors." Nor did he correctly recognize the author of these records, identifying the king called "Beloved of the Gods" with Devānaṃpiyatissa of Ceylon (473).⁵⁰

c11.s1.t1.39

⁵⁰ The correct identification of "Piyadasi" with Aśoka was made shortly afterward by George Turnour (*JASB* 6, 791), from his acquaintance with the Pāli historical texts of Ceylon.

c11.s1.t1.40

In the subsequent parts of *JASB* 6 and in the next volume (7, 1838), Prinsep published a series of articles giving complete readings and translations of various early Brāhmī inscriptions, including the Aśokan pillar inscriptions,⁵¹ the Nāgārjunī inscriptions of Aśoka's grandson Daśaratha,⁵² the inscriptions at Udayagiri and Khaṇḍagiri (including the Hāthīgumphā inscription),⁵³ the Aśokan rock inscriptions,⁵⁴ the Girnār (Junāgaḍh) inscription of Rudradāman,⁵⁵ and further inscriptions from Sāñcī.⁵⁶ In these articles (as also in his continuing series of articles on later inscriptions, mostly under the title "Facsimiles of Ancient Inscriptions," which had begun in *JASB* 5) the interpretations and translations continue to be uneven. In some passages, for example, in his translation of the Aśokan rock edict I,⁵⁷ Prinsep produced a reasonably accurate rendering, but elsewhere he strayed far from the mark. But here again it is his paleographic contributions which are lasting, as in *JASB* 7, 271–6, where he presented his "Completion of the Alphabet," a virtually perfect interpretation of Aśokan Brāhmī.

c11.s1.t1.41

⁵¹ *JASB* 6, 566–609.

c11.s1.t1.42

⁵² *JASB* 6, 676–9.

c11.s1.t1.43

⁵³ *JASB* 6, 1072–91.

c11.s1.t1.44

⁵⁴ *JASB* 7, 156–67, 219–82.

c11.s1.t1.45

⁵⁵ *JASB* 7, 334–56.

c11.s1.t1.46

⁵⁶ *JASB* 7, 562–7.

c11.s1.t1.47

⁵⁷ *JASB* 7, 249.

c11.s1.t1.48

While Prinsep's success as a decipherer looms large in Indian studies, his decipherment of Brāhmī usually gets little if any attention in general studies of decipherment of ancient scripts worldwide. This is at least in part because, as Prinsep himself realized from the outset (see n. 30), the decipherment of Brāhmī was relatively simple when “compared with the difficulties of the Persepolitan [i.e., Old Persian], or cuneiform character . . . or the more abstruse hieroglyphics of Egypt.”⁵⁸ Prinsep’s methodology was also less spectacular than that of the decipherers of cuneiform and Egyptian, mainly in that the decipherment of Bactrian and Indo-Scythic Coins”⁶¹ and by C. L. Grotefend (son of G. F. Grotefend, the famous decipherer of the Old Persian cuneiform script) in 1836 in “Die unbekannte Schrift der Baktrischen Münzen.”⁶² At this time the *e* and for the consonants *ma*, *ya*, *pa*, and *na* (330) and also came close to identifying a few others such as 𐨑 *kha*, corresponding to Greek X in the name of Antimachus (332). Although his immediate progress by this method was limited, it was ultimately to prove more fruitful.

c11.s1.t1.49

Meanwhile Grotefend, working along similar lines as Prinsep (though evidently not aware of the latter's article), made similar progress, correctly identifying eight characters (initial *a* and the consonants *ta*, *da*, *pa*, *ma*, *ra*,⁶⁷ *la*, and *ha*) from the correspondences between the kings' names in the Greek and Kharoṣṭhī legends. He also correctly identified some syllabic combinations such as *me* (in the name of Menander) and *li* (in Lysias).

c11.s1.t1.50

⁶⁷ Grotefend's reading of *ra* (op. cit., 311) was a pure stroke of luck. He identified it from the third consonant in the word which he read as *maha-rão* (correctly *maharajasa*) by a comparison with the word he read as *rao* in the Bactrian Greek legend on the coins of the Kuṣāṇa kings. This latter word, we now know, is not *rao* but the Iranian *shao*, with the Bactrian Greek character Ϸ representing *sh*. Thus Grotefend was lucky enough to correctly identify Kharoṣṭhī *ra* by comparing it with an incorrectly read Greek letter in an Iranian word for 'king'!

c11.s1.t1.51

Beyond this, however, both Prinsep's and Grotefend's efforts were largely blocked by their failure to realize that the language of the unknown script was an Indian one, and specifically a dialect of Prakrit. It was not yet clear at this time that the regions of the far northwest where these coins were found, corresponding to modern eastern Afghanistan and northern Pakistan, had been part of the Indian world in ancient times, and it was thus only natural that scholars of the time should have thought at first in terms of Iranian languages. As a result, Grotefend and Prinsep only vaguely and partially realized that the graphic system was of the Indic variety, with vowels attached diacritically to the consonants they followed, rather than left unindicated as in Semitic scripts or represented alphabetically as in Greek.⁶⁸ This led both of them into the crucial error of identifying the character 𐨯, which appeared as the final of every word in the Kharoṣṭhī coin legends, as *o*, which Grotefend (310) compared with the nominative ending in *-o* in the legends in Greek script on the Kuṣāṇa coins, and which Prinsep thought to represent a termination of either Semitic, Greek, or "Zend" (op. cit., 329–30); but both failed to recognize it as the Prakrit genitive suffix *-sa* (i.e., *-ssa*).

c11.s1.t1.52

⁶⁸ Cf. the comments by Lassen in *Zur Geschichte der griechischen und indoskythischen Könige in Baktrien*, 16.

c11.s1.t1.53

But in July 1838 Prinsep published his crucial article "Additions to Bactrian Numismatics, and Discovery of the Bactrian Alphabet,"⁶⁹ in which he announced his discovery of the true Indic character of the Kharoṣṭhī legends: "I threw off the fetters of an interpretation through the Semitic languages, and at once found an easy solution of all the names and the epithets through the pliant, the wonder-working *Páli*" (643).⁷⁰ He thus realized that the word-ending character 𐨯 was not *o* but the Prakrit genitive *sa* ("[coin] of . . ."),⁷¹ and this enabled him to read the first word of the legends correctly as *maharajasa* (instead of *malakáo*), and thence to correctly identify at least seventeen characters.⁷²

c11.s1.t1.54

⁶⁹ *JASB* 7, 636–55.

c11.s1.t1.55

⁷⁰ Here as elsewhere, Prinsep, like others of his time, uses "Pāli" to refer to middle Indic vernaculars, i.e., the Prakrits, in general; see section 3.1.5.2 n. 29.

c11.s1.t1.56

⁷¹ Prinsep mentions that he came to this realization by noticing "the same letter (affected with the vowel *i*) in two Greek names" (op. cit., 642) in nonfinal position. The names in question are presumably *lisiasa* (Lysias) and *pilasinasa* (Philoxenos).

c11.s1.t1.57

⁷² The initial vowels *a*, *e*, *i*, and *aṃ*, and the consonants *ka*, *kha*, *ja*, *ta*, *da*, *pa*, *ba*, *ma*, *ya*, *ra*, *la*, *ha*, and *sa* (op. cit., 639-43). Here, as elsewhere in connection with the decipherment of Kharoṣṭhī, it is difficult to give precise figures for numbers of characters identified, due to such complications as the high degree of graphic variability and ambiguity of several of the *akṣaras*.

c11.s1.t1.58

Besides identifying almost half of the basic consonants of Kharoṣṭhī, Prinsep at this point began to understand the graphic principles governing the script. He correctly noted that (unlike Brāhmī) the initial vowel signs were “formed by modifications of the alif as in the Arabic,”⁷³ while the postconsonantal vowels were indicated diacritically as in Brāhmī, and he correctly identified the diacritic forms for *i*, *e*, and *u* (but not *o*; 640). He did not clearly realize that there are no special signs for long vowels in the script but seemed to have some suspicions to this effect (640, 643). He also attempted, though with little success, to identify some consonantal conjuncts (643).

c11.s1.t1.59

⁷³ Op. cit., 639; see section 2.3.4.

c11.s1.t1.60

Although Prinsep is clearly the foremost figure in the decipherment of Kharoṣṭhī, a good deal of credit must also be given again⁷⁴ to Christian Lassen, who almost simultaneously⁷⁵ made major breakthroughs in his *Zur Geschichte der griechischen und indoskythischen Könige in Baktrien*. Working along the same general lines and from the same published materials as Prinsep, Lassen correctly identified eleven characters (*a*, *kha*, *ja*, *ta*,⁷⁶ *na*, *pa*, *ma*, *ya*, *ra*, *la*, and *ha*). He failed, however, to make the all-important discovery of *sa*, which he continued to read as a final *o*, for example, in *mahārājo* (33).⁷⁷ Lassen also made some important general observations about the script and language of the coin legends. He realized that the vowel *a* was inherent in all consonants, and that there was no *virāma* sign (18). He had a partial understanding of the vowel diacritics (21ff.), though less clearly stated than Prinsep’s; and like Prinsep, he also suspected the absence of long vowels (22). Moreover, he more accurately designated the language of the legends as “Prakrit” (as opposed to Prinsep’s “Páli”), noting similarities to the Prakrits of the classical dramas (55), though he arrived at this, ironically enough, for the wrong reason; since he still read *sa* as *o*, he read the epithet *mahatasa*, ‘great’ (Gk. MEΓAΛOY), as *mahato* (43), which he took as a Prakrit nominative (rather than genitive).

c11.s1.t1.61

⁷⁴ Cf. section 6.2.1 n. 44.

c11.s1.t1.62

⁷⁵ As noted by Windisch (*Geschichte der Sanskrit-Philologie*, 110), the foreword to Lassen’s book is dated August 1838, while Prinsep’s article appeared in July of the same year and had not yet been seen by Lassen.

c11.s1.t1.63

⁷⁶ Lassen (26-9) was unable to clearly differentiate *ta* and *da*, which are very similar in numismatic and some inscriptional varieties of Kharoṣṭhī. But as he was aware of the problem, it seems fair to credit him with at least one of the characters.

c11.s1.t1.64

⁷⁷ Lassen, however, did accept Prinsep’s correct identification of *sa* when it became known to him; see the notes in Roer’s English translation of *Zur Geschichte der griechischen und indoskythischen Könige in Baktrien*, JASB 9, 263 and 357 (= 13 and 44 of the separate reprint edition; see the bibliography for details).

c11.s1.t1.65

Despite the simultaneous discoveries by Prinsep and Lassen, the Kharoṣṭhī script did not fall into place at once as happened in the case of Brāhmī, where the breakthrough with the Sāñcī inscriptions almost immediately enabled Prinsep to compile a virtually complete alphabet of early Brāhmī. The discoveries from the Indo-Greek coins yielded less than half of the characters of the Kharoṣṭhī script, and the task of completing the inventory took far longer than with Brāhmī. Prinsep was not to make any further contributions to the subject in the short time left to him, although, as Cunningham remarked in 1840, “had James Prinsep lived, he would long before this have perfected what he had so successfully begun.”⁷⁸ The next major advance was thus left to Edwin Norris (1795–1872), who published in 1846⁷⁹ the first reliable transcription of a stone inscription in Kharoṣṭhī, namely Aśokan rock edict VII at Shāhbāzgaṛhī,⁸⁰ which had been discovered by A. Court in 1836. The text, presented synoptically with the corresponding Brāhmī versions at Girnār and Dhauli (306–7), is not perfect but is on the whole remarkably accurate given its early date and the imperfect reproductions then available. It is presumably on the basis of this text that Norris was able to compile his chart of “The Alphabet” (before 303), which is a remarkable tribute to his paleographic skill; it contains 35 simple characters, plus a few variants and conjuncts, nearly all correctly identified, and is thus the first reasonably accurate presentation of the script as a whole. Only the sign given for *tha*, [Figure:], is incorrect, this character being really an alternative form for *ṭha*.⁸¹ Norris also included a brief but accurate account of the system of vowel notation and of the formation of conjuncts with following *r*. Thus although Norris’ edition of the complete Shāhbāzgaṛhī edicts, promised for the next number of the JRAS, did not appear,⁸² his brief article nonetheless constituted a major step toward the complete decipherment of the Kharoṣṭhī script.

c11.s1.t1.66

⁷⁸ JASB 9, 887.

c11.s1.t1.67

⁷⁹ “On the Kapur-di-Giri Rock Inscription,” JRAS 8, 303–14.

c11.s1.t1.68

⁸⁰ Then known as Kapur-di-Giri; see n. 37.

c11.s1.t1.69

⁸¹ The precise phonetic value of this character has still not been definitely determined (see Brough, GD, 75–7, for the problems involved with this and related Kharoṣṭhī characters), so that Norris can surely be forgiven his error.

c11.s1.t1.70

⁸² This was finally accomplished by H. H. Wilson in JRAS 12, 1850; see section 6.2.3.

c11.s1.t1.71

In the meantime, Alexander Cunningham had been pursuing in a series of articles in the JASB the study of the bilingual coins and their legends, to the point that he was able to claim in 1845 (before Norris’ alphabet!) that he had “found the Ariano-Pali equivalent for every letter of the Sanskrit alphabet.”⁸³ In 1854 he claimed for himself “the discovery of the true values of eleven letters, or of just one-third of the Ariano-Pali alphabet,”⁸⁴ as well as of the value of several consonantal conjuncts. To Prinsep he credited the discovery of seventeen of the consonants, and to Norris only six. Much later in his life Cunningham voiced similar claims, adding, with a hint of pique, “It is perhaps curious to note, that though all these readings have now been generally adopted, scarcely one of them has been acknowledged as mine.”⁸⁵ Cunningham’s claims for a prominent role in the decipherment of Kharoṣṭhī are, however, difficult to evaluate. The “discoveries” he claims for various characters are scattered through his various numismatic studies (to which he often gives only vague references) and are usually presented there in readings of coin legends without any paleographic explanation. They are nowhere collected or discussed in any detail, and some of them are clearly wrong, for instance, *ng* (i.e., *ṅa*) based on a misreading of *ṇa* in the Māṇikiāla silver disk inscription (JASB 14, 431); in fact, Kharoṣṭhī has no character for *ṅa*. His claim of having discovered the characters for *ca* and *cha*⁸⁶ is typically misleading. In the original article to which he is presumably referring⁸⁷ the two characters are transliterated identically (*Chatrapasa apratihatachakrasa* . . . ), without accompanying commentary or explanation.⁸⁸ In any case, it is now known that the character [Figure:] identified by Cunningham as *chh* (i.e., *cha*) actually represents, or at least corresponds to, Sanskrit *kṣa* rather than *cha* (see 2.3.4).

c11.s1.t1.72

⁸³ JASB 14, 430.

c11.s1.t1.73

⁸⁴ JASB 23, 714.

c11.s1.t1.74

⁸⁵ “Coins of the Indo-Scythians,” *Numismatic Chronicle*, 3d ser., vol. 8, 1888, 201–48 (quoted 204; reprinted in *Coins of the Indo-Scythians, Sakas, and Kushans* [Varanasi: Indological Book House, 1971]). Cunningham also expressed similar sentiments in ASIR 1, xxxvii and xl–xli.

c11.s1.t1.75

⁸⁶ "Coins of the Indo-Scythians," 203; ASIR 1, xl.

c11.s1.t1.76

⁸⁷ JASB 23, 1854, 690.

c11.s1.t1.77

⁸⁸ In ASIR 1, xl, he apparently tries to clarify the matter after the fact by noting that "*Ch* is found in *aprati-chakra* [*sic*], 'invincible with the discus', *chh* in *chhatrapa* or Satrap."

c11.s1.t1.78

Thus while it is undoubtedly true that Cunningham did discover several of the letters which he claimed (e.g., *ga*, correctly read on the coins of Gondophernes and in the Māṇikiāla bronze casket inscription), his sketchy presentations and questionable interpretations make it difficult to confirm many of his claims. One therefore cannot help but suspect that his repeated claims to a major role in the decipherment of Kharoṣṭhī are somewhat inflated. Surely he played a significant part, for which he has been accorded due credit in subsequent publications (e.g., BIP, 18), but he cannot be said to have equaled the brilliant insights and fundamental contributions of Prinsep and his other predecessors. In balance it may be fairest to say that the decipherment of Kharoṣṭhī was a combined effort in which Prinsep again takes the place of honor, with Lassen and Norris making important contributions and Cunningham, Grotefend, and Masson playing significant secondary roles.

c11.s1.t1.79

By the sixth decade of the nineteenth century, the Kharoṣṭhī script had been substantially deciphered, although much remained to be done by way of refining the understanding of the script and applying it to the interpretation of the inscriptions, of which a good number had by then been discovered. This was a gradual process which continued at a slow pace, not only through the rest of the nineteenth but on into the twentieth century, and indeed up to the present day; for by no means have all of the intricate problems of Kharoṣṭhī paleography and orthography yet been solved (see n. 81). It will thus be clear that the decipherments of Brāhmī and Kharoṣṭhī took very different courses, and that the latter was ultimately a far more difficult problem. Although the first steps might have seemed relatively simple, given the presence of not one but thousands of little Rosetta stones in the form of the bilingual Indo-Greek coins, there were also some serious obstacles. First of all, the script has no modern descendants, such as were so helpful to the decipherers of Brāhmī; and though its roots in Semitic scripts were more or less clear from the beginning, the connections were not so paleographically transparent as to be of much help for the initial decipherment. Second, the decipherers were hampered by the linguistic situation; for even after the early incorrect assumptions about Pehlevi, Bactrian, and so on were discarded and it became clear that the language of the inscriptions was an Indian Prakrit, only very gradually did it become clear that the dialect in question, namely, what is nowadays known as Gāndhārī (3.1.4.2.1), was significantly different in grammatical and, especially, phonetic structure from the Prakrits known from literary sources. Still further obstacles were presented by the aberrant orthographic characteristics of Kharoṣṭhī, such as the lack of vowel quantity notation and the peculiarities and ambiguities in its formation of consonantal conjuncts (2.3.4). Moreover, reflecting its origins as a "clerk's script," Kharoṣṭhī has a high degree of graphic ambiguity, such that several common characters like *ta*, *da*, and *ra* may be difficult to distinguish even for those who are familiar with the script (see n. 76); such problems were noticed but not solved by early decipherers such as Prinsep.⁸⁹ The problem of graphic ambiguity was further aggravated for these pioneer scholars by their necessary reliance on numismatic texts, which are often imperfectly written, incomplete, or otherwise unclear; and while the repetitiousness of coin legends was helpful at the initial stages, the amount of data provided by a series of personal names and a small repertoire of titles and epithets ultimately proved to be limited. It is thus no coincidence that a tolerably complete picture of the alphabet began to emerge only when scholars turned to the study of recently discovered Kharoṣṭhī inscriptions (6.2.3 and 6.3).

c11.s1.t1.80

⁸⁹ JASB 7, 641.

6.2.3 Other developments during the era of decipherment

c11.s1.t2.1

After the pioneering works of Prinsep and Norris, further progress in the study of the Aśokan inscriptions continued at a slow but steady pace in the fifth and sixth decades of the nineteenth century. The Bairāṭ (or Bhābrā, Calcutta-Bairāṭ, etc.; see 4.3.1.1.2 n. 69) rock inscription, discovered by T. S. Burt in 1840 and published by M. Kittoe in the same year,⁹⁰ was the first known example of the separate or minor rock edicts. In 1850⁹¹ H. H. Wilson published a detailed study with synoptic texts of the fourteen rock edicts in the three versions then known, namely, Girnār, Dhauli, and Kapur-di-Giri (i.e., Shāhbāzgaṛhī). Though far from authoritative, Wilson's edition represents a major step in the reading and understanding of the Aśokan inscriptions.⁹² Several of the Aśokan edicts known at the time were also published by Eugène Burnouf in Appendix 10 of *Le Lotus de la bonne loi*.⁹³

c11.s1.t2.2

⁹⁰ JASB 9, 616–9.

c11.s1.t2.3

⁹¹ JRAS 12, 153–251.

c11.s1.t2.4

⁹² Despite his superior understanding of the text, however, Wilson (op. cit., 234–51) still refused to accept Turnour's identification of King Piyadassi of the inscriptions with Aśoka of the Ceylonese Pāli tradition. But he did at least retract his arguments against the Buddhist character of the inscriptions in his revised edition of the Bairāṭ inscription in JRAS 16, 1856, 357–67.

c11.s1.t2.5

⁹³ Paris: Imprimerie Nationale, 1852.

c11.s1.t2.6

During the same period progress also continued to be made in the study of the Kharoṣṭhī script and the inscriptions written therein. An important discovery was the biscript Brāhmī and Kharoṣṭhī inscription from Kanihiāra, first published by E. C. Bayley in 1854⁹⁴ and reedited by Edward Thomas in 1858 in his edition of Prinsep's *Essays on Indian Antiquities* (I.159–61). Among his extensive additions to Prinsep's articles in the same work, Thomas provided facsimiles and transcriptions of a few other Kharoṣṭhī inscriptions, including the Māṇikiāla stone inscription (I.143–6), the Bīmarān reliquary (I.105–9), and the Wardak vase (I.161–5), as well as a “Review of the Bactrian Alphabet” (II.144–67); the accompanying plate XI (facing page 166) summarizes the Kharoṣṭhī script as then known, with minor improvements in detail on what had been presented by Norris in 1846.

c11.s1.t2.7

⁹⁴ JASB 23, 57–9.

c11.s1.t2.8

In tive, mainly because he did not realize that the system involved was additive, rather than place-value as in later numerical notation (2.4.1.3); he therefore arrived at several wrong readings, interpreting, for example, the symbol for 300 as 3. Edward Thomas in 1850¹⁰¹ was the first to correctly recognize the additive nature of the system, that is, that "the figures employed were incapable of ... attaining any accession of power from relative position" (33), and he further explicated it in 1855 in his article on "Ancient Indian Numerals."¹⁰² Important supplementary information on this subject was gathered from the Nāsik inscriptions, which contain many dates and other numbers written in numerical symbols; this enabled J. Stevenson in 1853¹⁰³ to correctly identify the hitherto unknown figures for 10 and 20 and to recognize the basic character for 1,000, although he did not correctly interpret its various combinations. From this material, Thomas was able to publish an essentially accurate table of the early Brāhmī numbers in his edition of Prinsep's *Essays on Indian Antiquities*, plate XLa (facing page 84). Further advances were added in 1862¹⁰⁴ by Bhau Daji,¹⁰⁵ who gave a nearly complete chart of the system (231), and in 1877 by Bhagvānlāl Indrajī,¹⁰⁶ who clarified the values of the remaining problematic signs.

c11.s1.t2.9

¹⁰¹ "On the Dynasty of the Sáh Kings of Suráshtra," *JRAS* 12, 27–47.

c11.s1.t2.10

¹⁰² *JASB* 24, 551–71. See also his editorial notes in Prinsep's *Essays on Indian Antiquities*, II.80–4.

c11.s1.t2.11

¹⁰³ "On the Násik Cave-Inscriptions," *JBBRAS* 5, 35–57; see esp. fig. 18.

c11.s1.t2.12

¹⁰⁴ "The Ancient Sanskrit Numerals in the Cave Inscriptions, and on the Sah-Coins, Correctly Made Out," *JBBRAS* 8, 1863–66, 225–33.

c11.s1.t2.13

¹⁰⁵ But according to Bühler, "the results of [this] article belong chiefly to Bhagvānlāl Indrājī, though his name is not mentioned" (BIP 77 n. 5).

c11.s1.t2.14

¹⁰⁶ "On the Ancient Nāgarī Numerals," *IA* 6, 42–8.

c11.s1.t2.15

The first edition of Christian Lassen's *Indische Alterthumskunde* (1847–62) marked the first comprehensive effort at incorporating epigraphic evidence into a study of ancient Indian political and cultural history. In II.44 Lassen discussed the function and significance of inscriptions in a study of this type, following the famous remarks of Colebrooke quoted earlier (p. 202). Lassen's work set the pattern for future studies of early Indian history, with its heavy reliance on epigraphic and numismatic sources.

Linked texts (11)