Pramāṇa · Search · About · API

Indian Palaeography c17: 9. South-East Asia

Indian Palaeography c17

Language
English

Contents

9. South-East Asia

c17.2

*Scope*

c17.3

As early as 1877 K. F. Holle produced his *Tabel van Oud- en Nieuw-Indische Alphabetten* illustrating the different forms of writing, past and present, known from South-east Asia. His object was mainly to acquaint his readers with the scripts, but he had realized the significance of studying the writings of this vast area in relation to Indian palaeography. For this purpose he used not only Prinsep's tables¹ but also A. C. Burnell's *Elements of South Indian Palaeography*. However, not much advance could be made at this time because of the paucity of accurate reproductions of the inscriptions. Naturally the letter-forms given by him appear to be much affected by retouching. This point was brought out by H. Kern² in his review of the book. The real advance in the comparative study of these scripts was made with the publication of Bühler's *Indische Palaeographie*. Barth, Bergaigne, and Coedès discussed the problem in their editions of the inscriptions from Indo-China and Siam, Vogel contributed on the early Sanskrit inscriptions from Java and Borneo, and Blagden commented on the inscriptions from Burma. The late E. H. Johnston edited some Sanskrit inscriptions from Arakan. Early inscriptions from Malaya are published in the *Malayan Journal of the Royal Asiatic Society*,³ and those of Sumatra are in volume xxx of *B.E.F.E.O.* Some inscriptions have been re-examined by Dr. B. Ch. Chhabra in his thesis, ‘Expansion of the Indo-Aryan Culture’.⁴ All these materials are utilized here, and in this chapter an attempt is made to trace the origin and development of the different scripts of South-east Asia to the end of the seventh century A.D., stopping at a point when the proto-regional scripts had almost taken the field in the different areas. Accordingly I shall include all the early Sanskrit records as well as the earliest Cham, Mon, and Pyu inscriptions.

c17.4

¹ See above, p. 2.

c17.5

² *Eene Bijdrage Tot de Palaeographie van Nederlandsch Indie.*

c17.6

³ Vol. xviii. 1940.

c17.7

⁴ *J.A.S.B.*, 1935, pp. 1-64.

Approach

c17.s1.1

As the writings of South-east Asia originated from those of India, my approach to the problem will be twofold. Firstly, I shall attempt to trace contacts with different parts of India on the basis of palaeography, and secondly a study will be made of the inscriptions in order to find the diffusion of the different types of writing in South-east Asia and how they lead on to the development of the regional scripts.

Contacts with India

c17.s2.1

(i) The countries of South-east Asia lie on the route of eastern commerce and trade. It is quite natural therefore that the earliest contacts should have been commercial. Of such contact the oldest

c17.s2.2

[Figure: OC-ÈO SEALS FIG. 18]

c17.s2.3.1

evidence comes from Oc-èo in Southern Cambodia, where a number of inscribed stone seals have been found. One such seal is also known from Malaya.¹ These seals were apparently not manufactured locally, but were probably imported from India, as was also the case with other types of seals found in this place. From what part of India could these seals come? The only evidence to decide this question is palaeographical. In Fig. 18 the letter-forms are divided into four groups. (A) This has only two letters, reading *deva*. Both letters have line head-marks. *Da* shows the double-curved form, but the lower curve is rather squat. *Va* is triangular. The medial *e* is a top slant. These features were known only in North India and are also to be seen in the later inscriptions of the Western Kshatrapas of Gujarat and Malwa. In terms of the palaeography of this latter region a date in the third to fourth century A.D. will not be far amiss. But on the basis of only two letters it is wrong to be dogmatic. (B) This group of inscriptions is characterized by notched head-marks. Only two medial vowels are used: the long *ā* has a downward tick attached to the end of its horizontal stroke, and the *anusvāra* is marked by a line.

c17.s2.3.2

The characteristic letters are the broad type of *bha* with a notch in its horizontal bar, *ta* with its two lower arms rounded at the upper end, the round-backed *da*, the triangular *va*, the tripartite *ya* with a loop at the left vertical, and *sha* with a bend at its left side. These features leave hardly any doubt that they belong to a region where the South Indian forms were used along with North Indian letters. This again points to Gujarat. They may be dated in the fifth century A.D. (C) This group has the largest number of seals. The letters are characterized by solid square head-marks. The medial vowels are all northern—the medial *ā* and *o* being top slants, except in *tā* where the medial *ā* is a horizontal stroke. The medial *i* is a left curve, and the medial *ī* has an additional initial prominence. The letter-forms tend to angularity, and they are easily comparable with the first group of the Vākāṭaka inscriptions (Pl. XV. 4). The forms are almost identical except in the case of *a*, which suggests a later date, and *ta* which points to the northern region of the Vākāṭaka dominions. They may be assigned to the fifth century A.D.¹ (D) To the last group belongs a seal reading *Vishṇumitrasya*. Another seal reading *Śrīmagaya* has been doubtfully assigned to this group. These letters have solid triangular head-marks, characteristic of the northern scripts from the fourth century A.D. onwards.

c17.s2.3.3

The forms are rather ornamental, with notches in the lower sides of *va*, *ma* and *sha*. The subscript *ra* is of the southern rounded form. *Ta* is northern, while *ma* has a looped base. The medial *i* is an upward stroke ending in a curl, as seen in the Western Indian inscriptions of the fourth to fifth century A.D. The nearest comparable material comes from Malwa (Pl. IX. 7). They may be dated in the fifth century A.D. The Malayan seal belongs to this group. Thus all these seals point to the region of Gujarat and Malwa between the fourth and fifth centuries A.D. It is natural therefore to suppose that the first contact was established with Western Indian seaports. As the detailed analysis given below will show, it is from this region that the earliest writing style of South-east Asia was derived. It is found again in Indo-China. Two types of writing are known—one represented by the Vo-Canh inscription, and the other, the square-headed type, seen in the inscriptions of Bhadravarman. These styles are quite in keeping with the two styles in use in Malwa between the fourth and fifth centuries A.D. A crude local copy of the first type of writing is also seen on a stone inscribed with the Buddhist creed from site no. 1 at Kedah in Malaya, but that inscription must be dated to the end of the fifth century A.D., as it shows the later form of *tha*, and the open-mouthed type of *ṇa*.

c17.s2.4

¹ Chhabra, op. cit., *J.A.S.B.*, 1935, pl. 5.

c17.s2.5

¹ Professor V. V. Mirashi has published similar seal-stamps from the Central Provinces (now Madhya Pradesh) in *J.N.S.I.*, vol. iii, pt. ii, pp. 100–2, two of which read *apramāda* and *jitaṁ bhagavatā* respectively. They are exactly in the same character, and Professor Mirashi attributes them to the fourth century A.D.

c17.s2.6

(ii) From the close of the fifth or the beginning of the sixth century A.D. onwards writing spread to most of the regions of South-east Asia. The style which became current at this time bears the closest resemblance to the writings known from the records of the Madharas, Vishṇukuṇḍins, and Pallavas. The earliest examples known are from the *yūpa* inscriptions of Mūlavarman from Borneo, which, though showing ornamental developments and new types of medial vowels, are related to the style of writing known from the Bhadravarman's inscriptions in Indo-China. This relation is established by the simplicity of the long verticals, which generally do not have curved ends. On the other hand, the new developments link the script with the ornamental style of writing seen in almost all the inscriptions of the sixth century A.D. The best-known examples in this style are the inscriptions of Pūrṇavarman from Java, the records of the kings from Fou-nan, the Srideb inscription from Siam, and the inscription of the *Mahānāvika* Buddhagupta from Malaya. These inscriptions borrow not only the Pallava medial vowels but also some of the forms of their letters. It seems that contact with the south-eastern coast of India was greatest at this time. But it is wrong to think that the western coast was totally forgotten. The influence from that coast is traceable in a few letters of this time, and much more so in those of the seventh century A.D. In fact the dated inscriptions of this last century from Fou-nan and Sumatra exemplify a happy mixture of the sixth-century writing of the eastern coast of India and the new features of the seventh century from the western coast. It is strange to note that the Pallava Grantha writing of the seventh century is not traceable in these inscriptions at all. On the other hand, we find a local development towards simplification, first recognizable in the forms of *ka*, *ra* and *la*, which later characterized the growth of the local scripts in South-east Asia. These dated inscriptions are very helpful in relating the undated sixth-century inscriptions with them, and thus indirectly they confirm my dating of the Pallava inscriptions on the basis of the Indian evidence.¹

c17.s2.7

¹ See above, p. 200.

c17.s2.8

(iii) Burma appears to have been in contact by the overland route with Eastern India and by the sea-route with the eastern coast of South India. The Eastern Indian influence is seen in the inscriptions found in Arakan and as far down as Hmawza near Prome. These inscriptions are datable from the seventh century A.D. onwards. The South Indian influence is traceable in the Maunggun gold plates and the Kawgun cave inscription, both of which belong to the sixth century A.D. We may be able to detect another contact if we can decide about the origin of the Pyu writing. The medial vowels used in this writing are exclusively northern, though the medial *i* shows a variation in the local style. Letters such as *u*, *kha*, *ta*, *ya* and *la* suggest influence from Western India, possibly Chālukyan. But the letters *a*, *tha*, and the tripartite form of *ya* point to an earlier date. On the whole it seems that the original writing from which these characters are derived is connected with the source of the Vo-Canh inscription, but in course of time this writing became stylized in this region. The examples that we actually have are seventh-century versions of the original writing.

Indo-China

c17.s3.1

Apart from the Oc-èo seals already dealt with, the early inscriptions from Indo-China fall into two geographical zones: (A) Lin-i, and (B) Fou-nan. Lin-i has produced important early inscriptions, the exact dating of which has long been debated. These inscriptions can be divided into four groups.

c17.s3.2

A. (i) In the first group falls the badly preserved Vo-Canh inscription (Pl. XX. 1) which refers to a king called Śrīmāra. To this inscription have been assigned various dates ranging from the second to the fourth century a.d. The advocates of the earlier date point to the tripartite form of the subscript *ya*. But this form was preserved in Gujarat as late as the middle of the fifth century a.d., as is attested in the Junagadh rock inscription of Skanda Gupta (Pl. XIV*b*. 1). D. C. Sircar¹ argues for a later date, as the record is in Sanskrit. But this argument is hardly valid when it is realized that Sanskrit was generally used in the records of the Western Kshatrapas from the middle of the second century a.d. onwards. We will first analyse the letters before assigning a date.

c17.s3.3

¹ *Select Inscriptions*, Calcutta, 1942, p. 471, n. 2.

c17.s3.4

The letters have distinct head-marks, which are generally notched, but sometimes angular. The lower verticals of *a*, *ka*, *ra*, and the vertical type of the medial *u* do not end in curves. The only initial vowel used is *ā*, which has a hook appended below its sloping upper left side. The medial vowels are very instructive. The medial *ā* is marked both by a top slant as in *yā* and by a horizontal stroke as in *gā*. The medial *i* is a stroke curved to the left, while the medial *ī* has an additional initial prominence. The medial *u* in *ru* is a stroke sloping downwards, but in *stu* it is an upward curve going outwards. The medial *ṛi* is a left curve of the southern type. The medial *o* in *ko* has one of its strokes in the form of a top slant, but in *lo* it has the cursive form of the southern type. This mixture of northern and southern medial vowels points to the region of Gujarat and Malwa, where they were in actual use in the fourth century a.d. *Ka* has its central bar curved. *Ga* has equal arms with rounded and flat-topped varieties. *Ṅa* in the conjunct *ṅga* has its vertical bent. *Cha* narrows the loop on the left side. *Ja* is of the three-armed type, with its vertical slightly curved in the middle. *Ña* in the conjunct *ñcha* maintains its proper form with a hook on the right. *Ṭa* in the conjunct *shṭa* has a round back. *Ṭha* in the conjunct *shṭhā* makes a rough oval. Both *ṇa* and *na* have bent bases. *Ta* has its lower two arms meeting in an upper curve. *Tha* in the conjunct *sthā* has a dot within a circle. *Da* is round backed. The left sides of *pa*, *ba* and *ha* are bent. *Bha* is of the broad type. *Ma* has a triangular looped base with two slanting uprights. The tripartite form of *ya* has its third left arm replaced by a loop, but in the conjuncts the three uprights of this form are well preserved. *Va* has a triangular loop below a stem, and *sa* a hook on the left.

c17.s3.5

On the basis of these mixed features it is difficult to connect the script with any particular style of writing in India. The absence of curves at the lower ends of the verticals would point to North India, but this feature was also known in Malwa (Pl. XIII. 1, 2; Pl. IX. 6, 7). The equal arms of *ga*, the broad type of *bha*, the looped base of *ma*, the curved bases of *ṇa* and *na*, the left curved stroke of the medial *ṛi*, and the cursive form of the medial *o* in *lo* would connect this with the southern scripts, but all these features were also known in Central India in the fourth to fifth century A.D. The form of *ya* used here is known only from Malwa at this time. When we know that the writings of Malwa and Gujarat were connected in the fourth century A.D., because of the rule of the Western Kshatrapas, it is reasonable to suggest that the style of writing originating from this region affected the Vo-Canh inscription. The cursive form of *lo* suggests that the dating must be between the latter half of the fourth and the first half of the fifth century A.D. The use of the Sanskrit language in this inscription is well explained by contact with Gujarat.

c17.s3.6

(ii) This group includes three inscriptions—the Myson inscription of Bhadravarman (Pl. XX. 2), the Hon-Cuc inscription (Pl. XX. 3), and the Cho-Dinh inscription of Dharmamahārāja Bhadravarman (Pl. XX. 4). All these inscriptions are characterized by the square or rectangular head-marks added to the top left of the verticals. This square mark is also seen over *ga*, *tha*, *dha* and *śa*, but not over *kha*. In style there is a close connexion between these inscriptions and that from Vo-Canh, the main difference being in the head-marks. The square head-marks of these inscriptions have also given angularity to the forms of the letters. As in the Vo-Canh inscription, the lower ends of the verticals do not curve, but show a sharp cut-end, especially in the Myson inscription. The medial vowels are of the same type. We may also note the additional medial *au*, which is of the curved variety of the south. These features link this square headed type with the Deccani style of Malwa (Pl. XV. 3). Other links with Gujarat are seen in the forms of *i*, which is of the roofed type; *u*, which has its lower horizontal curved upwards; *kha*, with a broad looped base; the broad type of *cha*; *ja* having its upper curve smaller than the lower; the looped varieties of *na* and *na*; and *śa* having a short mid-stroke attached to its right arm. The invariable use of the hooked variety of the subscript *ya* suggests that the date of these inscriptions should be placed slightly later than that from Vo-Canh. The middle fifth century A.D. would be the earliest possible date for them. In the Cho-Dinh inscription we get the looped variety of *ta*, which is first seen in the late Śātavāhana records in India. Henceforward the looped varieties of *ta*, *na* and *ṇa* become common in South-east Asia. This northern link of these inscriptions is further confirmed by the title *Dharma-mahārāja* adopted by Bhadravarman—a title obviously borrowed from the Vākāṭakas.¹

c17.s3.7

¹ D. C. Sircar, *Select Inscriptions*, Calcutta, 1942, p. 407.

c17.s3.8

(iii) In this group is included the earliest Cham inscription from Dong-Yen-Chau (Pl. XX. 5). It is connected with the second group by its square head-marks and the use of similar medial vowels, except the medial *o*, which here has horizontal strokes, but the medial *au* in *tau* is of the northern variety of three distinct strokes. The letters also are of the same forms with one fundamental difference, that here the lower ends of the verticals are curved. As this style was known in Gujarat and Malwa, there is no reason to look for another source for it. The inscription may be dated to the latter half of the fifth century A.D.

c17.s3.9

(iv) In this group falls the Myson inscription of Śambhuvarman (Pl. XX. 6). Its style is much later and is connected with the early inscriptions from Fou-nan. It will be discussed along with them. B. The early inscriptions from Fou-nan can be divided into three main groups:

c17.s3.10

(i) In this group are included three series of inscriptions: Pl. XX. 7 is taken from Thap Muoi inscription of Guṇavarman; Pl. XX. 8 is derived from Ta Prohm inscription of Rudravarman, with which also agrees the Neak Ta Dambhang Dek inscription of Jayavarman² and the Myson inscription of Śambhuvarman; and Pl. XX. 9 is taken from the Tham Prasat inscription of Mahendravarman and the Phnom Banteai Neang inscription of Bhavavarman. No. 9 represents the latest in the series. There is a marked difference in the style of writing noticed in these inscriptions from that of group A. In these inscriptions only line head-marks are seen, and the influence of stylus writing is noticeable in the frequency of the notches or waves in the lines. The medial vowels are entirely southern. It is further possible to localize the origin of these medial vowels to one of the proto-regional scripts of South India. The most distinctive are the circle-type of the medial *i* and *ī* and the broad curved type of the medial *u*. These are invariably found in the Pallava records beginning from the sixth century A.D. onwards (Pl. XVIII. 5, 6). In the Kadamba records they are used optionally, while the Gorantla plate of Attivarman (Pl. XVII. 11) shows them as an influence from the Pallavas. Other medial vowels show much local development. The medial *ā* has an exceptionally long vertical, as can be seen in *sā* and *hā*, the medial *e* drops down and ends in a curl, while the medial *ai* has one stroke on the top and another below the letter, as in *lai*. No. 9 has a new type of medial *o* in *no*. It has one top stroke on the right and a lower sloping one on the left. The broad type of the medial *u* is applied not only to *bhu* and *gu* as in the Pallava records, but also to *tu*. No. 9 also shows a new type of the medial *ū* in *bhū*, which has a double curved stroke attached to its right vertical. Of the initial vowels we get only *a* and *e*. *E* has a stem over its foot-form. *A* is of two varieties: one with a small curve at the lower end of the right vertical, as in no. 7, and the other with an extended curve, as in no 8. This difference is also maintained in the case of *la*, no. 7 having a small upward curve and no. 8 showing an extended curve. This extension of the curve in no. 8 suggests the continuation of the influence from the western coast. This influence is also to be seen in the recurrence of the looped varieties of *ṇa*, *na* and *ta*, the broad looped base of *kha*, and the narrow upper ends of *tha* and *dha*. However, it should be noted that the looped varieties of *ṇa*, *na* and *ta* were fairly well known on the eastern coast as well. The study of the consonants shows a gradual development of the letters from 7 to 9. The lower curve of no. 9 is extended farther, while *ra* in nos. 8 and 9 has doubled its vertical. *Ga* in no. 9 is of the horseshoe type. All three arms of *ja* are waved. *Ṇa* in no. 9 has double loops as in the Pallava records (Pl. XVIIIa, 9, 11). Another remarkable development in the looped *ṇa* and *na* is seen in their right-hand hooks, which are now turned up. The cursive form of the medial *o* in *lo* is extended below the letter. Other distinguishing forms are those of *śa* and *sha*, which have short mid-strokes attached to the right verticals and are definitely derived from the Pallava records. The same is true in the case of the subscript *sha* in *ksha*. The hooked type of the subscript *ya* has an ornamental bulge on the left. On the whole the writing shows highly ornamental forms ultimately traceable to the style known from the Pallava records of the sixth century A.D.

c17.s3.11

² *Journal of the Greater India Society*, iv, no. 2, pp. 117–21.

c17.s3.12

(ii) This is an intermediate group represented by the Han Chey and Phnom Prah Vihar inscriptions of Bhavavarman (Pl. XX. 10). It is distinguished by three features. The ornamental forms of the earlier group continue here side by side with a tendency towards simplification, seen optionally adopted in the forms of *ka* and *la*, both of which discard the superfluous curve. New forms are seen in *a*, which develops a left turn at the lower hook—a practice which recalls the fifth- to sixth-century forms from Malwa (Pl. XIII. 2–7). The form of *i* with a tail below a double curved roof also comes closer to the Malwa type. Another northern type is seen in the open-mouthed form of *ṇa*. The northern forms are also seen in the medial *i* and *ī*, which have extended curves on the left and right respectively. The medial *ū* in *gū* and *bhū* shows the new double curved form, though the Pallava form of the broad curved type is seen in *tu*. The medial *ṛi* in *kṛi* makes a broad curve to the right. The origin of these northern features is traceable in the records of the Gurjaras of Broach (Pl. XVI. 9, 10). The date of these inscriptions lies midway between group (i) above and group (iii) below, the latter being dated in the Śaka era corresponding to A.D. 624–67. Hence (ii) may be dated towards the close of the sixth century A.D.

c17.s3.13

(iii) This is the last group of inscriptions dealt with here. Pl. XX. 11 is taken from the Bayang inscription, dated A.D. 604 and 624, and the Vat Chakret inscription of Īśānavarman. Of these the latter shows the simplified version of the writing. Pl. XX. 12 is taken from the Ang Pou inscription of the time of Īśānavarman and the Ang Chumnik inscription, dated A.D. 628. Pl. XX. 13 is derived from the Ang Chumnik inscription of Jayavarman, dated A.D. 667. In this group the two new tendencies seen in (ii) are more clearly brought out. The northern forms of the letters all continue here. The simplification is further marked in the forms of *ka* and *la*, where all the verticals are reduced to equal height—a tendency that may also be seen in the doubling of the vertical of *ra*. A new feature seen in nos. 12 and 13 is the development of an optional loop at the left corners of the open-mouthed *ṇa*, *la* and *kha*. The extended curves are also sometimes seen, as in *a*, the medial *ā* in *jā* and *ṇā*, and the cursive *o* in *lo*. The right hook of *ña* in *jña* is extended downwards. The medial *ū* in *bhū* and *gū* is generally of the double-curved type, though the broad curve of the Pallava style is also occasionally seen. The medial *ā* in *yā* is attached to the third right vertical instead of the middle one. The medial *o* now becomes standardized with one top stroke on the right and another stroke turned downwards on the left. The curve of the medial *au* is extended downwards. On the whole the tendency towards linear simplification becomes more and more pronounced in this series of inscriptions. This tendency affects the development of the later regional scripts.

Borneo

c17.s4.1

Several *yūpa* inscriptions have been found in Koetei in East Borneo, four of which have been edited by Vogel. They refer to a king called Mūlavarman. Pl. XXI. 1 is taken from the illustration accompanying the article by Vogel. As we have seen, the style of writing represented here bears some connexion with that of the Bhadravarman’s inscriptions from Lin-i. The resemblance can be marked in the absence of the curves at the lower verticals of *ka*, *ra*, and the vertical type of the medial *u*. Only very rarely is the curve seen in the case of *ka* and *ra*. But there is another development. At the ends of these verticals a left slant is added. The medial *i* and *ī* are of the same type, though here they are occasionally extended downwards, the most exceptional being in *vi*. One more link is seen in the use of the looped forms of *ṇa*, *na* and *ta*, and the broad looped base of *kha* seen in the conjunct *khyā*. A still closer link is established by the rectangular head-marks seen in these letters, but these head-marks are apparently derived from the south rather than from Malwa. Here the head-marks are not added to the top left of the verticals, but they are centrally placed over them, as is seen in the inscriptions of the Madharas and the Gaṅgas of the eastern coast of India. The link with the south is further established by the medials *ā*, *e*, *ai*, *o* and *au*. The medial *ā* (except in *shā*, where it is a top slant) and the medial *o* have their strokes extended downwards. The medial *e* and *ai* also fall down, and end in a curl. These medial vowels agree with those given in the Pallava records of the fifth to sixth century A.D. (Pl. XVIII. 4, 5). The connexion with Pallava writing is further strengthened by the forms of *ga*, the left limb of which is curved inwards, *tha* which has a mid-line in place of the central dot, and *śa* and *sha*, both of which have a short mid-line attached to the right vertical. The Pallava form is also seen in the conjunct *ksha*, in which the subscript *sha* is broadened. The forms of the other letters agree with the southern type. Another important clue is the ornamental form of the subscript *ya* in *khyā*, which is of the hooked type. This ornamental form, with the extended forms of the medial vowels, provides a link with the inscriptions of Pūrṇavarman from Java, and further suggests that there should not be a long gap between them. The inscriptions of Mūlavarman may be dated towards the close of the fifth century A.D.

Java

c17.s5.1

The inscriptions from Java fall into two geographical zones: (A) Western Java, and (B) Central Java.

c17.s5.2

A. All these inscriptions refer to King Śrī Pūrṇavarman. They fall into two groups. The first (i) includes Pl. XXI. 2, 3, and the second (ii) Pl. XXI. 4. No. 2 is based on the Ci-Aruton inscription, and no. 3 on the Jambu and Kebon Kopi inscriptions, while no. 4 is derived from the Tugu inscription. The main difference between the first and second groups is in their head-marks, the first having those of the rectangular or thick type and the second of the notched variety. There is no difference at all in the forms of the letters. The first group establishes a link with the inscriptions of Mūlavarman. Besides the head-marks, we find a similar use of the broad-looped base in *kha* and the looped varieties of *ṇa*, *na* and *ta*. But there are other fundamental differences. The most easily noticeable is the frequency of waves in the lines. The next is recognized in the new forms of the medial vowels, derived from the Pallava records, identical with those used in B (i) group of the inscriptions from Fou-nan. Another feature is the emphasis laid on the extended curves at the lower verticals of *a*, *ka*, *ra*, *ña*, and the vertical type of the medial *u*. The letter-forms are also similar to those of the inscriptions from Fou-nan. But minoı differences may be noted: here the subscript *ya* is of the ornamental type as used in the inscriptions of Mūlavarman, but no ornamental curve is seen in the medial *o* attached to *lo*; the double curved type of the medial *ū* as applied to *bhū* in the later inscriptions in Fou-nan is not found here; similarly, the later type of the medial *o*, having a top stroke on the right and a sloping one on the left, is also absent. In other words the new tendencies seen in the Tham Prasat inscriptions of Mahendravarman (Pl. XX. 9) are not found here. The Chālukyan type of *a* with an extended curve of its lower right vertical is also not met here. The closest resemblance is to the inscription of Guṇavarman from Fou-nan. These features suggest a date in the first half of the sixth century A.D.

c17.s5.3

B. The inscriptions (Pl. XXI. 5, 6) discovered in this zone do not go back to a very early date. Those found are in the simplified version of the style seen in Western Java and in Indo-China.

Malaya

c17.s6.1

Apart from the single seal already described, we have three groups of early inscriptions from Malaya.

c17.s6.2

(i) Pl. XXI. 8 is taken from a stone inscription recording only the Buddhist creed. It was found in Kedah site no. 1. Here the letters are very crudely drawn. The initial *e* appears to be of the foot type. *Ga* has its right limb slightly longer than the left. *Cha* has a broad beak. *Ṇa* is of the open-mouthed type of the north, and *ta* also is of the northern form with its two lower arms meeting in an upper curve. In *tha* the central dot is replaced by a line drawn with the same sweep of the hand as the circle. *Da* is of the double curved type but the lower curve is squattish. *Na* has a bent base, and *ma* a triangular looped base. *Ya* is of the tripartite form, and *ra* is a straight vertical. The left side of *śa* is bent. The medial vowels recall those of the Vo-Canh inscription. The writing is probably derived from Gujarat and Malwa. The use of the open-mouthed *ṇa*, *tha* with a mid-line, and the hooked type of the subscript *ya* suggests a date in the latter half of the fifth century A.D.

c17.s6.3

(ii) Pl. XI. 10 is taken from the inscription of the *mahānāvika* (great navigator) Buddhagupta. The letters have notched head-marks, and their forms agree with those of the inscriptions of Pūrṇavarman from Java. The medial vowels are also of the same type except that the ornamental extensions are not found here. The inscription may be dated in the early sixth century A.D.

c17.s6.4

There is another inscription from Kedah,¹ found in the ruins of a brick building near Bukit Mariam. Its first part is a Buddhist creed. No photograph is now available as the inscription is lost, but if the drawing is representative of the writing it should be dated in the sixth century A.D. We have here looped forms of *ṇa*, *na*, and *ta*, but *ṇa* also shows double loops, as in the Tham Prasat inscription of Mahendravarman (Pl. XX*a*. 9). The medial *i* and *ī* are of the circle type. *E* is of the foot type with its mouth opening upwards. The medial *ā* and *e* are extended downwards. Notches are seen in the sides of *pa*, *ma*, and *va*.

c17.s6.5

¹ *J.A.S.B.*, 1849, p. 247, pl. X.

c17.s6.6

(iii) Pl. XXI. 9 is taken from the tablet inscriptions found in Kedah site no. 2. The head-marks are not properly given. Wherever found, they are of the notched type. The forms of the letters show a simplified version of the seventh-century style already known from the region of Fou-nan. Important to note are the short verticals of the initial *a*, *ka*, *ra*, and *ña* in the conjunct *jña*. Western Indian forms are recognizable in the initial *i*, which has a double roof with a tail below, the open-mouthed type of *ṇa*, *dha* with its upper end narrowed, and *la* with its upper curl extended as in the inscriptions of the Maitrakas of Valabhī (Pl. XIV*b*). Another Western Indian form is the initial *ā*, which has a long vertical added on the right of the form of *a*.

Sumatra

c17.s7.1

Two groups of the dated inscriptions from Sumatra have been selected here. Both represent the simplified style of the late seventh century A.D. Pl. XXI. 11 is taken from the Kedukan Bukit inscription, dated Śaka 605 (A.D. 683), and Pl. XXI. 12 is derived from the Kotakapur inscription, dated Śaka 608 (A.D. 686), the missing letters having been supplied from the Talang Tuwo inscription, dated Śaka 606 (A.D. 684). The simplified forms of *ka*, *gha*, *ra* and *la* are worth noting. *Ka* and *la* are comparable with the late forms of the Gurjara inscriptions of Western India. The broad looped base of *kha* is also derived from the same source. Undoubted Western Indian influence is observable in the following: the initial *a*, which copies the northern form in having its left hook curving to the left; the initial *ā*, which has a long vertical added to the form of *a*; the initial *i*, which has a double roof and a tail below; the initial *u*, which has its lower horizontal curved upwards; the open-mouthed type of *ṇa*; and *śa* and *sha* with full length mid-lines. The medial vowels are of seventh-century style, in which the important change is the replacement of the Pallava type of the broad curved medial *u*, as applied to *bhu*, *gu* and *tu*, by the simple form as seen in no. 12. Of the other forms of the letters we may note *ṅa*, which has a notch in its vertical, *ña*, which has also a notch in its right hook, *ḍha* with its base line extended and waved, and *na* with its loop closed. The left sides of *pa*, *sha* and *ha* are bent. On the whole the style shows a development towards simplification in which the tendency to reduce the letters to linear forms is evident.

Burma

c17.s8.1

The early styles of writing in Burma can be divided into three groups: (A) The writing adopted in the Sanskrit inscriptions found mostly in Arakan. Only one bilingual text, the Sanskrit being in this character, has been found at Hmawza. This style is connected with the proto-regional scripts of Eastern India. (B) The writing adopted in the Pyu inscriptions, the origin of which is discussed below. (C) The characters employed in the Pali or Prakrit inscriptions. These are derived from the Pallava records and resemble the sixth-century writing in South-east Asia. The Kawgun cave inscription has been placed in this last group, though it shows some North Indian letter forms. We have no example of early writing in the Burmese language. The later Burmese script is not derived from these characters.

c17.s8.2

A. Pl. XXII. 1 is taken from the Vesali Bell inscription; Pl. XXII. 2 from the Sandoway stone inscription; and Pl. XXII. 3 from the Sanskrit text of the Hmawza bilingual inscription. All these belong to one school of writing, though the first two inscriptions are very badly written. The nearest comparable material comes from Bengal, but it is difficult to connect them with any particular period of writing there. The inscriptions may be compared with Pl. XI. 2, 3, which represent the writing styles of the fifth and sixth centuries in Bengal. But this date cannot hold good for the Burmese inscriptions, which, though preserving the older forms, show acquaintance with tendencies that are hardly known before the seventh century in Bengal. Hence the date of these inscriptions is to be determined by this last factor. Some of the letters, especially *ga*, *dha*, *pa*, *ma*, *ya* and *sa* (very clear in no. 3) exhibit the *kuṭila* forms, while in nos. 1 and 2 we find the left triangle in *sa* with its mouth open. No. 2 also shows a rude copy of *bha* with its triangle opening upwards. Similarly, the form of the initial *u* in nos. 2 and 3 has the curve of the horizontal extended upwards. Again the medial vowels in no. 3 are all of the seventh century. The medial *ā* is given by a *kuṭila* short vertical on the right of the letters; for the medial *o* another top slant is given over the stroke of the medial *ā*; the medial *i* is invariably a left curve extended downwards to the bottom of the letters; the medial *ī* is a right curve but not extended down; the medial *u* is also a long curve to the left, but in the case of *bhu* it is turned upwards; the medial *e* is a top slant; and the medial *ṛi* is a hook turned to the right. The vowelless consonant is shown by a mark below the letter as in *yām*. In the other two inscriptions these medial vowels are not properly given. In no. 2 the medial *i* in *pi* is a short left curve and the medial *o* is shown by two top curved strokes, as in *yo* and *ro*.

c17.s8.3

We may also note the older features in these letters. They are all characterized by line, or at best notched, head-marks, instead of the triangular head-marks of the seventh century. The initial *ā*, as seen in no. 3, has preserved the old form with its hook turned to the right. The initial *i* in no. 2 is indicated by two dots on the right and a vertical on the left, though in the correct old form the position of the dots and the vertical is reversed. *E* in no. 2 is triangular with its base downwards. *Ka* has its central bar curved. In no. 3 foot-marks can be seen in *ka*, *ta*, *bha* and *ra*. *Ga* in no. 2 has straight verticals. *Cha* is of the beaked type. *Ja* in no. 3 has three arms slightly bending downwards. In no. 1 the conjunct *jña* shows a developed form of the two-armed *ja*, and *ña* with only three teeth. *Ṇa* is of the open-mouthed type. Its late form is seen only in no. 3, in which the outer right hook is angular. The two lower verticals of *ta* fall from the head-mark. *Tha* is oval with a mid-line. *Da* is of the double-curved type. *Dha* in no. 2 is either a circle or a square, but in no. 3 has a developed *kuṭila* form. *Na* is of the looped type. *Bha* in nos. 1 and 3 has preserved the older angular form of the north. In nos. 1 and 2 the open-mouthed *ma* has a tail on the left. *Ya* is of the tripartite form with a loop at the third left vertical. *La* in no. 3 and *ha* in no. 2 are of the hooked type. *Va* is triangular with its base sloping towards the right. *Śa* has a mid-line, and *sa* in no. 3 has a triangular loop on the left. *Ṅa* in the conjunct *Ṅga* has a slight bend in its vertical. *Tha* in the conjunct *sthi* is laid flat. All these older features survive along with the new tendencies of the seventh century.

c17.s8.4

The first two inscriptions are to be dated in the later half of this century, as they show the late forms of *bha* and *sa*.

c17.s8.5

B. Blagden was the first person to decipher the Pyu inscriptions and comment on their palaeography.¹ Later, in *Epigraphia Birmanica*, vol. i, he edited the Pyu face of the Myazedi inscription, dated A.D. 1112–13 (Pl. XXII. no. 7). In *Epigraphia Indica*, vol. xii, he published the urn inscriptions from Payaggyi Pagoda near Prome (Pl. XXII. 5), which, according to his interpretation, is dated in the seventh century A.D. The one record which is definitely of the seventh century is a bilingual inscription from Hmawza, which is fairly well dated by the characters used for writing the Sanskrit words (Pl. XXII. 4). We have included one more inscription from Halingyi (Pl. XXII. 6), which appears to be slightly earlier in date. But in these inscriptions, which range from at least the seventh to the twelfth century, hardly any development in the letters is to be seen. It seems that very early a particular type of writing became associated with the Pyu language. This assumption is confirmed by the bilingual text from Hmawza, where the Pyu script shows no influence at all from the *kuṭila* character of the Sanskrit words. When was this writing derived? Blagden answered the question on the basis of the tripartite (which he calls ‘anchor-shaped’) form of the subscript *ya*, which he compared with the similar form in the Vo-Canh inscription. Accordingly, he maintained that the origin of the Pyu script cannot be dated later than the fourth century A.D. But it is hardly necessary to rely on the evidence of a single letter. We have to consider the general character of the writing and reduce it to a particular system. The other point which is relevant in this connexion is the use of different letter-forms in some inscriptions. This question will be discussed below.

c17.s8.6

¹ *J.R.A.S.*, 1912, pp. 365–88.

c17.s8.7

The Pyu letters are characterized by line head-marks; very rarely do we notice the notched type. The verticals of *a*, *ka*, *ra*, and the vertical type of the medial *u* are extended downwards, ending in a loop—a feature which is hardly known in India but most probably developed from the lower curves of the verticals known in the southern scripts. Though this writing does not favour waves in the lines, acquaintance with them can be inferred from the form of *ṅa*, which has its vertical waved, *ba*, which has its left side notched, and *ha* with a bend in its left arm. The medial vowels are further instructive. What has been read as medial *i* is used profusely. It is of two types: the Halingyi inscription shows a long twisted stroke turned to the left, which recalls the stroke for *e* used in the sixth- to seventh-century inscriptions of North India; the other inscriptions uniformly have a right curve, which actually stands for long *ī* in North India. This second type of stroke occurs only once in the Halingyi inscription with the consonant *va*, which has been doubtfully read as *vī*. The medial *o* is two top curves, a form common in North India. The only other medial vowel known is *u*, which in the cases of *bhu* and *gu* is a sloping stroke added to the right vertical. In the late Myazedi inscription we find the curved type of the southern *au*, which is hardly known in India before the end of the fourth century A.D. Only two initial vowels are known. The most commonly used is the South Indian *o*-type, but, as the evidence from the Hmawza bilingual text suggests, here it stands for the initial *u*. One form of *kra* was read by Blagden as *o*, but this is hardly justified. From the late Myazedi inscription we find the form of the initial *a*, which has an upper angular hook with a line below attached to the right vertical—a form known from the Kushāṇa inscriptions (Pl. VIII. 5) and the manuscript of *Kalpanāmaṇḍitikā* (Pl. XII. 10). The following consonants are known from these inscriptions:

c17.s8.8

*ka*, *kha*, *ga*, *gha*, *ṅa*
*cha*, *chha*, *ja* . .
. . *ḍa*, *ḍha* (?) .
*ta*, *tha*, *da*, *dha*, *na*
*pa*, *pha*, *ba*, *ḅa*, *bha*, *ma*
*ya*, *ra*, *ḷa*, *va*, *sa* and *ha*

c17.s8.9

*Kha* is of the southern type, resembling a fish-hook with a horizontal line attached to its lower right vertical. Only no. 4 has a small triangular loop in place of the horizontal line. This form is known from the Kadamba and early Western Chālukyan inscriptions. *Ga* is of the northern type with a foot-mark at the left limb and the right limb longer. *Gha* shows the simple form with three uprights. The loop of *cha* below a stem is slightly tilted to the left. *Chha* has the double loop below a stem. *Ja* has three arms to the right of a vertical, the lowest one sloping down as in the North Indian forms after the fifth century A.D. *Ḍa* is round-backed, with its tip falling downwards. There is hardly any difference between the forms of *pa* and *ḍha*, and hence one is inclined to read the latter letter as *pa*. *Ta* shows a development from the Deccani form and agrees fully with what I have called the Chālukyan *ta*. *Tha* is a dot within a circle. *Da* is either of the double-curved type of the north or of the simple form with a longer stem and short lower curve to the right. *Dha* also is of the northern oblong form. *Na* has its left loop closed. The left sides of *pa*, *pha*, *ba*, *ḅa* and *ha* are bent. The new letter *ḅa* is formed from *ba* by the addition of a mid-line. *Bha* is of the broad type. *Ma* has a looped base, but the letter is drawn in outline by first forming the outer bend of the arms and then inserting the mid-line. The lower portion of the bend is slightly tilted to the left. *Ya* is of the tripartite form but its third left arm is replaced by a loop—a form known in Malwa in the fifth century A.D. In the late Myazedi inscription we find the third left arm of *ya* with an outer curl—a form which was known in Rajasthan and Eastern India in the sixth century A.D. What has been read as *ḷa* actually adopts the hooked form of the northern *la* known between the fourth and fifth centuries A.D. Only in the late Myazedi inscription do we find the true form of *ḷa*, with a curve below the shape of *ṭa*. *Va* is an elongated loop. *Sa* has a hook on the left. In these inscriptions we also find a symbol at the beginning and at the end. Both resemble the form of *oṁ*, the first curving upwards as in the northern inscriptions, and the second downwards as in the later southern inscriptions of India. The first symbol is followed by two verticals with looped ends.

c17.s8.10

In these letter forms we find a mixture of the northern and southern types, and we also see a deliberate attempt to adapt these forms to write the Pyu language. The latest forms copied belong to the sixth century A.D. If we leave aside the form of *a*, which may be the result of simplification, there is nothing in the other letters which can suggest a date earlier than fifth century A.D. The presence of so many northern forms as well as those of the Chālukyan type suggests that the person who adopted this writing knew these different characters. The evidence further suggests that this adaptation could hardly have been made before the sixth century A.D.

c17.s8.11

As remarked before, in some Pyu inscriptions¹ each line of the Pyu alternates with another line of writing that is definitely in different Indian characters. As far as we have been able to read them, they do not make any sequence. It appears that the letters have been used as pure alphabets, and it is doubtful whether any medial vowel is used with the consonants, except the inherent *a*. Most of the letters repeat themselves, like *ma*, *ma*, *ma* or *ta*, *ta*, *ta* or *va*, *va*, *va*, &c., or we find them in combinations with others. It is for the epigraphists to decide whether they are to be taken as musical notes. But the letters have clearly preserved certain older forms. The most remarkable is the form of *a*, which has on the right a vertical with two curved strokes on the left—a form which is seen from Aśokan times to the 1st century a.d. in India, but which in Ceylon is preserved till the fourth century a.d. Another striking feature is the form of *ya*, which is of the ‘anchor-shaped’ type—a long vertical within a crescent below, a type also preserved in Ceylon till a late period. All other forms, such as *ta*, *na* with a curved base, the round backed *da*, *ma* with a looped base, the straight vertical of *ra*, and *va* with a circle below a stem, are later southern forms of about the fourth to fifth century a.d. All the letters are characterized by notched head-marks. These notches are of the same style as those in the main Pyu inscription. There is no doubt that these letters were written at the same time as the Pyu inscription, as the ornamental curved flourishes of the Pyu writing had been arranged in line with these letters. The Pyu inscription itself cannot be dated earlier than the seventh century a.d. It is difficult to account for the preservation of the older forms unless these were used for some definite purpose, musical or otherwise.

c17.s8.12

¹ See *Inscriptions of Burma*, University of Rangoon, Oriental Studies Publication, no. 5, Oxford, 1956, portfolio IV, pl. CCCLVa.

c17.s8.13

C. Pl. XII. 8 is taken from Maunggun gold plates, with which the writing of the Kyundawzu plates agrees. Here the letters are characterized by line head-marks. The lower verticals of *a*, *ka*, *ra*, and the vertical type of the medial *u* have curved ends, but the right limb of *la* is reduced in height. Waves are seen in the arms of several letters. The medial vowels and the forms of the letters agree with the sixth-century style of South-east Asia, except that in these inscriptions the ornamental extensions are not to be seen. They may be assigned to the late sixth century a.d.

c17.s8.14

Pl. XXII. 9 is taken from the Kawgun cave inscription. The writing is very crude. Except for the open-mouthed type of *ma*, which must have been derived from the Sanskrit writing known in Burma, all other forms including the medial vowels are southern, and do not appear in South-east Asia before the sixth century a.d. The date may fall in the sixth to seventh century a.d.

Siam

c17.s9.1

Three early inscriptions from Siam have been published. Pl. XXII. 11 is taken from the Pra Pathom inscription, which is in the Mon language; Pl. XXII. 10 is derived from the Srideb inscription; and Pl. XXII. 12 from the Vat Maheyang inscription; both the latter are in Sanskrit. The characters used in all of them are similar. No. 10 is in the sixth-century style of South-east Asia, showing the Pallava forms of the letters and the medial vowels, with the only difference that it does not favour ornamental extensions. The inscription may be placed in the first half of the sixth century a.d. Nos. 11 and 12 prefer the ornamental style, especially remarkable for the extended length of the verticals of the medials on the right. In no. 10 the medial *au* has been read as *o*, and the initial *o* as *u*. Here the initial *a* has a short curve at the end of the right vertical, and the hook on the left has its tip turned outwards. No. 11 has *i* with a tail below the notched roof. *Ka* is of two types—one with its lower vertical almost doubled, and the other in which the vertical is equal in length to the bends of the central bar. *Ṇa* shows a loop at the left corner of the open-mouthed form, and the subscript *ya* is of the ornamental type as in the inscriptions of Pūrṇavarman. Except for the last form, which is probably due to the preference for ornamentation, other features place the inscription in the first half of the seventh century a.d.

First 56 of 110 passages. Read on

Linked texts (1)