Indian Palaeography c18: 10. The Kharoshṭhī Script

Indian Palaeography c18

Language: English

c18.s1 Nomenclature — 2 parts
c18.s2 Area — 18 parts
c18.s3 Origin — 18 parts
c18.s4 Writing Styles — 6 parts
c18.s5 Conclusion — 15 parts

10. The Kharoshṭhī Script

Nomenclature

c18.s1.1

OF all the scripts of ancient India one stands alone, by virtue of the direction of its writing, which is from right to left instead of from left to right. This script was first noticed on the coins of the Bactrian Greek rulers, and hence it was called the 'Bactrian' character.¹ But as the rulers themselves, because of their connexion with India, were sometimes referred to as 'Indo-Bactrian', the same term was also applied to the script. However, it was later ascertained that 'not a single example of this writing has yet been found to the north of the Hindukush. Its use was confined to Ariana on the west of the Indus, and to the Panjab on the east.'¹ Lassen therefore preferred to call it 'Kabulian', and Wilson 'Arianian'. On geographical grounds Cunningham objected to the use of the latter two terms, and he suggested the name 'Gandharian' alphabet, so as to embrace the regions east as well as west of the Indus, where this script was current. As the language used in these inscriptions is some form of Pali or Prakrit, the script also came to be known as 'Bactro-Pali' or 'Ariano-Pali'. The present name 'Kharoshṭhī' is due to Bühler, who found evidence for this name in the *Lalita-vistara* (composed about the third century a.d.) and in the Chinese encyclopaedia, *Fa-Wan-Shu-Lin*, compiled in a.d. 668. But the meaning of the term is by no means certain.

c18.s1.2

¹ Cunningham, *Coins of Ancient India*, London, 1891, p. 31.

Area

c18.s2.1

Sten Konow says:

c18.s2.2

The area within which we can prove Kharoshṭhī to have been regularly used belongs to the north-west. The easternmost limit is, in the Panjāb, at Māṇikiāla. There are two inscriptions from Kangra, where Kharoshṭhī is used in addition to Brāhmī, and there is another record from Karnāl, which shows that the alphabet was known farther to the east, and foreign conquerors from the north-west used it in a well-known inscription from Mathurā on the Jamnā, where Brāhmī was the common alphabet, also in inscriptions and coins. We even possess a Kharoṣṭhī record from Patna. But the plaque on which it is written has evidently been left there by a person who came from north-west. We do not know exactly how far the use of Kharoṣṭhī extended towards the west. Coins with Kharoṣṭhī legends have been found in Seistan and Kandahār, but the westernmost Kharoṣṭhī inscriptions which have been found are from Khawat in Afghanistan and, side by side with Brāhmī records, from the Thal valley in Baluchistan. And even here we have every reason for assuming that the alphabet was brought and used by immigrants from the east. For it is little suited for the requirements of Iranian languages, and we have nothing to show that the dialect in which most Kharoṣṭhī records are written was ever spoken as a vernacular much farther east than Jalālābād. The northernmost Kharoṣṭhī records come from Tīrath in Swāt and Khalatse in Ladakh, and in the south we have some fragments from Mohenjodaro in the Lārkāna district and Kharoṣṭhī legends on the coins of some of the oldest of the Western Kshatrapas. But such stray instances do not prove anything more for the proper Kharoṣṭhī area than the Kharoṣṭhī word Lipikareṇa in the Siddhāpur edicts of Aśoka. The Kharoṣṭhī area proper may be defined as extending from about 69° to 73° 30′ E. and from the Hindu Kush to about 33° N., and there can be little doubt that its place of origin was Gandhāra, perhaps more especially Taxila.¹

c18.s2.3

¹ Konow, *C.I.I.* vol. ii, pp. xiii–xiv.

c18.s2.4

To this area of the Kharoṣṭhī inscriptions must now be added Chinese Turkestan, from the discovery there of a Prakrit *Dhammapada* in 1892 by the French traveller, Dutreuil de Rhins, and later of numerous Kharoṣṭhī documents by Sir Aurel Stein. This limited area within which Kharoṣṭhī was used helped in maintaining the unitary style of this writing, and there was hardly any chance for its branching out into provincial schools, as was the case with the Brāhmī.

c18.s2.5

*Period*

c18.s2.6

The earliest² inscriptions in Kharoṣṭhī are those of Aśoka (middle of the third century B.C.) found at Mansehra and Shahbazgarhi, and the latest are those known in the inscriptions of the later Kushāṇas (third to fourth century A.D.). A few private records may have been written slightly later, but already Brāhmī was adopted by the Indianized rulers in this region. In Chinese Turkestan Kharoshṭhī was kept up by the local people for some time more. But in India, when the Hūṇas appear in the fifth century A.D., no trace of Kharoshṭhī is seen.

c18.s2.7

² For the alleged early coins see above, pp. 59–61.

c18.s2.8

*Nature*

c18.s2.9

As Bühler has shown, the Kharoshṭhī script has been found (i) in stone inscriptions, (2) on metal plates and vases, (3) on coins, (4) on cameos, and (5) on a small piece of birch-bark from a stūpa in Afghanistan and on the *bhūrja* MS. of the *Dhammapada* from Khotan.¹ To this list must be added the documents on wood, leather or paper discovered by Stein in Chinese Turkestan. In all these writings Kharoshṭhī maintains its unified character. Hardly any evolution is seen in the forms of the letters. The main variation is in the manner in which a given form is reproduced on a material, the change in the latter affecting the form. The particular manner followed in a period or on one material became a convention. This convention did not develop into different schools of writing as it was controlled by a common technical process. It is the recognition of the conventionalized strokes, or the way of writing followed in the different materials, that roughly enables us to place the different inscriptions in chronological order. The invention of the diacritic strokes for new sounds is most important in Kharoshṭhī. As this script has been used to express sounds needed in different languages, it was necessary to devise ways by which a given alphabet might be adopted to the particular need. But despite these additional strokes the main forms of the nucleus letters hardly show any evolution. The Kharoshṭhī script has only one type, but it has complete power to express any sound of Sanskrit or foreign origin. The view propounded by Bühler² that it was a popular writing as opposed to the Brāhmī, which he thought to be literary, has been rightly corrected by Rapson³ on the evidence of the Niya documents, and he has conclusively shown that Kharoshṭhī could be used for writing correct Sanskrit. But the example of classical Sanskrit quoted by him is limited to only four verses. On the other hand, some form of the local Prakrit is the chief medium through which Kharoshṭhī is now known, and it seems that it was originally created to express this language. For while the nucleus letters and even the medial short vowels have separate signs for themselves, the additional letters are formed by a common diacritic stroke (an oblique stroke added to the foot of the vertical on the right), and a uniform procedure of adding a sloping stroke at the right bottom of the letters is adopted for the long vowels. This regularity suggests a later adaptation.

c18.s2.10

¹ *Indian Palaeography*, Eng. tr., p. 18.

c18.s2.11

² Ibid., p. 19.

c18.s2.12

³ *Kharoshṭhī Inscriptions*, pt. iii, p. 295.

c18.s2.13

*Technique*

c18.s2.14

The Kharoshṭhī is a pen style of writing, in which one type of pen remained in use throughout. The manner of its use is easily detected. In the Aśokan inscriptions it leaves behind a foot-mark in the form of an upward slant at the lower end of the verticals on the left. This foot-mark must be attributed to the style of the writer rather than to the engraver, as is proved by its presence in the hand-written documents from Chinese Turkestan (Pl. XXIII. 13). When it is not found on stone or metal, it is partly due to the engraver’s negligence¹ and partly to another style of writing seen in the documents. In this second style the verticals gradually thin out into curves, as is natural to a pen. It is these curves which predominate in the inscriptions of the Kushāṇa period. In such examples the medial *u* assumes the form of a loop. This loop in stone engraving sometimes becomes a triangle. In the ink style of writing seen in some potsherds from Baluchistan² and in the documents from Chinese Turkestan the tendency is to draw the lines and the additional strokes with one sweep of the hand. It is this process which has resulted in the cursive forms of some letters, especially *sa*. However, the writer’s hand may not be faithfully copied by the engraver or the metal worker, and sometimes the latter’s technique or convention produces a different effect; e.g. on metal plates and vases the letters are formed by rows of dots (not shown in our plate), in steatite vases the letters are scratched with a stylus, and in the coins and medals many other superfluous lines or dots are usually seen in the forms of *ma*, *ha*, and *ja*, and rarely in *ga*, *da*, *na*, and *sa*. Percy Gardner¹ took these lines and dots for lengthened vowels, and hence read *māhāraja* for *maharaja*, and *menadrāsa* for *menaṁdrasa*.²

c18.s2.15

¹ It must, however, be pointed out that this foot-mark in the Aśokan inscriptions is generally seen with the initial *a* (optionally) and with those consonants which have the medial *a* implicit in them. From this factual observation some (including Rapson) have taken this slant to stand for medial *ā*. But in view of the evidence from the hand-written documents this is hardly justified (see *Kharoshṭhī Inscriptions*, pt. iii, p. 299). Secondly, an altogether different stroke is used for lengthening the vowels in these documents. If this slant was adopted in the Aśokan inscriptions, it is difficult to understand why it was given up.

c18.s2.16

² Konow, *C.I.I.* ii, pl. xxxv.

c18.s2.17

¹ *Catalogue of Indian Coins in the British Museum, Greek and Scythian Kings of Bactria and India*, London, 1886.

c18.s2.18

² See also below, p. 264.

Origin

c18.s3.1

The very fact that Kharoṣṭhī was written from right to left led to the easy recognition of its relationship³ with the Aramaic script, which was the main branch of the North Semitic alphabet.⁴ This relationship was strengthened by the identity of letters such as *na*, *ba*, *ra*, and *va*, and the close similarity of a few other features, the most important being the employment of one basic form for the initial vowels in Kharoṣṭhī, as is also the case in Aramaic. Isaac Taylor⁵ explained the circumstances under the Achaemenians, which helped in the origin of Kharoṣṭhī and its evolution as seen in the Aśokan inscriptions. But Cunningham⁶ rightly pointed out that the expansion of the alphabet ‘took place after its contact with the more fully developed system of the Indian alphabet’. It seems that Cunningham did not agree with Taylor when the latter proposed a gradual evolution of the Kharoṣṭhī letters from the Aramaic alone or through its influence. However, the confusion could not be removed, mainly because no distinction was made between the script and the alphabet. Bühler accepted the proposal of Taylor and argued for the original and derivative signs and further added: ‘with respect to the origin of the mark of aspiration I can agree with Dr. Taylor, who explains it as a cursive form of *Ha*.’⁷ Bühler further develops the idea of Taylor and explains the circumstances of the origin of Kharoṣṭhī. He calls it a popular script, and remarks: ‘If thus the Kharoṣṭhī appears to be an alphabet, framed with particular regard to the wants of clerks, that agrees with and confirms the assumption, put forward above, according to which it arose out of the official intercourse between the scribes of the Satraps and those of the native chiefs or other authorities.’¹ Though Bühler’s definition of the character of the Kharoshṭhī script has been modified by us in the previous paragraph on the evidence of the documents from Chinese Turkestan, it is nevertheless true that Kharoshṭhī was originally adopted to write the local language spoken by the common people. This necessity must have been felt by the foreign conquerors who desired to establish relations with their subjects—relations demanded by administrative necessity. The local language had to be adopted for dealings with them, but the problem was to find a vehicle through which this language could be written in such a way that it was intelligible to the local people as well as convenient to learn by the administrative staff who were familiar with Aramaic. It was this necessity which created Kharoshṭhī from Aramaic, and it is mainly in the records of foreign rulers that this script was used. The script was carried as far as Mathura by the Śaka Kshatrapas and Gujarat by the Western Kshatrapas. It survived for as long as the foreign rulers were able to maintain their individuality against the growing Indianizing influences.

c18.s3.2

³ E. Thomas, *Indian Antiquities*, ii. 144–68.

c18.s3.3

⁴ D. Diringer, *The Alphabet*, 2nd ed., p. 255.

c18.s3.4

⁵ *The Alphabet*, London, 1883, ii. 252–62.

c18.s3.5

⁶ *Coins of Ancient India*, London, 1891, p. 33.

c18.s3.6

⁷ Bühler, *Origin of the Indian Brāhma*, p. 109.

c18.s3.7

¹ Bühler, *Origin of the Indian Brāhma*, p. 98.

c18.s3.8

While Aramaic script was suited to the Aramaic language, it could hardly be used for writing the local language without making necessary changes in the values of the letters to represent the local sounds. It would have been possible to develop Aramaic by the addition of diacritic marks, but this would have led to much complication. Kharoshṭhī is not the result of this process. The letter-forms have not gradually evolved from the original Aramaic to Kharoshṭhī. In order to understand its basic character we must analyse it into an alphabet as following a certain phonology and a script as conforming to certain general principles of form and shape. Its phonology, as has been shown in Chapter 3, is identical with that of Brāhmī, and both have the same rules underlying the conjuncts and vocalized consonants. But the forms are entirely different. The whole system of Kharoshṭhī follows the pattern of Aramaic. The resemblance is not so much in the identity of forms, though a few letters are the same, but in the way in which these forms are produced. The difference in the phonology and the grammar of the local languages stood in the way of adopting Aramaic forms wholesale. The same Aramaic letters could have been taken to stand for different values of sounds, but this would have caused greater confusion, *at least in the minds of the Aramaic-* knowing Satraps, for whose convenience the script was primarily evolved. Hence it was necessary to create the Kharoshṭhī script on the basis of Aramaic. It is only on such an assumption that we can explain the differences in the letter forms of Aramaic and Kharoshṭhī. There is no necessity for assuming a long duration between the introduction of the original Aramaic and the Aśokan Kharoshṭhī, as is done by Taylor. In fact the writing, as seen in the rock edicts of Aśoka, hardly shows any variation in forms. Minor differences seen in them are due to different hands being used and to the engravers’ slips.

c18.s3.9

As we have seen, Kharoshṭhī follows the principles of Brāhmī as far as the language demands. It was the necessity of grammar that brought it closer to Brāhmī and helped in the evolution of similar vowel notation and an equal number of consonants. The conjuncts are also formed in similar ways. But the main difference is seen in the vowels. While Brāhmī has three basic forms of vowels, *a*, *i* and *u*, Kharoshṭhī has only one, the forms of the remaining vowels being obtained by the addition of diacritic strokes. In this respect Kharoshṭhī is nearer to Aramaic, but while Aramaic does not need to express the medial vowels, Kharoshṭhī has to do it according to the needs of the grammar. Kharoshṭhī also differs from Brāhmī in the application of the signs for *dīrgha*, *guṇa* and *vṛiddhi*. As shown in Fig. 19, in Kharoshṭhī an extra stroke is added to the right bottom of the letters, but Brāhmī usually doubles the stroke.¹ This difference suggests that the indication of long vowels in the Kharoshṭhī script is of later origin, as it appears only in the documents from Chinese Turkestan, where it is mainly used in Sanskrit words.

c18.s3.10

¹ See above, p. 46.

c18.s3.11

The debt of Kharoshṭhī to Aramaic is not in borrowing the latter’s signs wholesale for writing the local language, but in making them the basis for the evolution of its own signs. This basis will be clear from the following analysis. Out of the twenty-two letters in Aramaic one has been adopted, as shown above, as the base of the vowels. The remaining letters are distributed as follows (Fig. 19):

c18.s3.12

(1) *ka* and (2) *ga* from *kaph* and *gimel*;
(3) *cha* and (4) *ja* from *ṣad* and *zayin*;
(5) *ta*, (6) *da*, and (7) *na* from *taw*, *daleth* and *nun*;

c18.s3.13

ORIGIN OF THE KHAROṢHṬHĪ

c18.s3.14

(8) *pa*, (9) *ba*, and (10) *ma* from *pe*, *beth* and *mem*;
(11) *ya*, (12) *ra*, (13) *la*, and (14) *va* from *yod*, *resh*, *lamed* and *waw*;
(15) *sha*, (16) *sa*, and (17) *ha* from *shin*, *samekh* and *he*.

c18.s3.15

Out of the remaining four Aramaic letters, *'ayin* and *ṭeth* have been discarded by Bühler. He derives *kha* from *qoph* and *śa* from *kheth*. Here *kha* is an aspirate, but it is derived separately from the Aramaic, though the other aspirates are formed by the addition of an extra stroke to the unaspirated consonants. The only exception is *tha*, which on no account can be derived from *ta*, but, on the other hand, is connected with the cerebral consonants. It seems that *tha* and the cerebral consonants (except the nasal) are to be derived from *ṭeth*. The original nasal letters are *ma* and *na*. From *na* is formed *ṇa* by extending the head forwards. Bühler says that *ña* consists of 'two *na* joined together'. But this does not appear to be so. This relation is not maintained through all the inscriptions. *Ña* appears to be formed from either *ja* or *jha*, as is perfectly clear from its later form. This same relation exists in Brāhmī.¹ The nasal *ṅa*, seen in the conjuncts *ṅka* and *ṅga* only in the Sanskrit words, appears to be influenced by the Brāhmī form.

c18.s3.16

¹ See above, p. 37.

c18.s3.17

Thus we get: (1) the evolution of the basic form of the initial vowel, and from it other vowels with the addition of the diacritic marks; (2) one diacritic stroke for *dīrgha*, *guṇa* and *vṛiddhi*; (3) seventeen letters based on Aramaic; (4) six aspirates, *gha*, *chha*, *jha*, *dha*, *pha* and *bha*, formed by the addition of a diacritic stroke; (5) three letters, *kha*, *śa* and *tha*, again based on Aramaic; (6) the four cerebral letters, *ṭa*, *ṭha*, *ḍa* and *ḍha*, based on the form of *tha*; (7) two nasals, *ma* and *na*, based on the Aramaic; (8) the nasal *ña*, based on the form of *ja* or *jha*; and (9) *ṅa*, suggested by the Brāhmī form. Besides these two further letters evolved; these are shown on Pl. XXIII as *ṭh'a* and *chh'a*. The first is formed by adding a stroke to the form of *ṭha*, and the second is derived from *chha* by dropping its horizontal stroke. And finally we have another diacritic stroke, an upward slant at the lower ends of the verticals on the right, to make other sounds like *g'a*, *ch'a*, *ḍ'a*, &c. A similar stroke added to *sa* I have read as *s'a* instead of *sa*. Rapson argues for another form of *k'a*, in which the stroke is added to the top of the vertical upwards instead of at the bottom. This form is also met in the Kurram casket inscription.¹ From these practices it is clear that the original letter forms of Kharoṣṭhī were based on Aramaic as far as the Indian alphabet could allow. The remaining letters were evolved by the addition of the diacritic marks.

c18.s3.18

¹ Konow, *C.I.I.* ii, pl. XXIX.

Writing Styles

c18.s4.1

The stylistic changes in the Kharoṣṭhī letter forms, as seen through the centuries, are described in the following groups.

1. Aśokan Rock Edicts

c18.s4.t1.1

Pl. XXIII. 1 is taken from the rock edicts of Aśoka found at Mansehra in Hazara district and Shahbazgarhi in the Peshawar district, West Pakistan. Round features predominate in the letter-forms, though angularity is by no means absent. It is possible that the emphasis on angularity in some letters is due to the nature of the engraver's technique. As already noticed, an upward slant at the lower end of the long verticals on the left is usually seen in some letters. This will be henceforward called the foot-mark. This foot-mark has been observed only in the case of the initial *a* and some of those consonants which have the medial *a* implicit in them. Rapson² argues that it may stand for the long *ā*. But such a foot-mark is also seen in the late Brāhmī inscriptions of North India³ and in the documents from Chinese Turkestan, where it is quite clear that it is due to the pen style. In these documents we have a separate sign for the long *ā*. Therefore this foot-mark cannot be taken for *ā* in the Aśokan inscriptions.

c18.s4.t1.2

² *Kharoṣṭhī Inscriptions*, pt. iii, p. 299, n. 1.

c18.s4.t1.3

³ See above, p. 149.

c18.s4.t1.4

The initial *a* is of the fish-hook type, with the upper hook rounded or angular. The long vertical sometimes has the foot-mark. The other initial vowels do not have the foot-mark at all. The initial *i* has an additional sloping stroke across its head. The initial *u* has a stroke, either straight or slanting, added to the left bottom of the long vertical. The initial *e* has a head stroke on the top of the round curve. Another form of *e* with a horizontal stroke on the right of the vertical, as shown by C. C. Dasgupta,⁴ is not found at this time, but is seen in later inscriptions. The initial *o* has a stroke falling from the lower half of the vertical on the left. The *anusvāra* added to *a* takes the form of the miniature *ma*, appended at the end of the vertical, or just above the end. The medial vowels have diacritic marks similar to those of the initials. Minor differences are discussed along with the consonants.

c18.s4.t1.5

⁴ *Kharoṣṭhī Script*, pl. I. 4. 2.

c18.s4.t1.6.1

There is only one form of *ka*, a straight vertical with a line stroke on the top left and an angular hook facing downwards on the right just below the upper end. Variation is seen in this form according to the writer’s preference for the round or angular shape, or because of the foot-mark. *Kha* apparently has two forms: the first resembles the form of the capital ‘G’ minus the short horizontal with its upper hood drawn forward, and in the second form the hood of this ‘G’-form is missing. Actually in the second example the curve of the letter is straightened and drawn at an angle, and hence the difference between the two forms is merely stylistic. However, even in the later inscriptions this stylistic difference is kept up. The foot-mark is also occasionally seen. The stroke of the medial *o* runs parallel to the lower vertical. *Ga* has an upper hook which makes a loop when joined to the vertical. Variation is seen according to the writer’s preference for the round or angular form, or in the presence of a foot-mark. The stroke of the medial *e* in *ge* is applied in the middle of the hook. *Gha* has an extra stroke added to the right of the form of *ga*. This is either a slant hanging down from the loop or an angular hook added to the vertical on the right. *Cha* consists of an upper curve standing on the fish-hook type of leg. The upper curve and the fish-hook are sometimes angular, and if there is any space left between the two a bar is added to link them.

c18.s4.t1.6.2

Sometimes the fish-hook assumes the form of a lower curve. The foot-mark is sometimes seen. Fundamentally there is no difference in the actual form, and the variation is due to the manner of drawing a given shape. *Chha* has the upper curve on a vertical which is cut by a cross bar. We find another letter formed from this *chha*. In this case the cross bar is missing. Bühler read this form as *chha*, while Hultzsch and Konow in their readings took it for the conjunct *ksha*. Though the latter reading is justified by the context, it is not warranted by the form. In this form we must recognize a new sound coming closer to *chha*, which probably stood in this region for *ksha*. Hence I transliterate it as *chh’a*.¹ *Ja* is a vertical at an incline, with an oblique stroke to the left of the vertical at its upper side. This oblique stroke is sometimes curved. The long vertical also shows a foot-mark. C. C. Dasgupta¹ gives another variety of *ja* with a horizontal line added to the lower end of the vertical. This type is indistinctly seen in Mansehra R.E. V, line 19, R.E. VIII, line 35, and R.E. XII, line 1, but a careful observation shows that this horizontal stroke is nothing but the engraver's attempt at giving the foot-mark. We have no example of *jha* in the Aśokan inscriptions. *Ña* appears to be derived from *ja* or from the possible form of *jha*. In this case the strokes above the stem of *ja* are both bent.

c18.s4.t1.6.3

The bend over the vertical resembles the form of *na*, and hence Bühler believed that *ña* was actually double *na*. But the form as seen in Pl. XXIII*a*. 8–10 is derivable from the form of *jha* as seen in Pl. XXIII*a*. 11. This later form of *ña* is occasionally seen in the Aśokan inscriptions (e.g. Shahbazgarhi, R.E. I–III), with only this difference, that the vertical is below the right bend instead of the left. Stylistically we can distinguish two varieties of *ña*, one in which the left bend is a variation of the oblique stroke, and the other in which this bend is broadened. *Ṭa* consists of a vertical with a horizontal on either side, the left one towards its upper half and the right one towards the lower half. Sometimes the right stroke is missing, and occasionally a foot-mark is seen. *Ṭha* has two strokes on the upper left of the vertical. It is due to Boyer and Rapson² that we are now able to recognize the form of *ṭh\textsuperscripta*, which is made by adding a short vertical to the end of the lower horizontal of *ṭha*. *Ḍa* resembles the form of *ṭh\textsuperscripta* but with the upper horizontal stroke dropped. *Ḍha* has a horizontal stroke standing on a vertical, which sometimes has the foot-mark. The forms of *ṇa* and *na* are very close. *Na* consists of a vertical with a bent top, and *ṇa* extends the curve of the bend to the right. The foot-mark is also seen in both these letters.

c18.s4.t1.6.4

*Ta* has a horizontal stroke with a short vertical at either end, the left one going upwards and the right downwards. Very rarely the horizontal is inclined and the verticals slightly bent. The foot-mark is occasionally seen. *Tha* is a cross with its lower vertical slightly prolonged. In a number of places the lower vertical of *tha* has an oblique stroke added on the right. Hultzsch reads it as *thra*, but this oblique stroke is a diacritic mark and is to be distinguished from *ra* in a conjunct. Hence this letter should be read as *th\textsuperscripta*. *Da* resembles the form of *ta* but its horizontal dips towards the right and the two verticals are curved. *Dha* is actually a cursive form of *da* with its lower vertical prolonged and an extra horizontal stroke added to the top. The foot-mark is occasionally seen. *Pa* consists of a vertical with a hook on the right at the upper end. The variation is marked according to whether the hook is curved or angular, or drawn at an angle to the vertical. The foot-mark is also seen. *Pha* has an extra stroke added to the form of *pa*. *Ba* resembles the form of *ta* but the horizontal has its right end tilted upwards and the right vertical is slightly prolonged. *Bha* shows many variations. The main form consists of a vertical with a top horizontal stroke and an angular hook at its lower half on the right.

c18.s4.t1.6.5

This hook is occasionally curved, and at times the vertical is inclined, with the right hook taking the form of a sloping stroke. The foot-mark is also seen. *Ma* is a semi-circle with its mouth upwards. C. C. Dasgupta,¹ following Bühler, gives variant forms of *ma*, which has either a straight or curved line on the left side of this semi-circle. These forms have been given according to the reading of Hultzsch. But there is no justification for these separate lines unless they stand for diacritic marks. However, a close examination of the words in which they are seen makes it clear that the reading is not above doubt. In Shahbazgarhi R.E. VIII and XIII these lines are seen in the word read as *nama*, but *ma* in other words does not have this line. In Shahbazgarhi R.E. XII, line 2 it is seen in the word *mañati*, and in Shahbazgarhi R.E. VI, line 16 in the word *parakramatu*. A comparison with the other words where lines are seen in *ma* suggests that the actual reading should be *namaṁ*, *maṁñati*, and *parakramaṁtu*. The real difficulty faced by the writers was in distinguishing *mma* and *ṁma*, and hence these attempts were made. Subsequently we shall see that the hook became the usual form for the *anusvāra*. *Ya* consists of two oblique lines meeting at an angle upwards. The foot-mark is occasionally given only at the left leg. The *anusvāra* is indicated by upward slants, one at each leg. The medial *e* in *ye* is added to the top of the left leg.

c18.s4.t1.6.6

*Ra* resembles the form of *ta* with its lower vertical extended farther. The *anusvāra* is sometimes indicated only by a short horizontal line below the lower vertical of *ra*. *La* is exactly the reverse of *pa*. Here the hook is added to the left of the vertical. The variation occurs in the position of the hook. either at an acute or at a right angle to the vertical, and in the presence or absence of a foot-mark. *Va* has a short horizontal stroke added towards the left at the upper end of a vertical. The palatal *śa* has a vertical falling from either end of a horizontal line. The foot-mark is seen only at the left vertical, and the *anusvāra* is applied in the same way as in *ya*. The cerebral *sha* assumes the form of an open umbrella, while dental *sa* has a tailed head on a stem. *Ha* is the reverse of the Roman ‘S’. Here the lower curve is sometimes straightened.

c18.s4.t1.7

¹ See remark by Rapson, *Kharoṣṭhī Inscriptions*, pt. iii, p. 302.

c18.s4.t1.8

¹ *Kharoṣṭhī Script*, pl. I, 13. 2, 3.

c18.s4.t1.9

² *Kharoṣṭhī Inscriptions*, pt. iii, p. 304.

c18.s4.t1.10

¹ *Kharoṣṭhī Script*, pl. I, 30. 2–5.

c18.s4.t1.11

In the conjuncts, where the lower ends of the verticals have some other strokes, the *anusvāra* is given in the middle of the vertical, as in *kraṁ* and *vraṁ*. As we have seen, the system of forming conjuncts is the same as in Brāhmī, though the cursive forms are found here more frequently, as in *mma*, *mya* and *bhye*. When *ra* is the second element of the conjunct, it is shown by a horizontal line on the right at the lower end of the vertical, as in *tra*, *dra*, *kraṁ*, *pri*, *śru* and *vraṁ*, but when it is the first element, the usual form is shown cursively in the middle of the vertical. The form of *va* in *tva* is very cursively drawn. Similarly *sa* in *spa* shows only its upper tail. The most difficult conjunct to recognize is *sti*, in which *sa* is shown only by a horizontal line.

2. Indo-Greek Coin Legends

c18.s4.t2.1

Pl. XXIII. 2 is taken from the coin legends of the Indo-Greek rulers, except those of Hermaeus and other later kings. Only the form of *vṛi* is taken from the coin of the *Vṛishṇi Janapada*.¹ The palaeography of these coins is influenced by the metallurgical technique. The letters generally show angular features. In a few letters there are minor changes from those seen in the Aśokan inscriptions. The *anusvāra*, as in *aṁ*, is indicated by a hook placed at the end of the vertical. The additional stroke of the initial *i* is marked almost horizontally in the middle of the vertical and that of *u* slants downwards. The stroke of the medial *u* in *thu* is turned up. The foot-mark is seen in *kha*, *da* and *ta*.² The most difficult features to explain are a line or a dot below *ma* and *ha* in the word *maharajasa*, a line below *da* probably to distinguish it from *ta*, a double or single stroke at the left end of the long vertical of *ja*, and a line sometimes inserted below *ga*, *na* and *sa*. It seems that the lines below the verticals are remnants of the foot-marks seen in the inscriptions. But in some cases the marks appear to imply some special phonetic value. The clearest example is that of *da*, which has the form identical with that of *ta* but for this horizontal line. It may therefore be assumed that this line is added to distinguish the two letters. In the case of *na*, which occurs only in the word *mena*(*ṁ*)*drasa*, probably the correct reading is *naṁ* and not *na*. Similarly, in the case of *ja*, *ga*, *sa*, *ma* and *ha* some phonetic difference may be implied. If this suggestion is not accepted, we must assume that these marks became conventional in the coins, as they are not found in the inscriptions at all.

c18.s4.t2.2

¹ J. Allan, *Catalogue of the Coins of Ancient India in the British Museum*, London, 1936, pl. XVI. 5.

c18.s4.t2.3

² P. Gardner, *Catalogue of Indian Coins in the British Museum, Greek and Scythian Kings of Bactria and India*, London, 1886, pl. IX. 8–12.

3. Indo-Greek Inscriptions

c18.s4.t3.1

Pl. XXIII. *3* is taken from the Bajaur casket inscription of the reign of Menander (only inscription nos. A, A 1 and A 2), and from the Swat relic vase inscription of Meridarkh Theodoros. The letters are scratched on steatite and show a style that was in vogue in the second to first century B.C. The forms are similar to those seen in the Indo-Greek coins, but the additional strokes below the verticals are absent here. Angularity is particularly marked in the forms of *a*, *ka* and *ṇa*. The new letters *g'a* and *d'a* occur. The foot-mark is also seen in *da*, while it is usually cursively drawn. The most important change is seen in the form of *mu*, in which the right limb is extended upwards but the left sharply bends down with the stroke of the medial *u*. Henceforward this becomes the common form. *Ra* in the conjunct *rkhe* is just a curved line drawn across the lower vertical. The initial *u* has its left stroke drawn as if it were a bend of the long vertical.

4–8. Scytho-Parthian Period

c18.s4.t4.1

There are a large number of inscriptions falling in this period, some dated and others undated. There has been a great controversy as regards the era or eras to which these inscriptions are to be assigned. The chart shown in Pl. XXIII. *4*–*8*, has been made in order to draw attention to the different hands that were responsible for the varying modes of writing seen in this period. In general the inscriptions fall into two categories: (*a*) those represented in nos. 5 and 6 show a great similarity to no. 3 above, but at the same time mark a new stage in the development of the Kharoshṭhī writing, and no. 6 especially brings us to the early years of the Christian era; (*b*) no. 7 is taken from the Takht-i-Bahi inscription of the year 103, which definitely falls in about the middle of the first century A.D. No. 4 shows a late development. Its letters are much more advanced, and the characteristic use of the new letters *kh'a*, *g'a*, *t'a*, *d'a*, *dh'a*, *y'a* and *s'a*, which are formed by the addition of an oblique stroke at the right side of the lower end of the vertical, places it in line with the inscriptions of the Kushāṇa period. In no. 4 we also find the medial *u* shown by a loop and *sa* with the mouth of its head wide open. No. 8, which is taken from the Mathura lion capital inscription,¹ is difficult to date. Here the medial *u* is formed by a loop, and very rarely this loop has an open mouth, but sometimes we also meet with a simple bend. The right hook of *ka* generally slopes downwards, though occasionally we also meet with the angular form. *Kha* has no hood on the top. *Ña* is of the later type, with a crooked hook attached to the upper half of a vertical on the right. *Da* has two forms, one in which the horizontal is less marked, as in no. 6, and the other showing the older form with a significant horizontal line. The lower end of *dha* is drawn at an incline to the left—a form seen in the later period. These double forms can also be seen in *na*, *ba*, *bha*, *la* and *śa*. *Ya* is of the angular type. *Sa* again shows the mixed varieties. The most common is decidedly later, with its head having its mouth wide open. *Ha* occasionally has the lower curve straightened but more often it is awn farther, as in the Kushāṇa period. The most important new cure is the use of new letters *k'a*, *g'a*, *ḍh'a*, *t'a*, *bh'a*, *r'a*, *ś'a*, *ṣ'a* and *h'a*, all of which are not marked in the chart given. Some of the letters are very loosely drawn, showing a wave in the vertical, and the others show a gentle curve towards the left as the verticals descend. These features are strange, and are not seen in any other inscription of this period. I would therefore suggest that this inscription should not be taken into consideration while determining the style of the Scythian period, because its features remind us at once of the writing known in the inscriptions of the Kanishka group of the rulers. Therefore the genuineness of this inscription is not above suspicion.

c18.s4.t4.2

¹ For a detailed analysis of this inscription see A. H. Dani, ‘Mathura Lion Capital Inscription (A Palaeographical Study)’, *Journal of the Asiatic Society of Pakistan*, v. 128–47.

c18.s4.t4.3

The inscriptions of group (*a*) show some distinguishing features: the fish-hook form of the vowel base sometimes has an angular head. The initial *i* has its horizontal stroke in the middle of the vertical. The initial *o* in no. 5 has its horizontal stroke added to the left middle of the vertical. The medial *u* turns its stroke upwards. The *anusvāra* is always indicated by a hook at the bottom of the letters. The consonants also show some changes. *Kha* has an upper hood in nos. 5 and 6, and in them we also see the old form of *cha*. The lower left horizontal of *ṭha* is optionally inclined. The right hook of *bha* is now applied at the upper half of the vertical—a style which henceforward becomes common. *Śa* has its verticals equal in length to the horizontal. The lower curve of *ha* is invariably straightened. We also meet a new conjunct *śpa*, in which *śa* is cursive, shown only by an oblique line falling down from the upper end of the vertical of *pa* on the left. In (*b*) we have the medial *u* shown by a loop. *Śa* optionally has a rounded top. *Sa* has the mouth of its head wide open. The most difficult problem raised by these inscriptions is to decide about the phonetic value of what has been read as *vha*. Rapson¹ prefers to read it as *ph a*. But, as we have seen before, the phonetic difference from *pha* would normally be shown by adding an oblique stroke on the right at the lower end of the vertical. This clear departure, in which we find a combination of a *va* with an angular hook on the right with its mouth upwards, argues for a conjunct. Rapson's reading was influenced by the Greek legend on the coin of Gondophares, but from the palaeographical point of view this reading is not justified. Although, of course, the form of *ha* is not at all clear in the conjunct, *vha* appears to be a better reading.

c18.s4.t4.4

¹ *Kharoshṭhī Inscriptions*, pt. iii, p. 307.

c18.s4.t4.5

*9–12. Kushāṇa period*

c18.s4.t4.6

The inscriptions of this period can be classified in four groups: (*a*) Those inscriptions which mention the name of Kushaṇa, Khushaṇa, or Gushaṇa, or those dated after the year 122, beginning from the Panjtar inscription (Konow, no. B. XXVI). Pl. XXIII. 9 represents this group. The Takht-i-Bahi inscription, described above, shows a writing very similar to that of the present group. (*b*) This group includes the stone inscriptions dated in the new series and referrable to the time of the Kanishka group of the rulers. The writing here generally continues the style of (*a*), though some new influences from (*c*) are also traceable. It is represented here by Pl. XXIII. 11. Of the inscriptions falling in this group the Mamane Dheri pedestal inscription (Konow, C. LXXXVIII) shows great similarity to those of the next group. (*c*) This group is represented by Pl. XXIII. 10, 12, which introduce the true Kushāṇa style of the time of the Kanishka group of the rulers. To this group belongs the Bimran vase inscription (Konow, B. XVII), and its date lies in this period. No. 12 is the latest in the series and has some connexion with the writing known from Chinese Turkestan, discussed below. The dominant feature of this writing is the characteristic bend of the lower end of the verticals towards the left. The medial *i*, as in *niṁ* in no. 10, is occasionally given at the left end, in place of the middle as before. Rapson¹ correctly recognized the form of *visarga*, as in *naḥ* in no. 12. Here the two dots placed side by side are given on the top of the letter. The medial *u* in no. 11 assumes the form of a triangle. In no. 12 *ka* joins the left horizontal and the right hook in a single curved stroke—a form which is commonly found in the manuscript writing from Chinese Turkestan. *Kha* here always has an upper hood. In no. 11 *jha* is the reverse of *ña* as seen before. This form of *jha* with a vertical on the right and a curved stroke on the left is the usual form in the manuscripts. Generally the right horizontal of *ṭa* is not shown. In no. 12 *ḍ'i* is drawn cursively in a single sweep of the hand, and *ṇa* has an insignificant tick in place of the head curve. In nos. 10 and 11 *da* is just a straight line with the characteristic bend at the lower end. *Ya* in nos. 11 and 12 has a flat top, and it is hard to distinguish it from *śa*. The left vertical of both *ya* and *śa* have the usual bend. *Sa* is drawn very cursively with the mouth of its head wide open. *Ra* in the conjuncts such as *rva*, *rya*, and *rtha*, is drawn with a loop at the junction of the vertical. Of all the new conjuncts the most remarkable are those with *ya* as the second element. It is of one type in no. 10, as in *sya* and *shya*, which have a curved stroke beginning with a loop and turning to the left, but in no. 12 the additional hooked stroke is turned to the right, as in *pya*, *sya*, *kya* and *lya*. (*d*) The last group of the Kushāṇa period includes those inscriptions which are dated in the series 303 to 399, beginning with the relic casket from Charsadda² of the year 303. As these inscriptions continue the style of the stone inscriptions of (*b*), they have not been represented in the chart. But their dates fall after the time of the Kanishka group of rulers. Here also we get the flat-topped *ya*, the open-mouthed *sa*, and the looped type of the medial *u*.

c18.s4.t4.7

¹ *Kharoṣṭhī Inscriptions*, pt. iii, p. 300.

c18.s4.t4.8

² *Epigraphia Indica*, xxiv. 8–10.

13. The Manuscript Style

c18.s4.t5.1

Pl. XXIII. 13 is taken from pl. I of the *Kharoshṭhī Inscriptions from Chinese Turkestan*, edited by Boyer, Rapson and Senart. These letters are given in order to show the manuscript style, as it was known during or immediately after the Kushāṇa period. As has been pointed out by Rapson, the style bears great similarity to the writing seen in no. 12 above, but it is in this pen style that we learn all those minute details which are the natural results of writing with the pen, but which are not faithfully reproduced in no. 12 above. No attempt is made here to distinguish different hands in the manuscript, nor have I tried to reproduce all the letter forms from the manuscript. For this second purpose I reproduce the chart¹ prepared by Rapson, who has collected in one place the different letters occurring in the manuscript (Fig. 20). Here my purpose is simply to emphasize the important aspects of the manuscript writing. Some general characteristics are to be noted first. The verticals either end in a slight upward slant on the left, or gently curve towards the left. The horizontals also show slight curving, and the additional strokes, as in *e*, *g'a* or *ñe*, are drawn with one sweep of the hand, leaving a closed loop at the end of the verticals. The medial *i* is sometimes given at the left of the letters instead of across the middle. The general tendency is to simplify writing rather than to develop an ornamental form. As a result of this last tendency the medial *e* is added, as in *ñe*, at the lower end of the vertical; both *ya* and *śa* have rounded tops; both *ta* and *va* sometimes lose their horizontal (generally one); *ta*, *da* and *na* have almost identical forms—a double-curved vertical; the head of *sa* is very cursively drawn, and *ha* develops a tail at the end of the lower curve. On the whole the manuscript style shows that the writing traced in the inscriptions was not far removed from the pen style known at the time.

c18.s4.t5.2

¹ Some of the signs were wrongly interpreted by Rapson, e.g. no. 221 is now correctly read as *ly'a*, no. 182 as *tg'a*, no. 192 as *ts'a*, no. 250 as *l'sa*. The chart is thus partially out of date.

Conclusion

c18.s5.1

In the above description and the chart given in Pl. XXIII an attempt is made to point out different tendencies, as they developed through the centuries during which Kharoṣṭhī was in vogue. The inscriptions selected, with one exception, are those which are definitely dated. The forms given in the chart are my own reproductions of the letters as finally ascertained after examining the different examples. It has been my endeavour to reconstruct the most usual form of the letters, out of which the individual writers created their own style of writing. In these general forms Kharoṣṭhī maintains a uniformity throughout the periods that it was in use. This uniformity of the Kharoṣṭhī, as opposed to the changing forms of the Brāhmī script, speaks of a wider cultural unity in this region.

c18.s5.2

The following list follows the arrangement of the plate.

c18.s5.3

1. (i) Shahbazgarhi and Mansehra R.E. of Aśoka, E. Hultzsch, *C.I.I.* i, *Inscriptions of Aśoka*, Oxford, 1925, pp. 50–84.

c18.s5.4

2. (i) Coin legends of the Indo-Greeks, except those of Hermaeus and other late rulers, Percy Gardner, *Catalogue of Indian Coins in the British Museum, Greek and Scythian Kings of Bactria and India*, London, 1886.

c18.s5.5

3. (i) The Bajaur casket of the reign of Menander by N. G. Majumdar in *Epigraphia Indica*, xxiv. 1–8 (only nos. A, A 1 and A 2). For correction see Sten Konow in *New Indian Antiquary*, ii. 647.
(ii) Swat relic vase inscription of the Meridarkh Theodoros, Konow, *C.I.I.* ii. 4 and pl. I.

c18.s5.6

⁴ (i) The Bajaur casket inscription (the remaining inscriptions), *op. cit.*
(ii) Taxila copper plate inscription of a Meridarkh, Konow, *C.I.I.* ii. 5 and pl. I.

c18.s5.7

⁵ (i) Shahdaur inscription of Damijada, dated year 60, *ibid.*, p. 16 and pl. III.
(ii) Shahdaur inscription of Śivarakshita, *ibid.*, p. 17 and pl. III.
(iii) Mansehra inscription of the year 68, *ibid.*, p. 20 and pl. IV.
(iv) Fatehjang stone inscription of the year 68, *ibid.*, p. 22 and pl. IV.

Linked texts (8)

Dhammapada — Pāli
Shahdaur Inscription of Śivarakshita — Gāndhārī
Shahdaur Inscription of Damijada — Gāndhārī
Aśoka
Epigraphia Indica
Corpus Inscriptionum Indicarum
Kharoṣṭhī inscriptions
Mathurā lion-capital inscriptions — Prakrit