Pramāṇa · Search · About · API

On the Origin of the Indian Brahma Alphabet c7: IV

On the Origin of the Indian Brāhma Alphabet c7

Language
English

Contents

IV

c7.1

As the literary evidence points to the common use of writing in India during the fifth and perhaps in the sixth century B. C., and as the palaeographic evidence proves the Brāhmī lipi to be the oldest Indian alphabet and to be probably identical with the script referred to in the Buddhist Canon, it is a matter of course that its source must be found in the more ancient Semitic characters. A short time ago such a result would have precluded the possibility of all attemps to make Southern Arabia the country from which the parents of the Brāhma letters came. But at present, pretensions to a high antiquity are put forward on behalf of various epigraphic documents from the latter country. The theories, it is true, are still conflicting. Some ascribe certain Sabaean inscriptions to the tenth century B. C. or an earlier period, while others declare those of the Minaeans to belong to very ancient times and deny the antiquity of the Sabaean documents. More light on these questions is required, but even as matters stand at present, it seems to me hazardous to make use of the old argument, that the derivation of the Brāhma letters from a South-Arabian source is *a priori* impossible, because the inscriptions, found there, are of too modern dates.

c7.2

[52] Nevertheless the theory of a South-Semitic origin of the Brāhma alphabet appears to me untenable. What has been brought forward in its favour by the two chief advocates, is to a great extent far from convincing. The ocular evidence, it seems to me, speaks against many of their identifications.¹ Further, in some cases late or at least secondary forms of Indian letters have been used for comparison.² Again in other cases the most extraordinary and absolutely unnecessary changes in the phonetic value of the signs are assumed to be quite natural and hardly worthy of an explanatory remark. When one is told that the Hindus utilised the Semitic *Qoph* to express the sound *cha* (Deecke) or *cha* and *ka* (Taylor), that they employed the Semitic *Shin* for *ja*, *śa*, *ṣa* and *sa* (Deecke) or at least for *ja* and *ṣa* (Taylor), that the Semitic *Wāw* furnished both the signs for *ya* and *va*, or that *Lamed* became *ra* and *Resh* was turned into *la*, one can only say that such assertions are hard to believe. As the Hindus are always very particular, even pedantic, in matters connected with phonetics, and as the framers of the Brāhmī lipi evidently have been careful with respect to the formation of many derivative signs, duly deriving *ḍha* from *ḍa*, *pha* from *pa*, *bha* from *ba* and so forth, it seems incredible that they should have had no regard for phonetic affinities in utilising the signs, which they borrowed. And the assumed vagaries appear absolutely inexplicable if one considers that the Semitic alphabet has the *Tsade*, the phonetic value of which comes close to *ca* and *cha*, and that it possesses separate signs for *Wāw* and *Yod*, which might be used to express *va* and *ya*.

c7.3

¹ That is also Sir A. Cunningham's statement, Coins of Ancient India, pp. 39—40.

c7.4

² To the former class belong Dr. Deecke's initial *A* and Dr. Taylor's broad-backed *śa*, to the latter Dr. Taylor's initial *A*, *kha* and *ra*.

c7.5

As long as the theories regarding the derivation of the Brāhma alphabet contain equations like those just mentioned, and as long as these theories do not take into account all the oldest forms of the Indian letters, the problem, it seems to me, has not been solved and the task has to be redone. Really trustworthy results can only be attained under the following conditions:—

c7.6

[53] (1) The comparison must be The identifications are shown in the accompanying Table, kindly drawn for me by Mr. S. Pepper and etched by Messrs Angerer and Göschl of Vienna. The signs of the first two columns have been taken from M. Ph. Berger's Histoire de l'Écriture dans l'Antiquité, pp. 185 and 202 and those of Col. III with the exception of the Tāw from Sinjirli (22. III. b) from Professor Euting's Tabula Scripturae Aramaicae (1890). Among the forms of column IV without asterisks, those opposite the twelfth and the seventeenth Semitic characters [54] have been taken from a plaster cast of Sir A. Cunningham's Eraṇ coin (see also Coins of Ancient India, plate XI, 18) and that opposite the sixteenth character from the word *eṣā* in the Kālsī version, Ed. XIII, 1, l. 37.

c7.7

Among the signs in column V and VI a few are from Sānchi, viz. the lingual *ḷa*, the fifth of Col. VI opposite the fourth Semitic character, and the first of Col. V opposite the sixteenth Semitic character. Further, in Col. VI the fourth sign opposite the sixth Semitic character, *Ū*, comes from Bharahut, the first opposite No 15 from the Eraṇ coin the third opposite the fifteenth Semitic character in Col. VI has been taken from the Ghasundi slab, and two signs of the two rows in the same Col. opposite the sixteenth character, viz. the second in the first row, *AI*, and the last in the second row, *Ī*, belong to the Hathigumphā and Nānāghāt inscriptions. All the remaining letters have been drawn according to Dr. Burgess' facsimiles of the Rock and Pillar Edicts,¹ and of the Bhaṭṭiprolu inscriptions.

c7.8

¹ The facsimiles of the Girnār and Kālsī versions, on which I have chiefly drawn, have been published, Epigr. Ind. vol. II, p. 447 ff. The Pillar Edicts have appeared *op. cit.*, p. 245 ff. and the Bhaṭṭiprolu inscriptions *op. cit.* p. 323 ff.

A. The Borrowed Signs

c7.s1.1

Before proceeding to the details, it will be desirable to state the principles, on which the twenty-two signs of the Semitic alphabet seem to have been turned into Brāhma letters. Even a superficial examination of the Brāhma alphabet shows the following chief characteristics:—

c7.s1.2

(1) The letters are set up as straight as possible, and they are with few exceptions made equal in height.

c7.s1.3

(2) The majority consists of vertical lines with appendages attached mostly at the foot, occasionally at the foot and at the top, or rarely in the middle; but there is no case where an appendage has been added to the top alone.

c7.s1.4

[55] (3) At the top of the characters appear mostly the ends of vertical lines, less frequently straight horizontal lines, still more rarely curves or the points of angles opening downwards, and quite exceptionally in *ma* (No be turned topsy-turvy or to be laid on their sides, while the triangles or double angles, occurring at the tops of others had to be got rid of by some contrivance or other. A further change in the position of the signs had to be made, when the Hindus began to write from the left to the right. They had, of course, to be turned from the right to the left, as in Greek. [56] Instances, where the old position has been preserved, are however met with both in borrowed and derivative signs.

c7.s1.5

With respect to the single borrowed letters I have to add the following remarks.

c7.s1.6

No 1. That the Hindus, like the Greeks used the Semitic *Aleph* in order to express the vowel *a* has been suspected by Professor Weber. His successors have asserted the identity of the first letters of the two alphabets as a fact, though the Indian forms of the initial *A* with two curves, which have been mostly used for comparison,¹ do not agree very well. The doubtlessly oldest form with the angle to the left of the vertical (Col. V) which, as stated above, is peculiar to the Southern versions of the Edicts, agrees almost exactly with the oldest Semitic signs. Only the vertical has been placed at the point of the angle in order to make the appearance of the letter more regular, and the whole sign has, of course, been turned from the right to the left. On a Patna seal it occurs, however, once in its original position, see the Grundriss, Taf. II. 1. I, and p. 8.

c7.s1.7

¹ M. Halévy, however, correctly compares the angular form. His attempt to derive it from the Greek Alpha is not intelligible to me. The two signs possess a minimum of resemblance to each other.

c7.s1.8

No 2. I can only agree to the identification of the Brāhma *ba* with the Semitic *Beth*, which has been asserted on the strength of the Sabaean and Ethiopian forms² by Professor Weber and his successors, with the exception of M. Halévy, who derives it from the dissimilar Greek *Beta*. But I think that it is a development, derived by the Hindus immediately from the old Phoenician and Mesa signs. As the Hindus did not tolerate heavy tops, they opened the triangle and made the central stroke hang down from the end of the angle in the manner shown in the hypothetical form in Col. III.¹ The next result was a rhomboid form (Col. V, a), a few examples of which are found occasionally in various versions of the Edicts, where we have *e. g.* [Figure:] *bo* in *apalibodhāye*, Kālsī Edict V, l. 15. The liking of the [57] Hindus for straight top-lines produced further the square and the oblong *ba*, which latter is the commonest form in Kālsī and the regular one in the Bhaṭṭiprolu inscriptions. It may be noted, that this letter is frequently, though not regularly, made smaller than those with verticals. Regarding the want of the lower line in the derivative *bha* (Col. VI) see below.

c7.s1.9

² The Sabaean form is no doubt very similar to the Brāhma sign but, I think, due to a different modification of the ancient North Semitic letter. It would seem that the triangle at the head of the letter has been turned into a square and the stroke on the right has been made straight. Hence arose first the 月 and by a simplification ㄇ.

c7.s1.10

¹ An apparently analogous modification of the old Semitic Beth which one might be inclined to see in the b² [Figure:] of the Orkhon inscriptions (V. Thomsen, Insr. de l'Orkhon déchiffrées, p. 24) is disposed of by Prof. Donner's remarks on the sign in his essay, sur l'origine de l'Alphabet Turc, p. 57 f.

c7.s1.11

No 3. The identity of the Indian *ga* with the Semitic signs is so evident that it has been always recognised. I will only add that there are numerous instances in the Edicts where one of the sides of the angle is shorter than the other. The formal looking very pointed angles with perfectly straight equal sides are characteristic of the Western and Southern versions.

c7.s1.12

No 4. The identity of the Brāhma *dha* with the Semitic *Daleth* has also been generally admitted, Dr. Taylor, who derives it from *tha* (No 9, Col. V) alone dissenting. To me it seems that the first form, given in Col. V, with the vertical on the right, is the immediate offspring of the oldest Semitic forms, the letter having been made higher and the right hand stroke straightened in accordance with the general principles of formation, stated, above. The round back is probably due to a cursive development. In the Edicts this form is rare, and the second, given in Col. V, which is younger and due to the change in the position required by the change in the direction of the writing, is found in the great majority of cases. The older one (Col. V, a) is used exclusively in the Bhaṭṭiprolu inscriptions and usually in the Western cave inscriptions of the Nānāghāt, Nasik, Karle, Junnar and so forth, which range from about 150 B. C. to 200 A. D. It is also the parent of all the later northern forms of the letter, including the present Devanāgarī *dha*. In the caves mentioned we find also in the Andhra and other inscriptions of the first and second centuries A. D. numerous triangular forms¹ like ⨞ or ⨞. They may be possibly ancient survivals, especially as the Edicts too occasionally offer forms with half angular backs like [Figure:].

c7.s1.13

The difficulty with respect to the phonetic value of the Brāhma sign, which does not exactly correspond to that of [58] the *Daleth*, may perhaps be explained by the suggestion that the letter was used for than the year 725 B. C. As this sign belongs to the eighth century B. C., when the very archaic forms of the *Aleph*, *Daleth*, *Cheth*, *Theth*, *Wav* and *Qoph*, represented by the corresponding Brāhma letters, still existed,¹ it may be considered the prototype of the Indian *ha*. The aversion of the Hindus against heavy tops, of course, caused the sign to be turned round, and their liking for regularity caused the bar to be attached to the stroke which then became the base-line, and the whole letter to be set up straight. All things considered, this explanation seems [59] more probable than attempts to connect *ha* with the later more exactly agreeing sign in Col. III, b, or with the still Edicts, given in Col. V, b—d, are all cursive, and represent attempts to make the letter with a single stroke or at least with two. The third form with a dot at the junction of the two curves, is of course a development from the second with the loop. And the substitution of a dot for a loop indicates that the persons who invented it wrote with pen and ink. A man working with a stylus would not think of such a substitution, nay could hardly effect it.¹

c7.s1.14

¹ Regarding the use of ink in India, see above p. 22. The ancient Indian term for ink is *maṣi*, often spelt *masi*, and now pronounced *makhi*, which is derived from the obsolete verb *maṣ*, *hiṃsāyām*, compare *maṣa* and *maṣmaṣā*. Etymologically and originally it means 'powder' (of charcoal and the like), used for the preparation of ink, see the larger Petersburg Dictionary under *masi*, and the Grundriss, I, 11, p. 91.

c7.s1.15

No 8. The discovery of the angular *gha* (Col. V, a) which occurs a few times in the Kālsī version² and is constant in the cave inscriptions, makes its identification with the three-barred [60] *Cheth* of the archaic Phoenician, of the Sinjirli and other ancient inscriptions, very probable. The Semitic sign, which often slants towards the left, has been laid on its side (see the hypothetical form in Col. IV) and the long upper bar, covering the letter, has been raised upwards and converted into the long vertical on the left.³ The usual form of the Aśoka Edicts with the curve is of course cursive. In the Bhaṭṭiprolu alphabet this letter has been discarded and a derivative from *ga* (No 3, Col. VI) takes its place. The Sabaean *Cheth* 𐩢, from which Dr. Taylor derives the Indian *gha*, is no doubt an analogous development from the old North-Semitic sign.

c7.s1.16

² The sign given in the Table has been taken from *upaghāte*, Ed. XIII, 1, l. 37.

c7.s1.17

³ It may be mentioned as a curiosity, that from the twelfth century A. D. the Nāgarī *gha* is again placed upright and has become almost identical with its Semitic prototype, see the Grundriss, Taf. V. 13. XX—XXII. This is, however, not due to a survival, but merely to the desire of equalising the breadth of the Nāgarī letters.

c7.s1.18

No 9. The identity of the Brāhma *tha* with the Semitic *Theth* has always been recognised and is evident enough. I may mention as a curiosity that a small cross appears inside the circle instead of a dot in the Pallava landgrant, Epigraphia Indica, Vol. I, p. 1 ff. In this document all dots are replaced by minute crosses. On the other hand, a *Theth* with a dot occurs on an Assyrian weight, see Col. III.

c7.s1.19

No 10. The derivation of the Brāhma *ya* from the ancient Semitic *Yod* has been asserted by Professor Weber, who however could only compare the later round Phoenician form, which opens downwards.¹ The oldest Indian form is probably the notched one (Col. V, a stroke slants sharply to the left and the angle on the left is attached very close to the top, compare also the signs of the Siloah inscription, in Prof. Euting's Tabula Script. Aram. Col. 3. In setting the sign up straight, the Hindus took the upper stroke of the angle for the top of the down-stroke, and a cross-like figure was the natural result, which, on being made more regular, yielded the dagger-shape.

c7.s1.20

No 12. For the identity of the Brāhma *la* (Col. V, a) with the Semitic *Lamed* (Cols. I and II), which has been recognised by Professor Weber and accepted by Dr. Deecke and M. Halévy, the *la* of the Eraṇ coin (Col. IV), the Bhaṭṭiprolu *la* (Col. V, b) and the Bhaṭṭiprolu *ḷ* (Col. VI) furnish new illustrations. The most ancient survival is found in the last mentioned character, which, if the little bar denoting the lingualisation is removed, almost exactly represents the Semitic sign. Next comes the letter of the Eraṇ coin, which has preserved the original direction of the *Lamed*, but shows on the right the little bar, no doubt intended to mark the end of the line. This bar is wanting in the Bhaṭṭiprolu *l*, which however has been turned from the right to the left and has developed a curious long tail, hanging down from the top of the main line. Originally this appendage no doubt was a flourish, marking the end of the vertical.

c7.s1.21

No 13. The identification of the Brāhma *ma* with the Semitic *Mem* is also due to Professor Weber. He, as well as M. Halévy, derive it from the Aramaic sign of the seventh century¹ B. C., 𐡌. Though this derivation is no doubt tempting it seems to me that it is equally possible to connect the Brāhma *ma* with a sign like that of the Mesa inscription, which likewise shows a curve at the end of the down-stroke. A little stronger development of this curve would necessarily lead to a sign, like the hypothetical one, entered in Col. IV, which is almost the same as the first Pehlevī *Mem* of Col. 58 in Professor Euting's Tabula Script. Aram. And in support of this view I may appeal to the development of the looped *ka*, Northern ꦏ and Southern ꦏ from the dagger-shaped *ka* with a curve at he foot ꦏ, as well as of the round ad Southern *ra* 𑀭 from 𑀭, which show analogous changes on purely Indian ground. The superposition of the angle, which originally remained at the side, over *ña* (Col. VI, a), which consists of the hypothetical form, turned from right to left, and of a small bar at the top, indicating that it belongs to a different class or Varga.

c7.s1.22

¹ See M. de Vogüé, loc. cit.

c7.s1.23

Later only, I think, the Hindus substituted as a cursive development for the hook at the side, the single bar which protrudes on either side of the foot of the vertical.

c7.s1.24

No 15. Professor Weber doubtingly proposes to connect the Brāhma *sa* (Col. VI, d)

c7.s1.25

of the Edicts. The northern variety of the latter developed a new *sa* (Col. VI, d) with a curve instead of the straight limb at the side, and out of this the sign with two curves, one below the other (Col. VI, e), which occurs in the Kālsī version, in Daśaratha's Nāgārjunī cave inscriptions and so forth, and which probably is also intended for *ṣa*.

c7.s1.26

It would, of course, be possible to connect the Bhaṭṭiprolu *ṣ* with the cursive Aramaic *Samech* (Col. III) of the sixth century B. C. But then the cross bar of the Bhaṭṭiprolu sign would remain unexplained. Moreover, there would be the insuperable difficulty that the Aramaic *Beth*, *Daleth*, *Waw*, *Cheth*, *Ain*, *Qoph* and *Resh* of the sixth century are so strongly modified that they cannot be considered the prototypes of the corresponding Brāhma *ba*, *dha*, *va*, *gha*, *E*, *kha* and *ra*. It seems, therefore, advisable to assume, as must be done also in the case of *śa*, that the Hindus independently changed the form of the ancient sign, but in a manner analogous to that adopted by the Aramaeans.

c7.s1.27

[65] The derivation of the signs for *sa* and *ṣa* from the same original letter points to the influence of the Sanskrit grammarians or phoneticists, who are well aware of the intimate connection of the two sounds and duly teach that Sanskrit *sa* becomes *ṣa* through the influence of a preceding *i*, *u*, *ṛ*, *e*, *ai*, *o*, *au*, *k*, *r* or *l*.

c7.s1.28

No 16. Professor Weber's ingenious identification of the ancient *Ain* (Col. I and II) with the Brāhma *E*,—which is supported by the analogous proceeding of the Greeks, who also used it to express the gutturo-labial vowel, and of the which looks like a *dha* is found in the word *etina* of the Hathigumphā inscription of Khāravela, l. 8, and in the Sānchi inscription No 173 (Epigraphia Indica Vol. II, p. 375) where, not thinking of the Hathigumphā letter, I unfortunately have read *Dharakinā* for *Erakinā*. Professor Weber's conjecture has been accepted by Dr. Taylor alone, Dr. Deecke and M. Halévy, trying to connect the triangular *E* with *Aleph*. There was however a very good reason for giving up the round *E*. For it could not have been distinguished from the lingual *ṭha*, which the Hindus developed out of the corresponding dental. In my opinion the triangular *E* is a development, formed by the Hindus independently, and the angular forms for *Ain* in the later Semitic alphabets, Euting op. cit., Col. 8 b, 14 ff., are merely analogous, showing how easily a circular letter may be converted into a triangle or rhombus. It is just possible that phonetic considerations may have induced the Indians to select *Ain* for expressing their *e*. For according to a description of the Indian *e*-sound, which seems to go back to Śākaṭāyana and has been generally accepted by the later native grammarians,¹ it is half guttural and half palatal (*kaṇṭhya-tālavya*). As the *Ain* no doubt was a guttural, it may have appeared more suitable for *E* than for *I* or *U*.

c7.s1.29

¹ See Weber, Indische Studien, vol. IV, p. 119 f.

c7.s1.30

No 17. The fact that the Brāhma *pa* is the old Semitic *Phe* turned topsy-turvy, has been acknowledged by everybody. The new form of the Eraṇ coin, Col. IV, shows it in its original position.

c7.s1.31

No 18. Regarding the Brāhma equivalent of the Semitic *Tsade* I differ from all my predecessors. I believe that it was [66] used, as one would expect from its phonetic value, for the formation of the letter *ca*, the sound of which is and always has been, not the English *tsha*, but *tśa* almost like *tya*. The Semitic sign, of course, had to be turned topsy-turvy on account of its heavy top, and the small bar running to the right was turned towards the vertical. This process gave first a tailed [Figure:] and later with the direction to the left, [Figure:]. The angular shape of the letter has been well preserved in the *ca* of the Edicts, Col. V, a, which is not uncommon, and the tail appears in the Bhaṭṭiprolu form, Col. V, c.¹ The round *ca* (Col. V, b) is purely cursive, and not the parent of the later Indian letters, which mostly go back to angular or pointed forms.

c7.s1.32

No 19. With respect to the Semitic *Qoph* I must likewise differ from all my predecessors, who have compared it to the Brāhma *cha*, which is clearly an Indian derivative from *ca* and has a very different phonetic value. In my opinion the Semitic *Qoph* has its counterpart in the Brāhma *kha*, the oldest form of which (Col. V, a) consists of a circle with a superimposed vertical line ending in a curve. This is the oldest Semitic sign (Cols. I and II),² turned topsy-turvy on account of its heavy top. The curve at the end of the vertical has no doubt been added in order to distinguish the letter from *va*. The *kha* with the circle at the base occurs sometimes in Jaugada and is used frequently in the Kālsī version³ of the Edicts, where not rarely the circle is replaced by a somewhat irregular loop, sometimes attached to the right of the vertical line (see the *kha* figured above p. 41). It is the parent of most of the later Indian signs for *kha*, including the modern Devanāgarī form, which all show a loop or a triangle at the base of the vertical. In several versions of the Pillar [67] Edicts and elsewhere a well developed dot takes the place of the circle. This change In favour of this identification speaks also the fact that the modern Indians in loan words not rarely express the Arabic *Qoph* by their *kha*. Thus the Marāṭhī has खस्त, खस्सा, खुमास, खुरबाण, खूम and so forth for Arabic words with *Qoph*. This substitution shows that the Indian ear detects an affinity between the deep guttural *Qoph* and the harsh sounding *kha*.

c7.s1.33

No 20. I can only agree with Professor Weber's identification of *Resh* with the Brāhma *ra* or *repha*. But the original *ra* has not the form, consisting of a straight stroke, which is given in his table and in that of Dr. Taylor. The straight *ra*, evidently a late cursive development, is very rare in the Edicts. I know only of one instance in the Rūpnāth Edict in which a nearly straight stroke occurs. In the Girnār version, where *ra* is very common, it has at the top invariably one or two little angles, open to the right, or instead of the angle an irregular bulge towards the left, see Col. V, a—b. I take the form with the single angle to be the oldest and consider it to be derived from a Semitic form like that of the Mesa inscription, the triangle of which the Hindus opened, in order to avoid the heavy top, by attaching the vertical to the lower side of the base *i. e.* by putting [Figure:] for [Figure:]. The signs with two or more angles, Col. V, d—e, are no doubt artificial, ornamental developments. In the new Siddāpur Edicts this development has been carried to an extreme, and the whole letter has been converted into a wavy line, consisting of four or five little angles.

c7.s1.34

No 21. Professor Weber's identification of *Shin* with the palatal sibilant *śa*, which has been accepted by Dr. Deecke, seems also to me self-evident. It is only necessary to substitute for the late Indian form (given in Professor Weber's Col. II) the formerly not accessible, real old Brāhma signs, Col. V, a—c, which are found in the Kālsī, Rūpnāth and Siddāpur versions of the Edicts, on the Bhaṭṭiprolu prism, on the Ghasundi slab and in the Pabhosa cave inscriptions.¹ I con- [68] sider as their prototype the oldest Semitic form with two angles (Cols. I and II), not the very similar Aramaic *Shin* (Col. III) of the sixth century. For in accordance with the principles of the Brāhmī lipi, the Semitic letter had to be turned topsyturvy, and the double angle at the top had to be got rid of, which latter change could be most easily effected by placing the one angle inside the other. The Hindus may be well credited with the independent invention of this modification, as the later Phoenicians and the Ethiopians have likewise introduced it independently. Moreover, the same considerations, which make it impossible to accept the Aramaic *Samech* as the immediate source of the Bhaṭṭiprolu *ṣ* (see above p. 67), speak in this case against the assumption that the Semitic alphabet, adopted by the Hindus, contained the Aramaic *Shin*. Finally the Western letter-numeral *śu* for 100, still shows the two angles quite plainly, see App. II, pl. III and the Grundriss, Taf. IX, Cols. IV—V, XI.

c7.s1.35

¹ I enumerate the occurrences so fully, because, when the sign first turned up in Kālsī, it was considered to have been borrowed from the Kharoṣṭhī, which has the remotely similar lingual *ṣa* [Figure:]. I will add that in Bhaṭṭiprolu where we have *śamaṇudeśānaṃ i. e. śramaṇoddeśyānām*, in the Ghasundi inscription where we have *Parāśarīputa* and *śilā*, and in Pabhosa which offers *Śonakāyana* i. e. *Śaunakāyana*, this sign appears only in words which have *śa* in Sanskrit, while in the Edicts it occurs mostly and very irregularly in words, where the Sanskrit has *sa* or *ṣa*. The confusion is owing to the negligent pronunciation and writing of the clerk, who made the copy. His dialect probably contained two sibilants, the dental and the palatal, and his alphabet had three, the dental, the palatal and the lingual. His negligence in pronunciation made him pronounce *sa* for *śa* and vice versa, and his negligence in writing made him use *sa* and *śa* indiscriminately. As stated above, papers written in our times by half-educated clerks and others show exactly the same confusion and for the same reasons. They pronounce *e. g.* both *sac* and *śac* (*satyam*) and they write indiscriminately सच, शच and षच.

c7.s1.36

No 22. The oldest form of the Brāhma *ta*, which Professor Weber has recognised already as a modification of the Semitic *Taw*, seems to be that (Col. V, a) consisting of a long slightly inclined downstroke with a short straight bar slanting off to the right. Next comes the sign, given in Col. V, b, with the bar slanting off to the left, and this is due to the change in the direction of the writing. The *ta* consisting of a vertical stroke with an angle at the foot, Col. V, c, which is frequent in the southern and western versions of the Edicts and also on the Sohgaurā Plate, is probably [69] due to the formalism of the Hindus, their desire to set the letter up straight and to make it perfectly regular in appearance. The forms with a round limb to the right or left of the vertical (see above p. 41) of course are cursive. The oldest among these five forms for *ta* closely resembles the *Taw* from the earliest Sinjirli inscription (Col. III, b), which belongs to the beginning of the eighth century B. C., and still more that of Col. III, a, which occurs on an Assyrian weight of the reign of Salmanassar¹ and has therefore been engraved before 725 B. C. As it thus appears that Semitic forms, consisting of a long inclined downstroke with a crossbar very slightly protruding on the left or with a simple bar on the right, have been developed in very early times, it is not improbable, that one of them occurred in the alphabet which the Hindus borrowed. But the possibility that the Indian sign is an independent development from the straight Phoenician cross (Col. I), is not altogether excluded. For the cross could not remain in the Brāhma alphabet, because it would have been undistinguishable from *ka*.

c7.s1.37

¹ As Professor Euting kindly points out to me, the inscription is found Corp. Inscr. Sem., P. II, No 2 *c*.

B. The Derivative Consonants and Initial Vowels

c7.s2.1

The contrivances, by which the derivative signs, both primary and secondary, for consonants and initial vowels have been formed, are:—

c7.s2.2

(1) the transposition of one of the elements of a phonetically cognate borrowed sign,

c7.s2.3

(2) the mutilation of a borrowed letter or of another derivative sign of similar phonetic value,

c7.s2.4

(3) the addition of straight lines, curves or hooks to borrowed or derivative signs. If a hook is added the original sign suffers a slight mutilation.

c7.s2.5

Two cases of transposition have already been mentioned above under No 15, where it has been pointed out that the *sa* and the Ghasundi *ṣa* come from the Bhaṭṭiprolu *ṣ*. A second case occurs, as Professor Weber has been the first to recognise, in the dental *da* (No 4, Col. VI, a—b, f—g). The [70] two first forms, one of which occurs in Bhaṭṭiprolu as well as on a Patna seal ¹ and the other in the majority of the Edicts, are derived from the two *dha* in Col. V, by dividing the straight line and pushing the halves back to the ends of the semicircle which remained. In the third actual form (Col. VI, f) and the hypothetical one, Col. VI, g (required on account of the next following sign) the semicircle has been converted into a small square, left open on one side, and this change is due to the liking for angular forms, mentioned above p. 37.

c7.s2.6

The cases in which a borrowed or derivative sign has been mutilated, are those of the lingual *tenuis*, *tenuis aspirata* and *media*, all of which Professor Weber has already explained correctly. The lingual *ḍa* under No 4, Col. VI, c, which occurs occasionally in Kālsī and commonly in the later inscriptions of the Nānāghāt and the other Western caves etc., is derived, by the removal of the lower end, from the Bhaṭṭiprolu *da* (Col. VI, a), which had not yet been turned from the right to the left. The angular *ḍa* (Col. VI, h) comes in like manner from an angular *da* (not yet turned to the left), of which the known inscriptions do not offer any example. The *ṭha* (No 9, Col. VI, a) is of course the dental *tha* minus the central dot, and the *ṭa* (No 9, Col. VI, b) has probably been obtained by halving the *ṭha*, as Professor Weber conjectures. To a Hindu this process probably appeared very natural. For he formed several aspirates by adding curves. Hence he may be supposed to have considered a round sign, denoting an aspirate, as equivalent to an unaspirated letter *plus* a curve of aspiration. Thus the division of the sign would be quite legitimate. In the Edicts both *ṭha* and *ṭa* are frequently made smaller than the other letters.

c7.s2.7

Two other cases of the mutilation of borrowed letters occur in the signs for initial *I* and *U*. It has been recognised already by Mr. Prinsep that the three dots of *I* (No 16, Col. VI, a—c, lower row) indicate of *I* generally agrees with the position of the angles of *E*. To a Hindu phoneticist or grammarian the derivation of *I* from *E* would appear a matter of course, because *E* is very commonly [71] the representative of an *I* in strong forms or its *Guṇa*. Hence he expressed the latter by a lighter form of the former, just marking the corners of the triangle.

c7.s2.8

The case of *U* (No 6, Col. VI, a) is somewhat different. It has been customary to derive the sign directly from one of the later forms of the Semitic *Waw*. Considering the facts, connected with the linguals and with initial *I*, I would propose to derive it either directly from the old Semitic *Waw*, turned topsy-turvy, or from the Brāhma *va* by a bisection of the circle at the foot of the sign and the substitution of a straight line for the irregular pendant, which remained. The several steps were, therefore, 𑀉 or 𑀉, 𑀉, 𑀉, and finally¹ 𑀉. It is probable that phonetic considerations, the observation of the frequent substitution of *u* for *va* in weak verbal and nominal forms, as in *uhyate*, *upyate*, *supyate*, *upta*, *supta*, *uṣita*, from *vah*, *vap*, *svap* and *vas*, may have led to the adoption of this proceeding. The framers of the Brāhma alphabet were, as pointed out already, Brahmans, acquainted with phonetics and with grammatical theories.

c7.s2.9

¹ The former actual existence of the third form is proved by the Jaugada *O* (No

c7.s2.10

The last case of mutilation, I believe, occurs in the Anusvāra (No 13, Col. VI, b) which is represented by a simple dot. This is no doubt a substitute for the small circle (Col. VI, a), which reappears very commonly in certain later epigraphic and literary alphabets. I consider the circle to be a mutilated minute *ma*, the upper angle of which was omitted,² and I believe that this conjecture is supported by the following facts. In the Edicts and all other ancient inscriptions the Anusvāra is placed rarely at the top of the letter after which it is pronounced. Usually and regularly it stands opposite the middle, but in a number of cases it is found also to the right of the foot. Now vowelless consonants always stand at the foot of the preceding sign in the oldest Sanskrit or Mixed inscriptions from the times of the [72] Kuṣana kings down to the fifth century A. D., when the *Virāma* makes first its appearance. Moreover all such vowelless consonants are made exceedingly small, even after the invention of the *Virāma*, and they are very commonly mutilated at the top.¹ These facts would fully explain the use of a small circle for a vowelless *ma* of the Aśoka type, which then became the general sign of nasalisation in the Brāhma alphabet, just as the Kharoṣṭhī *ma* was turned into the Kharoṣṭhī Anusvāra. My theory, of course, rests on the assumption that the Brāhma alphabet was used from the beginning, not for Prakrit, but for Sanskrit, and this is made more than probable by the occurrence of *ṅa* in the Mahābodhi Gayā alphabet of the masons, as well as by the arrangement of its letters on phonetic principles, see above p. 33, likewise by the numerous indications that the alphabet was elaborated by phoneticists or grammarians, * in which *O*,¹ No 6, Col. VI, f—g, has been formed out of *U*, and in the first sign (Col. VI, f), the stroke stands in its [73] original position, because the letter (Jaugaḍa form) has not been turned round. The *AI* (No 16, Col. VI, b, upper row) appears to have been derived from *E* in the same manner, but the letter has been turned from the right to the left. The sounds *u* and *o*, as well as *e* and *ai*, appear to a Hindu and to a Sanskritist closely connected, because in numberless cases *o* is the *Guṇa* or representative of *u* in strong forms, and because *e* and *ai* both appear in the strong forms of roots with *i* and in derivatives from nouns with *i* and *e*. These phonetic or grammatical affinities no doubt influenced the formation of the signs.

c7.s2.11

Added to signs for consonants, either on the right or across the top, a straight stroke likewise denotes a change of quality viz. that the sign expresses the corresponding sound of a different class or *Varga*. The stroke has its original position in the Bhaṭṭiprolu *ḷ* (No 12, Col. VI), which has not been turned round. It appears on the left in the palatal *ña*,² because this has been turned. Its position is again the original one in the guttural *ṅa*, which in the Gayā alphabet and the Kuṣana (see above p. 31) inscriptions is [Figure:], and in which the foot of the *na* has also been modified. Finally in *ṇa* (No 14, Col. VI, b) the bar goes across the vertical. The peculiarities of the last two letters are probably due to a desire to avoid collisions with *nā*, *ne* and *O*.

c7.s2.12

Aspiration is expressed by the addition of a curve in the Bhaṭṭiprolu *gha* (No 3, Col. VI), and the ordinary Brāhma signs for *ḍha* (No 4, Col. VI, d), *pha* (No 17, Col. VI) and *cha* (No 18, Col. VI, a—b) are derived in the same manner from *ḍa*, *pa* and *ca*. In the sign for *cha* both ends of the curve have been connected with the vertical line of *ca*. There are numerous instances, like that given in Col. VI, a, in of a circle bisected by a vertical line is in my opinion cursive.

c7.s2.13

In the two signs for *bha* (No 2, Col. VI) and for *jha* (No 7, Col. VI) an angle or hook serves the same purpose as the curve of the other four aspirates, and in both cases the original sign is mutilated in order to make the new form less cumbersome. The *ba* has lost its base line and the *ja* its two [74] bars. It seems not improbable, as has already been suggested by others, that the angle or hook of aspiration may be a cursive development from the letter *ha*.

c7.s2.14

In the lingual *ḷa*, derived from the round *ḍa* (No 4, Col. VI, e) a small semicircle¹ has been added to the foot of the original sign in order to indicate the change of the phonetic value. Here also, I believe, we may recognise the influence of the grammarians or phoneticists. For the sounds *ḍa* and *ḷa* are frequently interchanged in the same word. Thus we find already in the Vedas regularly a *ḷa* for a *ḍa* between two vowels, as in *īḷe* for *īḍe*. In the later Sanskrit and in the Prakrits there are numerous variants like *nāḍī* and *nāḷī*, *nāḍikera* and *nāḷikera* and so forth, where it is often difficult to decide, which is the original form. As the principle, on which the ordinary *ḷa* has been formed, differs from the more general one, applied in the case of the Bhaṭṭiprolu letter, I consider it to be of later origin.

c7.s2.15

¹ Possibly a sign which occurs in Jaugada and in the Pillar Edicts [Figure:] *i. e.* *ḍa* with a dot a the foot, may have the same value, as it is used in words which have double forms with *ḍa* and *ḷa* or *la e. g.* *duḍī* or *dulī*, *eḍaka* or *eḷaka*, *Coḍā* and *Coḷā*

C. The Medial Vowels

c7.s3.1

Hitherto two systems for the notation of the medial vowels have become known, that of the Edicts and all the later Sanskrit and Prakrit inscriptions and that of Bhaṭṭiprolu. The first, which is by far the older one, shows clearly the influence of the grammarians and their ingenuity. As the vowel, expressed in the beginning of words by the representative of the Semitic *Aleph*, occurs in Sanskrit nearly as frequently as all the other vowels taken together,¹ it was not expressed by any sign, but considered to be inherent in all signs for consonants. This device fully agrees with the system of nearly all the phonetic and grammatical treatises, which, as Professor Max Müller has shown so clearly, do not refer to written letters but only to sounds.² They almost invariably speak of the *kakāra*, *gakāra* [75] and so forth. The commentators no doubt are right, when they assert, that the vowel has been added in order to make the pronunciation of the consonants possible, and the vowel *a* was selected for this purpose on account of the frequency of its occurrence. It seems impossible to assume that there is no connexion between the two facts, and, as the grammarians base their theories on spoken words not on written texts, I think that they are the men who also in this case influenced the formation of the Brāhma alphabet.

c7.s3.2

¹ See Professor Whitney's calculations in his Sanskrit Grammar, p. 73 (second edition).

c7.s3.3

² History Anc. Sansk. Lit., p. 507 ff. This assertion has been hotly contested by Professor Goldstücker in his Introduction to the Mānava Kalpa Sūtra, p. 13 ff. But Professor Kielhorn, who has studied the Vyākaraṇa during so many years, informs me that he does not know of a single passage even in the Bhāṣya, which indicates with certainty, that a written text of Pāṇini's grammar is referred to, or where the technical terms of the grammarians and their theoretical speculations refer to written signs; see also his remarks in Gurupūjākaumudī p. 29 ff. It seems, therefore, to be a fact that the grammatical and phonetic researches were begun either before the introduction of writing or independently of writing, and that even those ancient authors, who like Pāṇini, mention alphabets and clerks, continued to work on in the old manner. I must, however, add that Geheimrath von Böhtlingk dissents and believes that the phrase *yathānyāsam* in the Mahābhāṣya (e. g. I, 22, l. 3) points to a written text.

c7.s3.4

As regards the other vowels, medial *ā* is expressed by placing to the right of the consonant the same short horizontal stroke,³ used for the differentiation of the initial *Ā* from *A*, apparently because the other portion, the short *a*, is already contained in the consonant. The remaining ones are expressed by the signs for the initial vowels or by modifications thereof, placed above or below the consonants. A very clear case is that of the medial *o* in the syllable *ko*,¹ given in two forms under No 6, Col. VI, h—i. If the *k*, *i. e.* the portion of the sign below the second bar, is removed, there remains in the eighth sign a minute initial *o* of the type in Col. VI, f, and in the ninth one of the type in Col. VI, g. Now in the Jaugaḍa version of the Rock Edicts, where the initial *O* has the top bar to the right, the same is invariably the case with the medial *o*. It, therefore, would seem that the writer was perfectly aware of the connection of the two signs. But, in Aśoka's time this [76] feeling was dying out. For in Girnār, where the initial *O* with the top-bar to the left alone is used, the medial *o* is made in both ways, and in the second part of the Delhi Pillar Edict VII, 2, l. 2 we have once the cursive medial *o* in *nigohāni*, where both strokes are placed on the same level above the consonant.

c7.s3.5

³ Originally the stroke, of course, stood on the left, and it is found in this position on the Eraṇ coin, where the letters run from the right to the left, see the Grundriss, Tafel II. 28. I.

c7.s3.6

¹ Compare also the *go* [Figure:] in *mago*. Girnār, Ed. I, l. 11, where a distinct *O* stands above *ga*.

c7.s3.7

Equally clear is the case of the medial *u*, which is the initial *U*, put below the consonant. This is distinctly recognisable in the sign *dhu* (No 6, Col. VI, b) which occurs repeatedly in the Kālsī version.² Cursively it assumes the form, given in Col. VI, c or of D, with the omission either of the vertical stroke or of the horizontal bar at the end. On the same cursive principle *ū*, which in *kū* etc. retains the shape of *Ū* (Grundriss, I, 11, p. 37), is expressed by two strokes, placed either horizontally at the side of the consonant (No 6, Col. VI, e) or below the consonant, where they frequently form an acute angle but are also placed parallel side by side. These facts seem to indicate that Aśoka's clerks had lost the remembrance of the origin of the signs for medial *u* and *ū*, and that they considered the old forms, which they occasionally used, merely as permissible variants without any special significance. In later inscriptions,³ however, reminiscences of the origin of the subscript *u* and *ū* are found. Thus in the ancient Śāradā characters of the Baijnāth Praśasti and elsewhere *pū* is expressed by पु.

c7.s3.8

² *E. g.* in Kālsī, Edict III, l. 8 (twice).

c7.s3.9

³ Compare also the Grundriss, *gu* in Taf. VII. 10. 20; *bhu* in Taf. VII. 30. XII; *śu* in Taf. VII. 36. XXI; and *bhū* in Taf. VII. 30. XX—XXI.

c7.s3.10

As regards the medial *i*, the small angle to the left of the top of the consonant by which it is marked, seems to be the result of a connexion of the three dots of the initial vowel by means of two lines, see the *ki* No 16, Col. VI, d—e, lower row. As long as the Brāhma alphabet was written from the right to the left, the *i*-strokes, as well as all other vowel signs of course stood to the left of the consonant. In the Kālsī version, Ed. XIII, 2, l. 10, there are two signs, at the end of the words *anuvidhiyaṃti* and *anuvidhiyisaṃti*, where the vowel has this position. They look like *A*, because the vowel strokes have been added to the middle of the consonant. A better formed *ti* with the vowel on the left occurs in Allahābād, Ed. I (end), and a similar *hi* on the Sohgaurā copperplate, l. 4. The medial *ī* does not seem to be connected with the initial *Ī*, but to have been formed by the addition of the [77] straight stroke, denoting also in other cases the lengthening of vowels, which for convenience's sake in this case was made vertical. Cursively the angle of medial *i* is converted in Girnar into a shallow curve and medial *ī* is expressed by a curve bisected by a vertical stroke.

c7.s3.11

The sign for medial *e*, a straight or slanting line to the left of the consonant, I take to be the remnant of a triangular initial *E*, the outlines of which have been indicated by dots in the *ke*, figured under No 16, Col. VI, a, upper row. I may add that in the Edicts forms like [Figure:] are occasionally used for *ge*, where the vowel sign seems to consist not of a line, but of a hook put on the top of the letter. Such forms are perhaps ancient survivals, dating from a time when the vowel was represented by an angle, to which the triangle probably was reduced in the first instance. The position of the *e*-stroke is abnormal, as it stands to the left of, or before, the letter, after which it is pronounced. The cause is no doubt, that, if it had been placed to the right, it would have been undistinguishable from medial long *ā*.

c7.s3.12

In accordance with the form of the initial *AI* consisting of *e* and a horizontal bar to the left of the top, the medial *ai* is expressed by two bars to the left of the consonant, see the syllable *thai* from the Girnār version, given under No 16, Col. VI, c, upper row.

c7.s3.13

The absence of a medial vowel between two consonants is expressed by the formation of a ligature, in which ordinarily the second consonant or its most essential portion is attached to the foot of the first. In the Girnār and Siddāpur versions however, the ligatures *ṣṭa*, *tpa*, *vya*¹ as well as those containing a *ra*, like *tra*, *pra* and *vra* show the inverse order and are spelt *ṭṣa*, *pta*, *yva*, *rta* and so forth, while all the others like *sta*, *mha* etc. are formed regularly. Moreover, in the words *brāmhaṇa* (Girnār IV, l. 2) and *drahyitavyaṃ* (Siddāpur, I, l. 9) the first vertical of *ba* and the two vertical strokes of *da* have been converted into wavy *ra*-lines. These irregularities are no [78] doubt due to an artistic feeling and the desire of the clerks to produce regular, shapely signs. The combinations of the consonants with medial vowels furnish analogies. Properly all the vowels ought to stand to the right of the foot of the consonants. If the majority is nevertheless placed at the top, that has been done merely for the sake of convenience. Later inscriptions also furnish a few isolated cases of an inversion of the order of the elements of ligatures. Thus the name of Caṣṭana's father is spelt on the coins *Ysamotika* instead of *Syamotika* and Harṣa's Banskhera Plate of A. D. 628/9 offers *vaṇṇra* for *varṇṇa*.

c7.s3.14

¹ At least in *vyaṃjanato*, Ed. III, l. 6. The cases of the passive future participles in *tayva* are doubtful, as they may have been pronounced as they are written, compare Pali *mayham* and so forth. The Siddāpur version has *tavya* in *drahyitavyaṃ*.

c7.s3.15

The cause of the formation of ligatures in order to express the absence of vowels must again be sought for in the influence of the Sanskrit phoneticists on the development of the Brāhma alphabet. The Prātiśākhyas and the later works on phonetics and grammar all use the expression *saṃyuktākṣara* "a conjunct syllable" for groups like *kta*, *kra* and so forth. The combination of the signs in writing looks very much like a practical illustration of the meaning of the term.

c7.s3.16

The manner in which the absence of a vowel after a final consonant was probably expressed, has been stated in the remarks on the Anusvāra, above p. 74 f.

c7.s3.17

The Bhaṭṭiprolu system of vowel-notation differs from the ordinary one merely by marking the short *a* by the bar, which denotes *ā* in the Edicts, and the long *ā* by the same bar *plus* a vertical or slanting stroke, hanging down from it, *e. g.* [Figure:]

c7.s3.18

*na* and [Figure:] *nā*. This system, according to which the consonants have no inherent *a*, seems to have been invented in order to avoid the necessity of forming the ligatures, which make the ordinary Brāhma alphabet cumbersome and difficult to read in its later developments, and in order to express final consonants more conveniently. I believe it, therefore, to be of later origin, especially as the other Bhaṭṭiprolu vowel-signs do not differ from the ordinary ones. The invention must, of course, have been made for writing Sanskrit, as the Prakrits have few groups of dissimilar medial consonants and no final ones.

c7.s3.19

To sum up—the forty four letters¹ of the ordinary Brāhma alphabet, traceable in the oldest inscriptions as well as the variants of the Bhaṭṭiprolu inscrip- [79] tions, contain representatives of, and derivatives from, all the twenty two Semitic characters, viz:—

First 79 of 119 passages. Read on

Linked texts (26)