Pramāṇa · Search · About · API

On the Origin of the Indian Brahma Alphabet c6: III

On the Origin of the Indian Brāhma Alphabet c6

Language
English

III

c6.1

If we now turn to the consideration of the oldest Indian inscriptions, it is not difficult to show that the palaeographic facts fully confirm the results, which the preceding examination of the ancient literature has yielded. They likewise show that writing, and especially the Brāhmī lipi, had had a long history in India, before king Piyadasi-Aśoka caused his Edicts to be incised in the various provinces of his large empire. In addition they permit us to recognise that the Brāhmī lipi is the real old Indian alphabet, which was popularly used in the third century B. C. all over India, and that it was fully developed [34] before the introduction of the Kharoṣṭhī, whence we may infer that it is the writing which the Buddhist canonical works mention so frequently.

c6.2

That the Brāhmī lipi was in the third century an ancient alphabet with a long history is proved by the very considerable variations in the forms of its signs found in the several versions of the Edicts, as well as by the recent discovery, on Sir A. Cunningham’s Eraṇ coin, of a variety which runs from the right to the left, and of another one, on the Bhaṭṭiprolu relic caskets, which includes a number of more archaic forms and independent developments.

c6.3

As regards Aśoka’s Edicts, even a cursory inspection of good impressions or of trustworthy facsimiles, such as those of Drs. Burgess, Fleet and Hultzsch, reveals the existence of numerous differences in the formation of the several characters. A more careful study shows not only that almost every letter¹ has two, three, half a dozen or more sometimes widely divergent forms, but also that certain peculiarities are confined to particular districts, as well as that the great majority of the apparently or really more advanced forms, which appear more or less constantly in the inscriptions of the next three or four centuries, are found already in the Edicts. And it is significant that the local differences observable permit us to speak of a Northern and a Southern variety of the ordinary Brāhma alphabet, between which, as in later times, roughly reckoning, the river Narmadā marks, the boundary line.

c6.4

¹ According to Tafel II of my Grundriss der indischen Palaeographie the only exceptions are the letters *U*, *jha*, *ña*, *ṭa*, *ṭha*, *ṇa*, *tha* and *na*.

c6.5

In order to show, how considerable the divergences are, I give here the eight chief varieties of the initial *A*,² among which the first and the last show hardly any resemblance, though if all eight are placed side by side their connexion is easily recognised.

c6.6

² The same varieties are of course found also in the initial *ā*.

c6.8

It is not difficult to see, that the first seven varieties have been caused by two conflicting tendencies, a liking for angles and a liking for curves, which are observable also in [35] the divergent forms of other Brāhma letters. Nos 1—3 are purely angular. The first and the second differ only thereby, that in the one the obtuse angle touches the vertical line, and that in the other it has been first made separate and then connected by a short crossbar.¹ In the third form the right hand stroke is slightly bent in the middle and the whole letter resembles a Greek Χι. Nos 6 and 7 show on the left, instead of the straight sides of the angle, two well developed curves, and in No 6 they are united in the middle, while in No 7 they do not touch. Nos 4—5 appear to be mixed forms, as the former has on the left a curve below with a straight stroke slanting towards the left, while in the latter the curve stands above and the straight stroke below. This apparent mixture may be due to an incomplete change of the older, angular form. But the two varieties may also be explained as cursive developments from No 6, the writer not caring to make the more elaborate curve twice. No 8, finally, is a purely cursive development from No 6, a straight stroke being substituted for the notched line on the left.

c6.9

¹ There are also instances in which this crossbar is omitted and the angle stands by itself, see *e. g. amisā*, Siddāpur I, l. 3.

c6.10

Now the local distribution of these forms, which with the exception of Nos 2 and 3 are of very frequent occurrence, is as follows. The angular forms Nos 1—3, as well as the variety mentioned in note 1, are confined to the Southern versions of the Edicts. They appear only in Girnār and Siddāpur, Dhauli and Jaugada, and it may be noted that in Girnār and Siddāpur they are in the majority, Nos 4—7 appearing only occasionally, while in the two Southeastern versions (with the exception of the Jaugada Separate Edicts) the contrary is the case. On the other hand I know of no case where a purely angular form is found in the Kālsī, Delhi, Bairāt, Sānchi, Allahābād, Mathia, Radhia, Rampūrva Barābar, Sahasrām and Rūpnāth Edicts. In the majority of these documents Nos 4—7 alone are used. No 8 occurs very frequently in Kālsī (sometimes with small variations) and once or twice in Rāmpūrva. Hence it would appear that the angular forms are southern peculiarities and that, as they are undoubtedly the more ancient ones, they furnish an instance of the conservatism, frequently observable in the southern alphabets of later times. This inference is confirmed by the fact that other ancient inscriptions of the same or nearly the same period, like those on the Kolhapur and Bhaṭṭiprolu relic caskets and from the Nānāghāt cave, likewise show the angular forms, (mostly side by side with the curved ones), while the documents, found north of the Narmadā, such as those on the Bharahut and Sānchi Stūpas and in the Nāgārjunī caves, as well as the coins of Agathocles offer almost exclusively the *A* with two curves, rarely the mixed form No 4. A solitary exception in Mahābodhi Gayā (Cunningham, op. cit., plate V, No 2) may be explained by the consideration that, as the dialect of several Gayā inscriptions shows, Southerners came to visit that famous place of pilgrimage, and that the person who wrote the copy may have been a Southron.

c6.11

In addition to the local differences in the form of the initial *A* and *Ā*, there are also others observable in *kha*, *ja*, *ma*, *ra* and *sa*, which may be briefly noticed here, though the full details must be reserved for the discussion of the palaeography of the Edicts in my *Grundriss*.¹ The *kha* with a circle² or loop at the foot, a very ancient form, alone is used in Kālsī and besides occurs only in Jaugada together with the simplified forms showing a dot instead of the circle or no appendage at all. The *ja* with a loop in the middle³ is used only in Kālsī. It seems to be a form peculiar to the extreme North and the Northwest, as it is found also on the coins of Agathocles and of the Taxila merchants (see below p. 48 f.). The other northern versions have mostly a secondary development from it, the *ja* with a dot in the middle (plate I, No 7, Col. V, c) more rarely the independent form (given under No 7, Col. V, d), which is used exclusively in Girnār together with its derivative, the later angular *ja* with three horizontal bars (see below p. 41). The *ma* with the angle at the top, an ancient form, (plate I, No 13, Col. V) occurs in the southern versions (with the exception of Siddāpur), where an open square appears instead of the angle) [37] and is used exclusively in Girnār. All the northern versions of the Edicts have a semicircle instead of the angle. The angular and wavy forms of *ra* and their insertion in the verticals of consonants are also peculiar to the southern versions (Girnār and Siddāpur); the solitary *ra* in the Rūpnāth Edict consists of an almost straight stroke. Finally, the ancient *sa* with a straight limb on the left (plate I, No 15, Col. VI, b) is likewise confined to the South (Girnār and Siddāpur).

c6.12

¹ See now the Grundriss der Indo-Ar. Phil. und Alt. I, 11, p. 34 ff.

c6.13

² See plate I of this paper, No. 19, Col. V, a.

c6.14

³ See plate I, No. 7, Col. V, b.

c6.15

These facts, to which some more might be added, are sufficient to show, that the very common idea¹ of the homogeneousness of the characters of the Edicts and of the absence of local varieties, is erroneous. The differences between the writing of the northern and the southern versions are quite as considerable as those, found four hundred forty years later, between the letters of the northern and southern inscriptions of the first and second centuries A. D.² And it must be kept in mind that the circumstances, under which the Edicts were engraved, were not favourable to a full expression of the local varieties of the letters. Copies were sent out from Pāṭaliputra into the provinces, which were recopied and, as the dialectic differences in the language and occasional peculiarities in the wording prove, also recast by the clerks of the district-governors, before they were made over for engraving to the masons. It seems only natural to assume that the characters of the copies prepared at Pāṭaliputra influenced the writing of the provincial clerks, and caused the introduction of forms, otherwise not usual in the several [38] provinces.¹ Such a suspicion is the more natural, as the provincial clerks have in no case completely changed the language, but have always allowed some Māgadhisms to stand. But, however that may be, local differences are traceable in the writing of the several versions and they prove that the Brāhma alphabet had had a long history before the third century B. C. If the slowness of the change of the forms, and the comparative insignificance of the local variations which the inscriptions of the next four or five centuries show, are taken as the standard, it will be necessary to assume that the letters of the Edicts had been used at least during four or five hundred years.

c6.16

¹ See *e. g.* Burnell, El. South Indian Palaeography, p. 7, note 4, with whose remarks Dr. Taylor, M. Halévy and others agree.

c6.17

² I state this in accordance with the facts, shown by Tafel III of my Grundriss, which includes *inter alia* the signs from the inscriptions of the Śakas and Kuṣanas of Mathurā, Kāman and Sānchi, the Western Kṣatrapas, the Andhras and Ābhīras. Dr. Burnell's statement (loc. cit.) that "in the course of a few hundred years [after Aśoka] the alphabets used in Gujarat and Bengal had already become so different as to be very little alike in appearance", I fear, cannot be substantiated. Between 200 B. C. and 200 A. D. there are no inscriptions from Bengal proper. The inscriptions from Mahābodhi Gayā in Behar (given by Sir A. Cunningham) which may be assigned to this period, look very much like those of the Śakas and the Western Kṣatrapas.

c6.18

¹ As I have stated already in Dr. Burgess’ Arch. Surv. Rep. W. I., Vol. IV, p. 79 f. and in the Ind. Ant., Vol. XII, 190, the clerks who prepared the fair copies of the inscriptions, not the masons or coppersmiths who engraved them, are the men who influenced the formation of the letters. How the masons worked, may be seen from two passages of the Kālsī version. In the twelfth Edict, l. 31 six letters have been scored out and the corrections have been written above the line. The letters in the upper row are as large as those in the lower and the distance between lines 30 and 31 becomes from the beginning of the corrected passage twice as great as it was before. It is evident that the mistake and its correction occurred in the MS. given to the mason. If they were due to the latter, the line would run on straighter and the letters of the correction would be smaller. Again in the fourteenth Edict l. 20 the syllables *ṭite* of the word *ghaṭite* are corrections, one standing above the line and the other below it. But the distance between *gha* and the first letter of the following word, *ma*, is twice as great as those intervening between the other letters of the line. Here it is again certain that the MS. had the mistake and the correction. If the mason had skipped the two signs and added them afterwards, the *gha* would not stand further off from *ma* than from its predecessor. I think that these two instances are sufficient to prove that Aśoka's masons copied quite mechanically. It seems, therefore, impossible to attribute to them any other influence on the shape of the letters than such as may be caused by a slip of the chisel or by their accidentally overlooking a stroke in the MS. before them.

c6.19

As regards the second important point, to which attention has been called above, viz., the occurrence of numerous apparently or really advanced forms, identical or closely agreeing with those of the later times, its significance will become best apparent, if all such signs, found in the Edicts, are placed together. The subjoined little table, for the preparation of which [39] I have to acknowledge the help of Dr. W. Cartellieri, gives in the lines, marked *A*, the advanced Aśoka letters photolithographed according to cuttings from the facsimiles, and in those, marked *B*, the corresponding characters from the later inscriptions of Hāthigumphā, the Nānāghāt, Mathurā and the Western caves. The dates of the latter vary between the middle of the second century B. C. and of the second century A. D., and in every case the oldest available counterpart has been chosen.

c6.21

The table shows that the later signs for sixteen letters occur already in the third century B. C. Four of these forms, the dagger-shaped *ka* (common in Kālsī and occurring in most other versions except in Girnār), the angular *gha* (Kālsī) and *ha* (No 21, in various versions) as well as the curved *ḍa* (Kālsī) are in reality archaic,¹ while the corresponding common signs of the Edicts are advanced developments, which like the peculiar *va* (No 18, Girnār, Sohgaurā), and *sa* (No 20, Kālsī, Jaugaḍa), have left no trace in the later writing. The remaining ones are really cursive or derived from cursive forms. No 1, the initial *A*, has been discussed above, its counterpart in line *B* occurs in the Kuṣana inscriptions from Mathurā. No 3, the *kha* with a loop to the right (Kālsī) is a cursive form for the *kha* with the circle at the foot,¹ and itself the parent of No 4 [40] (Delhi-Sivalik) which preserves the twist in the down stroke caused by the loop, but substitutes a dot for the latter.² No 5, the *kha* with the triangle (once in Mathia) is a fanciful variant for the oldest form, its counterpart has been taken from the archaic Mathurā inscriptions. No 6, the *ga* with the round top³ is a cursive form of the pointed letter, the corresponding form is from Hathigumphā, but found in all inscriptions of the second century B. C. No 8, the *cha* with two loops⁴ is a tertiary development, immediately derived from the form with the bisected circle, which again is merely cursive.⁵ The same remark applies to No 9, the *ja* with three bars. It is derived from the notched Girnār form, which itself is a cursive development from the Bhaṭṭiprolu form.⁶ Nos 11 and 19, the stunted *ti* and *vi*,⁷ give examples of the triangulation of the lower portion of *va* and of the reduction of the vertical strokes so characteristic of the alphabets of the next centuries, which appears already with great regularity in the Nāgārjunī cave inscriptions of Aśoka's grandson. No 12, the *da* with a shallow curve and the tail twisted towards the right (Jaugaḍa, Girnār, etc.) is a transitional form, corresponding to those in the Nāgārjunī cave, the archaic Mathurā and the Pabhosa inscriptions (*B*), and leading up to the *da* of the first [41] and second centuries A. D. Nos 13, 14 and 16 give examples of the introduction of angles in the lower limb [42] To an earlier stage than the writing of the Edicts belongs the inscription on Sir A. Cunningham's Eraṇ coin, which runs

c6.22

¹ See below pp. 58f., 61, 73. The counterparts in line *B* are from the Hāthigumphā inscription.

c6.23

¹ See below plate I, No 19, Col. V, I.

c6.24

² The corresponding forms in line *B* have been taken from Hathigumphā. Better ones for No 3 are found in the inscriptions from the Western caves, see Tafel III, 8, VIII of the Grundriss.

c6.25

³ Once in Delhi Sivalik Ed. VII and in the mason's alphabet at Mahābodhi Gayā.

c6.26

⁴ Kālsī once and mason's alphabet from Mahābodhi Gayā. The form in line *B* is from Hathigumphā.

c6.27

⁵ See below p. 68.

c6.28

⁶ See below plate I, No 7, Col. V, a and the discussion on No 7, p. 60 f., The sign, given above, occurs in Girnār, Ed. IX, l. 1. The great length of the central bar is caused by the *ā*-stroke which it includes. Similar forms occur in Kālsī, where they represent the looped *ja*. The corresponding form in line *B* occurs already in Daśaratha's Nāgārjunī inscription.

c6.29

⁷ The stunted *ta* is very frequent in Kālsī and occurs occasionally also in other versions. The triangular *va* with the very short vertical is found once in a correction in Kālsī, Edict XIII, 2, l. 13. The forms in line *B* have been taken from the Kuṣana inscriptions. Some instances of *pa*, *pha* and *sa* with very short verticals occur likewise already in the Edicts. where writing is taught according to the Bārākhaḍī, described above, which contains no ligatures and more sibilants than the Vernaculars possess, and where composition received little or no attention, being at the best confined to the copying of a few forms for letters. Another cause was the want of a settled system for the minutiae of vernacular grammar, both in writing and in speech (compare my remarks, Ep. Ind., Vol. III, p. 136) all efforts in this direction being of quite recent date. A third cause is, I fear, the deeply rooted tendency of all Hindus to inaccuracy in small matters. Now the Aśoka Edicts are official papers, written by his *lipikaras* or clerks. As there is no reason for assuming that they had received a better education than the men in the modern offices of British and native India, and that the Vernaculars of the third century B. C. had been polished and perfected by grammarians, it seems difficult to ascribe the defects in their spelling to other causes than those which produce the same imperfections in the office work of the modern Karkuns. The probability that the causes are the same increases, if it is remembered, that even the Sanskrit landgrants, issued from the secretariats of later kings, show the same mistakes, sometimes in a very high degree. Thus the Valabhī grant, published in the Indian Antiquary, Vol. VII, p. 68 f., has no long *ī*, very few long *ū*, invariably *dha* for *ddha* and almost regularly *śa* for *ṣa*.

c6.30

very few allusions to the use of letters are contained in the literary works which date from the fourth century B. C. The answer to this statement is contained in the second part of this paper, and it need only be pointed out that Dr. Burnell could not know of the passages mentioning private and official documents, as the works, in which they occur, had not been published in 1878. His second and accessory argument is drawn from internal evidence furnished by the Edicts. He says, El. South Indian Palaeography, p. 2, "The inscriptions of Aśoka are also in themselves proofs that writing was about 250 B. C. a recent practice; for they present irregularities of every kind," and in the note to the passage, he quotes as instances, conflicting spellings like *āñapitaṃ* (correctly *ānāpitaṃ*) and *ānāpitaṃ*, *dasaṇa* and *dasaṇa* (from the Girnār version), the irregular insertion of nasals before consonants (which, as he admits, may be due to the negligence of the masons) and the constant neglect of the reduplication of consonants *e. g.* in *piyasa* for *piyassa*, *ārabhisante* (read *ārabhisare*) for *ārabbhissare* and so forth. These facts are indisputable, and other similar ones like the irregular employment of the signs for sibilants in Kālsī, Bairāt II and Siddāpur and the constant or nearly constant use of short *i* and *u* for long *ī* and *ū* in Kālsī, Bairāt and Rūpnāth may be added. But they do not prove the proposition, in support of which Dr. Burnell adduces them. The numerous double and even treble forms of the same words, which occur in one and the same version are not graphic. Some are, as the analogies in the ancient literary Prakrits show, real variants which occurred in one and the same dialect, and some are due to slips in the translation of the Māgadhī originals of the Edicts into the Western and Northwestern dialects. The irregular use of the Anusvāra before consonants may also be ascribed partly to the same cause, since the Pali too occasionally omits a nasal and then doubles the following consonant. In other cases it may be due to the carelessness of Aśoka's clerks, who treated their Vernaculars exactly as a modern Karkun treats his. Everybody who has had any experience of Indian office work, must know that the *Sheras* or official papers, prepared by the clerks of the older generation, who had received their elementary instruction in the indigenous schools and had afterwards been trained in the offices, show an extreme irregularity in the use of the Anusvāras, of the short and long *i* and *u*, of the three signs for sibilants and of the reduplication of consonants. During my service as Educational Inspector I have seen a great many, sometimes important, documents from British Government offices and from those of native princes, which in their spelling were quite as bad as, and even worse than, Aśoka's Edicts, and I do not recollect that any papers, except those sent by carefully trained schoolmasters, were quite exact. One cause of this state of things was the wretched instruction in the indigenous schools, [43] from the right to the left.¹ The letters agree exactly with Aśoka's and the *dha* has the position required for the writing from the left to the right. It dates, therefore, from a period during which the Brāhma characters were written in both directions. This period is probably not very far distant from the middle of the third century B. C., as the Edicts still show single letters, which belong to the writing from the right to the left, viz., the *dha*, given in the first Comparative Table No 4, Col. V, a, the *O*, ibidem, No 6, Col. VI, f, and the *ta*, ibidem, No 22, Col. V, l. The coin may, therefore, be somewhat younger than Sir A. Cunningham thinks, who assigns it to 400 B. C. The great antiquity of its find-spot, the town of Eraṇ, is attested by an inscription on the Sānchi Stūpa, where it occurs as *Erakina*.

c6.31

¹ Coins of Ancient India, p. 101 and plate XI, 18. According to a plaster cast, which I owe to the kindness of Mr. Rapson, the inscription is *Dhamapālasa.o*, the last consonant being illegible.

c6.32

A still more important palaeographic witness for the antiquity of the Brāhma alphabet is the variety, found in the inscriptions on the Bhaṭṭiprolu relic caskets, the value of which, I am sorry to say, I have somewhat underestimated in the introductory remarks to my edition in the Epigraphia Indica, Vol. II, p. 323 ff. Though the article has been published only in 1894, it was written more than two years earlier, before I had begun to ing the details, the necessity for which will become more fully apparent in the next part of this paper, naturally forces me to modify the general proposition that the Bhaṭṭiprolu alphabet does not teach us much regarding the history of the Brāhma [45] writing and regarding the conversion of the Semitic letters into Indian characters. It certainly furnishes us with valuable intermediate forms for four radical signs, *c*, *j*, *l* and *ṣ*, which latter appears to be the parent of the ordinary Brāhma *ṣa* and *sa*, and with one entirely independent derivative sign *gh*. Its separation from the ordinary Brāhma alphabet must fall not only before the third century B. C., but also before the time, when the Eraṇ coin was struck, and cannot have happened at a later period than the fifth century B. C., though it may fall much earlier.

c6.33

This estimate carries us back to the period, for which the passages of the Jātakas, the Pācittiya and the Mahāvagga, quoted above The first point, which is of considerable general importance for Indian palaeography, will best become intelligible by a brief review of the epigraphy of those districts where Kharoṣṭhī inscriptions occur. From the third century we have the two Kharoṣṭhī versions of the Aśoka Edicts, incised in the northwestern corner of the Panjab, at Shāhbāzgarhī and Mansehra. To the same or possibly a somewhat earlier period belong Sir A. Cunningham's coins from the site of the Taxila or Takṣaśīlā, which prove the contemporaneous popular use of the Brāhma characters in Gandhāra.

c6.34

These coins have been figured by Sir A. Cunningham in his Coins of Ancient India, plates II and III. He has pointed out, op. cit. p. 61, that they are partly punch-marked silver pieces and partly single or double die copper pieces, all of the standard peculiar to India, and he takes them for this reason and on account of the very archaic forms of the letters of the legends, "to be anterior to the Greek conquest of Alexander". It will perhaps be safer to say "anterior to the Greek conquest of Demetrius". From their inscriptions, which are partly in Brāhma characters of the Kālsī type¹ and partly both in Brāhma and in Kharoṣṭhī letters, Sir A. Cunningham has already drawn the obvious inference, that both alphabets were used in Northern India during the third century B. C. Something more, it seems to me, may be elicited from an analysis of the legends.

c6.35

¹ The Kālsī type is visible in the looped *ja*.

c6.36

On the coin, plate II, 17, the Brāhma legend *Vaṭasvaka* corresponds to Sanskrit *Vaṭāśvakāḥ* and probably means "the Vaṭa-Aśvakas" or "the Aśvakas of the Vaṭa or 'fig-tree' division". It is well known that there was an Aśvaka tribe in Northwestern India, whom the Greeks call Assakenoi and state (Arrian, Indica, I. 1) to have inhabited the country west of the Indus as far as the Kophen. It may further be mentioned that some old Indian tribes, like the Yaudheyas² were actually divided into sections or *gaṇas*, as well as that, as the case of the Audumbaras shows, tribes were occasionally named after trees. With this explanation the coin appears to have been issued by one of the subdivisions of a tribe, which occupied just those districts, from which so many Kharoṣṭhī inscriptions come, and a Brāhma inscription on a tribal coin would certainly indicate that the alphabet was in popular use.

c6.37

² See, Sir A. Cunningham, Arch. Surv. Reports, Vol. XIV, p. 141 and plate XXXI, where coins of their second and third *gaṇas* are described.

c6.38

[47] The latter point comes out still more strongly through some other coins, figured on plate III,¹ viz.

c6.39

¹ Though the letters are perfectly distinct, Sir A. Cunningham gives erroneously *Nekama* as the reading of the Kharoṣṭhī legend of No 9. On the obverse of No 8 he reads *Rālimata*, and he takes *negamā* as the equivalent of the Greek Νόμισμα, for which explanation there is no authority (see also Ind. Ant., vol. XXI, p. 346). He correctly points out that, on plate II, Nos 21 & 22 have the inscription *Kādasa*, i. e. *Kālasa*, in Brāhma letters of the type of the Edicts, and asserts that No 13 on plate III bears the Kharoṣṭhī legend *Paṃca Nekamma*. According to the autotype the third sign is not *ne* but *a* and the fourth looks like *kra*.

c6.40

| Obverse | Reverse |
| No 9 *Dujaka* (Kharoṣṭhī) | *Negamā* (Brāhmī) |
| No 10 *Dojaka* (Brāhmī) | *Negamā* (Brāhmī) |
| No 8 *[T]ālima[ta]* (Brāhmī) | *[N]ega[m.]* (Brāhmī) |
| No 11 *A[taka?]takā* (Brāhmī) | *Negam[ā]* (Brāhmī) |

c6.41

The word *negamā* is common enough in Pali and in the epigraphic Prakrits, and means always ‘the traders’. It shows here that the coins are mercantile money-tokens, issued by traders, and the words on the obverse may be either names of towns or of guilds.² The latter explanation is perhaps the more probable one. But however that may be, the use of both alphabets by traders proves indeed that both were in popular use in the heart of Gandhāra.

c6.42

² I am unable to explain *Dujaka-Dojaka*. *Tālimata* seems to be connected with the Sanskrit *Tālī* “Corypha Taliera” or “Flacourtia cataphracta”.

c6.43

The other coins of the same period, struck outside the Panjab, show legends in Brāhmī, even that from Mathurā, op. cit. plate VIII, 1, in which town also a votive inscription in Brāhma characters of the third century (Reports Arch. Survey, Vol. XX, plate VI) has been found. Only in one case Kharoṣṭhī letters have turned up further south, but under circumstances, which do not allow the inference that the alphabet was generally used or known. This case occurs in the Siddāpur Edicts where the writer Paḍa has added at the end his qualification *lipikareṇa* 'the scribe' in Kharoṣṭhī characters. This looks like a joke or a boast, as if Paḍa, proud of his accomplishments, had been anxious to make it apparent that he knew more than the ordinary characters. And, as he was [48] in the royal service, it is not unlikely that he may have acquired a knowledge of the Kharoṣṭhī during a stay in a northern office, or may have been a Northerner by birth.¹

c6.44

¹ See the Wiener Zeitschrift f. d. Kunde d. Morgenlandes, Vol. VII, p. 30 f. Dr. Grierson, Ind. Ant., Vol. XXIV, p. 247, points out an analogous practice of the modern Kāyasths of Bihār who sign the documents, written in the Kaithī character, with their names in Arabic letters.

c6.45

From the second and first centuries B. C. we have chiefly legends on coins, which were struck in the Panjab or in non-Indian countries further west. The Indo-Grecian kings generally use Kharoṣṭhī letters, but Agathokles and Pantaleon employ also Brāhma characters, showing thereby that this alphabet likewise continued to be used in the Northwest, side by side with the Kharoṣṭhī. The same fact is proved by the double legends in Kharoṣṭhī and Brāhma letters on Sir A. Cunningham's Audumbara and Kuninda coins, op. cit. pl. IV² and V, 1—6, which come from the same districts and probably belong to the same time. Further east in Kosambī, Ayodhyā (op. cit., pl. V, 7—18 and IX) and Pancāla (pl. VII) as well as further south in Ujjain (pl. X) none but Brāhma letters occur. And it is curious that even the Yaudheyas (pl. VI), who were settled on the lower Satledge, use only the latter characters. A few single letters on the gateway of the Bharahut Stūpa, among which there is a *sa* of the second century, (Cunningham, the Stūpa of Bharahut, pl. VIII), are probably marks of northern masons, who erected this additional portion of the monument for Dhanabhūti.

c6.46

² The *Vṛṣṇi*-coin No 15 is according to the evidence of the characters later and probably belongs to the period when the Śakas had carried the Kharoṣṭhī as far as Mathurā. A few votive inscriptions of the second and first centuries B. C. from the same town are in Brāhma characters, Epigraphia Indica, Vol. II, p. 195.

c6.47

During the period of the Śaka and Kuṣana kings, from the first century B. C. to the end of the second century A. D., the numerous inscriptions, incised in the Western Panjab, are in Kharoṣṭhī and those on Dr. Bhagvānāl’s Lion Capital prove that the Kharoṣṭhī penetrated during the reign of the Śaka Satrap and king Śuḍasa or Śoḍāsa as far as Mathurā, where however, as the enormous quantity of Jaina and Bauddha votive inscriptions shows, the Brāhma alphabet was at that time and later the usual one. [49] The coins of the earlier Śaka kings from the North like those of Mauos and Azes and of their Satraps, as well as those of Gondopherres and his brothers, have Greek and Kharoṣṭhī legends, while the later Kuṣana kings, Kaniṣka, Huviṣka and Bazodeo or Vāsudeva discard the latter, and their successors finally adopt the Brāhma alphabet. Further east and south the king and Satrap Rājubula and his son Śuḍasa, who ruled over Mathurā and perhaps over other portions of the eastern Panjab, either follow the same practice as Mauos or use Brāhma letters (Cunningham, op. cit., pl. VIII, 2—5

c6.48

The second point, the improbability that the Kharoṣṭhī was already developed or in general use even in its home as early as say 500 B. C. requires here only a few remarks as the details have been discussed fully below in Appendix I. Its derivation from the Aramaic alphabet has been generally accepted, ever since Mr. E. Thomas pointed it out, and the shape of its *ba*, *na*, *ra* and *va* makes a doubt impossible. According to Dr. I. Taylor's suggestion, The Alphabet, Vol. II, p. 261, which [50] is most probable on historical and palaeographical grounds, its immediate source is the Aramaic alphabet of the Akhaemenian period, and the introduction of the Aramaic letters into India dates from the time, when the Akhaemenian kings of Persia had conquered the Panjab, just those districts where the Kharoṣṭhī has its real home.¹ As the Persian con- [51] quest probably happened shortly before 500 B. C., it is impossible that the Kharoṣṭhī can have been developed before 450 B. C., and it is not to be thought of that it could have penetrated into Eastern India, where the Buddhist Canon was composed, during the fifth century, much less could it have been there in so general use, as the alphabet mentioned by the ancient Buddhists certainly was. Under the circumstances just discussed, the assumption that the alphabet, referred to in the Jātakas, the Mahāvagga and so forth is the Brāhmī—which, I repeat, the palaeographic facts contained in Aśoka's Edicts, the Indian letters on the Persian Sigloi, the legend of the Eraṇ coin, the Bhaṭṭiprolu inscriptions strongly suggest—undeniably gains a very high degree of probability.

c6.49

¹ M. Halévy, op. cit., p. 250 ff. and particularly p. 267, tries to establish a still later date for the elaboration of the Kharoṣṭhī. He seeks the prototypes of its characters in the Aramaic alphabet of the fourth century B. C. He believes that they were developed in Ariana not earlier than 330 B. C. after the appointment of Macedonian Satraps by Alexander, who, as he thinks, may have fostered the use of the already prevalent Aramaic writing and thereby may have induced their subjects to adapt it to their special wants. The obvious weakness of M. Halévy's arguments prevents my accepting his theory, which would be more favourable for my point. It is a matter of course that I am likewise unable to agree with M. Halévy's theory (op. cit., p. 280—286 and Plate II, A) according to which the Brāhmī lipi has borrowed from the Kharoṣṭhī six radical signs, *śa*, *jha*, *ḍa*, *ña*, *u* and *ra*, the Anunāsika, the system of vowel-notation, and the numeral signs 4—9. The reasons, why I do not consider the six Mātṛkās and the vowel-notation of the Brāhma alphabet as loans from the Kharoṣṭhī, will become apparent in the next section of this Essay. With respect to the Anunāsika, I must point out that the sign, 󰁛, which M. Halévy derives from the Kharoṣṭhī *ma* 󰁜 of the Indo-Grecian coins, appears in no Indian document written in India before 1200 A. D., while the Kharoṣṭhī went out of use about 200 A. D. As regards the assertion that the Brāhma numeral signs for 4—9 are the initial Kharoṣṭhī letters for the words *catur*, *pañcan*, *ṣaṭ* (old Prakrit, *sa*, *chha* or *saḍ*), *saptan* (Prakrit *satta*), *aṣṭan* (old Prakrit *asta*, *aṭṭha*) and *navan* (old Prakrit *nava*), which is also put forward by Dr. Taylor (The Alphabet, II, p. 266), it is unfortunate that in four cases the facts are absolutely against it. The Kharoṣṭhī *cha* 󰁝 no doubt looks like the Kṣatrapa sign for four. But the Indian words for four, all begin with the unaspirated *ca*. Again, the Aśoka signs for six (see below plate III) have on resemblance to the Kharoṣṭhī *sa* [Figure:]. Further the Kṣatrapa sign for seven [Figure:] or Gupta [Figure:] cannot be the Kharoṣṭhī palatal *śa* [Figure:], because the word for seven begins in all Indian dialects with a dental *sa*. Again, the Kṣatrapa and Andhra [Figure:] cannot represent the Kharoṣṭhī *Aś*, because no such ligature is ever formed in any Indian writing. Besides, if it had been formed, it would be wrong. The second part ought to be the lingual *ṣa* [Figure:], because the word is *aṣṭan* not *aśtan*. Finally, the nought of the ninth and later centuries O cannot be derived from a sign for 10 expressed by a circle, because no such figure exists, nor from the entirely dissimilar Kharoṣṭhī *da* [Figure:]. The Hindus call the nought explicitly *śūnyabindu* “the dot marking a blank”, and about A. D. 500 they marked it by a simple dot, which latter is commonly used in inscriptions and MSS in order to mark a blank (see the Grundriss, I, 11, pp. 78 and 86), and which was later converted into a small circle.

Linked texts (21)