Pramāṇa · Search · About · API

Indian Palaeography c13: 5. The Provincial Brāhmī Scripts to the Middle of the First Century A.D.

Indian Palaeography c13

Language
English

Contents

5. The Provincial Brāhmī Scripts to the Middle of the First Century A.D.

c13.2

AŚOKAN Brāhmī, described in the last chapter, had a unity of purpose which ultimately determined the style of writing. That script, in spite of minor differences, can be termed imperial. In contrast to the Aśokan inscriptions stand a great number of records, which, except for the donative inscriptions of Daśaratha, the grandson of Aśoka, show a variety of purpose and are entirely different in content and style from the Aśokan. Whether we take the administrative records, like the Mahasthan inscription or the Sohgaura plate, or study the royal rescripts, like the Nanaghat or the Hathigumpha inscriptions, we are struck by the great change that marks them out from the Aśokan mode of expression. To these we must add the relic casket inscriptions found within the monumental stūpas which were to develop into great store-houses of stone sculptures bearing numerous gift records, and also the short epigraphs mentioning the excavation of caves found far away in the south—all due to the influence of the Buddhist (and very rarely Jaina) monks. Last but not the least is the writing found for the first time on coins, which no longer follow the old technique of punching, but are produced by the techniques of incuse stamping as in the Negama coins from Taxila, die-striking, and even casting. All these mark a great advance from the world of the Mauryas. The Mauryan standard no longer held good in this new context, but unfortunately the old palaeographers, for lack of better evidence, took Aśokan Brāhmī as the time-scale for dating all sorts of writing. This was evidently not valid for the vast subcontinent where the imperial hand was no longer the moving factor; and further, when we take into consideration the diffusion of technical knowledge, we have to bear in mind the question of time-lag from one region to another. It is not necessary to think of a universal standard of writing for the whole of India and Pakistan, nor should we expect the evolution from Aśokan Brāhmī to take place in all the regions in the same manner. Aśokan Brāhmī, as we know today, was the creation of the royal scribes, but the inscriptions included in this chapter are of many kinds—royal, popular, and individualistic. They follow different standards reducible into various regions on the basis of technical achievement, and hence the title of this chapter—The Provincial Brāhmī Scripts. The middle of the first century A.D. completes the incorporation of a new technique of writing that came in with the intrusion of the Śakas into the subcontinent.

c13.3

*New approach to the problem*

c13.4

The shape of the Aśokan letters has usually been taken to be the basis for dating the inscriptions all over the subcontinent. While we hear of the Mauryan inscription from Mahasthan in the Bogra district, East Pakistan, we are also told, from purely palaeographical criteria, that the cave inscriptions from the south belong to the same period. In the determination of the age of an inscription the ratio of the length of the verticals to the breadth of the letters has been used most carelessly, without properly understanding its significance. Further important evidence given both by Bühler and Ojha is from the mistaken reading of the Maurya era in the Hathigumpha inscription, and accordingly it has been dated 'between B.C. 157 and 147'. Various synchronisms have been established on this false statement. Again Ojha alone speaks of the Mahāvīra era on the basis of the Barli inscription and dates it to the 84th year of this era. Dr. Jayaswal's correction makes no improvement, while the new reading given by Dr. D. C. Sircar is still hypothetical, though somehow he reaches the correct conclusion that the inscription should be dated to the first century B.C. His argument is not based on palaeography but on his supposed identification of Bhagavata mentioned in this inscription with King Bhāgavata in the Bhilsa record. Lastly palaeographers have used the word 'serif' very loosely without properly defining it, and have built up their chronological schemes on faintly recognizing it in some inscriptions.

c13.5

As we have seen in the previous chapter,¹ no reliance can be placed on the length of the verticals. There is no gradual reduction in the length from Aśokan Brāhmī onward. But there is a definite reduction of the verticals in later inscriptions, and the approximate equalization of length and breadth is seen for the first time in the Śaka records. It is not found in the coin legends of the Indo-Greek rulers Antialcidas and Pantaleon. What necessitated the change during the Śaka period? The answer is related to the change in Greek palaeography, as noted by several numismatists in the coinage. Without digressing into the question of Greek palaeography I must emphasize the well-known use of the square omicron on the Śaka and Parthian coins. Its adoption is not an isolated phenomenon. It had a fundamental effect on the style of writing, and it is this new tendency that simultaneously influenced the Brāhmī writing as seen in the Śaka coins and inscriptions. Even before this the metallic issues had begun to exert influence on the way of writing, and, as indicated by the coins of the Greek rulers, the angular feature was, no doubt, dictated by the Greek lettering as well as by the technique of die-cutting. But this was a slow and doubtful process, as is shown by the forms in the Besnagar inscription of Heliodorus. It became a fully established characteristic only in the Śaka records, and from this time onward dates one most important character of Indian scripts—the equalization of all the verticals. They assumed a new face producing an effect not dissimilar to the Greek lettering. Another great change was due to the introduction of a new writing tool. This was a broad or edged pen, the use of which is noticeable clearly in the drawing of the verticals, which begin with a thick top and gradually thin downward—an effect produced even today in the Punjab by the reed pen which is used by children for learning the Arabic alphabet, which they write on a wooden board. It is from the manner of holding the pen and its proper twisting as it moves that gradually thinning lines are produced. Some children, not fully conversant with this method, produce lines of equal thickness throughout, as we see in the Ghosundi inscription. This new tool gave a new face to the letters already recognized by the earlier palaeographers, e.g. the form of *ka*, which is vaguely described as 'dagger-shaped'. This top-head formation was not deliberately done. It was implied in this new process of writing itself. It also resulted in a beautifully curved flourish in the formation of the medial *i*. In some places this technique was properly known and used, but in backward regions it was only copied in a crude fashion. In the latter case the head-formation appears like a dot or a sign, as in Nanaghat II and Arikamedu II (Pls. VI, VII). However, this head formation should not be confused with the 'serif', a term which I reserve for the deliberate marking of the heads of the letters, as we shall see in the next chapter, and which, no doubt, was a further development from the new style of writing introduced in this period. In the next stage this serif mark developed into a full headline, and ultimately we get the continuous top line of the later Indian scripts.

c13.6

¹ See above, p. 33.

c13.7

*Classification*

c13.8

In order to verify the actual working of this new approach the inscriptions have been classified by regions. Four main regions stand out clearly: (1) Eastern India with the old centre in Magadha, and the new foci—one, south of the Ganges at Kauśāmbī, the influence of which went south-west to Bharhut and eastwards to Bodhgaya as far down as Hathigumpha in Orissa, and the other, north of the Ganges at Ayodhya. (2) North-west India with its centres at Taxila, Mathura, and a number of places in the Malwa plateau on the route leading from the north to the Deccan. (3) North-west Deccan, which is a continuation of the culture from the Malwa plateau. (4) South India, with which has been included Eastern Deccan. Amaravati and its neighbourhood saw a farther southward progress of the culture from Western Deccan, and from this centre writing trickled into the barbarous hill-caves of the extreme south. Sanskrit or Prākrit speakers from the north, as is shown from the Sanskrit legend on a potsherd from Arikamedu,¹ brought northern technical advancement into the slow-moving backward region of the south. For each of these regions a different standard was to be established according as the new technique of writing was variously practised.

c13.9

¹ See below, p. 74.

c13.10

*Development of the alphabet*

c13.11

Before we go into the details of the regional writings, we must catalogue the new letters and conjuncts that were introduced in these inscriptions.²

c13.12

² For Dravidian alphabets see below, p. 70.

c13.13

In the initial vowels we have three additions to those used in Aśokan Brāhmī. The long *ī* is formed by adding a fourth dot to the three dots of the short *i*, each dot being placed at the four angles of the quadrangle (Fig. 6. I). Ojha reads the broken sign [Figure:] in the Barli inscription as *vī* and argues that the curved addition over *va* should be taken for long *ī*, but if he is correct this is a unique example, and in any case the letter is defective. Dr. D. C. Sircar

c13.14

FIG. 6

c13.15

suggests that it is a compound of *dva*. But the form of *da* opening on the right is not seen in this period save for the exceptional script of Bhattiprolu. In its fragmentary nature it is difficult to suggest a definite reading. The long *ū* is formed by the same old principle, by adding a further horizontal stroke to the short *u* (Fig. 6. II). Similarly *ai* is formed by adding a stroke to the left of *e* (Fig. 6. III). Besides these, we also find the use of *visarga*. It is marked by adding two dots one above the other at the end of the letters (Fig. 6. IV). In the medial vowels we have the addition of *au*, formed in the fashion already anticipated in the last chapter,¹ and *ṛi* which is added as an inclined stroke to the bottom of the letter (Fig. 6. V).

c13.16

¹ See above, p. 46.

c13.17

The most important change took place in the formation of the conjuncts, especially in correcting the combinations formed with the consonant *ra*. In Aśokan Brāhmī there was no difference, whether *ra* was the initial or the second letter in the compound. In both the cases it was shown by giving a serpentine vertical to the straight verticals of the compounded letters. In these inscriptions the initial *ra* is always shown by a small vertical added to the top of the other letter (Fig. 6. VI), while *ra*, when following, is added below the other consonant (Fig. 6. VII). All other compounds are formed as required by Sanskrit grammar. In the Ramgarh inscription the sign [Figure:] has been read as *spha*, but such a combination, if correctly read, is unique. Elsewhere the second consonant is never added to the secondary loop of *sa* as in this case. The reading is doubtful. At Sanchi (inscription no. 3, Stūpa I) the sign [Figure:] has been read as *hnu* in order to form the word *hnusā*, derivable from *snushā* (daughter-in-law). But inscriptions nos. 284 and 286 prove that the reading should be *nhu*. It is possible that *hnusā* was locally pronounced as *nhusā*.

c13.18

There is also one addition to the consonants, found once at Sanchi and once at Mathura. This is the West Indian *ḷa*. At Sanchi it occurs in the place-name *Vāḷivahana*¹ which has been spelt in other places as *Vāḍivahana*; and at Mathura it is used in the word *kāla-vāḷa*,² which Bühler takes for *kāla-vyāla*. It is formed from the letter *ṭa* (Fig. 6. VIII). This letter resembles *ḍha*, but it is distinct. It should not be confused with the North Indian *ḍa*, but should be taken as representing similar sounds occurring in the Dravidian languages and is to be regarded as the parent of the West Indian *ḷa*, which survives in modern Marathi.

c13.19

¹ J. Marshall and others, *Monuments of Sanchi*, Calcutta, 1940, pl. CXXXI, no. 187.

c13.20

² *Epigraphia Indica*, ii, pl. opposite p. 397, no. 33.

Eastern India

c13.s1.1

In this region there are five odd inscriptions which, being isolated in far-flung areas, have proved difficult for palaeographical study. The first is a series of three inscriptions from the Nagarjuni caves in Bihar, found not far from the Barabar caves where Aśokan inscriptions are engraved. These have been securely dated as they refer to Devānaṁpiya Dasaratha, the grandson of Aśoka. Palaeographically the inscriptions of Aśoka and Dasaratha are closely comparable. Both of them have short verticals. Some words, such as *kubhā* and *(a)bhisitena*, are virtually identical in the two series of inscriptions, but there are also some differences, e.g. in the forms of *a*, *ya*, and *va*. The Dasaratha inscriptions show a more cursive hand than the Aśokan. In the reading of the Dasaratha inscriptions there has been one blatant mistake. The dental *sa* has been taken for cerebral *sha*, but the error is proved by the addition of the medial *ū* to the main curve of *sa*, e.g. [Figure:] (*sū*).

c13.s1.2

The next four inscriptions are markedly individualistic. The Piprahwa vase, found in the Basti district, U.P., has an inscription scratched on the steatite stone in a careless manner. As the inscription refers to the remains of the Buddha, it was originally dated to the pre-Mauryan period, but it has been brought down to the third century B.C. on a comparison with Aśokan Brāhmī. The style of writing is very poor, and there is nothing in it that speaks of the hand of the Aśokan scribes. We should not overlook the crude form of *ka* and *ta*, in which the vertical is unnecessarily lengthened. But here we do not find the later form of *dha* (i.e. the reverse of the Roman capital letter ‘D’), and the medial vowels are applied as in Aśokan Brāhmī. We should mark also the angular form of *pa*, the long oval of *va*, and the reduction of the vertical in *la* and *na*, when the medial *i* is added, as we shall see also in the Barli inscription.¹ On these grounds the inscription may be confidently dated to the earlier half of the second century B.C.

c13.s1.3

¹ See below, p. 63.

c13.s1.4

In contrast to the above inscription the Sohgaura (Gorakhpur district) copper plate shows an advanced technique of casting, in which the letters are in high relief. Quite naturally the writing is very stylized and some of the letters have marked angular features. The form of *ma* is very exceptional, recalling that of the optional form at Bodhgaya and of the Mysore inscriptions of Aśoka (Pl. IV*b*), but this is unique. It can be safely assigned to the earlier half of the second century B.C.

c13.s1.5

We will take up the next two inscriptions together: one is from Mahasthan in the Bogra district, East Pakistan, and the other from Ramgarh in Chhattisgarh, M.P. In the latter place there are actually two inscriptions, chiselled on the rock surface, the first using *ra* and dental *sa*, and the second replacing *ra* by *la*, and dental *sa* by palatal *śa*. Except for this linguistic difference there is close similarity in the style of writing. The letter *ya* in the Ramgarh inscriptions is wedge-shaped (Fig. 5. V*a*), an angular development from the crescentic *ya* of Aśokan Brāhmī. Angularity is also marked in *pa*, *la*, *sa*, and *ha*, but this is of a different type, produced by the shorter vertical which is drawn at an angle. *Ra* has two forms, the straight vertical and the serpentine vertical; the last resembles the form seen in the Besnagar inscription of Heliodorus (Pl. V*b*). But *bha*, with notched vertical and the two separate horizontal lines for medial *o*, suggest a date not later than that of the Besnagar inscription, possibly the middle of the second century B.C. The Mahasthan inscription bears a close resemblance to the second epigraph from Ramgarh, except for the palatal *śa*, which is replaced here by dental *sa*. The form of this *sa* is very similar to that in the Dasaratha record. Both at Ramgarh (only in the second epigraph) and at Mahasthan the straight vertical (*daṇḍa*) is used as a punctuation mark. The main difference is in the forms of *ya* and *kha*. At Mahasthan alone *dha* keeps up the true shape of the Roman capital letter ‘D’, suggesting that its date is earlier than the time when the reverse variety became the common fashion. It may therefore be placed in the first half of the second century B.C.

c13.s1.6

The remaining inscriptions from this region (i.e. from Bharhut, Bodhgaya, Hathigumpha, Pabhosa, and Ayodhya) will be taken together (Pl. VI *a*, *b*). The Bharhut inscriptions fall into two groups: Bharhut I representing those from the railings, and Bharhut II from the gateway inscription mentioning the Śuṅgas. All these inscriptions have certain common characteristics, the small difference being mainly due to different hands and to the varied knowledge of the use of the new technique. In all these examples the left-hand curves of *a* and *ā* meet the verticals separately (as in Fig. 4. I), though optionally in Bharhut I, Pabhosa, and Ayodhya they meet at the same point. In all these cases the vertical of *ka* is lengthened. *Kha* has below its vertical a dot, a circle, or a triangle. *Ga* is round-topped; only exceptionally an angular form is seen in Bharhut I and at Bodhgaya. *Gha* has assumed a perfect angular form. The semicircle of *cha* here makes a rough quadrangle. *Chha* shows two distinct circles on either side of the lower end of the vertical, though in Bharhut I we have optionally a simple oval. *Ja* in the majority of cases is three-armed, but it also has the double curved variety, especially in Bharhut I. *Ṭa* has both flattened and round forms. *Ḍa* has its top vertical shortened, while the lower is lengthened and makes a slight curve with the horizontal. Sometimes *ṇa* has its lower horizontal bent. *Ta* usually has the curved form, but the angular form is seen optionally in Bharhut I and at Bodhgaya. *Da* is developing the rounded form but it still opens to the left. The angular *da* is also occasionally met. *Dha* has the reverse form of the Roman capital letter ‘D’. The letters *pa*, *sa*, and *ha* have their lower curve changed into an angular form. *La* has a peculiar round cursive form, but at Pabhosa, Ayodhya, and Hathigumpha it shows a new angular form known elsewhere only from the Śaka Kshatrapa inscriptions of Mathura. *Bha* has a straight vertical with its additional hook slightly broadened. *Ma* has its lower circle turned into a rough triangle, though the circle is retained in Bharhut I and optionally at Bodhgaya and Hathigumpha. *Ya* has both the crescentic and the double-curved varieties, both sometimes giving place to angular forms. The vertical of *ra* is generally straight, but occasionally serpentine. *Va* has its lower circle changed into a definite triangle, which is occasionally very rough.

c13.s1.7

There is also some change in the application of the medial vowels. In Bharhut I the medial *ā* in *jā* is optionally attached separately to the top. Except in Bharhut I, the medial *i* shows a tall flourish, developing into a beautiful curve in the Pabhosa and Ayodhya inscriptions. This form is derived from Śaka Kshatrapa records. The medial *u* in *pu*, *bu*, *su*, and *hu* is a downward stroke drawn in line with the right vertical; only in Bharhut I does the older practice persist. The medial *o* is a single horizontal line drawn on the top of the letters.

c13.s1.8

These common features are new developments. But Bharhut I shows a number of differences from Bharhut II and other records. Only here do older forms optionally persist, suggesting that it is somewhat earlier than the other inscriptions. The date of these latter inscriptions is fixed with more or less certainty by Pabhosa and Ayodhya examples, where we find the full use of the new pen and equalization of the verticals as in the Śaka Kshatrapa inscriptions of Mathura. In Hathigumpha, Bodhgaya, and Bharhut II these forms are copied. It does not seem that the writers knew the use of the new pen. This is also the reason why there is great variance in the length of the verticals in the same inscription. It must also be pointed out that if one analyses these inscriptions from the morphological point of view, he will find the disparity in the forms very confusing; for example the inscriptions other than the Hathigumpha found in the Khandagiri and Udaigiri caves, which are all contemporary, show marked differences in the forms. These other inscriptions show a preference for archaic forms which are optionally met with in the Hathigumpha inscription. This is bound to be the case in such a remote region as Orissa, where the new technical tradition could not get a foothold for a long time. It is only by incidental letter forms that the palaeography of this region is related to those of other areas. This is an important point, but it is usually forgotten by Indian palaeographers, who are bewildered to find archaic forms surviving side by side with the newer types. All these inscriptions date from the same time as that when the Pabhosa and Ayodhya records derived the new technique from the Śaka Kshatrapa inscriptions of Mathura, i.e. in the first half of the first century A.D.¹ Hence Bharhut I may be dated in the later half of the first century B.C.

c13.s1.9

¹ Much controversy has been raised over the phrase, *Naṁda-rāja-ti-vasa-sata*, occurring in the Hathigumpha inscription. The phrase is ambiguous. Even if we accept the opinion that it refers to 300 years after Nandarāja, not much reliance can be placed on this round number 300. The actual time may be slightly more or less.

North-West India

c13.s2.1

Bühler vaguely hinted at a regional script for North-west India, in which he included the Kalasi R.E. of Aśoka and the coin legends of the Indo-Greek rulers. Though some of the letter-forms of Kalasi persisted in the later inscriptions, it is hard to maintain that a regional script existed as early as the time of Aśoka. The Aśokan P.E.s, now at Delhi, go against any such assumption. The regional script of North-west India started much later, when its letter-forms began to be influenced by the palaeography of the coins, especially of the foreign intruders. This change was natural, as much of the region was within the orbit of the foreign rulers, and we have good evidence for contact and mutual exchange.

Coins

c13.s2.t1.1.1

We will begin our study with the palaeography of the coins (Pl. V *a, b*). Allan was the first numismatist to make a detailed study of the palaeography of the Brāhmī coin legends. His chronological scheme is taken for granted by others. But it is clear from a perusal of his *Catalogue of the Coins of Ancient India* that his study was influenced by factors other than palaeography. When no other evidence was available he fell back on the style of writing. No systematic attempt was made by him to analyse the letters. He dated some coins to the third century B.C. merely because their script resembled Aśokan Brāhmī. But, as we have seen, no reliance can be placed on such apparent resemblance. Similarly, Rapson, while illustrating a plate of early Persian and Indian coins countermarked with Brāhmī letters, assigned them to the fourth or fifth century B.C. on the evidence of the original coins, but coins by their very nature have a longer span of life than is usually believed. When was the countermarking with Brāhmī letters started? What is the meaning of the word 'countermarking' as applied to these coins? A close examination shows that the word here does not mean 'putting a new symbol on the coins', usually assumed to be the mark of the money-changers or the mint-masters. In these cases the letters are stamped on the old figures and thus the latter are partly defaced. This practice was possible only after the original coin-figures had ceased to be recognizable.

c13.s2.t1.1.2

Even if this assumption is not accepted we have to take into account all the coins illustrated by Rapson. The plate includes six of the debased silver coins of the rectangular variety which are also countermarked in the same fashion. The date of countermarking must be later than the time when these issues were current, as this practice is limited to a few coins and not to be seen at any later time. This fact makes the date considerably later than that given by Rapson. If we look to the palaeography, if it can be relied upon at all, the letter-forms are highly suggestive. We find here two forms of *ya*, the crescentic, in which the verticals are equalized, as 𑀬, and the double-curved, in which the medial *o* is one horizontal line, as 𑀬𑁄 (*yo*); while *pa* has its short vertical drawn at an angle, *va* and *kha* still retain the round forms; but *ga* has the rounded form of the later time. These letter-forms agree with those occurring in the coins of Agathocles and Pantaleon. Hence there cannot be much difference in time between the countermarking and these Indo-Greek issues. In these cases fine workmanship is a clear contrast to the poor lettering on the Negama coins of Taxila. But the Negama coins are not a later degeneration of the Greek issues, because the Greek coins imitate not only the square shape but also the incuse-stamping of the local issues.

c13.s2.t1.1.3

This was done only to maintain the continuity of the local currency, and proves the earlier date of the Negama coins, which must be placed in the earlier half of the second century B.C. The study of these coins makes it clear how poor was the local craftsmanship of letter-cutting for coins. It was the influence of Greek writing and Greek technicians that gave a new face to Indian Brāhmī. Similar poor workmanship is seen in the coins of Kāḍa, probably found in the Punjab. These coins, which are round as well as square, are cast. In two of the pieces¹ the letter *sa* of the legend is reversed and shown as [Figure:] instead of [Figure:]. These coins may also be assigned to the later half of the second century B.C. Similar lettering is found in the earliest issues of the Yaudheyas, who continued their coinage still later, and then stopped to begin again after the Kushāṇas. Farther down in South-east Malwa we have another series of inscribed coins from Eran, the ancient Airikiṇa, spelt in the coins as ‘Erakanya’. Three coins bearing this placename have been published, one by Rapson² and two by Cunningham.³ In Rapson’s coin the letters are written one below the other, while in Cunningham’s coins they are written on the top margin from left to right. A fourth coin inscribed with the legend *dhama-pālasa* (published by Allan) has been the subject of hot discussion. This coin also shows an animal (a horse?—not described by Allan) just as the other coins.

c13.s2.t1.1.4

Here the legend is arranged on three sides of the coin, but all the letters are reversed and hence the writing appears to be from right to left. But it is clear from the coin that it was not done intentionally. Even the animal appears to be in reverse and is hence difficult to identify. A similar blunder of a reversed legend is seen in a coin of Sātakarṇi.⁴ The lettering in all these coins, which are either cast or die-struck, is of the same crude type. In no case can they be dated earlier than the issues of Kāḍa. The earliest issues of Mathura may also go back to this period, but the letters are very blurred and no definite conclusion can be reached.

c13.s2.t1.2

¹ J. Allan, *A Catalogue of the Coins of Ancient India*, London, 1936, pl. XIX, nos. 17 and 19.

c13.s2.t1.3

² *J.R.A.S.*, 1900, pl. VII, 7.

c13.s2.t1.4

³ *A.S.R.*, x, pl. XXIV, nos. 16-17.

c13.s2.t1.5

⁴ E. J. Rapson, *Catalogue of the Coins of the Andhra Dynasty, the Western Kshatrapas, the Traikūṭaka Dynasty and the ‘Bodhi’ Dynasty*, London, 1908, pl. I. 9.

c13.s2.t1.6

The majority of the coins fall into series II and III. The second series includes some of the coins of the Yaudheyas, the coins of the Arjunāyanas, the earlier issues of Kauśāmbī, some of the coins of Mathura local rulers, and the coins of Kanauj, of Tripurī, and of Upagoda and Upātikya. In these cases the angularity of the letters is well established and round forms of *ga*, *śa*, and *ta* are usual. In the palaeography of these coin legends we clearly recognize how great was the influence of the Indo-Greek coinage. Their dates must lie in the first century B.C. The third series of the coins is well defined at Mathura, where the local series is succeeded by the Śaka Kshatrapa issues. The Kshatrapa coins show two marked features: (i) equalization of the two verticals in letters such as *sa*, *sha*, *ha*, *pa*, *gha*, and *la*, and (ii) thickening of the top heads. The top flourish in medial *i* is also remarkable. These features are also noticeable in the coins of the Kuṇindas, the Pañchālas, the Vṛishṇis, and the Audumbaras, in the later issues of Kauśāmbī, and in the coins discovered at Almora and Ayodhya. These coins should be attributed to the first century A.D. Some of the still later coins of Kauśāmbī, like that of Dhanadeva, and of Ayodhya, must have been issued after the Kushāṇas.

c13.s2.t1.7

*Heliodorus inscription*

c13.s2.t1.8

Coin palaeography is rather suspicious, because firstly the legends are short, and secondly conservatism dies hard in coinage. However, its study has opened our eyes to the new technical knowledge that was gradually spreading in India. This conclusion is further supported by a very valuable pillar inscription from Besnagar (Pl. V) of the ambassador Heliodorus from the court of the Indo-Greek king, Antialcidas, ruling at Taxila, to King Bhāgabhadra in Malwa. This inscription provides an insight into the prevailing mode of writing and an inkling of the new tendency that was soon to affect this region. Remarkable forms, though optional, are the angular forms of *pa*, *va*, *sa*, *la*, and *ha*. The lower circle of *ma* faintly copies the triangular form. *Dha* is of the later regular variety. *Ra* has a new type of serpentine vertical, described by Bühler as the 'cork-screw' type, and also an optional curved form of the straight vertical. One variety of *ya* has exceptionally equalized verticals, as seen in the countermarked coins. *Bha* has a slanting vertical, and the medial vowels are attached in the older Aśokan fashion. The date of this inscription is towards the close of the second century B.C.

c13.s2.t1.9

*Barli, Ghosundi, and Bhilsa inscriptions*

c13.s2.t1.10

When we compare the Heliodorus inscription with three others (Pl. V) of this region—the Barli stone inscription from the Ajmer district, the Ghosundi inscription from the Chittorgadh district, and the Bhilsa inscription of King Bhāgavata—we notice a great difference. Not only do we find regularization of some new and old forms, but also a tendency to create the letters in a new style. Of these the Barli inscription is the crudest example, with exaggerated verticals, except a few as in *li*, while the two others show better penmanship with the lines drawn equally thick throughout, the Bhilsa one showing equalization of the verticals in letters like *pa* and *sa*. Of the common features mention must be made of *bha*, which has a straight vertical, *ka* with a lengthened vertical, and *ma*, the lower circle of which greatly varies: in Barli it is a loop appended to the upper strokes which join together in a curve; in Ghosundi it is triangular, and in Bhilsa oval. The circle of *va* in Ghosundi and Bhilsa has become almost triangular. In Ghosundi alone *ra* maintains a degenerate form of ‘cork-screw’ type known from the Heliodorus inscription. The application of the medial vowels is also instructive. The medial *u* in Ghosundi follows the old practice, though it has come nearer to the right vertical, while in Bhilsa it is marked by prolonging the right vertical of *pa* downwards. On the other hand, the medial vowel *i* is given in the old fashion. Only in Barli it optionally has the tall flourish. These features show that they are much later than the inscription of Heliodorus. They cannot be placed earlier than the second half of the first century B.C. In order of sequence we should have first Ghosundi, then Bhilsa, and finally Barli.

c13.s2.t1.11

*Mathura*

c13.s2.t1.12

Bühler, while editing these inscriptions in *Epigraphia Indica*, vols. i and ii, distinguished the inscriptions of the Śaka Kshatrapas from others which he termed ‘archaic’, obviously meaning that these are earlier than the Kshatrapa inscriptions. This classification produced a confusion regarding the date of the Kankali Tila at Mathura, from where most of these inscriptions come. It is hardly justified. The difference between the so-called ‘archaic’ and the Kshatrapa inscriptions is due to the scribes, one being a commoner and the other a royal protégé. Consequently the Kshatrapa inscriptions are ornate, and show greater skill in the handling of the new pen. The other epigraphs lack this skill. But there is no difference in the forms of the letters. This is also true in the case of the Mora inscription, though older forms persisted here. From the point of view of chronology they fall in the same period, i.e. after the coming of the Kshatrapas to Mathura.

c13.s2.t1.13

In the Kshatrapa inscriptions (Pl. VI *a*, *b*) the verticals are drawn in the new style, and even the curvature is affected by the use of the new pen. The thickness varies with the twist of the pen, e.g. *a*, *ā*, *cha*, *chha*, *sa*, *ḍa*, *ta*, *bha*, *ya*, *la*, *ha*, and *ra*. Then we find the equalization of the verticals, except in *la* which has a tall flourish. Angularity is a marked feature in letters such as *kha*, *gha*, *ja*, *ṭa*, *pa*, *pha*, the lower circle of *ma*, *ya*, the lower circle of *va*, *sha*, *sa*, and *ha*. Both *ga* and *śa* develop a beautiful curve. *Chha* has optionally an ellipse, but more usually a double loop. The lower half of *ḍa* shows curvature, and the two strokes of *ta* are also joined in a curve. *Ra* has both the serpentine form and the straight vertical; the latter has its lower end bent to the left. The medial *i* has a tall curved flourish, and the medial *o* in the case of *mo* is attached optionally to the middle of the letter, as is known from the Girnar R.E. of Aśoka, while in *po* and *gho* two upward slanting strokes are attached to the left vertical—a practice which became common in the next period.

c13.s2.t1.14

*Sanchi*

c13.s2.t1.15

Sanchi (Pl. VI *a* and *b*) has given us a great wealth of material for palaeographical study. Bühler, R. P. Chanda, and N. G. Majumdar have studied in detail the palaeography of these inscriptions. Their method is based on the old assumptions, and though Chanda and Majumdar have made many corrections on stylistic grounds, their conclusions do not stand the test of the new approach. As my purpose is not to write a memoir on the Sanchi inscriptions, this study is confined to selected inscriptions in order to throw more light on palaeography. The inscriptions up to the first century A.D. are divisible into three series. The third series includes the inscriptions from Stūpa III (except the casket inscriptions), illustrated by Marshall in pl. CXXXVII, lower half, and those on the eastern, southern, and western gateways of Stūpa I, illustrated in Marshall, pl. CXXXIV. The inscription of ‘Sātakaṇi’ is included in this series. There are others scattered among the railing inscriptions of Stūpas I and II, but it is not possible here to catalogue all of them. There is no fundamental difference between this series and the Mathura Kshatrapa inscriptions. They may be safely dated to the first half of the first century A.D.

c13.s2.t1.16

Series II includes the relic casket inscriptions of Stūpa II, illustrated by Marshall in pl. CXL, the railing inscriptions of Stūpa II, illustrated on pls. CXXXVI and CXXXVII, and those on the northern gateway of Stūpa I. The casket inscriptions are a class by themselves and show unity of hand, while the railing inscriptions are very varied. The angularity is clearly marked in most of the letters. The casket inscriptions have reduced verticals. The medial *o* is a single horizontal line, while medial *u* in letters such as *pu*, *su*, and *hu* is given by prolonging the right vertical downwards. Though these inscriptions do not show the use of the new technique, they illustrate to a slight degree the influence of the new style of writing. Hence they may be safely dated in the second half of the first century B.C.

c13.s2.t1.17

Series I includes the casket inscriptions of Stūpa III and most of the railing inscriptions of Stūpa I. Here we find some letters which resemble those in the Heliodorus inscription, but there are some features, such as the straight vertical of *bha* and the single horizontal line for medial *o*, which are so distinctive that they appear to occupy an intermediary position between the Heliodorus inscription and series II of Sanchi. Hence the inscriptions of series I may fairly be placed in the earlier half of the first century B.C.

Kangra Valley

c13.s3.1

At Kanhiara and Pathyar¹ small rock inscriptions in Brāhmī and Kharoshṭhī have been discovered. The Kanhiara Brāhmī inscription mentions the name of Kṛishṇayaśa, and the writing shows the angular feature of the Mathura Kshatrapa records. It may be dated in the first century A.D. The Pathyar inscription shows deliberate elongation of the verticals and the letters are also crudely drawn. Vogel places it in the third century B.C., but the accompanying Kharoshṭhī record indicates that the elongation is a freak of the writer. The inscription should be dated in the first century A.D.

c13.s3.2

¹ *Epigraphia Indica*, vii. 116–19.

North-West Deccan

c13.s4.1

The caves of North-west Deccan form a group by themselves. The inscriptions found in these caves have been analysed and assessed by Bühler and Burgess in the *Archaeological Survey of Western India*, vols. iv and v. Unfortunately the chart given by Burgess cannot be relied upon, as the drawings are marked rather by good draftsmanship than faithfulness to the original. The chronological scheme built up by these scholars is one of the main obstacles to the solution of many problems in the history of the Deccan. The scheme rests firstly on a comparison of the Nanaghat inscriptions with Aśokan Brāhmī, and secondly on the established contemporaneity of the Āndhra rulers with the Western Kshatrapa kings. While the second is incontestable, as it is based on the content of the inscriptions, the first is not reliable, as has been already pointed out. The second synchronism is supported by palaeography. As will be shown in the next chapter, these later inscriptions are united by a common character ultimately derivable from the script of the Kushāṇa inscriptions. Before this last script came in use, the older scripts were followed in this region. These are shown in Pls. VI and VII. The records so far found, except for the Nanaghat inscription of Nayanikā, are short epigraphs, very few in number, pertaining to the Buddhist caves, and the descriptive labels on the sculptures at Nanaghat. These inscriptions fall into two groups. In the first group we have the main long inscription of Nanaghat (here called Nanaghat I), inscription no. 1 from Bhaja (Bhaja I), inscription nos. 1 and 2 from Nasik, and inscription nos. 1 and 2 from Ajanta (all references are to the illustrations published by Burgess). In this same group falls the newly discovered wooden ceiling inscription from Bhaja.¹ In the second group we have the minor Nanaghat inscriptions nos. 3–8, Bhaja nos. 3–6, Pitalkhora nos. 1–7, the Kondane inscription, the early inscriptions from Karle, and Ajanta painting no. 6; the remaining inscriptions in these and the other caves of the region are later in date.

c13.s4.2

¹ *Indian Archaeology, a review*, 1955–6, p. 29, pl. XLIVA.

c13.s4.3

As the Nanaghat I is the longest inscription, I shall put its analysis first. In general no definite rule has been observed for the length of the verticals. In some places they are equalized, and in others they vary in length. The use of the new tool is nowhere traceable, but some of the forms are very suspicious as they recall those of the new tradition. Angular forms of *pa*, *pha*, *ma*, *va*, *sa*, and *ha* are seen here and there. *Dha* is of the later regular type. *Bha* shows a straight vertical, and a new type in which the upper vertical is placed on the right side of the lower appendage. This will be hereafter called Deccani *bha*. *La* is drawn cursively in almost all the inscriptions of this time in the Deccan. One type of *ma* is a direct descendant of that occurring in the Girnar R.E. of Aśoka, and in the same fashion the medial *ā*, *e* and *o* are attached to th middle of this letter, though optionally we also find them at the top right or left. In other cases the medial *o* is one horizontal line, while medial *u* in *pu*, *yu* and *su* is a prolongation downwards of the right vertical. The vertical of *ka* is lengthened; *kha* has three forms, with or without dot and with a circle; *ga* has both angular and rounded forms; *cha* is of three varieties, the semi-circular, the triangular, and the quadrangular; *chha* has an oval in place of the circle; *ta* has, besides the rounded appendage, a straight or curved stroke attached to the right of a long vertical. This second variety is the usual form in the Deccan and the south. *Da* has a rounded form. The letters in the Nasik inscriptions entirely agree with these. But Bhaja I is very crudely written. Its rounded *ḍa* and triangular *va* suggest a later date. The newly discovered Bhaja inscription should not be dated to the second century B.C., as is done by Dr. B. Ch. Chhabra. The angularity of the letters is a marked feature, and the letters *bha* and *ma* especially point to a later date.

c13.s4.4

On the other hand, when we compare Nanaghat I with Nanaghat II, the similarity of the letters is so close that it is difficult to distinguish them. But if we take the six short inscriptions of Nanaghat II together, as we are compelled to do by the contents and the circumstances, we find some altogether new forms. The letters *cha*, *ña*, *ta*, *ya*, and the angular forms of *va* and *ha* could not have been produced without knowledge of the tradition of the Kshatrapa writing of Mathura. Though they do not show evidence of the use of the new pen, they must nevertheless be taken as imitative forms. Thus Nanaghat II is a mixed complex of groups I and II. In the inscriptions of group II the influence of the new style is quite obvious. It seems that the inscriptions of Bhaja II were written with the new pen. But in Ajanta we again notice a difference between nos. 1 and 2. However, there is no reason to divide them chronologically. Far different is the brush technique followed in painting no. 6, which also must be placed in the same period. These features strongly suggest that the inscriptions of group II should be dated to the first half of the first century A.D. What should be the date of group I? Here the main obstacle is the evidence from Nanaghat II, but this very evidence is also suggestive. Here we have what in archaeological terminology should be called an overlap of the older forms with the newer ones.

c13.s4.5

The study of coin epigraphy is further instructive. The earliest Andhra coins have been found in the Malwa region and are described as of Malwa fabric.¹ They bear the words *Rañō Siri* *Sātakaṇisa* in characters which are hardly distinguishable from those of Nanaghat. But they (see especially Pl. I. 2) clearly show the head-formation recalling the new style of writing of the Mathura Kshatrapas. The coins, even of the Kolhapur fabric, from the time of Vāsishṭhīputra Viḷivāyakura onwards, all show this top head-formation, the flourish in the medial *i*, and angularity in the letters. This style of writing is due to influence from the Malwa region, where the latest Sanchi inscriptions are all in similar characters. Hence these coins date from the first half of the first century A.D. onwards. This new type of lettering continued in the coinage right down to the time of Vāsishṭhīputra Puḷumāvi and even later. But the epigraphy of the stone inscriptions reveals a different story. Here the influence is not so sudden. There is a gradual change from the Nanaghat inscriptions to those of Puḷumāvi. On account of the backwardness of the writing in the Nanaghat inscriptions the coins of Rājā Siri Sātakaṇi have been dated in the second century B.C., and an unnecessary gap has been assumed between his coins and those which follow. It is only for this reason that the Sātakarṇi of the Sanchi inscription¹ has not been properly identified. But this palaeographical anomaly must be understood on the basis of regional differences, and the apparent gap is filled up by the occurrence of occasional letters. If this is borne in mind, there does not seem to be any justification in differentiating between the Sātakarṇi of the Sanchi inscription, Rājā Siri Sātakaṇi of the coins of Malwa fabric, and Sātakarṇi of the Nanaghat inscriptions. The numismatic evidence completely closes up the gap between these inscriptions. Palaeography thus suggests that there is no chronological difference between group I and group II inscriptions of this region. They are to be placed in the first half of the first century A.D.

c13.s4.6

¹ Rapson, op. cit., pp. xcii–xciv.

c13.s4.7

¹ See above, p. 64.

South India

c13.s5.1

South Indian palaeography received early attention with the publication of a monograph by Burnell in 1874. But the paucity of materials from the early period left many problems unsolved. The discovery of the inscriptions from the Bhattiprolu stūpa in the Kistna district and their masterly handling by Bühler raised for the first time the question of the southern alphabet and the script. Bühler had already maintained the existence of a Drāviḍī script, including in it the character of the Peninsular inscriptions of Aśoka. Bhattiprolu gave him a new idea, that South India had a script of its own derived independently of the Aśokan. This statement was not borne out by the discovery of the 'early' inscriptions from Amaravati. The palaeographers did not pay heed to this discrepancy between the Bhattiprolu and the Amaravati scripts, probably because the Amaravati inscriptions are all in Prakrit and hence need not include characters suitable to the Dravidian languages. But a large number of cave inscriptions discovered in the districts of Madura and Tinnevelli and in the former state of Puddukottai, which are all, as has been shown by Subrahmanya Ayyar, in Tamil, once again raised the question of the Drāviḍī script. Ayyar has pointed out the differences that are found between the Bhattiprolu and the cave inscriptions, but even now C. Sivaramamurti has not touched the important question of the origin of this script in his recent memoir on *Indian Epigraphy and South Indian Scripts*.

c13.s5.2

In the previous chapter we did not accept the division of Aśokan Brāhmī into northern and southern characters as propounded by Bühler. Aśokan Brāhmī is mainly imperial and it is hard to distinguish regional varieties. The evolution of the South Indian scripts can be traced only from the early inscriptions of Amaravati and Bhattiprolu, and from the cave records. There is one common denominator to all of them—they are Buddhistic in content. Hence the introduction of writing into the extreme south, as far as we know, must be ascribed to the activities of the Buddhist monks. It is hardly likely that these monks were isolated from the Buddhist order of North India. Another link is provided by the Prakrit used in most of these inscriptions. It is no wonder that there is a general similarity at this time between the scripts of South India and those of the north. The main differences arose only when an attempt was made to represent the local sounds in writing. Even these sounds, as we shall see, do not bring in altogether new shapes. Almost all the new letters are derivable from those of the existing script of the north. Hence there is no necessity for propounding an original Drāviḍī script different from Aśokan Brāhmī and tracing it to some ultimate source in Western Asia.

Bhattiprolu Inscriptions

c13.s5.t1.1

The most puzzling of these scripts is that used in the Bhattiprolu inscriptions (Pl. VII *a*, *b*). Even here we find a difference between the inscription on the crystal and those on the three caskets. They agree in differing from Aśokan Brāhmī in having the curve of *da* open to the right, exactly the reverse of *da* in Aśokan Brāhmī, and in the northern characters of this period; otherwise the crystal script shows all regular forms. In the caskets we have a few new forms: *gha*, *ja* (optional), *bha*, *ma*, *la* and *sha*. One new letter, *ḷa*, is added to spell South Indian names. All other letters are of the regular type except that they show regional variation. In the medial vowels the casket inscriptions have a short bar to express the short vowel *a* compounded with a consonant when it is not followed by *anusvāra*, and for long *ā* a farther downward slant from this bar. This is peculiar to the Bhattiprolu caskets and is not found elsewhere in South India, not even in the crystal inscription. As the casket inscriptions have a style of their own, it seems likely that they were written by one person, and the fact that a Dravidian sound *ḷa* has been introduced suggests that the writer was a local Buddhist. If his peculiar fondness for expressing short *a* by a bar is set aside, it is not difficult to derive the new forms from those known in Northern Brāhmī. Most of them appear to be the reverse of the regular forms. Fig. 7. I–IV shows how the Bhattiprolu forms of *bha*, *ma*, *la* and *sha* are derived from the regular forms. In *ja* we find the middle bar omitted when short *a* is not to be shown, and it appears only with two horizontals attached to either end of a vertical. The only letter which is strangely derived is *gha*, where the second loop is added to the long vertical at the top (Fig. 7. V). The new letter *ḷa* can be derived from *la* (Fig. 7. VI). These changes must be attributed to the man who wrote these inscriptions, as they are found nowhere else. I would regard them as mistakes on his part rather than accept them as a new script.

First 61 of 102 passages. Read on

Linked texts (23)