Some Sanskrit Inscriptions of Arakan

By E. H. Johnston
Bulletin of the School of Oriental and African Studies, University of London, Vol. 11, No. 2. (1944), pp. 357-385.
(PLATES III-V)

INTRODUCTORY NOTE

PROFESSOR E. H. JOHNSTON'S sudden death in October, 1942, was a grievous blow to Oriental research. Sir Richard Burn, called in by the authorities of Balliol College to go through his papers, found amongst them a rough manuscript article and other material on the Sanskrit inscriptions of Arakan, together with correspondence on the subject with Mr. G. H. Luce and other scholars who had supplied him with rubbings and photographs. As will be seen from the following letter to Mr. Luce, dated the 15th August, 1942, Professor Johnston was contemplating the preparation of this work for publication:

“I have not yet heard what is your opinion about publishing some of the Arakanese results now. Life is short and uncertain these days, and it will hardly be possible to get anything out in Epigraphia Birmanica for years to come. What I should like to do is to publish Anandacandra's inscription, omitting all I have said in the draft about the paleography and just mentioning the others, with a brief account of the coins and of the historical conclusions. For one thing it might attract the attention of linguists and lead to some ideas about the vernacular names; e.g. what language is, Sevinren, assuming my reading to be correct? The fuller consideration of details could then be reserved till Epigraphia Birmanica is ready to come about. "

“I am hopeful that some day I may get more out of some of the other inscriptions than I have so far, e.g. I was looking the other day at what I called in my draft paper the `separate inscription ' and realized at once that it has a name Prabhacandra in the first line (I suspect that this is not the beginning of the inscription), and about the middle I can read bhupalah 8Sri Candakeyura varmma. Who were these people? Local lords in the interregnum between the Candra dynasty and Vajrasakti ? Quite possible palaeographically on my dating of it. One day I may get a few consecutive words out of several other lines. I suspect it of containing some sort of genealogical list.

“The coinage I am still in difficulty about. The typical Candra coinage is certainly connected with Vengi, and it is the earlier conch-shell and vardhamana coins that beat me. The two I put at the head of the photograph are certainly the original design, but the actual specimens may be much later. The type seems to occur all over Burma and even in Siam, and it seems, from Sir R. Temple, in the Indian Antiquary for 1927 and 1928, that similar coins were struck in Calcutta for a king of Burma some time before 1823 ; but I have not checked his references yet. Do you know a pamphlet he refers to by Captain C. H. White, printed at Akyab in 1892, `Notes and References to a Selection of Symbolical and Historical Coins of Arakan '? Temple says it is in the India Office, but probably it is quite ungetatable at present. What I have written about these coins wants much reconsideration. I have got evidence from Madras of the association of the vardhamana and conch-shell with Laksmi, unfortunately only about the ninth century A.D."

Professor Johnston had not put the finishing touches to his work, and it would not have been possible to publish his MS. but for the generous labour of Mr. Luce and Dr. Barnett. Mr. Luce deciphered the often difficult handwriting and made a clean typescript copy, querying anything .about which he was doubtful. Dr. Barnett checked Professor Johnston's reading of the inscriptions and provided English translation and necessary notes, which are all marked by enclosure in square brackets.

Professor Johnston's article, besides its palaeographic interest-his remarks on this subject have not been omitted-is of historical importance, as giving the first solid foundation for the study of ancient Arakan, and as indicating the valuable results likely to be achieved by full-scale archaeological excavation at Vesali, Mrohaung, and other sites.

J. A. S.

PART I. THE INSCRIPTIONS


The early history of Arakan is still a complete blank in the histories, all that is known of it being dynastic lists in late Chronicles, which, as will appear below, cannot be relied on for either names or dating, and the coins of a few kings which have yet to be arranged in chronological order. If scholars have neglected this province hitherto, that is largely due to its geographical position on the confines of India and Burma. Not unnaturally the historians of the latter country have hitherto directed their efforts to elucidating such facts as can be ascertained about Burma proper, and the scanty resources available to the Burmese Archaeological Service have not been adequate to undertaking serious excavations in Arakan. Indologists on the other hand have felt little interest for an area in which, even at the period dealt with here, Indian civilization was not a natural product, but was imposed on the country from without. The existence of one long inscription in Sanskrit, however, has been known for some time, though it has not yet been edited, and it is proposed to bring together here such epigraphic and numismatic material as is available for the period previous to A.D. 1000.

The following inscriptions in Sanskrit have been examined by me, all in rubbings belonging to the Rangoon University, except that from the Sandoway district, which I have read from the original stone :

(1) A votive inscription in two lines on a monastery bell found at Vesali near Mrohaung (Plate IV, 1); for establishing the reading of certain doubtful characters use has been made of photographs supplied by the Superintendent of the Archaeological Survey, Burma. The bell has been damaged in two places, so that the name of the donor is illegible and one other character is destroyed. It is said to be now at Akyab in the charge of the Honorary Archaeological Officer.

(2) A votive inscription on a stone obtained by Colonel G. E. Fryer in 1872 from a cave near Nga-lun-maw, Kwelu circle, Sandoway district, and published (but not read) by him with a very poor eye-copy in the Proceedings of the Asiatic Society of Bengal, 1879, p. 201. The stone, which measures approximately 17.4 x 14 cm. and is 2.4 to 2.7 cm. thick, passed at his death to the Royal Asiatic Society, who have recently deposited it on loan in the Museum of the Indian Institute, Oxford; another stone from the same neighbourhood with the ye dharma verse on it should also be with the Society, but has not been traced so far. The surface of the stone has scaled off in parts, making it impossible to be sure about the reading of the fifth line (Plate IV, 2).

(3) The ye dharmda verse inscribed round a small stone stupa at the northwest entrance of the Tejarama monastery, half a mile north of Mrohaung. Apparently the whole verse was inscribed in two lines, but only a portion of the first line is legible with certainty.

(4) A thirty-line inscription, now in the inscription-shed at Mrohaung, which was found half a mile east of that town, two furlongs east of the Middle Bazaar, at Wunhtitaung Hill. Part of the inscription is irretrievably lost, and I can only read occasional words of the rest with any certainty, not enough to determine the nature or object of the record; unfortunately no proper name has been identified by me, but an attempt to read it from the stone, instead of from a rubbing, might conceivably lead to greater success. (5) Most important of all are the inscriptions on a pillar now at the Shitthaung pagoda at Mrohaung, its original position being unknown. The oldest inscription, of about 100 lines in a small neat script, is on the east face; some of it is destroyed, and the rest is so rubbed that nothing consecutive can be made out from the rubbing, though prolonged examination of the actual stone might produce a tolerable reading for the lower part. Of about the same date are four lines on the top end of the north face. The west face has a prasasti of 71 ½ lines in honour of a king called Anandacandra, which is fairly well preserved and can mostly be read with certainty. 'A short account of it, based evidently on an inferior rubbing, was given by Hirananda Sastri in the Annual Report of the Archaeological Survey of India for 1925-6, pp. 146-8. I should acknowledge here the help I have received in reading this inscription from a reading of the last fifty lines prepared some years back by Dr. N. P. Chakravarti, Government Epigraphist for India, for the Archaeological Survey of Burma, which was unfortunately never revised for publication.- I have therefore thought it proper not to quote it when our results differ. This record is of capital importance for the early history of Arakan, as the correctness of the dynastic list which it gives is corroborated by the coins to a considerable extent; we thus have a reliable skeleton framework, going back 359 years from some date early in the eighth century A.D., with some information, possibly not equally sound, for the preceding 188 years. For reading this inscription, I have made use of two excellent rubbings belonging to the University of Rangoon; in a number of cases, where one is not clear, the other gives the doubtful character more legibly. I have also had at my disposal photographs taken many years back by the Archaeological Survey, but these appear to have been taken after the marks on the surface had been filled in with some white substance for greater legibility, and this process was not carried out with sufficient accuracy for the photographs to be usable with safety, except in a few instances, where their indications have corroborated readings of which I might otherwise have been in doubt. In a few lines the last few characters have been cut away at some later date, perhaps when the pillar was removed from its original site and had to be fitted into a new position (Plate III). Beneath this inscription are two short lines in a later hand (about tenth century?), whose purpose is not clear to me. On the north face the early inscription is followed by several later inscriptions, in an untidy hand of the tenth century, which have been considerably damaged, and though I have read a number of words, these are not enough to enable me to attempt a reading of them; as in the case of the east face, much more could probably be done by expert examination of the actual stone. The total number of lines in these is sixty-nine. Finally there is an inscription, almost entirely illegible, on the top of the pillar, but the very few readable letters on it do not justify any positive conclusions about its date, except that it is probably not distant in time from those just mentioned on the north face.1

The palaeography of these inscriptions is of considerable interest, and allows certain inferences to be drawn as to the nature of Indian influence in Arakan. When these inscriptions are compared with the few which have been published from Burma proper, the curious point arises that the scripts used in Arakan, unlike those in Burma, all find close analogies with those current in North-Eastern India, and I propose dates accordingly in this paper in consonance with the principles which would be applied to inscriptions from that part of India. The justification for this procedure lies in the results, namely, a logical development of the alphabets and a coherent scheme of dates, covering a number of centuries; the parallelism is so exact that we need hardly assume any substantial " time-lag " for the introduction of changes; that assumption would in the circumstances make it necessary to postulate a time-lag of similar length in all cases throughout a period of five centuries.

The first three inscriptions, which are of a purely religious character, belong to a tradition other than that which characterizes the remainder; the traces of this tradition in Eastern India have not hitherto been fully recognized.2 The bell-inscription, which I take first, is shown by its forms for

1 [Prof. Johnston succeeded in deciphering many letters of the inscription on the north face, but his results were too fragmentary to be published.]

2 For these three the plates in the excellent paper by Dr. S. N. Chakravarti, JASB., iv, 1938, pp. 352-391, on the development of the Bengali character should be consulted. But his view that inscriptions can be dated by consideration of one or two crucial characters is only valid when datable records are available in such numbers that development can be traced from

the letters ka and sa and its tripartite ya to be probably not later than A.D. 650. Other indications, particularly the forms of sa and ma, suggest the unlikelihood of its being much earlier, but the script has some unusual features, partly archaic. Da has the curve rounded, without the point on the left of the curve or the later short cross-stroke at the end of it, and dha is a small complete circle. The vowel u is added to t by bringing the end of the consonant round in a curve and up to the top line, a development of a form hitherto best known from earlier inscriptions in Central and Southern India, 1 but, as will be shown, characteristic of this epoch in Bengal. For ma the left-hand down-stroke turns to the right almost at once, and the short stroke at the bottom left-hand corner, which is usually horizontal, slopes upward so as to form a sharp re-entrant angle with the down-stroke. The letter sa has at the left a triangle, much larger than in Buhler, Pl. IV, cols. 11 and 12, so that the top of it reaches the cross-line at the top of the letter. Initial i consists of two dots or small nicks, one vertically above the other, which are clearly visible in the photograph but hardly to be distinguished in the rubbing, followed by a member closely resembling da. Unfortunately there are no examples of la and ha, which have marked peculiarities on the Sandoway stone, but the one occurrence of jna suggests the possibility that ja was beginning to undergo the change of shape which occurred in the second half of the seventh century.


This script is closely akin, both in general character and in respect of these peculiarities, to two of the Faridpur copperplates, Nos. 1722 and 1724 of Bhandarkar's list. The authenticity of the two latter, as well as of a third, No. 1723 of the same list, which was doubted at one time, appears now to be generally accepted ; I confess, however, to feeling doubts about No. 1723. No. 1722, issued by Dharmaditya in the year 3, and No. 1724, issued by Gopacandra in the year 19, should be genuine in view of their peculiar relationship to the bell-inscription, and Gopacandra's may perhaps be the earlier; but No. 1723, issued by Dharmaditya without any mention of the year, differs considerably in script, and I do not take it into consideration here, though I would not venture without a personal examination of the plate definitely to stigmatize it as a later forgery. 1722 has to in the form mentioned above, the two places where it should occur in 1724 being illegible in the reproduction, and the same form of u is found in 1722 for du and Au. In 1724 dha (11. 19 and 22) is circular, but larger in proportion to the other letters than on the bell ; but in 1722 it is usually slightly elliptical, e.g. 1. 21, but less often it is a semi-oval (both forms in 1. 2). Sa has the triangle on the left in both plates; but while it is normal in Gopacandra's, there are two instances, 11.14 and 19, in 1722, where the triangle is enlarged and the apex reaches right up to the top line, as in the bell. Ma is nearly normal in 1724, but shows the beginning of the

decade to decade ; this condition is not satisfied for the period in question, and in general it is desirable to take account of as many letters as possible.

1 Buhler, Ind. Palaeogr., pl. iii, cols. 17 aMd 19, and pl. vii, cols. 1 and 11 ; see also D. C. Sircar, Successors of the Satavahanas (University of Calcutta, 1939), pp. 328 if., a second century inscription from Ellura in H.H. the Nizam's Dominions.

process whereby the bell form was reached ; 1722, on the other hand, has it in a form even more exaggerated than the bell, and in 1. 20 for instance the point of the angle is only just below the main line, being thereby related to the occurrences on the Sandoway stone which will be discussed later. In both plates va is more or less triangular in shape; while, however, in 1722 the apex never reaches the top line, in 1724 occasionally, e.g. 1. 3, it not merely reaches the top line, but shows a tendency for the two sides not to meet on or before reaching the top line, thus resembling rvva of the bell, but in no case so pronounced as in the va of avaptaye. There are also certain differences : in 1724 ha has a normal form with rounded top (Biihler, Pl. IV, cols. 5 to 10), but in 1722 the central bar is curved upwards at each end, so as to give the top half of the character the shape of a horizontal oval. Initial i occurs in the archaic form of Biihler, Pl. IV, col. 5, consisting of two small circles followed by a downstroke with a short crosspiece at the top. The interesting forms of la and ha are discussed below in connection with the Sandoway stone.

It would have been desirable also to compare the scripts of the Damodarpur copperplates, Nos. 1272, 1286, 1307, and 1550 of Bhandarkar's list, bear dates running from 124 to 224 of the Gupta Era,,but the photographs published in Epigraphia Indica, xv, so seldom enable the letters to be seen clearly, that the use of this evidence is hazardous. Of letters whose shapes can be accurately determined, Plates I, III, and IV, all have to in the form discussed above, except that in Plate I the bottom of the u is square, not rounded, and the same to recurs in the Nandapur copperplate (G.E. 169, Monghyr district of Bihar, Ep. Ind., xxiii, pp. 52 ff.) ; the u of nu on Plates III and V is indicated, as in the bell and in 1. 18 of Dharmaditya's plate of the year 3, by prolonging the downstroke of the consonant in a straight line instead of making it curve left-handed. For the peculiar initial i of the bell I can only quote an inscribed brick (Ep. Ind., xxiv, pp. 20-2), recently discovered at Nalanda, which is dated 197 (evidently Gupta Era) and which proclaims its eastern origin by its forms of sa, sa, and ha, thereby differing from the similar bricks previously discovered at Nalanda and Gopalpur (Ep. Ind., xxi, 19f if., and JRAS., 1938, 547 ff.). Here the letter is identical with that on the bell. Tu does not occur on the brick, which, however, has a similar u in du, as in 1722. As is pointed out below, this initial i recurs on the Sandoway stone and in Anandacandra's prasasti, but with the members in the reverse order.

These comparisons are sufficient to prove that we are dealing in the bell with a script which was derived from Eastern Bengal, descending possibly from a variety slightly later than any of those described. No help can be derived from the shape or ornamentation of the bell in the present state of our archaeological knowledge, and I would suggest that, if the date is fixed on the palaeographical evidence as somewhere in the first half of the seventh century A.D., the margin of error is likely to be small.

The stupa-inscription is so badly preserved that caution is necessary in drawing conclusions, and it may have been incised by a somewhat careless workman. The letters ya, bha, and va appear to be exactly as in the bell; dha is probably a narrow upright semi-oval, and the exact shape of rmma is obscure on the rubbing. The letter pa was evidently rounded at the bottom left-hand corner, while in the bell it is rounded in two cases and squared in the other two. It is better to deal with ha under the Sandoway stone, but tu was probably as in the bell, if we may presume that the end of the u has been lost. The indications of letters in the second line are quite uncertain, and my suggestions in the reading below may be mistaken. The date is probably about the seventh century.

The Sandoway stone comes from an area which is and has always been less subject to the influence of India than the Akyab district, and this is apparent in the paleography of the inscription, whose script clearly descends from that of the bell or from a similar type, but has developed on lines of its own in a new environment. The letter dha remains the same small circle, the u of to is indicated in the same way, and there is the same alternance of the squared and rounded pa. Ta consists of two straight downstrokes proceeding from the top line and slightly sloping, a form which develops out of the letters ta, tra, and tna as seen in 1. 21 of Gopacandra's inscription, where the two lines meet at the top with a tendency to separate ; much the same shape is to be found in Niticandra's coins (Pl. V, Nos. 6-9). The peculiar forms of ma and sa evidently derive from the script of the bell, and ya is a hitherto unknown development of the tripartite ya, with the peculiarity that in 1. 1 the left-hand member has been turned the wrong way round, as may be seen by comparison with the occurrences in 11. 2 and 6. In 1.4 I read initial i, taking it to be the form found on the bell with the members reversed, as this development is found in the last line of Anandacandra's inscription. The two short horizontal strokes after yi in 1. 6 I understand as final m; at least there seems to be no other possible explanation of them. The one letter which shows a more modern style, and is not merely a deformation of an older form, is bha in 1. 1, which belongs to the type found in Anandacandra's inscription. The most interesting letters are ha and la. The four occurrences of the former are all slightly different: three have a straight downstroke, one with an upward hook to the left from the bottom, and the other two more rounded so as somewhat to resemble a capital J, while the fourth has a slightly curved downstroke with an upward curve to the left at the bottom rather like the two preceding cases. Evidently we have here a derivative of the archaic forms depicted in BUhler, Pl. III, col. 17, and Pl. IV, col. 1, where ha has very much the shape of the figure 5; it still survives not only in Bhimavarman's Kosam inscription (Ind. Culture, iii) pp. 177 if., G.E. 130), but also, somewhat modified, as late as the Nalanda brick of G.E. 197. The down-curve straightens out in the Paharpur copperplate (G.E. 159, Bhandarkar's list No. 2037), the Nandapur copperplate (Monghyr district of Bihar, G.E. 169, Ep. Ind., xxiii, pp. 52 ff.), the Gunaighar grant of Vainyagupta (G.E. 188, from Tipperah, Bhandarkar's list No. 2038), and the Faridpur copperplate, while Dharmaditya's plate of the year 3 has a normal western form (cf. Buhler, Pl. IV, cols. 7 and 8) ; Gopacandra's grant shows in 1. 24 a form in which the downward stroke is nearly straightened out like h in hya on the stone, and in 1. 22 one in which it is quite straightened out so as to resemble h in the first he of the stone. The lineage of this letter is thus quite clear, and the same shape has been recognized in the inscription on a Buddha-statue found at Hmawza (Ann. Rep. A.S.I., 1928-9, p. 108, and Pl. LI (b)). Similarly the letter la retains essentially the older form given by Buhler, Pl. IV, cols. 1-3 and 5-6, and S. N. Chakravarti, fig. v, col. 1. The Faridpur copperplates usually have a more modern form, but the older type occurs once at 1.12 of Dharmaditya's inscription of the year 3. The script of the stone thus clearly derives from scripts current in Eastern India during the Gupta period, and shows no novelty except the bha, which probably originated in India about A.D. 650; the difficulty in dating it lies in deciding how long it would have taken for the script in isolation to have developed its marked peculiarities. On the whole I consider it might be as late as A.D. 800, but hardly later, since the other stone from the same locality, now missing, shows according to the eye-copy a script which has been determined by later influence from Bengal, and there should be a substantial gap between the two.

The remaining inscriptions contain fewer peculiarities from the Indianist's point of view, but one curious feature is the tendency of the letters to grow larger in the course of time. Those on the west face of the pillar measure 8-11 mm. in height, and on the separate inscription about 9-13 mm., while the height on the earlier inscription of the north face is 12-16 mm., on Anandacandra's 16-19 mm., and on the later inscriptions of the north face, where the variation in size is considerable, about 21-29 mm. To fix the date of the earlier inscriptions by palaeographic considerations is not possible, as I cannot read enough of them to form a complete alphabet. The script of the west face seems fairly close to that of the Maukhari inscriptions as given by Buhler, Pl. IV, cols. 11 and 12, and it probably belongs to the sixth century A.D., though it might be as late as early in the seventh century. The same remark applies to the other two, though I should expect, if they could be read more fully, that they were slightly later. All that can be said about these inscriptions is that there appears to be nothing in them which would surprise us in an inscription originating from Eastern India; thus the inscription on the west face of the pillar and the separate inscription both write a tripartite ya with the left-hand member curling round outwards instead -of inwards as-in the , Rajshahi district, 188, from Tipperah, No. 2038), the Amauna grant of Nandana (G.E. 232, Gaya district of Bihar, Bhandarkar No. 1310), the Nalanda brick of G.E. 197 referred to above, and the Arakan coin of Dharmavijaya discussed below ; and tu, if rightly identified by me on the west face of the pillar, appears also in the same form as on the bell and the other inscriptions already discussed. The eastern bipartite form of ya with the bulge at the bottom evidently developed from a cursive form of the outward-curling tripartite ya, as appears from the form in line 29 of Gopacandra's copperplate. The short inscription on the top end of the north face seems to write ya with the left-hand loop curling inwards and with the same ma as on the bell; these two points and the larger size of the letters lead me to infer a somewhat later date for it than for the other two.

It is, however, fortunately easy to place the important inscription of Anandacandra. Obviously it is closely related to, but later than, the Aphsad inscription of Adityasena (Corpus Incr. Indicarum, iii, pp. 200 ff.); the substantial difference is in the form of ja, which at Aphsad shows the first beginning of the change effected in the second half of the seventh century, whereas on the Mrohaung pillar the change has been carried through. The script of Yasovarmadeva's inscription at Nalanda, which belongs undoubtedly to the first half of the eighth century (vide Bhandarkar's List, No. 2105), is almost entirely identical with that of Anandacandra's inscription, both in the form of the letters and in style of writing. The most substantial difference is in the form of jya, which in the Nalanda inscription is of the older type, whereas in the other the resemblance to an ordinary ya is most clearly brought out in the example at the beginning of 1. 45. This script, as has been pointed out by others, goes back to the Bodhgaya inscription of Mahanaman (CII., iii, pp. 274 ff. ; G.E. (?)269). Among noteworthy peculiarities are the peculiar forms given to consonants when they occur at the end of a hemistich, in place of the virama used at Nalanda. Thus k, 1.23; t, 11. 26 and 38; n, 1. 36; m, 1. 31. As precedents for this I can only quote the Bodhgaya inscription of Mahanaman, 11. 2 and 6, for final m, and 1. 14 for final t, and the Paharpur copperplate of G.E. 159 for final m. Initial i, 11. 29 and 39, consists of two small circles with what more or less resembles a modern Devanagari subscript u below and a crescent above the line. There is only one instance of initial i, namely in the closing word of the inscription, whose shape has already been described and appears to be a specialty of Arakan. These considerations would naturally lead us to place the inscription in the first half of the eighth century, and for other reasons probably fairly near the beginning of the century. For we have coins of a number of the kings mentioned in the inscription, and two of those, those of Devacandra and Dharmavijaya, can be approximately assigned on palveographic grounds to the first half of the fifth and seventh centuries respectively. Now Deva is recorded as ruling twenty-two years; the interval from his death to Dharmavijaya's accession is 177 years; Dharmavijaya ruled thirty-six years; and Anandacandra succeeded nineteen years after his death and had completed the ninth year of his rule at the time of the inscription. The maximum interval between the coins of Devacandra and Dharmacandra is 235 years, and between the coins of Dharmavijaya and the inscription of Anandacandra is sixty-four years. It seems, therefore, hardly possible to put the latter much later than A.D. 700. Further, if we compare this script with that on the coin of Dharmacandra, father of Anandacandra (Pl. V), the difference is such that we must infer that it was a recent importation into Arakan, probably direct from Nalanda. There are moreover four later coins, which used to be in the cabinet of Mr. Prafulla Nath Tagore, but whose whereabouts are unknown; they were published by R. D. Banerji in JASB., Numismatic Supplement XXXIII, vol. xvi N.S., 1920, with poor reproductions and readings which are decidedly speculative. So much of the script as can be clearly determined from the plate appears to be in direct continuation of the Anandacandra tradition.

The later inscriptions on the north face of the pillar are in a Bengali script of the tenth century A.D., very similar for instance to Biihler, Pl. IV, col. 23, which should be dated A.D. 931 in accordance with Bhandarkar's list, No. 53: some forms resemble those of Mahipala's Bangarh copperplate (Ep. Ind., xiv, 329 ff.), but in the main the script has a more antique appearance. I cannot, however, discuss it in detail, as I have been unable to form a complete alphabet from the inscription, in which also the letters are badly cut and vary considerably in shape. A characteristic letter of the inscriptions is ra, a downstroke with a sharp hook at the bottom to the left ; this form appears to have been in use for a relatively short period, as may be seen from S. N. Chakravarti, loc. cit., fig. v, col. 8.

The inferences to be drawn from the somewhat complicated paleography of these inscriptions are best considered at a later stage, after their language and contents have been treated. The bell-inscription follows a common Mahayana formula, which can be traced from the sixth century onwards for as long as Buddhism prevailed in India. It is unfortunate that the bell has been badly damaged just where the donor's name is written, but if I am right in reading the first syllable as Mya, it would seem unlikely that the name was Sanskritic in form. The stnpa-inscription has the well known ye dharma verse, and is not well enough preserved to show any peculiarities there might be in it. On the Sandoway stone the ye dharma verse is in Sanskrit, but so incorrectly written as to imply that Sanskrit was little known in South Arakan. The dedicatory lines are unfortunately difficult to read; the donors' names are uncertain, but clearly not Sanskritic in form, and it is impossible even to conjecture what the object dedicated was. The language of this part is presumably meant to be Pali, to judge from the verb akarayi ; but even the form of this word is uncertain, as the two short parallel horizontal lines after the last letter might indicate m, and if so, it should be conjectured that a su has been omitted. Kusala for kusala should also be noted. If the language is Pali, we can only conclude that South Arakan, unlike north, had derived its Buddhism from Burma proper.

Nothing to the point can be said about the other inscriptions earlier than Anandacandra's, except that the few words I have read show them to be in Sanskrit; these do not indicate their substance, and I have been unable to detect any certain proper names. In the separate inscription, 1.11, there may be a proper name before the words bhupa[tih] pr[thivz], which are legible, but I cannot make anything out of it that will hold water. Anandacandra's prasasti consists of sixty-five versese, in a rather doggerel style marred by several solecisms ; a prose sentence is interpolated between verses 45 and 46. Not enough of the first verse is legible to be worth recording; the second appears to be in ordinary anustubh sloka, as are all the rest, except verses 40, 61, 63, 64, and 65 in upajati, 52 and 59 in mixed indravamsa and varnsastha, and 32, 42, 44, and 62 in vasantatilaka. By a licence which Asvaghosa also allowed himself (Saundarananda, vii, 48 c), pada 2 of verse 44 in vasantatilaka metre ends in a short syllable instead of a long one. A short vowel remains short before sv in verse 24, before tr in verse 32 as emended by me in the notes, and before sr in verse 51. The first pada of verse 20 is hypermetric. In verse 21 nagara is treated as masculine, and in verse 30 upabhoga as neuter. In verse 18 the r in °kartrnam has been shortened for metrical reasons. There are some cases of double sarimdhi, in one of which a long syllable has also been shortened, again to suit the metre, namely verse 46, viharaneka for vihara anekah, verse 49, pasadarumaydnekd for °maya anekah ; and in verse 51 the compound °rajatanekan is presumably meant for °rajatan anekan.1 In two cases, again because of the metre, the nominative ending -o has been shortened to -a, verse 44 Anandacandra for °candro, and verse 62, ritamrapattananaradhipa for °-po ; in 42, therefore, one should probably understand Vajrasaktisuto vira° for °sutavira°, and I am not sure that the following line should not be divided 9ridharmacandra (i.e. for °candro) mahimaprathitaprabhavah, except that there is no authority for prath with a later than the Rigveda.2

For convenience of reference I have numbered the verses and assumed that the double danda after the eighth character in line 3 marks the end of the second verse ; if there were only one verse it is difficult to find a long metre ending - - - - -, and we should have to presume an introductory sentence in prose, followed by one invocatory verse. From the remains it seems that if I am right in counting two verses, the inscription started with two invocatory stanzas, one perhaps to the Buddha, and one to Hindu deities, as trilocana should refer to giva. The indications in the first line are consistent with reading Bodhisatva near the beginning of the first line, but only the letter sa is certain. The next forty-three verses give a list of the kings of Arakan, divided into. three periods, early (partly at least mythical) in sixteen verses, the thirteen kings of the Candra dynasty in fourteen verses, and the later kings in thirteen verses. The length of the early period is given as 1,060 years, taking saddasadhikam to mean " plus 60 " ; if " plus 16 " had been meant the reading would have been sodasadhikam, and we are authorized to take dasa in the sense of " decade " by Manu, vii, 116. In either case, however, it is impossible from what can be read to say how this total was arrived at. Much of the earlier part is illegible, including verse 3 dealing with the first reign. Then follow

1 [In v. 49a the poet may possibly have meant to write °maya naikah., and in v. 51a rajatan naikan. But in v. 46b, even if we correct to naika, we cannot save him from the reproach of barbarism.]

2 [See notes in loco and translation.]

five reigns of 120 years each ; the only certain names of kings are the fourth and fifth, Bahubalin and Raghupati. The next king appears also to have reigned 120 years. We then probably leave the region of myth with Candrodaya, who is said to have reigned twenty-seven years, and it is tempting to equate him with Candrasiirya of the Arakan Chronicles, which give his accession date as A.D. 146; if the length of reigns in the inscription is to be trustedand it should be in my view, at least from the Candra dynasty onwards-the date of his accession ought on the chronology accepted here to fall in the last quarter of the second century. The Chronicles may have preserved some recollection of fact at this point, but the names of Candrasnrya's successors, all beginning with Surya and reigning till 788, are clearly mythical. After Candrodaya the Annaveta kings reigned for five years; the name suggests indigenous rulers. The name of the next king is lost, and he is said to have reigned for the improbable period of seventy-seven years. Of the following names some are doubtful readings, and several are un-Indian, the list running
Rimbhyappa (?), 23 years.
Knverami or Knvera, a queen, 7 years.
Umavirya (?), husband of the preceding, 20 years.
Jugna (?), 7 years.
Lafnki, 2 years.1
If we have at present no means of checking the historicity of this list, the case is quite different with the Candra dynasty, the coins of six of them having been found as described in the Appendix, and it is to be hoped that when excavation is regularly undertaken in Arakan, the series will be completed. The founder of the dynasty had the curious name of Dven Candra, which perhaps survives in the word Taing, prefixed in the Chronicles' account to the name Candra for the first nine kings of the line, as its Sanskrit equivalent would apparently be Tuin. The Sanskritic names of all the kings, except 'the first, and the scripts used on their coins suggest that they maintained close contact with India. On the other hand the type of the coinage has no parallel among Indian coins, and the only changes that it shows are of minute details. The first king is recorded in the inscription as embellishing and fortifying the capital; but no name is given to it. If we are to follow the Chronicles, its Indian name was Vaisali, which is perhaps suggestive of the part of India from which the ruling family came. The probable date for the beginning of the dynasty is between A.D. 330 and 360; it may be therefore that we are dealing with a family of adventurers who left North Bihar when the Guptas had finally established their dominion there. The list of the kings is as follows:

1. Dvefn Candra, 55 years. He is said to have conquered 101 kings.
2. Rajacandra, 20 years.
3. Kalacandra, 9 years. Taw Sein Ko reported the existence in the Phayre

1 [See notes below.]

Coin Cabinet of a coin bearing this name (Ann. Rep. A.S.I., 1910-11, p. 92). The name is given as No. 5 of the Chronicles' list for this dynasty, where also he is said to have reigned 9 years.
4. Devacandra, 22 years. Coins of this king, inscribed Deva (misread by Phayre as Dama), are extant.
5. Yajiiacandra, 7 years.
6. Candrabandhu, 6 years. The name suggests some doubts about his legitimacy.
7. Bhnmicandra, 7 years.
8. Bhnticandra, 24 years.
9. Niticandra, 55 years. The coins of this king, inscribed Niticandra on the large coins and Niti on the smaller ones, occur more frequently than those of any other king. Probably he was the most powerful king of the dynasty.
10. Viryacandra, 3 years. On the coins his name is given as Vira.
11. Priticandra, 12 years. Coins of this and the next two kings exist.
12. Prthvicandra, 7 years.
13. Dhrticandra, 3 years.

The dynasty of thirteen kings thus lasted for 230 years, and though the only kings who can be tentatively identified in the account given by the Chronicles are Dven Candra and Kalacandra, yet we find there the same length given to the dynasty, 230 years, from 788-1018; twelve kings are named, but the title Candra is not given to the last three, in whom is possibly preserved a muddled recollection of later kings. It would seem that the Chronicles derived ultimately from an authentic list, which- has survived in a form corrupted beyond hope of restoration.
The inscription suggests that after the fall of the Candra dynasty conditions were confused in Arakan, with the rule reverting partly to indigenous kings. First comes Mahavira, king of Pureppura, who ruled for twelve years. The name of the town is of great interest, as it appears to give us the correct form of a place mentioned in the Pali Niddesa. The passages in question were discussed by S. Levi in an article in Etudes Asiatiques, vol. ii, pp. 1-55. Mahaniddesa, pp. 154 and 415, has a list of places, part of which Levi successfully identified by comparison with Ptolemy on the coast of Burma. One of these places, which he could not identify (p. 25), is given in the edition according to the Sinhalese MSS. as Naranapnra, 1 but the Burmese MSS. read Purapura and Parapura, and the Siamese have Parammukha and Parapura. It can hardly be doubted that this place is that called Pureppura in this inscription, and the retention by the Burmese MSS. of a form so close to the later name reinforces the view that insufficient value is often given by editors of Pali texts to the readings of Burmese MSS.2 The best authenticated form of the variants is Parapura, and that this was the real form of the name in earlier

1 Neither this name nor its variants are recorded in Malalasekhara, Pali Proper Names Dictionary.
2 For another case, also a proper name, see JRAS., 1939, p. 225, n. 2.

days is suggested by a comparison with Ptolemy. Levi (p. 22) took the view that the name of Ptolemy's river Katabeda survives in the name of the island of Kutabdia. Immediately below this on the coast Ptolemy places a centre of commerce called Barakura, which may reasonably be equated with Parapura ; an exact identification of the town is a matter for the excavator, who should look for a site between Akyab and Kutabdia (what about Pruma ?) with remains going back to the beginning of the Christian era.
Mahavira was succeeded by two kings whose names indicate a non-Indian origin, Vrayajap (or Brayajap), twelve years, and Seviinrein (?), twelve years. The latter is accorded the curious epithet of Mavukaghatin, for which a conclusive explanation is not forthcoming. I would suggest that Mavuka is a term of kinship and indicates the previous king, Vrayajap. The next king is called Dharmasura ; he reigned for thirteen years. There is no trace of any of these four kings or of their successors down to Anandacandra in the Chronicles, unless they have some connection with the last three kings of Vaisali, who do not bear the name of Candra. Or it may be that Vaisali ceased to be the capital of a large kingdom after the collapse of the Candra dynasty and that the later kings were merely petty local lords, the stronger of who asserted their right to issue coinage. In any case the Chronicles had allotted so much space to a fictitious dynasty of Dhanyavati that they had to place the Candra kings more than four centuries too late and left themselves no room for the later rulers.
The next king, Vajrasakti, sixteen years, is,the first of the family to which Anandacandra belonged. He is described as originating in the deva family, which is more fully described with reference to Anandacandra in verse 62 as the devandaja family, and in verse 63 as the Soridharmardjdndaja family. The exact meaning of these terms is far from clear, but it is natural to refer the last-mentioned one -to the egg of Brahma, and to deduce an origin for the family from Brahma and Manu, the latter being the traditional progenitor of the ten lines of kings. The epithets then would do no more than claim, a pure Ksatriya origin for the dynasty. 1 The word ddnasilddisamyukta recalls the six Paramitas and suggests that Vajrasakti was a follower of the Mahayana. Vajrasakti was succeeded by gridharmavijaya, 2 who ruled for thirty-six years; the coins with the mutilated inscription rmmavijaya should evidently be ascribed to him. He also is shown to have been a Buddhist by the allusion to his reverence for the Three Jewels in verse 40. The last line of this verse records that after death he went to the Tusita heaven, and possibly we ought to see here a suggestion that the king was a Bodhisattva incarnate; this idea recurs frequently in the Buddhist countries to the east of India. 3 The next king was Narendravijaya, son of the last, who ruled only two years and nine months and was succeeded by a son of Vajrasakti. Verse 42 leaves his name uncertain;

1 [This is an error. Andaja = Khacara = " bird ", and Andaja-varhsa is synonymous with Khacara-vamsa or Jimutavdhananvaya, a race on whom see E.I., xix, pp. 179 ff.]
2 [.Sri is only a prefix : the real name is Dharmavijaya.]
3 Cf. La Vallee Poussin, Melanges ch. et b., i, p. 378.

it might be either Viranarendracandra or Sridharmacandra.1 The latter, however, is indicated by the coins labelled Dhammacandra. The Prakritic form Dhamma is odd ; no sign of a superscript r is traceable, but Candra appearing in Sanskritic form does not correspond to Dhamma. He reigned sixteen years and died after nominating his son Anandacandra as his successor. The rest of the inscription recounts the virtues and good deeds of this king up to the ninth year of his reign, and presents a number of interesting points. He was evidently a Buddhist by personal religion and calls himself an upasaka in verse 54, but following the Indian tradition of religious impartiality he did not neglect the Brahmans in his display of liberality. Moreover the references to Bodhisattvas in verse 47 and to danapdramitd in verse 54 show that he was a follower of the Mahayana, as was clearly the case also with his grandfather Vajrasakti and with Sr! Dharmavijaya. This confirms the evidence of the bell-inscription that the Mahayana prevailed in North Arakan ; incidentally it should be noted that no information on this point is to be obtained from the Chinese pilgrims Hsuan-tsang and I-tsing. The orthography of the purely religious inscriptions suggests that this form of religion had been introduced from Eastern Bengal, and possibly the reference to the Tusita heaven, where Maitreya resides, in verse 40 should be held to show the influence in this area of the Vijnanavada school, who were especially devoted to him. The mention of Cunda and the rest in verse 47 probably refers to the chief srdvakas ; while Cunda is, significantly enough, omitted from the list in Anguttara Nikaya, i, pp. 23-6, he appears in Asvaghosa's list in Saundarananda, xvi, 91. Of the schools which know three or four chief Sthaviras, only the Sarvastivada includes Cunda in this group, 2 thus assigning him a place which justifies the expression of our text. It seems then that the Mahayana in Arakan was represented either by Mahayanist Sarvastivadins 3 or by a Mahayanist school which derived ultimately from that sect, such as the Vijnanavadins, who took from it much of their dogmatics. Anandacandra also followed a practice, which is well known from further east, in giving his own name to new foundations: thus the Anandodaya vihdras of verse 46, and the Anandamddhava and Anandesvara mathas of verse 56. The inscription recounts at length all the different materials of which he had Buddha images made; among these is a reference to countless images of clay, presumably of the same type as the votive plaques which have been found in such numbers in Burma. Verse 52 is not clear because the essential word, tandaka, is not known to have any meaning suitable to the context; the reference is clearly to something, in the course of which or after which there was religious teaching, and this suggests either some form of theatrical representation or religious dance, or else feasts given to the religious community. 4 The names of several places at which religious buildings were erected or tanks dug are mentioned
1 [Or rather, Dharmacandra.]
2 Przyluski, Le Concile de Rajagrha, p. 302.
3 Of. Przyluski's discussion of this point, op. cit., pp. 362-5.
4 [See, however, note in loco and translation.] VOL. XI. PART 2. 24

in verses, 56-9, which are mostly non-Sanskritic in form and presumably unidentifiable nowadays.
Two kingdoms are mentioned with which Anandacandra had relations. Verse 61 records gifts to the bhiksus in the country of king gilamegha, who is otherwise unknown, 1 and verses 62 to 65 describe twice his marriage to Dhenda, daughter of a ruler who is said to be king of Sritamrapattana in 62, but of the land of Sripattana in 65 ; his family is named as gaivandhra, possibly suggestive of an origin in the Deccan. 2 The manner in which this marriage is mentioned shows that it was a matter of exceptional importance, and makes it therefore possible that the. king of griksetra is meant ; but a definite identification cannot be put forward at present. Alternatively, is Ptolemy's Sambra a mistake
for Tambra, and if so, does it indicate gritamrapattana ?

Immediately below the inscription are two lines in a later hand, more or less contemporary. with that on the north face of the pillar. Whether they are
intended to have any connection with Anandacandra's prasasti is not clear. The form ekddasama is odd ; but similar forms are found in the other late inscriptions, and it possibly shows Prakrit influence.3

After Anandacandra there is a gap of at least two centuries before the inscriptions on the north face of the pillar, and the only material that falls within this period consists of the four coins already mentioned, formerly in Mr. P. N. Tagore's cabinet, which may follow fairly close on Anandacandra, and of the coin, No. 22 of Plate V, which shows a different type of script.
Unfortunately it cannot be read with certainty : Phayre (Coins of Arakan, Pl. II, No. 10) suggested Yarikriya, and V. Smith (Catalogue of Coins in the Indian Museum, Pl. XXX, 11) Yarikriya ; but the first letter cannot possibly be Yd. The next two letters may well be rikri and the last may be ya, as in my plate, or ma, as in V. Smith's reproduction. Till a better specimen is found it is best to leave the reading as an unsolved problem, and not to suggest any date for it. From the evidence, however, it does look as if there were a gap in the kings of Arakan at some time in this period, which may have been due to Nan-chao domination ; but this is not a point on which I am competent
to express a definite opinion.

Turning now to the remaining inscriptions, that on the top of the pillar has very few letters at all visible, and all that can be said of it safely is that the script is undoubtedly late. On the north face below the four lines of the earlier inscription, there is a series of inscriptions, amounting to sixty-nine

1 [This name seems to point to Ceylon, where Sildmeghavanna (in Sinhalese Salamevan) was a title borne by several kings.]

2 [See the translation below. The .4ri prefixed to the name of the city is unessential : the name is Tamrapattana, which conceivably may be Tamralipti. griksetra seems to be out of the question.]

3 Cf. Pischel, § 449, who says these forms are only authenticated so far in Jaina Prakrit
works. For other instances note the apparatus criticus of the colophons to Saundarananda, xi and xviii, and Varangacarita (ed. A. N. Upadhye, Bombay, 1938), colophons tQ_ xi, xii, and xiii.

lines, when counted on the right side. Parts of them are entirely gone; the rest is written in a slipshod hand, sufficiently rubbed to make reading of it chancy. There appear to be at least three inscriptions, all in much the same script. The first occupies the first seven lines, only the last few letters of each line being at all legible. The third line ends caturddasame, the fourth Simgha(or ha)vikramasura, and the last word of the inscription is krtarajyah. The next inscription begins in 1.8 with svasti sri and then apparently the name of a town, which with great reserve might be read as Avayapura. Line 9 has a proper name, the certain letters being tisuracandra(m ?) de . thara ; earlier in the line the remains suggest ,Sri Simghag(?)anap(? )a before ti. That we are dealing with another inscription about Buddhists appears from 1. 10, where can be read samghagata b(?)auddha, preceded perhaps by aryya. In the next line we have, certain but not very clear, caturddasame bde krtardjya, and in 1. 12 Siha(or gha)vikramasuracandr(aya vi ?). There is a certain parallelism in phrasing therefore between the two inscriptions, but the second is slightly fuller than the first. The remains are too small for exact inferences to be drawn; it is not clear if the phrase “in the fourteenth year ", in which the same irregular ordinal occurs as at the foot of Anandacandra's inscription, applies to Simghavikramasnracandra or his predecessor, Sirhghaganapatisnracandra. From this point very little can be read with certainty, and it is not clear where the second inscription ends. That a new inscription does begin some lines later seems a reasonable inference from the fact that from 1. 23, if not earlier, the writing is in three columns, each containing seven or eight characters in a line; the arrangement is perhaps clearest in 11. 34 and 35, where vinmukhikr ends 34 and tasatruh, the rest of the compound, is to be found in the letters seventh to fifth from the end in 35. The division of the text was not accurately judged, with the consequence that the right-hand column continues for five lines below the other two. A competent epigraphist, working on the stone instead of on rubbings, should be able to produce a fair reading of the right hand column from 1. 30, and of the other two from about 1.40. The contents of this last inscription may be somewhat unusual, as I read in lines of the left-hand column lupati I idam mayor krtam. In these circumstances all that can be deduced for the present from these inscriptions is that North Arakan again produced in the tenth century a dynasty of some importance, whose personal names ended in 95racandra, and two of whom were probably called Simghaganapatisnracandra and Simghavikramasnracandra.

A. Inscription on Western Face of Pillar at the Shitthaung Pagoda, Mrohaung, Arakan

Note.-Doubtful letters are enclosed in square brackets, and letters which cannot be read are replaced by a dot. Omitted characters, when restored, are shown in angular brackets : thus, < >. For convenience of reference, while the text is arranged and numbered marginally according to the verses, the line of the inscription is also noted in brackets in the body of the text. [To mark the elision of initial a- after a final -o a raised comma has been used, e.g. 'tinitiman (v. 23d).]

[Verses 1-3 are not transcribed.]

4…. (4) tato ri[sya ?] d…. jagata …..|
……..1 bhupalo varsa[m] viiinsadhikam satam ||
5. [.arvaryavi ?] 2 mahipalo (5) lokanugrahatatparah |
Rajyam tena krtam [tasm]ad varsavirinsottaram satam ||
6…. . nama tato raja lok . . jani ... (6) vat |3
Cakara4 ... rajyam varsa[ril] vimsadhikam satam||
7. Tasmad Bahubali bhnbhrt [pu]nar dhairyyavisaradah
Krtam ca krti- (7) na tena rajyam virnsa[bd]ikarn5 satam ||
8. Tato Raghupati[r]6 bhupah surupo nitivi[kramah]|
[Cakara] ... [tam] rajyam va[rsa]- (8) vimsottaraiii satam ||
9. Ta[sya] vi 7 ... amra[tya]po 8 mahabalah |
Vimsabda[ny] ... . [rajyam] ... . kr[tam] ||
10. (9) Tatas Candrodayo nama bhupalah sadhusammatah |
Saptavimsati varsani rajyam atmasatkrtam 9 ||
11. Annaveta- (10) mahipala danam datva tv anekadha |
Bhuvi lokasukham jiiatva [pamcabdani] 10 divam gatdh ||
12. Tatpascan nrpavara ... (11) caryasu visaradah |
Cakara mati[mam] rajyam abdani saptasaptatih ||
13. Rimbhyappo bhupatis tasmac caran dana ... (12) tih |
Tryadhikam vimsad abdani rajyam punyena nitavan||
14. Kuverami 11 tato devi danasila ... . |
(13) Saptabdani tato rajyam cakararivivarjitam 12 ||
15. U[mavirya]patis 13 tasyas tato bhupo 'tiniti
(14) jyam vimsati varsani cakara mahimakrti 14||
16. J[u]gnah[vayas 15. bhu]bhrt sarvasatvahitarthakrt |
Sa[pta] (15) tsarany evaril tads rajye pratisthitah ||

1 The name looks like Narappagmasva.
2 [Possibly the true reading may be Purvarthe 'pi.]
8 [For vat Professor Johnston gives an alternative kam.]
4 [A possible reading here is cakararimtapo. There is no clear trace of a long vowel after the first r, and ri would be a mistake for ri.]
5 Alternatively the reading in d is vimiadhikam, in which case there is no word for "year”.
6 Presumably read Raghupatir.
7 [Traces of the eight letters of the first pada survive. The second is ta, the third perhaps sya, the fourth vi, the last two apparently devah.]
8 [The first syllable of this pada is possibly vam, the second perhaps ka.]
9 [As the rubbing shows, the true reading is definitely tenatmasatkrtarh.]
10 [There seems to be no trace of a vowel i on the rubbing.]
11 [Possibly Knverapi.] 12 Read °vivarjitam in d.
13 [The first two letters of this pada look more like Orhppa-. Possibly, too, the stone-cutter has omitted a visarga before patis.
14 Mahimdkrti is an odd compound. [Comparison with verse 42d (see note on latter) suggests that the poet wrongly took mahima as a fem. vowel-stem.]
15 If the pillar has been correctly read in a, it should read Jugnahvayas. [It has Jugnahvayas tato bhnbhrt.]

17. Lankinama 16 tato raja krtva varsadvayarh krti |
Rajyarh v[ipu] 17 (16) virah kramena tridiv[ah]itah 18 ||
18. Kathyate varsasaThkhyatra devanarir kulakartrnam |
Etesarii bhubhrtam n[u]n[arn] (17) sahasraih saddasadhikam 19||
19. Tatpascad apare We punyalaksmiyuto bali |
Dven Candranamako dhimarh (18) yo 'bhnt bhubhrtam patih 20||
20. Nrpaikottarasatarir jitva punyato bahusalina
Prakarakhatasarhyuktarh (19) nagara[bhnsa]narh krtam 21 ||
21. Tena nispadya nagararh svarggasaundaryahasinam|
Parircaparircasad abda- (20) ni krtarir rajyam yasasvina 22 ||
22. Rajacandras tatah srima- it vimsavarsani rajyakrt |
Evarir svargasukharrt (21) jnatva divam yato mahipatih ||
23. Tasman navabdiko raja [K]alacandro 23 mahardhikah |
Krtva kirttimayI (22) ma[l]arir svargaih yato 'tinitiman ||
24. Devendreva Sa[kr]o 24 'bhnd Devacandro mahipatih |
Tato dvavirirsavarsani (23) rajyam krtva to svargabhak ||
25. Saptavarsikas tasmad 25 Yajnacandrah prakirttitah |
Candrabandhus tato loke satsa-(24)mvatsararajyabhak ||
26. Prthivyam uditas candro Bhumicandras 26 tato 'parah |
Sapta samvatsarany eva rajyam punyena (25) to nitavan 27 |
27. Caturvitsati varsani rajyam sambhujya nitiman |
Bhuticandras tato yato divyam sukham avaptaye||
28. (26) Niticandras tatah khyato nityutsaritavigrahah |
Pamcapamcasad abdani so 'bhnd raja Mahendravat ||
29. Abdatra-(27)yikas tasmad 28 Viryacandro naresvarah |
[Ta]to dvadasa varsani Priticandro mahipatih||
30. Saptarhvatsarany a-(28)smat Prthvicandrena bhubhuja |
Rajyopabhogarh sarnbhuktarir nityam dharmanuvarttina ||

16 [The rubbing seems to give Linki°. The first vowel is a short curve above the 1 to the right, somewhat like the i in sphita° in the inscription of YaAovarman, line 12 (E.I., xx, p. 43.)]
17 [The stone seems to have prapa-.]
18 In d possibly tridivan gitala ; in any case read tridivan gatala.
19 Saddasadhikaoh presumably means "+ 60 ", not " + 16 ". In either case it is not clear how the total is made up from what can be read of the inscription.
20 The reading in b looks like yo bhut bhu°, and it is not certain what the correct reading is.
[The stone certainly has yo bhut bhu°, but a syllable is lacking to make up the metre.]
21 Pada a is hypermetric. In d the consonant fourth letter is either bh or s, and the next syllable should contain either r or s because of the following na, but looks more like sva than anything else. The reading adopted seems the only possible one, though abhusana in this sense is unrecorded. [The rubbing is in favour of reading nagarasutranaiix, which may be an error for nagarasutranaoh.]
22 Read yasasvina in d.
23 The k in Kalacandro is badly formed, but the reading is definitely not Balacandro; Ralacandro is just possible.
24 Read Devendra iva Sakro. 25 [A syllable is lacking in this pada.]
26 Read Bhumicandras.
27 Line 26 begins tanitavan, but to is marked above for erasure. [Line 26 begins naaitavIcdi.]
28 [This pada is a syllable short.]

31. Jagaddhrtirii karoty a[sma] Dhr-(29)ticandro naradhipah |
Prajam apalayat tasmat trbhir 29 varsair divath gatah||
32. Isanvayaprabhavarh sodasa (30) bhupatinam
Candrabhiramayasasam 30 iha Candranamnam|
Trirnsa[dh]ikarh praganitani satadvayam 31 syad
Varsani rajya9ubha-(31)bhogakrtani nunam||
33. Tatah pascan Mahavirah Pureppuranaresvarah 32|
Tena dvadasavarsani dharmarajyarh (32) krtaiii tada ||
34. Vrayajapnamapi [so] 33 raja dvadasabdani bhnvibhuh 34|
Bhuktva rajyasukhaiii viras tata svargopa-(33)bhogabhak||
35. [Sevirirern] bhupatis tasmat smrto 35 dvadasakarsikah 36
Rajyasampatsukharir tena bhuktam Mavukaghatina 37 ||
36. (34) Ksitirh raraksa dharmena Dharmasuras tato nrpah |
Trayodasabdasaihpurnne svargam ya[t]o maharddhikah ||
37. Bhaktiman iva bhaktya (35) vai 38 yo vajriva mahibhrtdih |
Vajrasaktis tata 39 [kh]yato raja devanvayodbhavah ||
38. Pratipalya jagat sarvath rajyaiii so-(36)dasa 40 vatsaram
Danasiladisaihyukto devalokaih sa yatavan ||
39. Sridharmajayasarnyukto lokanugrahatatparah I
(37) Tatpascad abhavad dhirah Sridharmavijayo nrpah ||
40. Sattririrsad abdany upabhujya raj yam
Dharmena nitya ca jayena caiva
Ratna-(38)trayanusmaranabhiyogat
Sa devalokarii Tusitaih prayatah 41 ||
41. Narendravij ayenapi tatputrena mahipat 42|
(39) Navamasadhikaih rajyaih bhuktam varsadvayaih sata ||

29 Read tribhir.
30 A character has been erased after isanvaya. In a read Isanvaya prabhavatrayodasabhupa-tindm, and note a short before tr. In b read °yaaasam. [We should read °sodasa. The emendation -trayodaga- would gratuitously introduce a short syllable before tr- and make the pada a syllable too long. The better course is to read sodasa and risk the possibility that the author's reckoning was wrong; moreover, he may have intentionally omitted the names of some kings who were too insignificant for mention.]
31 [Read satadvayam.]
32 Just possibly one should read Purempura-, instead of Pureppura-. [The rubbing definitely gives Purempura-.]
33The inscription in a has Vrayajapnamapi so, with marks above pi to cut it out. It should probably read Vrayajapnamako. [So would be ungrammatical. Read °ndmdpi yo rdjd. The pada is hypermetric.]
34 [The rubbing gives bhovibhula.]
35 [In a the name may be read as Dovinren. In b the letters on the stone may be read as either smrto or smrta ; they should be smrto.]
36 Read °varsikah.
37 lllavuka may be a proper name or a word indicating kinship.
38 Possibly yo, not vai.
39 Read tatah. 40 Read sodasa.
41 [The stone has praydt, with a final t.]
42 Mahimat. is also possible, and there may be a r or u below p or m. One would expect mahibhrta. [The rubbing gives mahipat-, with possibly faint traces of -eh.]

42. Tsanvayah samabhavad vijitarivargah
Saktitrayapra 43-(4O)ya1abdhamahapratapah|
Yo Vajrasaktisutaviranarendracandrah
Sridharmacandramahimaprathitaprabhavah 44 ||
43. Srima-(41)n sodasa 45 varsani bhuktva rajyasriyam nrpah|
Datva sutavare rajyat pascat svargath prayatavan ||
44. Yas tatsu-(42)tarn46 pranatabhupatimaulimala
Ratnadyutiprasarararnjitapadapadmah |
Anandacandra bhuvanaikaya9o-(43)'titunga
Anandayarn jayati vairitamovibhuma 47 ||
45. Dane Karnnasamo raja satyenapi Yudhisthirah |
(44) Pradyumnaiva 48 rupena tejasa bhanuvad bhuvi ||
Tena maharajadhirajena parahitotsukadhiya svara-(45)jyaprathamasamvatsaratah prabhrti yavad a navamabdat svakrtakaritanumoditani sucaritani sa-(46)tvanarn darsanaprabodhanumodanapunyavistaram icchata pravaksyamte
46. Anandodayanamano vihara-(47)neka karitah|
Dasadasibhih sampannah ksetragomahisaih saha ||
47. Sugatabodhisatvanarir (48) Cundadinarh ca saktitah|
Pratimadhatumaccaityah karita raukmarajatah ||
48. Ritimayani bi-(49)mbani kansatamramayani 49 ca|
Karitani munindrasya bharasarnkhyapramanatah ||
49. [Pasa]daru-(50) mayaneka 50 pusta[s]ailas 51 tathaiva ca |
Sugatapratimah saumyah karitas [sa]dhucitritah 52 ||
50. [Mrtpa]. 53 (51) krtasaihkhyani bimbani caityakarmanah |
Saddharmapustakas capi lekhita bahusah sata ||
51. Sau-(52)varnnarajatanekan padman sadratnakarnnikan|
Nityarir sridhatupujartham adad bhupo 'tisraddhaya ||
52. (53) Dadau prahrstah suvisuddhacetasa 54

43 [At the end of line 39 there seems to be a faint trace of na.]
44 [See note on v. 15d above, p. 374. Apparently our poet treated O'ridharmacandra as a nominative. Cf. notes on v. 62c, p. 379, and on v. 64c, ib.
45 Read sodasa.
48 It should presumably be tatsutah.
47 [Apparently to be corrected to -vibhium¢.]
48 Read Pradyumna iva.
49 Read Kar sya°.
50 Pasadaru may be the name of a particular kind of wood, or it may be a compound implying images made of leather (?) and wood. [Read in a mayanekah or mays naikah (see above, p. 367 n.1).]
51 Pusta is presumably “plaster " here, and saila " stone ".
52 [The stone has karitasadhu°. Read karitah.]
53 The rubbing shows mrtpaha with the ha marked for omission; the following character is only faintly indicated on the rubbings. Perhaps nlrtpakva°. [The reading of the rubbing seems to be mrtsarhha°.]
54 [The rubbing seems to give suvisuddha° by error.]
Saddharmapujam prati tandakan 55 bahun |
Dine dine sarvajana-(54)numoditan
Naradhipo dharmakathanuragatah ||
53. Lauhapatrany anekani sannetracivarani 56 ca |
Na-(55)nade6agatanam ca bhiksunam gauravad dadau ||
54. Danaparamita hina ma me bhavatu jantusu|
Tasmad upasa-(56)kenapi sarvasatvahitesina 57 ||
55. Patncasadbrahmanavasaih ksetrabhrtyasamanvitath|
Vadyavadakasamyuktam ka-(57)ritaih mathacatustayam ||
56. Somatirthadvijavase mathas canandamadhavah|
Anandesvaranamapi 58 (58) Naulakk[e] ca matha smrtah 59 ||
57. Pilakkavanak[u]hve 'pi Domaghe purvanamakau 60 |
Vithika vividharam 61 (59) karita setusamkrama 62 |
58. Pratyaham bhaktasalayarn sada satrarh pravarttitam|
Ativadhyas ca karunya- 63(60)t pranino mocitasadah 64 ||
59. Da[ih]kanga margafiga d[u]varasamjnite
Bhurokanaulakkalavarakahvaye |
Pundi-(61)ngasomahvayasarnghasamjnikau
Manapavapy[au] 65 nijakhana bhupatih 66 ||
60. Purvarajakrta ye'pi devapra-(62)sadatirthika 67|
Nasta nispaditas tena sarvatha dhimata punch ||
61. Dharmasanarh hastiniko-(63)ttamaika
Bhupena netrojvalacivarani |
Bhiksvaryasamghasya hi na[y]itani 68
Dee gild-(64)meghanaradhipasya ||

55 None of the recorded meanings of tandaka fits here; possibly for tandavan. [The actual reading of the rubbing seems rather to be vantakan, though the letter below n is not clear. Vantaka, "share" (found in Sanskrit and Kanarese lexx. ; from -,/vant, whence Hindi bant) occurs in the sense of a holding or portion of land forming part of an estate in the Yadava Ramacandra's Thana grant of 8aka 1194 (E.I. xiii, p. 199).]
56 The first certain instance of netra in the sense of silk ; cf. Raghuvamsa, vii, 36.
57 Read °hitaisina.
58 [The p has been almost entirely out out.]
59 [Read mathah.] Apparently these two mathas are in addition to the four of the previous verse.
60 In a presumably read °vanakahve, and in b read °namake. [The reading of the rubbing is possibly °vadakuhve and Daumaghe.]
61 [The -a of the last syllable is fairly certain ; and the stone-cutter probably did not add -h to it.]
62 Samkrama, feminine, is odd ; or should it be °sarhkramah ? [The latter alternative is preferable.]
63 -A- is missing through breakage of stone.
64 Read sada, as mocitasadah as a compound is hardly possible. [Also read mocitale.]
65 [The rubbing gives -vapyo.]
66 Read nicakhdna.
67 Read °tirthikah in b. [Tirthikah can only mean .1 heretics ". Probably the poet meant to write tirthakdh, in the sense of tirthah.]
68 I do not know. if hastinika is to be taken literally. In b read netrojjvala°. Nayitani in c is odd, and dapitani would be better.


62. Vikhyatasauryagunadharmayasonuragad |
Devandaja-(65)nvayasamudbhavaparthivasya|
Sritamrapattananaradhipa bhaktinamro 69
Dhenda[rh] dadau sva-(66)tanayam paramadarena |
Apararin ca |
63. Anandacandraksitiparthivasya
Sridharma (67)rajandajavarhsajasya|
Srutva vaco dharmahitarthayuktam|
Sauryanvayatyagaguna[dh]i-(68)kam ca ||
64. Bhaktipranamena prakurvatajfiam
Saivandhravamsodbhavaparthivena|
Srima . . 70 (69)nodhiramaharddhikena
Kalyanamitratvam upagatena ||
65. Vapiviharau tvaritena (70) krtva
Sripattanatmiyamahipradese |
Striratnadhenda svasutatibhaktya
Sarhpresite-(71)hasamabhutiyukta|| -iti 71
(line 72) sri [ki]rttisarhpu[rna] vijaya
(line 73) ekddasame 'bde
[TRANSLATION
(Verse 4) ... the king [reigned] 120 years. (V. 5) [There was] a king… zealous in doing kindness to the world; he reigned afterwards for 120 years. (V. 6) Then ... a king named... reigned for 120 years. (V.7) After him again [was] King Bahubali, eminent for stoutheartedness ; that able man ruled for 120 years. (V. 8) The king Raghupati, fair of form, heroic in policy, reigned ... 120 years. (V. 9) His... puissant ... reigned [1]20 [years]. (V. 10) After him [was] a king, Candrodaya by name, approved by the good; he held the kingship for 27 years. (V. 11) The Annaveta kings, bestowing bounty in manifold wise, after experiencing worldly pleasure on earth for 5 years went to heaven. (V.12) After that an excellent king ... eminent in religious practices, possessing wisdom, reigned for 77 years. (V.13) After him, King Rimbhyappa, practising bounty reigned in righteousness for 23 years.

69 One must understand that °naradhipo bhaktinamro, which is impossible metrically, is indicated.
70 No doubt Srimanmano°. [Apparently Manodhira is the king's name (see translation, below). This name is rare; but it was borne by, e.g. the composer of the Veliirpalaiyam plates.]
71 The letter before iti is not clear, but is similar to the character found in the same position in the bell inscription. [How and where Professor Johnston found the word iti is not clear, for it is not on the rubbing. Anandacandra's inscription ends with the word bhutiyuktd followed by three double dandas, between the first pair of which there is a Garuaa-symbol (see Plate I), which is appropriate in the edict of a king claiming to belong to the "Bird-tribe ". Another example oi, this symbol occurs, e.g. in Govindacandra's Saheth-Maheth plate (E.I., vol. xi, pp. 20 ff.).]

(V.14) Then Queen Knverami (? Kiivera), [practising?] bounty and good deeds, for 7 years after him made the kingdom free from foes. (V. 15) Then Umavirya (?), her husband (?), a very politic king skilful in his majesty (?), ruled for 20 years. (V.16) After him a king named Jugna, who benefited all beings, was then so established. on the throne for 7 years. (V.17) Then the able king named Lank! (? Linki) after reigning 2 years, … a valiant man, in course of time went to heaven. (V. 18) Here is stated the number of years of the ancestral monarchs: [the number of years] of these kings verily is 1,016 (? 1,060). (V. 19) After them [there was] in later time one who possessed righteousness and fortune, puissant, sage, Dven Candra by name, who was a lord of kings. (V. 20) He, strong of arm because of righteousness, conquered 101 kings, and built a compact 1 (?) city furnished with walls and moat. (V. 21) He, possessing glory, having constructed the city, which laughed at the beauty of Paradise, reigned for 55 years. (V. 22) Then the fortunate Rajacandra reigned 20 years ; having thus known the pleasure of Paradise, the king went to heaven. (V. 23) After him, Kalacandra (?), a very prosperous and exceedingly politic king, who reigned 9 years, went to heaven after making [for himself] a garland of glory. (V. 24) Like Sakra the Lord of the Gods was King Devacandra, who then after reigning 22 years enjoyed heaven. (V. 25) After him Yajiiacandra was renowned, reigning 7 years. Then Candrabandhu a reign of 6 years in the world. second moon, on earth ; he reigned with righteousness for 7 years. (V. 26) Then arose Bhnmicandra, a (V.27) The politic Bhrticandra, after enjoying kingship for 24 years, then departed to gain celestial happiness. (V. 28) Then there was the renowned Niticandra, who removed strife by policy; he reigned like Mahendra for 55 years. (V. 29) After him King Viryacandra reigned for 3 years; then -King Priticandra [ruled] for 12 years. (V. 30) After him King Prthvicandra, constantly following religion, enjoyed the pleasures of kingship for 7 years. (V.31) King Dhrticandra after him supported the world ; he protected [his] people, and then after 3 years went to heaven. (V. 32) The years spent in happy enjoyment of kingship by the 16 monarchs sprung from the lineage of the Lord (Isa = Siva), who bore the name Candra and had glory delightful as the moon, when counted up will verily be 230. (V. 33) Afterwards Mahavira [was] king of Purempura (?); he then had a godly reign for 12 years. (V. 34) Also the king named Vrayajap, a valiant lord of the earth, after tasting the happiness of kingship for 12 years, thereupon enjoyed the pleasures of Paradise. (V. 35) After him King Sevilireli (?) is remembered as having reigned 12 years; slaying Mavuka (?), he enjoyed the happiness of prosperity in kingship. (V. 36) Then King Dharmasura protected the earth in accordance with religion; highly prosperous, on the completion of 13 years he went to Paradise. (V. 37) Then [was] the indeed devout famous king sprung from the gods' lineage, Vajrasakti, who because of devotion was like a Vajrin (Indra) among monarchs. (V. 38) Possessing bounty, virtue, and other [qualities], he went to the world of the gods after
1 See above, note on text.

protecting the whole universe in a reign of 16 years. (V.39) After him there was a brave king, the fortunate Dharmavijaya, attended by fortune, religion, and victory, zealous in doing kindness to the world. (V. 40) After enjoying kingship for 36 years because of religion, policy, and victory, zealous in doing kindness to the world. (V.40) After enjoying kingship for 36 years because of religion, policy, and victory, through practising remembrance of the Three Jewels he passed away to the Tusita heaven. (V. 41) That king's good son Narendravijaya enjoyed the kingship for 2 years and 9 months. (V. 42) There arose one belonging to the lineage of the Lord (Isa = Siva), a conqueror of troops of foemen, gaining great majesty by manifestation of the three powers,1 a moon of valiant kings, son of Vajrasakti, the fortunate Dharmacandra, having majestically illustrious puissance. (V. 43) The fortunate king, after enjoying a prosperous reign for 16 years, made over the kingdom to [his] excellent son, and afterwards passed away to Paradise. (V. 44) His son Anandacandra is victorious, having the lotus-flowers of his feet ruddied by the outpouring of gleams from gems in the garlands of diadems of reverently bowing monarchs, exceedingly lofty in glory unique on earth, causing gladness, potent over (?) the darkness of foemen. (V.45) [He is] a king equal to Karna in bounty, likewise a Yudhisthira in truthfulness, like Pradyumna in beauty, in splendour like the Sun on earth.
The good deeds done by that Emperor, whose thought yearned for the welfare of others, from the first year of his reign until the ninth year, whether done by himself or caused to be done and approved [by him], because he desired for living beings an abundance of merit through enlightenment of vision and acceptance, will be declared.
(V.46) There have been built many monasteries named Anandodaya, provided with men and women slaves, together with lands, kine, and buffaloes. (V.47) There have been made golden and silver chapels containing images and relics of the Buddhas and Bodhisattvas and of Cunda and others according to power. (V.48) There have been made images of the Lord of Sages (Buddha) composed of brass, bell-metal, and copper according to the measure of weight and number. (V.49) There have been made many pleasing [and] well decorated effigies of the Buddha composed of wood, plaster, and stone. (V. 50) Innumerable (?) clay ... effigies of a chapel-structure [have been made], and also books of the Holy Law have been caused to be written by the good [king] in large numbers. (V.51) The king with exceeding faith has constantly given for the purpose of worship of the blessed relics many lotus-flowers made of gold and silver and having pericarps of goodly gems. (V. 52) The king, rejoicing with very pure spirit, because of his delight in religious discourses bestowed day after day many shares [in land-estates ?] approved by all people for the purpose of the worship of the Holy Law. (V. 53) He has out of reverence given many copper bowls and robes of good silk (?) to friars coming from divers places. (Vv. 54-5) "Let not the perfection of bounty towards creatures fail me”: [with this intention] therefore he, seeking the welfare of all beings, though he was only a lay-worshipper, caused to be built four monasteries

1 Viz. lordship, counsel, and enterprise.

lodging 50 Brahmans, provided with lands and servants, furnished with musical instruments and musicians. (V. 56) The monastery [named] Anandamddhava at the residence of the Brahmans of Somatirtha and also the monastery called Anandesvara at Naulakka are recorded. (V. 57) At [the place] called Pilakkavanaka, formerly named Domagha (?), also there have been constructed streets, various pleasances, causeways and passages. (V. 58) Every day a session has constantly been carried on in the dining-hall; and because of his mercifulness capital offenders have always been released. (V. 59) At [the place] styled Dankangamargaingaduvara (?) [and] that named Bhnrokanaulakkalavaraka (?) the king has dug two delightful wells entitled Pundiinga and Somasafngha (?). (V. 60) Gods' temples and holy places built by former kings which had perished have also been completely restored by this wise [king]. (V. 61) A pulpit, an excellent cow-elephant, [and] brilliant robes of silk [?] have been dispatched by the king to the noble congregation of friars in the land of King Silamegha. (V. 62) From love for the renowned quality of valour, religion, and fame of the monarch sprung from the divine Bird-lineage, the king of the fortunate Tamrapattana, making devout obeisance, gave [to him] with the highest respect his daughter Dhenda.
Moreover:-
(Vv.63-4) On hearing the speech of Anandacandra monarch of the earth, scion of the Bird-lineage of fortunate righteous kings-[speech] fraught with meaning helpful to religion and abundantly marked by the qualities of valour, [high] descent, and bounty-the king sprung from the Saiva-Andhra lineage, the fortunate highly prosperous Manodhira,l fulfilling his command with devout obeisance, entered into happy friendship [with him]. (V. 65) Having promptly made a well and a monastery in the district belonging to his fortunate city, he sent here with extreme devotion his daughter Dhenda, a gem among women, provided with peerless ornaments.]

B. Inscription on Bell from Vesali

This is a votive inscription on a small bell found at Vesali, near Mrohaung, Akyab district, Arakan, and now preserved at Akyab, in the keeping of U San Shwe Bu, Hon. Archaeological Officer for Arakan.
1. deyadharmmo `yam Sakyabhikso ... yac atra 2 punyam tad bhavatu matapitfpurvvaingamam krtva
2. caryyopadhyayanam sarvvasatvanan ca anuttarajnanavaptaye ti 3
[Translation.-" This is a pious offering of the Buddhist friar ... May the merit that is therein be for the gaining of supreme knowledge by teachers, tutors, and all beings, in company with [his] mother and father."]

1 It seems necessary to take manodhira thus as a proper name, regarding maharddhikena as qualifying it (cf. narddhipa bhaktinamro, above, verse 62).
2 [Read yac cdtra.]
3 [The letter before ti is certainly i, though of an unusual type, resembling a u. A similar letter occurs in Incr. C below, line 4.]
C. Inscription on Stone from Sandoway

This is a votive inscription on a stone obtained by Colonel G. E. Fryer in January, 1872, in the cavity of a hill near Ngalunmaw, Kwelu circle, Sandoway district, Arakan (vide Proc. As. Soc. Bengal, 1879, p. 201), and now on loan from the Royal Asiatic Society at the Museum of the Indian Institute, Oxford.

1. ye 1 dharmma hetuprabhava hetu
2. tesam Tathagata by avocat 2 tesaii ca yo
3. nirotha evarimvadi mahasrana ||
4. upasak[a] 3 Ma[i]ga upasa
5. k. [Sa]koma[vamma]m akara
6. yi = matapitaku
7. Sala ||

[Translation.-" The Buddha has declared how the consciousness-moments arising from causes are caused, and how they are to be suppressed : thus spake the Great Ascetic. The lay-worshipper Maiga (?) [and] the lay-worshipper Sakomavamma (?) caused to be made [this object] for the welfare of [their] mother and father."]

PART II. EARLY COINS OF THE KINGS OF ARAKAN


Only small numbers of early Arakanese coins in situ have been discovered hitherto, and the best collection is in the British Museum, a representative selection of whose coins is reproduced in Plate V by permission of the Department of Coins and Medals. The majority of them, figs. 5-21, are of the same general type : obverse a humped bull lying down with the name of the king above, placed in a circle with a row of beads outside ; reverse a pattern, of which more later, with the sun and moon above, again in a circle with a row of beads outside. The bull, except in figs. 5 and 20, has a row of beads round its neck, the number varying for each king. Similarly there is a line of a varying number of beads below the pattern on the reverse. The shape of the pattern on the reverse shows a continuous development which, even without the dynastic list in Anandacandra's prasasti, would enable them to be placed in chronological order. This list, palaeographic considerations and the details alike prove Deva's coin (fig. 5) to be the oldest of the series. In this case, and in this alone, the bull has been placed in the centre of the obverse, and leaves insufficient place for the inscription, which accordingly has the appearance of having been added as an afterthought. Further, there is no chain of beads round the bull's neck, but the symbols of the sun and moon appear, though now much worn and indistinct,■bn the reverse. There remain to be placed f i gs. 1-4. At the top of the obverse, squeezed in between the beading and
1 [This formula should read thus: Ye dharmma hetuprabhava hetum te95m Tathagato by avocat tesaf ca yo nirodhah evamvadi maha3ramanah.]
2 [The stone has only avoca.]
3 [The stone reads upasaka.]

the central figure on the last of these, there is an inscription of four characters, hitherto unnoticed, which is not as clear in the plate as on the coin itself, but is far more obvious in Phayre's reproduction (Coins of Arakan, Plate II, fig. 11). The first two letters give Deva, in exactly the same script as in fig. 5; the last two are not clear, but are probably candra. The reverse has the sun and moon symbol of the Candra dynasty. The manner in which the inscription is inserted on figs. 4 and 5 makes it probable that Devacandra was the first king of Arakan to place his name on the coinage. This reason alone is enough to justify the conclusion that Taw Sein Ko was wrong in reading Kdlacandra on the coin mentioned above (p. 368 f.). Fig. 3 is closely related to fig. 4 in general pattern, but has no inscription on the obverse ; the symbols of the sun and moon appear to be discernible on the reverse. This coin must therefore be very close in date to No. 4, and may belong to Kala or Deva Candra. Discussion of Nos. 1 and 2, very similar coins and in particularly good condition, is more speculative. The obverse has a conch-shell in a circle of beads ; in No. 1 the turn-in of the shell has teeth-like markings, absent in No. 2, and there are two loops of what is apparently a ribbon at the top of the shell, this feature being almost obliterated in No. 2. The reverse has a pattern, which is familiar to us on many of the more ancient Indian monuments, such as Sanchi, and which I have tried to show elsewhere 1 to be a vardhamdna. In the centre is an ankusa (?) with a disproportionately long hook, and No. 2 has also a small crescent on the right of it, resembling the moon in the other coins. For the vardhamdna in this form I may particularly compare the specimen in the right hand quarter of the dydgapatta figured on Pl. IX of V. Smith's The Jaina Stupa and other Antiquities of Muttra, 2 which shows in the interior something like a bud on a stalk in place of the ankusa. It may be more than a coincidence that a good relief of a conch-shell was found also at Muttra, illustrated ibid., Pl. LXXI, fig. 7. If we compare the reverses of Nos. 1 and 2 with that of No. 3, is it not clear that the latter has developed out of the former ? The relationship is obvious, and the close connection between the reverses of No. 3 and Nos. 5 if. do not allow room for Nos. 1 and 2 to have developed out of the remaining coins, as can be seen from the latest stage of the evolution, as known to us at present, in Nos. 20 and 21. With this result assured, we can then realize that the obverse of Nos. 3 and 4 is merely the shell of Nos. 1 and 2, turned the other way up and stylized with much extra embellishment. Accordingly coins Nos. 1 and 2 must be the earliest of the series. This conclusion is corroborated by a coin which is intermediate in type between Nos. 1 and 3; there is no specimen of it in the British Museum, and I only know it from the indistinct illustration in Phayre, _ibid., Pl. II, fig. 9, and cannot therefore discuss it in detail.
As regards the origin of this coinage, it is remarkable how few Indian coins present any analogy with it. One might compare the reverse of No. 5
1 JRAS., 1931, 588 ff. ; 1932, 393 ff. ; and 1933, 690.
2 Also Vogel, La Sculpture de Mathura, Pl. LIV b.


with the appearance of a similar symbol on the Kulnta coin of Virayasas, reproduced on Pl. XVI, 4, of Allan, British Museum Catalogue, Coins of Ancient India ; but if the latter is rightly assigned to the first century A.D. there is a gap of some three centuries between them. The recumbent bull, presumably a gaivite emblem, is also rare on Indian coins ; of early ones I can only quote the Malaya coins, particularly V. A. Smith, Catalogue of Coins in the Indian Museum, Calcutta, i, Pl. XX, 18, while those of gasanka (ibid., Pl. XVI, 12, and Allan, B.M. Cat., Coins of the Gupta Dynasty, Pl. XXIII, 14, 15, 16, and Pl. XXIV, 1) are much too late.


The coins shown are as follows :
Nos. 1-2. The earliest coins, discussed above.
No. 3. Early coin of the Candra dynasty.
No. 4. Deva Candra. Similar to preceding.
No. 5. Deva. Earliest coin with the bull on obverse.
Nos. 6-9. Niti Candra and Niti. There are numerous types, with two main divisions, those with the bull headed to the left on the obverse, and those with the bull headed to the right ; the emblems of the sun and moon are similarly transposed on the reverses. There is a large silver coin resembling No. 6 in the possession of the Archaeological Survey, Burma, with the bull headed to the right (P1. V, No. 22).1 In the coinage of the other kings the bull is invariably headed to the left.
No. 10. The smallest coin of the series, with apparently no inscription. From the type of the reverse it seems to belong to Niti Candra.
Nos. 11-13. Vira Candra and Vira. The moon is to the right on the reverse in Nos. 11 and 12, and to the left in No. 13.
Nos. 14-15. Priti. Moon to the right. No. 16. Prthvi. Moon to the right.
Nos. 17-18. Dhrti Candra and Dhrti. Moon to the left.
No. 19. -rmmavijaya (read Dharmmavijaya).
No. 20. Dhamma Candra. Note the increased stylization, the absence of the bead necklet on the bull, and the curious shape of the hump. Moon to the left.
No. 21. Reading of inscription uncertain. Note the curious shape of the bull's hump, only paralleled in No. 20; this is also the only coin showing the bull's tail turning back at the end, instead of under the rump as in all other cases. Moon to the right.
[No. 22. Niti Candra. See above.]
1 [This coin was presented to the Phayre Museum by Maung Kyaw.]