WPS 37 Champa Revised - [PDF Document] (2024)

ARI Working Paper No. 37 Asia Research Institute ● Singapore

Asia Research Institute

Working Paper Series

No. 37

Champa revised


Michael Vickery


March 2005



ARI Working Paper No. 37 Asia Research Institute ● Singapore

The ARI Working Paper Series is published electronically by the Asia Research Institute of the National University of Singapore. © Copyright is held by the author or authors of each Working Paper. ARI Working Papers cannot be republished, reprinted, or reproduced in any format without the permission of the paper’s author or authors. Note: The views expressed in each paper are those of the author or authors of the paper. They do not necessarily represent or reflect the views of the Asia Research Institute, its Editorial Committee or of the National University of Singapore. Citations of this electronic publication should be made in the following manner: Author, “Title,” ARI Working Paper, No. #, Date, www.nus.ari.edu.sg/pub/wps.htm. For instance, Smith, John, “Ethnic Relations in Singapore,” ARI Working Paper, No. 1, June 2003, www.ari.nus.edu.sg/pub/wps.htm. Asia Research Institute Editorial Committee Geoffrey Wade Tim Winter Shen Hsiu-Hua Manjit Kaur Asia Research Institute National University of Singapore Shaw Foundation Building, Block AS7, Level 4 5 Arts Link, Singapore 117570 Tel: (65) 6874 3810 Fax: (65) 6779 1428 Website: www.ari.nus.edu.sgEmail: [emailprotected] The Asia Research Institute (ARI) was established as a university-level institute in July 2001 as one of the strategic initiatives of the National University of Singapore (NUS). The mission of the Institute is to provide a world-class focus and resource for research on the Asian region, located at one of its communications hubs. ARI engages the social sciences broadly defined, and especially interdisciplinary frontiers between and beyond disciplines. Through frequent provision of short-term research appointments it seeks to be a place of encounters between the region and the world. Within NUS it works particularly with the Faculty of Arts and Social Sciences, Business, Law and Design, to support conferences, lectures, and graduate study at the highest level.



ARI Working Paper No. 37 Asia Research Institute ● Singapore

Champa revised* Michael Vickery

As the title of this paper implies, I consider that the history of Champa, which, as a whole,

has hardly been given critical study since Georges Maspéro’s 1928 book, is in need of revision.1

The important points which require revision are the following:

(a) The origins of the Austronesian-speaking Cham who now live in Vietnam and


(b) The Linyi problem. Was Linyi identical with Champa, from the beginning of records

concerning it, or from a later date, or if not, what was it?2

(c) Relations with Vietnam, in particular the notion that Champa, including Linyi, was

always a victim of expansionism by its northern neighbor.

(d) The narrative of the history of Champa as conceived by Maspéro. Although his book

received critical attention soon after its publication and more thoroughly later on from Rolf Stein,

his main conclusions passed literally into the famous synthesis by Georges Coedès, and have

continued to exert strong influence on further work, including total acceptance by some linguists

within the last decade.3

This revision includes examination of the political-administrative status of the areas

inhabited by the Cham as identified by architectural and epigraphic remains and extending from

*The assistance of Bruce Lockhart in preparing this manuscript for publication is gratefully acknowledged—ARI Working Papers Series editor. 1 Georges Maspéro, Le Royaume de Champa (Paris: École Française d’Extrême-Orient reprint, 1988); the original imprint was Paris and Brussels: Éds. G. Van Oest, 1928. 2 I have decided on this spelling of ‘Linyi’. Usage in Vietnam today drops the hyphen which formerly separated the elements of two-term geographical names. It has usually been spelled Lin-I or Lin-Yi, as in Rolf Stein, Le Lin-Yi, sa localisation, sa contribution à la formation du Champa et ses liens avec la Chine, in Han-Hiue, Bulletin du Centre d’Études Sinologiques de Pékin, 2 (1947). Toponyms and names of Champa temples will also be spelled as in current Vietnamese literature. 3 Louis Finot, review of Maspéro, Le Royaume de Champa, Bulletin de l’École Française d’Extrême-Orient (henceforth BEFEO), 28 (1928): 285-92; reprinted in the 1988 edition of Maspéro, Royaume de Champa; George Coedès, Histoire ancienne des états hindouisés d’Extrême-Orient (Hanoi: Imprimerie d’Extrême-Orient, 1944); Coedès, Les états hindouisés d’Indochine et d’Indonésie (Paris: Éditions E. de Boccard,1948); Coedès, Les états hindouisés d’Indochine et d’Indonésie, rev. edn (Paris: Éditions E. de Boccard, 1964); Coedès, The Indianized states of Southeast Asia, ed. Walter F. Vella, tr. Sue Brown Cowing (Honolulu: University of Hawaii, 1968); Graham Thurgood, From ancient Cham to modern dialects, two thousand years of language contact and change (Honolulu: University of Hawaii, 1999).


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Quảng Bình to south of Phan Rang. That is, was there a single unitary state/kingdom of Champa,

as depicted in the standard classical scholarship, a federation of polities dominated by the

Austronesian-speaking Cham, or two or more quite distinct, sometimes competing, polities?4


There are three types of sources for Champa history: (1) physical remains – brick

structures considered to be temples, associated sculpture, and materials obtained from

archaeological excavation; (2) inscriptions in Old Cham and Sanskrit; and (3) references in

Chinese and Vietnamese histories about relations between those countries and the various

polities south of the Chinese provinces in what is now northern Vietnam and, after the late tenth

century south, of territory claimed by Vietnam.

Physical remains

The physical remains above ground, the temples, show through their architecture at least

three regions which began their development at approximately the same time – roughly, on that

evidence, the eighth to ninth centuries. However, it is certain that there was earlier architecture

which has disappeared and that the real beginnings were earlier. From North to South these

regions are (1) Quảng Nam, in particular the Thu Bồn valley, the location of Mỹ Sơn, Trà Kiệu

and Đồng Dương; (2) the region of Nha Trang with the Pô Nagar complex, and (3) the region of

Phan Rang, where parts of Hòa Lai may date back to the eighth century, and which should

perhaps be considered to include the structures of Pô Dam and Phan Thiết farther South. 5

4 For the early historians and epigraphers of Champa it was assumed that there had been a single Champa from the appearance of Linyi. See Étienne Aymonier, ‘Première étude sur les inscriptions tchames’, Journal Asiatique (henceforth JA), série 8, 17 (1891): 5-86; Louis Finot, ‘Stèle de Śambhuvarman à Mi-so’n’, BEFEO, 3 (1903): 206-11; Louis Finot, ‘Notes d’épigraphie VI: Inscriptions de Quảng Nam’, BEFEO, 4 (1904): 83-115 (see p. 113); Finot, ‘Notes d’épigraphie XI: Les inscriptions de Mi-so’n’, BEFEO, 4 (1904): 897-977; Finot, ‘Les inscriptions du Musée de Hanoi’, BEFEO, 15 (1915): 1-19; Paul Pelliot, ‘Deux itinéraires de Chine en Inde à la fin du VIIIe siècle’, BEFEO, 4 (1904): 131-385; and the sources cited in the previous footnote. More recent studies which treat Champa as a federation can be found in Actes du séminaire sur le Campā, organisé à l’Université de Copenhague le 23 mai 1987 (Paris: Centre d’Histoire et Civilisations de la Péninsule Indochinoise, 1988). 5 William Aelred Southworth, ‘The origins of Campā in Central Vietnam, a preliminary review’ (Ph.D. diss., School of Oriental and African Studies, University of London, 2001), pp. 416-7. Table 3 shows the architectural sequence proposed by Philippe Stern, L’art du Champa (ancien Annam) et son évolution (Paris: Adrien-Maisonneuve, 1942), and Table 4 Southworth’s revised proposal. He has omitted any reference to Pô Nagar of Nha Trang, even though his table is


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Another region where the temple remains are quantitatively very significant is around

Quy Nhơn, but the structures there have been dated to the eleventh to thirteenth centuries, with

no earlier remains.

All of these regions were located at very good river-mouth ports or on rivers not far from

the sea. One ancient site in the Thu Bồn valley where complete above-ground structures had

disappeared by the time the subject received modern interest, but where impressive sculpture was

still found, is Trà Kiệu, some 20-30 km from Mỹ Sơn; its importance since perhaps the first

century has been revealed by archaeology.6

There are two more rivers which must have been important in early Champa, although so

far they have not received adequate attention. One – which, as I shall show, has never been given

the attention it deserves – is the Trà Kúk in Quảng Ngãi, with two citadels: Châu Sa (apparently

a large fortified city) and Cổ Lũy (where some important sculpture, perhaps dating from the

seventh to eighth centuries, has been found). The citadels are on either side of the river near its

mouth, along with the remains of a temple with interesting sculpture, Chánh Lộ, dated to the

eleventh century. The relative neglect of Châu Sa by the early explorers was probably because

they found no impressive temples, and only one inscription.7 The other river is the Đà Rằng

(Sông Ba), which flows into the sea at Tuy Hòa, between Quy Nhơn and Nha Trang; it is the

largest river valley in Vietnam. Remains from different periods have been found along its length,

a fifth-century Sanskrit inscription at its mouth, and the later citadel of Thành Hồ (larger than

Châu Sa) about 15 km inland. It was no doubt an important route far into the interior.8

One more neglected area is that of Champa’s northernmost expansion in Quảng Trị and

Quảng Bình, apparently during the ninth to tenth centuries, when Indrapura and the temple of

Đồng Dương were flourishing and when impressive related Mahayana sites developed in

Ron/Bắc Hạ, Đại Hữu, Mỹ Đức and Hà Trung.9 The area was probably included in the domain

entitled ‘Localised sequence of the architecture of Campā up to the beginning of the eleventh century CE’, a time period from which parts of Pô Nagar may not be omitted. 6 Ibid., citing earlier work by Claeys and Glover. 7 This is the case even when the inscription (C61) is cited, as in Édouard Huber, ‘L’épigraphie de la dynastie de Đồng Dương’, BEFEO, 11 (1911) 282; Finot, ‘Inscriptions du Musée’, p. 11; Southworth, ‘Origins of Campā’, pp. 149, 151. For Chánh Lộ see Jean Boisselier, La statuaire du Champa (Paris: Publications de l’École Française d’Extrême-Orient, 1963, p. 214. 8 Charles Higham, The archaeology of mainland Southeast Asia (Cambridge: Cambridge University Press,1989), pp. 304-5; Ngô Văn Doanh, Chămpa ancient towers: Reality & legend (Hanoi: Institute for South-East Asian Studies, 2002), pp. 281-92. 9 Boisselier, Statuaire du Champa, pp. 118, 101, 133-9, 141, 147-8, 276.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

of Maspéro’s ‘sixth dynasty’, but its monuments had not been studied when he was writing, and

the region has never been given the importance it deserves. Moreover, its significance has been

obscured in the literature by assigning its work to ‘styles’ named for centers farther south, such

as Đồng Dương, Mỹ Sơn, etc.10 Requisite attention to this area will force reinterpretation of the

history of events in the tenth to eleventh centuries.

It must be emphasized that the dating of many of the architectural remains and surface

finds of sculpture must still be considered tentative because it is based on reasoning no longer

accepted. Examples include Mỹ Sơn E-1, dated by reference to a Cambodian sculpture over 700

km distant with a vaguely associated inscription; the edifice at Hòa Lai (Phan Rang), whose

dating has been influenced by Damrei Krap (on Phnom Kulen in Cambodia), ultimately dated by

a possibly mythical (i.e., not truly historical) section of the Cambodian Sdok Kak Thom

inscription; the Tháp Mắm style (in Bình Ðịnh) based on one of the more fictional treatments in

Maspéro; and the Mỹ Sơn A1 style based first on a prejudice of Henri Parmentier and a misused

inscription and now not clearly distinguished from Trà Kiệu, about which there is now much lack

of agreement resulting from different Trà Kiệu styles discovered through field archaeology.11

Thus, at this stage of research, architecture and sculpture are of little help in more than loose

relative dating of the history of Champa.


10 Examples are L’Association Française des Amis de l’Orient, Le Musée de sculpture Caü de Đà Nẵng (Paris: Éditions de l’AFAO, 1997), p. 142, no. 121, from Hà Trung (Mỹ Sơn style); p. 144, no. 124, from Đại An (Mỹ Sơn style); and p. 175, no. 192, from Mỹ Đức (Đồng Dương style). 11 Stern, Art du Champa, p. 70, in spite of having claimed not to have used any inscription in his dating of the monuments, dated Damrei Krap to the early ninth century (around 802) on the basis of the story of Jayavarman II in the Sdok Kak Thom inscription of Cambodia (AD 1052), but with the date 802 attributed to Jayavarman from other late inscriptions. For Mỹ Sơn A1, Stern pointed out (p. 94) that Parmentier ‘is working on the principle that the more perfect of two art forms is the older’. It was this preconception which resulted in the mistaken very early dating of Mỹ Sơn A-1, which then seemed supported by an early inscription which was found nearby but was not really related to the monument. For Tháp Mắm Boisselier’s treatment was awkward, perhaps because he too tamely follows Stern’s denomination of it as Bình Ðịnh style. Boisselier first suggested that the style should have come soon after the capital allegedly moved to Bình Ðịnh (around 1000, according to Maspéro), but then, seeing the difficulty of that, he placed it – still awkwardly – in the twelfth century, where it certainly does not fit with the main standing monuments of that area, the towers around Quy Nhơn (Boisselier, Statuaire du Champa, pp. 223, 256-274, 308-9). Its monsters, nicely preserved in the Đà Nẵng museum, show Vietnamese/Chinese influence (Boisselier, pp. 291-3) and must be later, say fourteenth or even fifteenth century. As William Southworth (personal communication, 10 Nov. 2004), has also noted, ‘The entire period and the whole of the Bình Ðịnh architectural and art historical sequence needs to be re-examined in much greater detail ... [and] the Tháp Mắm site itself may in fact date from the late fourteenth to fifteenth centuries’.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

The inscriptions of Champa are in two languages, Cham and Sanskrit. The inscription

considered oldest on palaeographic grounds is that of Võ Cạnh, from a site near Nha Trang. It

has been dated between the second and fourth centuries, and gave rise to a lengthy controversy

as to whether it belonged to Champa or to a ruler of Funan occupying that region which was

subsequently part of Champa. Coedès’ final opinion, shared by Maspéro, was that it belonged to

Funan, that the apparent chief named Śrī Māra was the Funan ruler known to the Chinese as ‘Fan

Shi Man’, and that the religion of the time was Buddhism. That opinion prevailed until 1969,

when Jean Filliozat argued that the title ‘Māra’ more probably derived from a Pandyan royal title,

and that the content of the inscription could just as well imply Hinduism as Buddhism. This last

treatment seems to show that the Võ Cạnh inscription may not be definitely ascribed to either

Funan or Champa, and certainly not to Linyi.12

Now, however, William Southworth has called attention to certain features of the

inscription’s contents, which seem to reflect Austronesian society; if his argument is accepted, it

may be reinstated as the first inscription of a Cham entity, although not of Linyi. As Southworth

notes, following the translations of Filliozat and Claude Jacques,

the author of the inscription may not be a descendant of Śrī Māra at all, but a son-in-law who had married into the ruling matrilineage. The pivotal focus of this lineage is clearly the daughter of the grandson of Śrī Māra, to whose family the author belongs, and this female hierarchy is suggested by the content of the inscription.

That is, the type of donations described in the inscription are ‘common in matrilineal societies’,

and the inscription was ‘primarily motivated by indigenous social concerns’. The name Śrī Māra

could still be of Tamil origin as proposed by Filliozat – learned by the Cham on their voyages to

India and adapted for a time until the Sanskrit varma became popular in later times. Nha Trang,

as Southworth describes it, was a port on ‘the main maritime trade route through Southeast Asia’

from India to northern Vietnam and southern China, ‘providing a logical geographic and

historical context for the carving of the Vo Canh inscription’.13 The inscription considered next

oldest – from the fourth century, again based on palaeography – is the first Cham text, from

12 Coedès, Indianized states, p. 40; Jean Filliozat, ‘L’inscription dite de “Vỏ [sic]-Cạnh”’, BEFEO, 55 (1969): 107-16; Claude Jacques, ‘Notes sur la stèle de “Vỏ [sic]-Cạnh”’, pp. 117-24 of the same issue; Michael Vickery, ‘Funan reviewed: Deconstructing the ancients’, BEFEO, 90-91 (2003-4): 101-43; and the discussion in Southworth, ‘Origins of Campā’, noted below. 13 Ibid., pp. 204-5.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Đông Yên Châu near Trà Kiệu. Found in the Thu Bồn area, not far from Mỹ Sơn, it is also the

oldest writing in any Austronesian or Southeast Asian language.14

Both of these early inscriptions are isolated and may not be integrated with the rest of the

corpus, surviving inscriptions of which are not distributed in total conformity with the physical

remains. The first coherent group of inscriptions is linked with the early development of the Thu

Bồn valley, site of Mỹ Sơn, at a time when there are only isolated texts elsewhere. From the fifth

to the late eighth centuries, there are 20 inscriptions, all in Sanskrit and all but two located in or

near Mỹ Sơn. According to statistics compiled by Southworth, 19 inscriptions with 279 lines of

text are in Quảng Nam, with 12 inscriptions and 258 lines in Mỹ Sơn and only three inscriptions

with 13 lines elsewhere.15 Then from the mid-eighth to mid-ninth centuries – between 774 and

854 – there is a coherent group of eight inscriptions in the South. Most of these are in Phan Rang

but some are in Nha Trang; five of them are entirely or partly in Cham.

Following that, from 875 (or perhaps slightly earlier) until 965, there are 25 inscriptions

ascribed to the Indrapura/Đồng Dương dynasty – again in the North in the Thu Bồn area, but

distinct from Mỹ Sơn. These inscriptions delineate a coherent area from Quảng Nam to Quảng

Bình and include the only epigraphy in the published corpus found north of Huế. Four

inscriptions of this group are in the South, and 16 are entirely or partly in Cham.16 One more

Cham-language inscription, possibly related, is from Mỹ Sơn, dated 991 (on this see below).

Thereafter inscriptions are rather equally distributed between North and South until the

early thirteenth century, after which there are 32 in the South and only six in Mỹ Sơn, the last

dated in 1263. After 991, of the 75 known inscriptions until the last in 1456, only five are in

Sanskrit (all of them before 1263), and the rest are in Cham. During the same period, there are 18

inscriptions from Mỹ Sơn until the last dated one in 1263 and another from the end of the twelfth

14 This earliest Cham inscription is C174, unknown to Maspéro and omitted in Anne-Valérie Schweyer, ‘Chronologie des inscriptions publiées de Campā, Études d’épigraphie cam-1’, BEFEO, 86 (1999): 321-44. See George Coedès, ‘La plus ancienne inscription en langue chame’, in Eastern and Indian Studies in honour of F. W. Thomas, C. I. E. (Bombay: New Indian Antiquary Extra Series I, No. 48), pp. 39-52.. 15 Southworth, ‘Origins of Campā’, 241. 16 Anne-Valérie Schweyer, ‘Le dynastie d’Indrapura (Quảng Nam, Viet Nam)’, Southeast Asian Archaeology 1998: Proceedings of the 7th International Conference of the European Association of Southeast Asian Archaeologists, ed. Wibke Lobo and Stefanie Reimann (Hull: University of Hull Centre for South to East Asian Studies, 1998), pp. 205-17; Schweyer, ‘Chronologie des inscriptions’. Schweyer, ‘La vaisselle en argent de la dynastie d’Indrapura (Quảng Nam, Việt Nam), Études d’épigraphie cam - II’, BEFEO, 86 (1991): 345-55 proposed that the area of the inscriptions delineated the area under control of that dynasty. William Southworth objected very strongly to the use of Indrapura for more than the site of Đồng Dương, and he apparently did not accept Schweyer’s view of a dynasty; see Southworth, ‘Notes on the political geography of Campā in central Vietnam during the late 8th and early 9th centuries AD’, in Lobo and Reimann ed., Southeast Asian Archaeology 1998, pp. 237-44.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

century śaka, two hundred years after the Champa kings had allegedly – according to Maspéro –

been driven south to Vijaya under Vietnamese pressure, a circ*mstance which should force

reconsideration of the relations between the two polities.

Peculiarly, Bình Ðịnh/Quy Nhơn, in spite of its obvious importance as revealed by brick

towers and the apparent attention given it in Champa and Cambodian sources, has produced only

seven very short inscriptions – all very late, and only one of which contains much of historical

value (C53 and C54/1178-1278, C58/1259, C55/1265, C52/twelfth century ś., C47/1401,

C56/1456). All major inscriptions by rulers believed to have controlled Bình Ðịnh before the

thirteenth century were engraved at Mỹ Sơn. (On this see further below in the section on


The largest coherent and detailed body of inscriptions, at least a dozen texts, concerns the

twelfth to early thirteenth century relations, almost entirely bellicose, with Cambodia. Detailed

discussion of the inscriptions can be found in the section on narrative history below. The first

work on Champa inscriptions began toward the end of the nineteenth century. Abel Bergaigne

began summarizing information from the inscriptions in 1888, and published the Sanskrit texts in

1893. The first work on Cham-language inscriptions was by Étienne Aymonier in 1891. Then in

a series of articles, Louis Finot treated both Sanskrit and Cham inscriptions, modifying on some

points Aymonier’s interpretations of the latter. Further important work was done by Édouard


There are still problems with the literal interpretation of some of the Cham-language texts.

For most of the inscriptions treated in Aymonier’s 1891 article, he did not publish the text itself,

nor try to offer a full translation, but only gave a summary of important details. Obviously, some

of these will require new interpretation when eventually given attention by a competent Chamist.

When Louis Finot continued the work of publishing and translating the Cham inscriptions, he

usually chose texts which Aymonier had not treated; and since he was not a specialist in the

language, it is not possible to accept all of his translations without question. Because of this

uncertain quality of translations from Old Cham – in particular the work of Aymonier, but also

that of Finot – all interpretations of historical events based on them must be presented with the

caveat that better translations may ultimately force revisions of some details. 17 Aymonier, ‘Première étude’; Finot, articles cited in footnote 4; Abel Bergaigne, ‘L’ancien royaume de Campā dans l’Indochine, d’après les inscriptions’, reprinted from Journal Asiatique (Paris: Imprimerie Nationale, 1888); Bergaigne, Inscriptions sanscrites de Campā, Inscriptions sanscrites du Cambodge (Paris: Imprimerie Nationale, 1893); Huber, ‘Épigraphie de la dynastie’.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

There is now new work by Anne-Valérie Shweyer on the inscriptions of Champa which

serves as a guide to all of the publications. It purports to list all the inscriptions about which there

have been publications in order of their corrected dates, with columns for registry number,

location, names of persons and gods mentioned and the related publications. 18 Except for

Schweyer’s work, there has been no new original study of the Cham-language inscriptions since

before 1920, and the standard bibliographical list of Champa inscriptions, both Cham and

Sanskrit, dates from 1923.19

Chinese and Vietnamese histories

The Chinese sources used by Maspéro were the standard dynastic histories plus extracts

from Ma Duanlin’s Wenxian tongkao, translated into French by Hervey de Saint-Denys as

Ethnographie des peuples étrangers and cited by Maspéro as Méridionaux.20 Recently Geoff

Wade has translated the Champa sections of another text not used by Maspéro – the Song huiyao,

which differs in certain significant details from the Song shi cited by Maspéro, as will be noted

below. For some periods of Champa history these Chinese histories were compiled long after the

event, and obviously at second hand. As described by Wade, the Song huiyao was compiled ‘in a

process which extended from the early eleventh century until the middle of the thirteenth ... but

was never printed’. Then it was used as a source for the Song shi compiled in the fourteenth

century, after which nearly half of the Song huiyao was lost, the remaining portions included in a

fifteenth-century encylopaedia and later reworked into its present form in the nineteenth

century.21 Given these conditions, it requires religious faith to insist that all their details for

Champa in, say, the tenth to eleventh centuries must be accepted as factual, and their

inconsistencies require close attention.

18 Schweyer, as noted, missed the very first Cham-language inscription, C174 from Ðông Yên Châu, published in Coedès, ‘Plus ancienne inscription’. It is also missing from the list in Études épigraphiques sur le pays cham, ed. Claude Jacques (Paris: École Française d’Extrême-Orient, 1995). Schweyer has also prepared some new transcriptions and translations of the Cham inscriptions of Pô Nagar, Nha Trang, which appear in Aséanie, 14 (2004): 109-40 and 15 (2005), forthcoming. These will be cited here as ‘Po Nagar’. I wish to thank her for sharing them with me. 19 The list is in George Coedès and Henri Parmentier, Listes générales des inscriptions et des monuments du Champa et du Cambodge (Hanoi, Ecole Française d’Extrême-Orient, 1923). 20 Ma Duanlin, Ethnographie des peuples étrangers à la Chine, tr. Léon Hervey de Saint-Denys (Farnborough: Gregg, 1972 reprint); the original edition was published in 1876 by H. Georg in Geneva. 21 See Geoff Wade’s chapter in Lockhart and Tra (eds.), New Scholarship on Champa, 2006.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

The Vietnamese sources which have been used for Champa history are the Ðại Việt sử ký

toàn thư (Tt), Khâm định Việt sử thông giám cương mục (Cm), and Việt sử lược (Vsl).22 A

problem in the use of the Vietnamese histories which is obscured in Maspéro and not given

attention in subsequent work is the existence of two main chronicle traditions which diverge with

respect to some of the events involving Champa (and others as well). The first is Tt, followed by

Cm, and the second is Vsl. Maspéro adopted the versions of one or the other, apparently

according to his arbitrary decision of what ‘should have happened’. Examples will be given


These Chinese and Vietnamese histories begin, for the region in question, with references

to a polity named Linyi, situated south of Jiaozhi/Giao Chỉ, the Vietnamese provinces which

were considered to be under Chinese administration; it was first noted with that name in 220-230,

and the last reference appeared in 757. Linyi was a problem area because of its aggressive

activities against Jiaozhou to its north. After the disappearance of ‘Linyi’ from Chinese records,

the official Chinese histories for a century referred occasionally to a polity named Huanwang

(Viet. Hoàn Vương), apparently in the old Linyi area, until in mid-ninth century they recognized

Champa in the name Zhan Cheng (Viet. Chiêm Thành), ‘City of the Cham’.

Throughout the Linyi period until the mid-seventh century – that is, a century before the

name itself disappeared – Linyi chiefs in the Chinese histories were entitled ‘Fan’ followed by

one- to three-syllable names, with the last Fan chief, Fan Zhenlong, named around 645. In

general it is impossible to reasonably identify the Fan names with names of rulers in the

contemporary Champa inscriptions, although this was attempted by Maspéro and has been

followed by later historians. During the mid-eighth to late ninth centuries, when all the

inscriptions are in the South, the Chinese record no Fan titles and their references to Huanwang

do not show names of rulers of that polity in any form.

Names in Chinese and Vietnamese histories apparently attributable to rulers of Champa

begin again in the 860s and continue – though with significant breaks – until the late twelfth

century, after which for nearly one hundred years during the time of Angkorean intervention in

Champa, the Chinese and Vietnamese apparently paid it little attention. Those sources begin to

22 Ngô Sĩ Liên, Ðại Việt sử ký Toàn thư (henceforth Tt), tr. Hoàng Văn Lâu, ed. Hà Văn Tấn (Hà Nội: NXB Khoa học Xã hội, 1993); Quốc sử quán Triều Nguyễn, Khâm định Việt sử thông giám cương mục (henceforth Cm) (Hà Nội: NXB Giáo dục, 1998); Việt sử lược (henceforth Vsl), ed. Trần Quốc Vượng (Hà Nội: NXB Văn Sử Địa, 1960).


ARI Working Paper No. 37 Asia Research Institute ● Singapore

provide names for Champa rulers again from the early fourteenth century, and continuing

through several decades when there are no Champa inscriptions.

As was already emphasized by Stein, it is impossible to identify most of the names in the

Chinese and Vietnamese texts with the names in inscriptions. Most of the latter are titles in -

varma, for which there was a standard Chinese transcription with ‘-ba-mo’ (跋摩). During the

entire history of Champa from the first -varma (perhaps in the fourth century) until 1471, there

are only four instances of Chinese transcriptions in ‘- ba-mo’, all in the sixth to seventh centuries,

and very few instances in which other elements of the Chinese transcription may be identified

with a local title; one of these rare exceptions is Chinese lu-tuo-luo (庐陀罗) and lu-tuo (庐陀)

for Rudra-[varman] in mid-eighth century. There is also one instance of a credible Chinese and

Vietnamese transcription of the Cham title yāï po ku vijaya śrī (see below). As I have argued

with respect to one case in an earlier paper, and shall argue further below, the reason for this

discrepancy is probably because the Chinese and Vietnamese were not dealing with the same

Champa polity that was recorded in the inscriptions.23

Champa history and its revisions

I. The origins of Champa and the Cham

Until relatively recently, the Cham were considered to be part of one of the ‘waves’ of

population movement which were believed to have proceeded successively out of China into

mainland Southeast Asia, with some of them continuing on to the islands of Nusantara. These

population groups were generally defined in terms of physical anthropologically. First came the

Australoid-Melanesoid, resembling the Australian aborigines and the peoples of Papua New

Guinea, then two successive groups of ‘Indonesians’ (Proto- and Deutero-). The Cham – for

those who recognized their language as ‘Indonesian’ (now called ‘Austronesian’), an affiliation

which was not at that time universally accepted – were considered to be remnants of these

‘Indonesians’ who had remained on the mainland after the rest had moved on to the islands under

pressure from the last wave of arrivals, the Mon-Khmer. In this view all of these movements

would have been completed before the beginning of the Common Era. Implicitly, then, Champa

23 Michael Vickery, ‘Cambodia and its neighbors in the fifteenth century’, National University of Singapore Asia Research Institute Working Paper Series No. 27, www.ari.nus.edu.sg/docs/wps/wps04_027.pdf, ‘Appendix to note 15’, pp. 49-52.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

was a land power, as was, in that view, its southern neighbor Funan. With occasional minor

variations, this was accepted by all the early scholars, including Coedès, Maspéro and Stein; the

continental origin of the Cham has most recently been retained by Jacques Népote, and is

implicit in the ‘Copenhagen papers’ of 1987.24 It was also recognized that the Tai/Thai had

moved into their present habitat much later than the migrations outlined above, and it was

believed that the Vietnamese, thought to be a Sinitic branch, had moved into what is now

northern Vietnam independently of those migrations.

Starting around thirty years ago, when physical-anthropological groupings were no

longer in vogue and linguistics had become more advanced, it was determined that the

Austronesian languages (including Cham) spread not via peoples migrating overland out of

China across mainland Southeast Asia and then into the islands, but via the sea – probably

starting in Taiwan, then to the Philippines, Indonesia, the Pacific islands, Madagascar, and

among the latest moves, the ancestors of the Cham from Borneo to the coast of what is now

Vietnam, at a time estimated as somewhere between 500 BCE and the early years of the

Common Era, although such dating is still very approximate.25 It has long been held by some

archaeologists that the Sa Huỳnh Culture of the central coast is the first material evidence of the

Cham, although ongoing research may modify this, and it is probable that there was more than a

single landfall by the first seaborne Cham. 26 There is no longer doubt that Cham is an

Austronesian language, closely related to Achenese and Malay and more distantly to other

languages of Indonesia, the Philippines, Polynesia and Madagascar.

24 Jacques Népote, ‘Champa, propositions pour une histoire de temps long’, Péninsule, nouvelle série, 26 (1993): 3-54 and 27 (1993): 65-119. For the ‘Copenhagen papers’, see Actes du séminaire, notably Tâm Quach-Langlet, ‘Le cadre géographique de l’ancien Campā’, pp. 28-47, who treats Champa as a land-based economy and shows no awareness of its maritime background or activity. 25 Peter Bellwood, ‘Southeast Asia before history’, in The Cambridge history of Southeast Asia, ed. Nicholas Tarling, Vol. I, pp. 53-136; see also Bellwood, ‘Cultural and biological differentiation in Peninsular Malaysia: The last 10,000 years’, Asian Perspectives, 32, 2 (1993): 50, where he refers to the ‘differentiation of Malayo-Chamic [still in southeast Kalimantan] commencing in the third or fourth century BCE’; Robert Blust, ‘The Austronesian homeland: A linguistic perspective’, Asian Perspectives, 26 (1984-5): 45-67; Blust comments elsewhere that ‘probably during the last two or three centuries before the Christian era, the Chamic languages and Malay became established in mainland Southeast Asia along the coasts of the South China Sea’; Blust, ‘The Austronesian settlement of mainland Southeast Asia’, in Papers from the second annual meeting of the Southeast Asian Linguistics Society 1992, ed. Karen L. Adams and Thomas John Hudak (Tempe, AZ: Arizona Sate University Program for Southeast Asian Studies, 1994), p. 30. Graham Thurgood posits a Chamic arrival on the mainland about 2000 years ago (From ancient Cham, p. 5). 26 Peter Bellwood, Prehistory of the Indo-Malaysian Archipelago (Orlando: Academic Press, 1985); Bellwood, ‘Southeast Asia’; Higham, Archaeology of mainland Southeast Asia, pp. 230-97.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

As the prehistorian Peter Bellwood has put it, ‘the old idea, so often repeated in popular

works today, that the Austronesians migrated from the Asian mainland through the Malay

Peninsula or Vietnam, is absolutely wrong’. Bellwood, moreover, considers that the

archaeological remains of Sa Huỳnh may be identified with the earliest Cham, who would have

arrived in the first millenium BCE. Or, again quoting Bellwood, ‘Heine Geldern (1932) was

clearly wrong ... when he suggested that early Austronesians migrated from Mainland Asia

through the Malay Peninsula into Indonesia. The true course of Austronesian expansion was in

the other direction’.27 Of course, there may well have been more than one landfall along the

coast, but the earliest known Champa remains suggest that they would have been south of Huế.

When they arrived there, the region was already inhabited by people speaking Mon-Khmer

languages. In fact, mainland Southeast Asia at that time would have been linguistically a solid

Mon-Khmer block.28

Rolf Stein, whose major study of Linyi disagreed strongly with Maspéro on many details

and seems to show definitely that before the fifth century, at least, Linyi and Champa may have

been distinct, also examined evidence for Linyi’s linguistic position, which he believed to be

Mon-Khmer. We must emphasize again that Stein still believed that the Cham had moved

overland and that they had thus been in the region of southern China – northern Vietnam for

hundreds or even thousands of years. Stein did not say this explicitly, but it is clear from his

remarks that he accepted it as a basic fact which did not need to be evoked anew. Thus, assuming

that the Cham had been in contact with the Chinese for so long, he analyzed the Archaic (eighth

to third centuries BCE) and Ancient (sixth century CE) pronunciations of the Chinese characters

used for ‘Linyi’ and ‘Cham’ and found that they were nearly identical, beginning with a cluster

‘KR-’ or ‘PR-’. Through this analysis he sought to discover the original pronunciation of the

relevant Chinese characters, and he considered that Linyi, and Cham as well, were Mon-


Now, however, it is understood that the Cham did not arrive on the coast before the first

millenium BCE, and in a region which would have had little or no contact with the Chinese by

the time the latter were taking notice of and writing about them. The Chinese character used to 27 Bellwood, Prehistory, pp. 124, 275; Bellwood, ‘Cultural and biological’, p. 53. The ‘old idea’ is still repeated in some scholarly work, such as Népote, ‘Propositions’. 28 ‘Peninsular Malaysia is one of the few places in the Austronesian world (the other major ones being southern Viet Nam and western Melanesia) where Austronesian settlers found agriculturalists, in this case Austroasiatic speakers, in prior occupation’ (Bellwood, ‘Cultural and biological’, p. 51). 29 Stein, Lin-Yi, p. 221 gives definitions of Ancient and Archaic Chinese.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

denote them would therefore have acquired something close to its modern pronunciation and

would probably have represented – as Stein, following Bergaigne, recognized – some

designation the Cham were using for themselves, but about which we can now have no

knowledge. Stein’s study of the archaic pronunciations of the characters is thus of no relevance,

as already noted by Paul Demiéville.30

This new understanding of Cham origins based on linguistics means that the Cham must

be understood as one of the great navigator peoples of prehistoric Southeast Asia, and that the

political-economic status of Champa must be revised accordingly. There is now a consensus that

until the twelfth century, navigation and maritime trade within Southeast Asia and between

Southeast Asia and China and India were dominated by local groups, prominent among whom

were the speakers of Austronesian languages, including Cham. This interpretation of Cham

origins and the approximate date of their arrival on the coast of what is now Vietnam casts doubt

on the insistence of some researchers on the polyethnicity of ancient Champa, at least in its

description of the ethnic groups as including the separate (but closely related) Cham, Jarai,

Rhadé, Chru and Roglai as well as various Mon-Khmer groups. It is now understood that Jarai,

Rhadé, Chru, Roglai and any other Austronesian languages in late Champa and Vietnam

developed out of Cham, and were probably not distinct languages during the period of classical

Champa up to the fifteenth century. Since the first Cham inserted themselves into Mon-Khmer

territory, no doubt the Champa polities always included some of them, and because Champa

consisted of port cities at the mouths of major rivers, some of the intervening territory may

always have had a more Mon-Khmer than Austronesian population.31

As the Mon-Khmer specialist Gérard Diffloth describes the process:

The linguistic evidence suggests that what actually happened was that speakers of (Old) Cham moved into a territory (the Highlands) which was at the time entirely occupied by Mon-Khmer speakers (specifically speakers of early forms of Bahnar, Sre, Mnong, perhaps also Sedang and others), established political control over

30 Ibid., p. 234, n. 223, citing Bergaigne, ‘Ancien royaume’ ; Paul Demiéville, ‘R.A. Stein, “Le Lin-yi, sa localisation, sa contribution à la formation du Champa et ses liens avec la Chine”’, T’oung Pao, 40 (1951): 345, quoted in Southworth, ‘Origins of Campā’, p. 373. 31 Bernard Gay, ‘Vue nouvelle sur la composition ethnique du Campā’ in Actes du séminaire, pp. 49-57; Gay is also obsolete (p. 51) in his treatment of the Vat Luang Kau inscription as showing Champa expansion up to the Mekong near Vat Phu in the fifth century, and in his acceptance of old Cambodian legends without historical value. On Vat Luang Kau see Claude Jacques, ‘Notes sur l’inscription de la stèle de Vǎt Luong Kǎu’, JA, 250 (1962): 249-56 and the discussion in Michael Vickery, Society, economics and politics in Pre-Angkor Cambodia: The seventh to eighth centuries (Tokyo. Centre for East Asian Cultural Studies for Unesco, Toyo Bunko, 1998), pp. 73-4.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

them, and eventually made them switch language, forsaking their original Mon-Khmer language and adopting Chamic, which has now become Jarai, Rhadé, etc... This is made very clear by the fact that these Chamic languages are structurally typically Bahnaric and not Austronesian, and their vocabulary contains hundreds of Bahnaric items which are not borrowings but retentions from the earlier languages. The Mon-Khmer substratum in Mountain Chamic is so visible and striking that it fooled the early researchers, including Schmidt, who included Chamic in the Mon-Khmer family [also Stein, Le Lin-Yi, see below], considering them ‘mixed-languages’, a concept which is no longer used’.32

The Copenhagen crowd has also maintained an old idea of Finot that the term

‘Cham/Cam’ was not the name of an ethnic group, but only an apocope of ‘Champa’. This is a

peculiar position to hold. The name ‘Champa’ may be understood as an imitation of the Champa

in India, but its choice for a region on the Southeast Asian coast, as Stein recognized, would have

been because the name of the people for themselves sounded something like /cam/. The name

was probably chosen by themselves after their voyages to India and not imposed by arriving

Indians, as Finot no doubt believed. The 1906 dictionary of Aymonier and Antoine Cabaton

gives čaü as the current name for themselves in their own language, as does the 1971 dictionary

of Gérard Moussay (same spelling in Cham script, transcribed there as căm). Their modern

neighbors, no doubt innocent of Indological preconceptions, call them čam (Rade), cam (Jarai,

Chru), cap (Roglai), etc.; and what was the Vietnamese ‘Chiêm’ in their historical accounts of

ancient Champa but their version of ‘Cham’, just as the official Chinese designation for Champa

after the ninth century (Zhancheng) was ‘city of the Cham’, not ‘city of Champa? Even Po

Dharma, a Cham himself, when not concerned with the ideology of his Paris group, can unself-

consciously entitle his book Quatre lexiques malais-cam anciens, treating not really ancient

Cham, but that of the nineteenth century.33

II. The Linyi Problem

For the first Europeans interested in the subject, Champa history began with a polity

named ‘Linyi’ (ancient pronunciation *Liem-.iep) which was first noted in the Chinese histories

32 Gérard Diffloth, ‘The outward influence of Chamic into Mon-Khmer speaking areas’, Symposium on New Scholarship on Champa, Asia Research Institute, National University of Singapore, 5-6 August 2004 ; and personal communication. 33 Finot, review of Maspéro, p. 286; Gay, ‘Vue nouvelle’; Étienne Aymonier and Antoine Cabaton, Dictionnaire čam-français (Paris: Ernest Leroux, 1906), p. 116; Gérard Moussay, Dictionnaire cǎm-vietnamien-français (Phan Rang: Centre Culturel Cǎm, 1971), p. 39; Thurgood, From ancient Cham, pp. 2, 336; Po Dharma, Quatre lexiques malais-cam anciens rédigés au Campā (Paris: École Française d’Extrême-Orient, 1999).


ARI Working Paper No. 37 Asia Research Institute ● Singapore

as having revolted against the authorities in Giao-Chỉ in the third century CE. Throughout the

following centuries until the name ‘Linyi’ disappears from the Chinese (and later Vietnamese)

histories after 757, it was depicted as an aggressive entity constantly pressing northward against

the Chinese provinces in what is now northern Vietnam. The first Chinese records thus situated it

in the North, but historians such as Maspéro believed that the Võ-Cạnh inscription near Nha

Trang belonged to the same polity, and therefore that the original Linyi extended from the far

North to the mid-South of modern Vietnam. The Sanskrit and Cham inscriptions which began in

the fifth century in the Thu Bồn valley – most importantly in Mỹ Sơn – and continued in number

and importance to the eighth century were also attributed to the same polity named Linyi by the

Chinese, and this assumption reinforced the view that Linyi and Champa were one.

There is nothing in the Chinese accounts of Linyi, at least as described in current

literature, which provides specific indication of ethnicity or language. If the archaeological

remains of Sa Huỳnh indeed represent the Cham, they are far south of the region which seems to

be reflected in the Chinese records of Linyi. Moreover, other than archaeological remains there

are no other local contemporary records (such as inscriptions) until the first Cham-language

inscription dated hypothetically to the fourth century, and other slightly later Sanskrit

inscriptions associated with Champa, but all of which seem also to be south of Linyi as described

by the Chinese. Stein took note of these matters and, after a close reading of the relevant Chinese

sources, proposed that the center of early Linyi was at Qusu/Badon, north of what seemed to be

the main Cham center, and that amalgamation of the two – if it occurred – was not until the sixth

century. He accepted, however, that the Linyi capital which the Chinese sacked in 605 was

probably Trà Kiệu, which archaeological investigation seems to show as belonging to Champa,

and which was probably a Champa center from the first or second centuries.34

In an earlier publication I proposed that Linyi was linguistically Mon-Khmer, but I would

accept that part of its area may have been absorbed by Champa at a time when the Chinese were

still using the name Linyi, and that this is reflected in the invasion of 605 as analyzed by Stein.35

I would still prefer to argue that the main ethno-linguistic group of Linyi was Mon-Khmer,

perhaps of the Katuic branch, or even a branch of Vietic or Viet-Mường moving northward when

the Chinese first became aware of them, until they merged into what eventually became the first

Viet Nam. Linguists now consider that the area of origin of the Viet-Mường languages was in

34 Southworth, ‘Origins of Campā’; Stein, Lin-yi, pp. 317-8. 35 Vickery, Society, economics and politics.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Nghệ An, and that the Urheimat of Katuic was in central Laos.36 Along with this process the

Cham, whose first center may well have been at Trà Kiệu, also expanded northward into what

had been old Linyi territory and the Chinese – unaware of, or unconcerned with, the ethno-

linguistic complexities – continued to call the region ‘Linyi’ until the mid-eighth century.

My initial suggestion that Linyi was Mon-Khmer was based on the title Fan used by the

Chinese for both rulers of Linyi and some early chiefs of Funan, and which, for Funan, I

considered to represent the local chiefly title poñ, found often in seventh-century Khmer

inscriptions but never after the mid-eighth century. Poñ is the only Khmer title which bears any

resemblance at all to Fan, ancient pronunciation b’iwAm. 37 The first Fan in Linyi – Fan

Xiong/Phạm Hùng – is recorded in the fourth century and the last in the seventh century.

Thereafter no Fan appears in the Chinese and Vietnamese lists of Linyi and Champa chiefs.

Just as some Fan in the Chinese records of Funan are described as ‘generals’ but not

reigning chiefs, and some poñ in the seventh-century Khmer inscriptions were other than

paramount figures in their areas, so in Linyi as well there were Fan who were not considered to

be its rulers. A certain Wen, of poor background but who eventually became Fan, is described as

having served a chief in Xiquan (Tây Quyên), not a ruler of Linyi, named Fan Chui; there was a

Fan Jian, a general of the Linyi chief Fan Huda; two more non-king Fan are listed in the reign of

Fan Yang Mah; an envoy of Fan Yang Mah II was entitled Fan Long Pa; and a certain Phạm (=

Fan) Côn Sa Đạt ....., not mentioned by Maspéro, is recorded in a standard Vietnamese history.

There was also a second Fan Xiong who seemed to be competing with the Linyi ruler Fan Huda

by sending an envoy to China, leading Maspéro to concoct a bit of historical fiction.38

From the fifth until the mid-eighth centuries there were two sets of records – inscriptions

in Cham and Sanskrit and Chinese records of the activities (usually bellicose) of Linyi under

leaders entitled Fan. For Maspéro, and for most later historians, the two sets of records

concerned the same polity, and the Indic names in the inscriptions were identified with the names

of the Fan as seen through Chinese. The seventh- to eighth-century architectural remains south

36. On Viet-Mường and Katuic see Gérard Diffloth, ‘Vietnamese as a Mon-Khmer language’, in Papers from the first annual meeting of the Southeast Asian Linguistics Society, ed. Martha Ratliff and Eric Schiller (Tempe, AZ , Arizona State University Press, 1992), pp. 125-39 and a paper by the same author presented at the Institute of Linguistics (Viện Ngôn ngữ học) in Hà Nội, cited in Nguyên Hữu Hoành, Tiếng Katu cấu tạo từ (Hà Nội: NXB Khoa học Xã hội, 1995), pp. 21-2, 219. 37 Vickery, Society, economics and politics; Vickery, ‘Funan reviewed’; Paul Pelliot, ‘Le Fou-Nan’, BEFEO, 3 (1903): 248-303, gives the ancient pronunciation.. 38 Maspéro, Royaume de Champa, pp. 54-55, 62, 72, 74, 61 respectively; Phạm Côn Sa Đạt is in Tt, Ngoại kỷ, 4:11b (v. 1, p. 175).


ARI Working Paper No. 37 Asia Research Institute ● Singapore

of Huế in the Thu Bồn valley, obviously belonging to an entity which could reasonably be called

‘Champa’, supplemented the inscriptions, while Chinese and then Vietnamese histories

continued to provide names which Maspéro identified with the same rulers, often with a large

input of imagination.

Serious confusion in Maspéro’s treatment of the Chinese histories and the Champa

epigraphic records starts early, in his ‘second dynasty’ (336-420) for which he lists the first

Bhadravarman (known, apparently, from three inscriptions dated on palaeographic grounds to

the fifth century) followed by a certain Gaïgārāja whom he identifies with a certain Di Zhen in

the Chinese records of Linyi; Gaïgārāja, however, is only mentioned retrospectively in the 658

inscription C96 of Vikrāntavarman. Then, during the entire period from the early fifth to the

seventh centuries and the C96 inscription, there are no inscriptions and thus no authentic

Sanskritic name which may reasonably be construed to represent a Fan ruler until Rudravarman I,

named retrospectively in the 658 inscription and whose name is convincingly transcribed by the

Chinese as Lu-tuo-luo-ba-mo in a record of 529.39 Another varma (Chinese ba-mo) name, Pi-

cui-ba-mo appears in a slightly earlier Chinese record, dated in Maspéro to 526-7; he construed it

as Vijayavarman – not unreasonable, but unknown from epigraphy. The reconstituted name of

Vijayavarman’s predecessor, however – ‘Devavarman’, from Fan Tiankai, is not acceptable,

even if not totally rejected by Pelliot.40 Thus for what Maspéro calls the ‘third dynasty’ (420-

530), I would say that what we have is Linyi as seen by the Chinese, but perhaps nothing at all

about Champa. For the ‘fourth dynasty’ (529-757) all the Sanskritic names are of predecessors

named in Vikrāntavarman’s 658 inscription, except the second Rudravarman.

The most thorough new treatment of this subject is in Southworth. He accepts the

Chinese textual evidence that the original Linyi was north of Huế and presents a new discovery

from the Chinese sources – ‘one of the great surprises of this thesis’, namely ‘the existence of

specific Chinese references to ten kingdoms along the central coast of Vietnam, including the

kingdom of Xitu [Sino-Vietnamese Tây Ðô]…the foremost among the independent states located

close to Lin-yi’, 200 li (100-120 km) to the south, which ‘strongly suggests the placement of this

country in the Thu Bồn valley system’.41 Then, during the fifth to sixth centuries (the period

39. There are other seventh-century inscriptions without year dates naming some of those listed in C96 (Schweyer, ‘Chronologie des inscriptions’, p. 326). 40 Maspéro, Royaume de Champa, pp. 245-6; Pelliot ‘Deux itinéraires’, p. 384, n. 3; in this particular case they argued that the Chinese term represented a translation rather than a transliteration of the alleged Cham name. 41 Southworth, ‘Origins of Campā’, pp. 335, 292-3, 291 respectively.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

when Stein and Boisselier agreed that the original Linyi might have merged with early Champa),

‘the earlier distinction between Linyi and Xitu in the Chinese histories gradually becomes

confused and their historical traditions combined’. This is seen in the contradictions between two

different Chinese histories of the period. In the Nan ji, the Linyi king Fan Hu-da, for whom the

last recorded date is 413, was succeeded by a son Fan Yang-mai, whose first date – that of a

mission to China – was in 421. In the Liang shu, however, Fan Hu-da was succeeded by a son Di

Zhen who abdicated in favor of a nephew who was killed, after which a brother of Di Zhen was

made king with the name Wen Di. He in turn was killed by a son of the king of Funan; another

Fan, Fan Zhou-nong, then became king and was succeeded by his son Fan Yang-mai who sent an

envoy in 421.42

As Southworth notes, there is hardly time for all of this between the dates of 413 and 421,

and the Chinese must have erroneously inserted a story from another polity. The name-title

Di/Địch (Chinese/Sino-Vietnamese) is entirely outside the Linyi tradition as recorded by the

Chinese, and their report that Di Zhen/Ðịch Chân abdicated and went on a pilgrimage to the

Ganges in India identifies him with the Gaïgārāja of Mỹ Sơn inscription C96. Thus the Ðịch

were from the Thu Bồn valley, probably Xitu, and this is the ‘earliest clear correspondence

between a King of Linyi known from Chinese sources and a ruler listed in epigraphy from the

Thu Bồn valley’.43 I would prefer to consider Gaïgārāja not as a historical king but as a mythical

ancestor of the first Thu Bồn lineage, who is mentioned not only in C96, but also in C73A as

Gaïgeśa and in C81 as Gaïgeśvara. Thus the Chinese in their story of Di Zhen were recording

Thu Bồn mythology. This does not lessen the value of Southworth’s conclusion about a

correspondence between Chinese sources on Linyi and Thu Bồn epigraphy.

From then on, according to Southworth, the name Linyi may be taken as referring to the

Chamic Thu Bồn valley; and his conclusion is that the previously recognized state of Xitu

absorbed ‘the former territory, political traditions, and trading concessions of its northern

neighbor, Linyi’.44 This does not help, however, with the eight Fan chiefs registered by the

Chinese from around 420 to the mid-sixth century at a time when there is no Champa epigraphy;

and if it is true that Linyi and the title Fan were Mon-Khmer, it is difficult to accept that Fan was 42 Ibid., pp. 303-4. See also the summary of these events and sources in Pelliot, ‘Deux itinéraires’, pp. 382-3, n. 9, where the family relationships are more clearly expressed than in Southworth. 43 Southworth, ‘Origins of Campā’, pp. 302-4. Maspéro, Royaume de Champa, pp. 63-5 already saw the identity of Di Zhen (his ‘Ti Tchen’) with Gaïgārāja, but because of his conviction that Linyi and Champa were one did not give it any special significance. 44 Southworth, ‘Origins of Campā’, p. 318.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

taken over by Cham who were already adopting Sanskritic titles in -varma and -dharma. It is

true, of course, that yang (a title indicating sacred or royal) mai/mah (gold) sounds Austronesian,

and the record of C96 – allegedly going back to the fifth century – indicates that the Cham may

not yet have entirely adopted a tradition of varma (Chinese ba-mo) names for their rulers.

Southworth is not concerned with a possible ethno-linguistic difference between Linyi

and Champa. He accepts that they may have been the same – that is, Austronesian – but does not

give any attention to the question, saying merely that when the Chinese sacked the capital of

Linyi in 605 they found books in the Kunlun language, ‘almost certainly a proto Caü or

Austronesian dialect’.45 This is not certain. The references to ‘Kunlun’ show only that it meant

something in maritime Southeast Asia, and could have referred to Mon-Khmer as well as

Austronesian. In fact, if Kunlun was a Chinese representation of kuruï, as some have argued, it

was almost certainly Mon-Khmer rather than Austronesian. Southworth was assuming that the

sacked capital was Trà Kiệu, the earliest archaeologically studied Cham site.

The first instance in which the Chinese records of Linyi seem to literally agree with a

name of a Champa ruler in an inscription is Lu-tuo-ba-mo for Rudravarman early in the sixth

century. Following this the Chinese recorded three more Fan, whose names, pace Maspéro, do

not agree at all with the Champa inscriptions. Thereafter, except for the three mentioned above,

there are no more Fan in Linyi or Champa, but the Phạm (Vietnamese transcription of Fan, same

Chinese character) were important thereafter in Vietnamese history. Until the seventh century all

Fan/Phạm, with possibly one exception, were Linyi rulers, generals, or other high officials. The

last Linyi king with that title was Fan Zhenlong in the mid-seventh century; thereafter the title

was never used in Vietnamese histories for Champa, but only for Vietnamese. The single unclear

case serves to prove the point: this was the mid-sixth-century Phạm Tu, a general of the

Vietnamese rebel Lý Bí (Lý Bổn), described in Maspéro as an ‘Annamite’ who had overthrown

the ‘legitimate’ Chinese governor. Phạm Tu defeated an attack by Rudravarman of Linyi.46

Phạm Tu could very well have been someone of old Linyi serving a Vietnamese faction,

consistent with old Linyi’s bắc tiến (Northward movement), while the Chinese were applying the

45 Ibid., p. 312. 46 Maspéro, Royaume de Champa, p. 81; Keith W. Taylor, The birth of Vietnam (Berkeley: University of California Press, 1983), pp. 136ff. Maspéro uses the form ‘Lý Bổn’, an alternate reading which has evolved because some areas of northern Vietnam avoid pronouncing the name ‘Bí’ out of respect for his memory.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

name ‘Linyi’ to another power center, that of Champa.47 My proposal is that by the time of

Rudravarman, the Mon-Khmer (Katuic or Viet-Mường) chiefs of Linyi had moved north and

integrated with the Vietic peoples of Giao Chỉ, maintaining themselves as a high status group.48

Thereafter, from the mid-eighth century, Linyi disappears, new epigraphy appears in the

South, and there are no Chinese records of rulers there or in what they called Huanwang. This

period will be discussed below. In its introduction, the Song Huiyao (new translation by

Geoff Wade) says, ‘in previous dynasties, this country [Zhancheng] rarely had contact with

China’, which seems to indicate that the Chinese historians of the Song dynasty did not consider

Champa to have been Linyi, the history of which is based entirely on Chinese records of

relations with it.49 Further doubt about the identity of early Champa and Linyi is implied in

the seventh-century writings of Xuanzang and Yijing, who spoke of ‘Zhan-po’, clearly an

attempt at a phonetic rendering of ‘Champa’. This suggests that the official Chinese histories,

writing then only of Linyi, ignored the already existing and separate Champa, perhaps because

the latter centers were not yet sufficiently developed to become of interest to China as trading

partners or subordinate polities.50

All of the earlier historians who identified Linyi as early Champa believed that the Cham

had moved overland into their modern habitat and had thus been present in what is now northern

and central Vietnam for hundreds, perhaps even thousands, of years. Given the new consensus

based on linguistics that the Cham only arrived by sea in the last centuries BCE, probably in

relatively small numbers, in an area already populated by other ethno-linguistic groups, and then

expanded westward into the highlands, it seems incredible that by the early centuries CE they

47 Southworth, ‘Origins of Campā’, p. 309, where he has mistranslated Maspéro concerning the result of the fight between Rudravarman and Phạm Tu, saying that the former was victorious and that it marked the final victory of Xitu over Lin-yi. Maspéro (Royaume de Champa, p. 81) wrote, ‘Il [Rudravarman] y rencontre un général de Ly Bon, Phạm Tu, est défait et réintegre son royaume’ (‘he met a general of Lý Bổn, Phạm Tu, was defeated and regained/returned to his kingdom’. Southworth (p. 309) rendered this as ‘Rudravarman ... attack[ed] one of the generals of Lý Bôn named ... Phạm Tu, whose territory was defeated and reintegrated by Linyi’. Taylor, Birth of Vietnam, p. 138, notes correctly that ‘the Lin-i army was defeated and withdrew’. 48 It must be emphasized that the terms ‘Mon-Khmer’, ‘Vietic’, and ‘Viet-Mường’ are linguistic terms and that all are now included by linguists within the linguistic grouping Mon-Khmer. This means that even if the name ‘Việt’ derives from the ‘Yue’ of ancient southern China, the latter were not the linguistic ancestors of modern Vietnam, the languages and dialects of which (except for the Tai of the Northwest and some small Tibeto-Burman and Kadai groups) had their origins in Nghệ An and neighboring regions of Laos. No doubt, taking this into consideration, the prehistory of Vietnam must be revised. 49Geoff Wade, translation of the Song Huiyao. As Wade puts it, ‘earlier polity names which are today [emphasis added] associated with Champa, such as Lin-yi and Huan wang, were not associated by the SHYJG compilers with this polity of Zhan-cheng’. It must not be forgotten that the modern association of the names Linyi and Huanwang with Champa dates only from the time of European scholarship. 50 Stein, Linyi, pp. 234-5.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

could have dominated the region farther north which Chinese accounts of Linyi imply. Much

later they did expand into former Linyi territory in Quảng Trị and Quảng Bình and for a time the

Chinese seem to have used the old name ‘Linyi’ for what was really the new ‘Zhan-po’.

III. Relations with Giao Chỉ and Vietnam

The standard view of relations between Champa and its northern neighbors (including

Linyi) has been that of constant aggression from the north. As Nayan Chanda quoted Paul Mus,

the Vietnamese ‘flowed across Indochina like a flood carrying off other peoples wherever they

occupied lowland rice field[s] or where it could be put under rice’. For many years Michael

Cotter’s treatment of ‘the Vietnamese Southward Movement’ (Nam tiến) was the standard.51

Until very recently, when there has been progress among Vietnam specialists, no historian took a

critical view of this mantra about the political economy of ancient Vietnam.52 On the contrary,

the Chinese records of Linyi, which are the only original records of that entity, complain

constantly of Linyi aggression against Giao Chỉ, a real Bắc Tiến (northward movement) rather

than the Nam Tiến imputed to the Vietnamese as their constant policy. As will be seen below,

this continued into the time of the first attempt at an independent Vietnam by Lý Bí, and the first

war between fully independent Vietnam in the tenth century and Champa began with the latter’s

intervention into the former’s internal politics.

Thereafter the conflicts between the two parties were mostly between equals, and in the

last quarter of the fourteenth century (1360-90) the Cham very nearly conquered all of Vietnam.

Only after the failure of that adventure was Đại Việt clearly dominant; thus the term Nam Tiến, if

accurate at all, may only be applied from the beginning of the fifteenth century. Indeed, a new

generation of scholars of Vietnam reject entirely the concept of Nam Tiến. As Li Tana puts it, it

was a ‘series of different episodes responding to particular events or opportunities’; and Keith

Taylor, one of the leading historians of Vietnam, has written, ‘I do not believe that such an event

[Nam Tiến] ever took place’. Like Li Tana, he writes in more detail of a series of episodes.53

51 Nayan Chanda, Brother enemy (New York: Harcourt Brace Jovanovich, 1986), p. 49, quoting Paul Mus, Viêt-nam: sociologie d’une guerre (Paris: Éditions du Seuil, 1952), p. 17; Michael G. Cotter, ‘Towards a social history of the Vietnamese Southward Movement’, Journal of Southeast Asian History, 9, 1 (1968): 12–24. 52. For an extended discussion of this and other such mantra see Michael Vickery, ‘Two historical records of the Kingdom of Vientiane’, in Contesting visions of the Lao past: Lao historiography at the crossroads, ed. Christopher E. Goscha and Søren Ivarsson (Richmond: Curzon Press, 2004), pp.3-35. 53 Li Tana, Nguyên Cochinchina, Southern Vietnam in the seventeenth and eighteenth centuries (Ithaca: Cornell Southeast Asia Program, 1998), pp. 19, 21, 28; Keith W. Taylor, ‘Surface orientations in Vietnam: Beyond histories of nation and region’, Journal of Asian Studies, 57, 4 (1998): 951, 960.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

IV. The Narrative History of Champa

This section is concerned with a criticism of the history of events as described in

Maspéro, a history which has been adopted by most later writers. One, who issued an early

warning on Maspéro, was Rolf Stein, who said that

historians who are not Sinologists and who have access only to Maspéro’s work can easily be led astray, not only by a certain number of errors in [its] translations but also, most importantly, because Maspéro’s descriptions are always reconstructions: the [Chinese] texts are almost always incomplete and often contradictory. Maspéro has taken the various components [of these texts] and, without any critical comparison, combined them into a continuous story which seems to be backed up by the texts when in fact [this is the case] only for its various elements coming from different sources of different dates.” 54

That is, Maspéro’s narrative history of Champa is sometimes historical fiction.

Although Stein was then studying only the Linyi problem, his criticism of Maspéro is

valid throughout the latter’s Le royaume de Champa, with respect not only to his use of Chinese

sources but also to his treatment of inscriptions; one of the objectives here will be to sort out the

evidence behind Maspéro’s overenthusiastic synthesis. Except for the Linyi problem,

however, I shall be concerned only with those periods of Champa history for which there is

adequate evidence from Champa inscriptions to compare with Maspéro’s syntheses based on the

use of Chinese and Vietnamese sources. There are long periods – for example, Maspéro’s

chapter 8 (‘Struggle with the Mongols’) and much of chapter 9 (‘Zenith’), such as the late

fourteenth-century invasion of Vietnam by the Cham and events leading up to the Vietnamese

occupation of Vijaya in 1471 – for which there are no inscriptions and the history to date has

been compiled entirely on the basis of Chinese and Vietnamese histories, on which new work

must still rely. There are no doubt errors there, too, but correction will depend on thorough

revision of those sources by competent specialists.

In particular, I intend to show that there was no single kingdom or state of Champa, and

that the regions distinguished by their epigraphy (discussed above) and corresponding to their

geography, were often quite separate, even rival, polities. I shall take it as given, following the

discussion above, that Linyi was not Champa – except possibly from the early seventh century, 54 Stein, Linyi, p. 72. Further critiques of Maspéro are found throughout Stein’s study, and, where relevant, will be noted below.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

when an expanding Champa polity occupied former Linyi territory and was given that name in

Chinese reports on the region. The first objection to that interpretation in a solid historical work

of which I am aware was by Keith W. Taylor: ‘Champa is a generic term for the polities

organized by Austronesian speaking peoples along the central coast of Vietnam... an

archipelagically-defined cultural-political space’. Taylor’s assertion is based on research and

writing by Nora Taylor, whose ‘basic argument is that Champa was never a unified kingdom but

rather pockets of power that competed with each other’ – the position adopted here. Among

currently active Champa specialists, William Southworth has also stated without hesitation that

‘Champa’ consisted of independent entities.55

As noted above, the first two inscriptions – Võ-Cạnh near Nha Trang and Đông Yên

Châu – are isolated records which cannot yet be integrated into Champa history, and the first

Champa in which continuing development is revealed by inscriptions was situated in the Thu

Bồn valley centered on Mỹ Sơn, in the modern province of Quảng Nam.

The first kings named in inscriptions

Except for the misconceived identification with Linyi chiefs and the further

interpretations resulting therefrom, there is little that is controversial in Maspéro’s treatment of

the earliest Champa chiefs. The next three inscriptions, dated on palaeographic grounds in the

fifth century, offer devotion to a god Bhadreśvarasvāmin; the author of two of them was a

Bhadravarman, who has been assumed as the author of the third as well. Although the first two

are in Sanskrit and do not name ‘Champa’, their locations, following the Đông Yên Châu

evidence, permit their identification with Champa. One was found at Mỹ Sơn, one nearby, and

the third at the mouth of the Đà Rằng river near Tuy Hòa, far in the South. This distribution fits

well with the identification of Champa as a seafaring society – though not, however, with Finot’s

conclusion that ‘the three inscriptions…in Bhadravarman’s name prove, moreover, that the

Cham people formed a unitary State and not a series of small independent kingdoms’.56 Finot

was certainly wrong about Champa unity.

The next important record of early Champa royalty has been considered to be inscription

C96 dated 658, in Sanskrit; it is interesting not only for Champa history, but for the sojourn of a 55 See Keith W. Taylor,‘The early kingdoms’, in Tarling ed., Cambridge History of Southeast Asia, p. 153 and Southworth, ‘Notes on the political geography’; the quotation on Nora Taylor’s argument is from a personal communication with her. 56 Louis Finot, ‘Deux nouvelles inscriptions de Bhadravarman Ier’, BEFEO, 2 (1902): 185-91.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Champa prince at the major seventh-century center of Cambodia where he married a daughter of

King Īśānavarman – a rare instance of amicable relations between regions which were later so

often at war. This is the first local record of the name ‘Champa’ (in the Sanskrit terms

Campāpura and Campānagara) and it is approximately contemporary with the Cambodian

Sanskrit inscription K.53, recording in 667 CE an envoy sent to the ‘ruler of Campā’

(Campeśvara). Inscription C96 starts with a king Gaïgārāja who abdicated in order to go and see

the Ganges. Then there were kings named ‘Manorathavarman’ (?, a name restored hypothetically

by Coedès), Rudravarman, Śambhuvarman, Kandharpadharma, and others, with the last,

Prakāśadharma-Vikrāntavarman, following the intercalated story concerning Cambodia.57

Part of the same Champa royal genealogy seems to be found in the badly damaged, and

untranslated, C73A, also from Mỹ Sơn, which lists Gaïgeśa, Rudravarman, Śambhuvarman,

Kandharpadharma, and Prakāśadharma, along with an apparently retrospective reference to

Bhadravarman (not named in C96). 58 Reflections of the same story are also in the very

fragmentary C81 at Mỹ Sơn, in which a Champa king Prakāśadharma, possibly also named

Vikrāntavarman, appears to be called ‘of the lineage of Gaïgeśvara’ (Gaïgeśvaravaïśajaþ).

Named also are a Bhadravarman and a Rudravarman.59

This of course conflicts with the interesting proposal by Southworth, noted above, that a

Chinese story of a certain Di Zhen/Ðịch Chân who abdicated and went on a pilgrimage to the

Ganges in India identifies him with the Gaïgārāja of Mỹ Sơn inscription C96, and is the first

concordance of an individual recorded by the Chinese with a king in a Champa inscription.

Maspéro also made this identification, but for him it was of no special significance given his

assumption of Linyi-Champa identity. For Maspéro Bhadravarman was the Linyi chief Fan Huda

and Di Zhen- Gaïgārāja was his son. 60 I would say, however, that the concordance is as

important as seen by Southworth, but that the Chinese were recording a Champa tradition known

to them – probably mythical and not necessarily factual events in the life of a real king.

In contrast to Maspéro, who identified Bhadravarman with the Fan Huda of Linyi,

preceding Gaïgarāja, Southworth considers that the ‘religious content, regnal titles, and

palaeography of the inscriptions of Bhadravarman do not support such an early datation’ and that

57 Coedès’ reconstruction is in his ‘Note sur deux inscriptions du Champa’, BEFEO, 12, 8 (1908): 15-7. 58The correspondence is improved by Finot’s redating of the inscription from fifth century to sixth century śaka; Finot, ‘Stèle de Śambhuvarman’, p. 207 and Finot, ‘Inscriptions du Musée’, p. 5. 59 For Inscription C81, see Jacques ed., Études épigraphiques, pp. 110-1, where it is numbered “VI’. 60 Maspéro, Royaume de Champa, pp. 64-5.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

his ‘reign can be placed in the temporal displacement between Manorathavarman and …

Rudravarman’. This does not agree with the apparent readings of the fragmentary C73A and C81,

not given attention by either Maspéro or Southworth, and which (especially C73A) seem to refer

back to Bhadravarman as an ancestor and the founder of the temple of Śrīśānabhadreśvara at Mỹ


Some of the important works of architecture and sculpture are believed to date from this

period; in this respect Trà Kiệu , the importance of which has been shown by archaeology, may

have been more impressive than Mỹ Sơn, although no buildings have survived until modern

times, and the site lacks inscriptions. Following this long period of predominance of the Thu Bồn

valley, which for the Chinese was still Linyi, the historical scene shifts abruptly to the South,

Phan Rang (Pāõóuraïga ) and Nha Trang, which Maspéro termed – incorrectly, as I shall show –

‘Huanwang and Pāõóuraïga Hegemony’.62


Huanwang appears in Chinese histories in the mid-eighth century, just when the name

Linyi drops out of use; in fact, some official histories are explicit that the name ‘Huanwang’

replaced that of ‘Linyi’. The name has never been successfully explained, but its literal meaning

is apparently ‘ring’, ‘circle’ [of] ‘King(s)’.63 Because Huanwang appeared just when epigraphy

ended in north Champa and continued for over 100 years in the South (Phan Rang and Nha

Trang), the early students of the area (in particular Maspéro), continuing in their belief that Linyi

and Champa were one and singular, identified Huanwang with Champa, and more particularly

with the ‘hegemony of Pāõóuraïga’, in spite of the great distance between the latter and what

appears to have been the region of Huanwang in the Chinese sources.

Pelliot looked at the location of Huanwang in his study of the route taken by Jiadan, and

deduced that its frontier was at Đồng Hới, with six more days travel’ to its capital. Stein,

however, considered that the frontier which Pelliot located at Đồng Hới was farther north on the

Gianh River. They were also in disagreement about the capital of Huanwang, several days to the

south. Pelliot, believing that Huanwang was Champa, wished to place it in Quảng Nam on the

basis of the names of Cham capitals which Aymonier had found in their chronicles. Stein was 61 Finot, ‘Stèle de Śambhuvarman’, pp. 206-11; Finot, ‘Notes d’épigraphie XI’, pp. 928-9. 62 Maspéro, Royaume de Champa, ch. IV. 63 See ibid., p. 95, n. 1 for some of the sources. The literal translation of ‘Huanwang’ is from Geoff Wade, personal communication.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

more circ*mspect – favoring, it would appear, a location between Quảng Trị and Huế. Both,

however, were overly concerned with the supposed identity of Huanwang and Linyi and whether

the capital of one was the capital of the other, and also a capital of Champa.64 Accepting that

Huanwang was a region in the North, quite distinct from Cham Pāõóuraïga, means that very little

may be said about it, for the Chinese notes on it are sparse and short, with no names of chiefs,

but indicating that Huanwang was a source of trouble like earlier Linyi and the northern Champa

regions. In either case, this would make Huanwang a fit with the area of Maspéro’s later ‘sixth

dynasty’, that of Đồng Dương/Indrapura, which left important monuments as far as Quảng Bình

(see below).

The official Chinese histories stopped using the name Huanwang by 877, saying that it

had been replaced by ‘Zhan-cheng’ (city of the Cham), not the phonetic transcription ‘Zhan-po’

known since the seventh century; in one Chinese source, however, that replacement had already

been made in 809. Noting the Song Huiyao remark quoted above about the paucity of China-

Champa contact in earlier centuries, it may be conjectured that Huanwang, although perhaps

ethno-linguistically Cham, was not identical with the polity which the Chinese had known as

‘Zhan-po’ since the seventh century.65

As Boisselier cogently put it,

‘Maspéro has rather artificially connected the new ‘dynasty’ [Panduranga] to the ‘Areca clan’… contrasting it with the ‘Coconut clan’, from which the Northern dynasties had supposedly sprung ‘more directly’, he believes that this Huanwang period corresponds to ‘Panduranga hegemony’.…Nothing, in any case, indicates that Panduranga…exercised any kind of hegemony during the Huanwang period.66

Here Boisselier was referring to a firm belief by Maspéro that the ruling families of

Champa were divided into two clans, the Areca clan and the Coconut clan; ‘these two clans’,

Maspéro suggested, ‘fought for supremacy down through the centuries…The Areca clan ruled

the State of Panduranga [while] the Coconut clan dominated the North’. He evoked them in

several contexts in his history of Champa – stating, for example, that ‘the inscriptions allude to

64 Pelliot, ‘Deux itinéraires’, pp. 184-6; Stein, Linyi, pp. 79-81. Pelliot, of course, did not know of the inscriptions which give importance to Pāõóuraïga (Nha Trang and Phan Rang) in the eighth century. 65 On the date 809 see Stein, Linyi, p. 234 and Pelliot, ‘Deux itinéraires’, p. 196. This detail was brought to my attention by Népote, ‘Champa, propositions’, part 2, p. 87. Together with his citations of seventh-century ‘Zhanbo’ from Xuancang and Yijing, Stein added, ‘I have given up trying to explain the transitional but short-lived name of Huanwang’. 66Boisselier, Statuaire du Champa, pp. 61-2.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

mythical traditions…there are two traditions, one of which seems to belong to the country of the

“Areca clan” and the other to that of the “Coconut clan”’. Uroja, Maspéro added, was the

‘mythical ancestor’ of the latter clan.67 In fact, however, there is only one inscription which

mentions these clans at all: C90A from Mỹ Sơn, dated 1080, in which it is said that a king

Harivarman, Prince Thãï, descended from both clans, the Coconut on his father’s side and the

Areca on his mother’s. There is nothing in this inscription which provides evidence for a

geographical location of either clan. As Huber had already noted, ‘the inscriptions give us no

information on these Areca and Coconut clans; and as Boisselier reemphasized, ‘the references

to the “Areca”…and “Coconut” clans…only appear fairly late in Cham epigraphy and almost by

accident’.68 Thus, along with the supposition that Huanwang represented Pāõóuraïga hegemony,

all of Maspéro’s remarks about the importance of these clans in intra-Champa relations must be



Contrary to the standard interpretation, we have seen that Huanwang probably did not

represent all of Champa, and given the logistics of the time it is aberrant to assume that

Huanwang, clearly in the North according to Chinese records, could have been equated with

Pāõóuraïga in the far South. Huanwang could have been seen by the Chinese as replacing Linyi,

in terms of political geography as well as in name, and to that extent Maspéro may have been

correct in accepting the Chinese statements that the name Huanwang replaced the name Linyi.

However, a major objective here has been to demonstrate that Linyi was not Champa and thus

neither was Huanwang if it was just a replacement for Linyi.

All of the Champa inscriptions of this period (with dates from 774 to 854) are in the

South, either in the region of Phan Rang (most of them) or at the Pô Nagar temple in Nha Trang.

Although three of the rulers named asserted that they ruled over all of Champa

(Pçthivīndravarman and Indravarman in Phan Rang) or over Campa-pura (Harivarman in Nha

Trang), given the absence of all remains from their time in the North, there is no need to believe

that they had effective control over territory much beyond Nha Trang, any more than one needs

67Maspéro, Royaume de Champa, pp. 18, 43, 154. 68On the inscription see Finot, ‘Notes d’épigraphie XI’, pp. 933-40; Jacques ed., Études épigraphiques, pp. 115-22. Quotations are from Édouard Huber, ‘Le clan de l’aréquier’, BEFEO, 5 (1905): 170-5; and Boisselier, Statuaire du Champa, pp. 61-2.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

to believe the claims of Angkor kings that they held sway as far away as China. All we may

accept about the claims of the Pāõóuraïga rulers is that they considered their region to be part of

Champa, as seen in the inscription of Harivarman in his Campa-pura at Nha Trang, giving the

charge of Pāõóuraïga -pura (presumably the Phan Rang area) to his son Vikrāntavarman.

Maspéro began this chapter, after listing the kings, with a major excursus into historical

fiction, at the time of the shift of the important Champa center from Mỹ Sơn to the South. He

wrote that ‘on Rudravarman’s death, around 757, the powerful men [les Grands] gave the crown

to one of their own’. First, however, the name of that Rudravarman is nowhere attested in an

inscription. The last inscriptions in Mỹ Sơn are of a king Vikrāntavarman (C77, C80, C97, C99,

all dated only in the seventh century śaka [AD 678-778], and C74/741. The first inscription in

the South is C38/774, 784, referring to Satyavarman and a Vikrāntavarman – who, given his

genealogy, was not the Vikrāntavarman of Mỹ Sơn. Around 749, the Chinese referred to a

certain Lu-duo-luo or Lu-duo (which may reasonably be interpreted as ‘Rudra’) as ruler of Linyi,

apparently the last whom they recognized. 69 The date 757 in Maspéro is merely a guess,

apparently based on the fact that after 758 the Chinese no longer used the name Linyi.

The ‘Grand’ who was then given the crown and who was assumed by Maspéro to have

originally been in the North, was Pçthivīndravarman, the name of an ancestor with which the

Pāõóuraïga inscriptions begin their genealogy. Maspéro wrote that ‘it was probably [sic]

following his [Rudravarman’s] death that Pçthivīndravarman shifted the royal residence to the

South. The new ruler, Maspéro believed, ‘probably [sic] belonged to the great princely family of

Pāõóuraïga, the “Areca clan”, and he continued to live in the South’.70 Of course, there is no

record of Pçthivīndravarman belonging to the Areca clan, nor any reference to such clans at that

date; and prior to that there is no information about a ‘great princely family of Pāõóuraïga’. All

of Maspéro’s imaginative text was due to his assumption that there was a unity of Linyi and

Champa as far as Phan Rang; that therefore the successor of the last Linyi king known to the

Chinese must have assumed power in the North; and that because the following local records are

from the South, that must have been the original clan home of that successor

(Pçthivīndravarman), named there as uncle (Bergaigne) or father (Maspéro) of the second king to

leave an inscription, Indravarman.

69 Maspéro, Royaume de Champa, p. 93 and note 3 and p. 246, table. On the inscriptions see Bergaigne, ‘Ancien royaume’, pp. 242-60. Maspéro referred to Vikrāntavarman II and III. 70 Maspéro, Royaume de Champa, pp. 96-7 and note 7. See above on the Areca and Coconut clans.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Maspéro continued on the status of Pçthivīndravarman, justifying in a peculiar way the

assumption that he did not inherit the throne but was named by the ‘grands’ – by citing the first

dynastic inscription of the following Đồng Dương dynasty dated 875. Here, however, he may be

excused for following an interpretation of Finot. This inscription is for the glorification of

another Indravarman over a century later, but at one point refers to him as Pçthivīndravarman

(Pçthivī-Indravarman, just as still another Indravarman was named at one point Rājendravarman


This point requires elaboration. This inscription, C66 – ‘the first stela from Ðồng-Dương’

according to Finot – was erected by King Indravarman to the glory of himself and his ancestors.

Like the records of this dynasty in general and the architectural remains of the site, it indicates

devotion to a type of Mahāyana Buddhism, unlike the records of either earlier Mỹ Sơn or

Pāõóuraïga. There is also a list of ancestors beginning with Bhçgu, different from the tradition of

their predecessors and at least two of whom are quite legendary: Parameśvara (Śiva) and Uroja

(‘breast’), then Dharmarāja, Rudravarman, Bhadravarman (son of Rudravarman) and

Indravarman (son of Bhadravarman). Finot commented that the name ‘Uroja’, although Sanskrit,

is peculiar to Champa mythology and unknown in India, and that ‘Dharmarāja’ sounds peculiar

for the name of a Champa king. He proposed that they were both fictitious and that

Indravarman’s dynasty only began with his grandfather Rudravarman. Certainly ‘Uroja’ is

mythical, but ‘Dharmarāja’ does not sound peculiar in a polity devoted to Buddhism.

Although Buddhist, Indravarman also praised the traditional Mỹ Sơn lingam of

Śaübhubhadreśvara. Finot was surprised that he did not repeat the ‘true story’of that shrine and

instead inserted a new mythical lineage, and he therefore described Indravarman as a ‘usurper’.

This is because Finot, like Maspéro later, assumed a single unified Champa from Linyi

throughout, so that kings who could not be attached to previous rulers must have usurped the

throne. What we see in C66 is a new mythology of a new chiefly family applied to an already

traditional site. If we accept instead that Champa was never a unitary polity, there is no problem

with chiefs in Đồng Dương – whose religion and prehistoric legends were different – becoming

important separately from other kings in the South. It is not necessary to go along with Finot’s

idea that ‘here we are perhaps seeing once again the reluctance of a usurper to proclaim the

71 See inscription C25 in Bergaigne, ‘Ancien royaume’, number XXII, A, ii: ‘Srīmān Rājendra(va)rmmā’, translated by Bergaigne as ‘this king Śrī Indravarman’. This type of double naming was recognized as metri causa by Schweyer (‘Dynastie d’Indrapura’, p. 207).


ARI Working Paper No. 37 Asia Research Institute ● Singapore

works of a dynasty whose place he had taken’ (meaning the earlier dynasty of Mỹ Sơn), or his

astonishment that Indravarman ‘mentions the name of Pçthivīndravarman, who reigned around

[778]’.72 Given the structure of the text, it is preferable to take the title Pçthivīndravarman here

as belonging to the Indravarman of Đồng Dương around 875, not the Pçthivīndravarman of


Returning to the problem of Pāõóuraïga, Maspéro, engaged in some clear historical

fiction in identifying Huanwang with Pāõóuraïga and in pasting together the Chinese reports on

Huanwang and the Champa inscriptions from the far South (Phan Rang and Nha Trang). ‘Having

scarcely taken power’, he begins, Harivarman (successor of Indravarman according to the

inscriptions) in 803 invaded the Chinese provinces of Hoan and Ái, located in present-day Nghệ

An and Hà Tĩnh. This is because the Chinese sources recorded that Huanwang had attacked but

without mentioning any names of rulers. It is true that Harivarman’s Sanskrit inscription says

that he ‘burned the Chinese’, but in that context it must be taken as the same type of hyperbole as

seen in Angkor kings’ exaggerated claims of the extent of their borders and subordinate

kingdoms (which, in one inscription of Jayavarman VII, included Vietnam and Java). The

logistic possibilities of the time would not have permitted a Pāõóuraïga attack on Hoan and Ái,

unless it was an expedition by sea. Besides the distance, Pāõóuraïga is separated from those

provinces by the three great physical barriers along the coast of Vietnam: Mũi Đại Lãnh (the

French Cap Varella); Ðèo Hái Vân (Pass of the Clouds or Col des Nuages), just north of Đà

Nẵng; and Ðèo Ngang at Hoành Sơn in Hà Tĩnh. Tthe Chinese records of invasions by

Huanwang cited by Maspéro, however, imply attacks overland.73

What, then, can we say about eighth- to ninth-century Pāõóuraïga on the basis of its own

inscriptions? Very little, except for a list of rulers, and the emergence of a new mythical lineage

ancestor, Vicitrasagara. Boisselier remarked that the Phan Rang inscriptions were by dignitaries,

while those in Pô Nagar were royal.74 This does not seem to be accurate. C38/774, 784 is indeed

royal, but so are C25/799, C24/801, and C14/854 from the Phan Rang area. The inscriptions

which give particular importance to an official – C19/seventh century ś. (Phan Rang), C37/813

(Pô Nagar), C31/817 (Pô Nagar) – relate the exploits of a famous warrior, Senāpati Pār, who

claimed victories over the Cambodians, but in form they too are royal inscriptions.

72 For inscription C66 see Finot, ‘Notes d’épigraphie VI’, pp. 84-99. 73 Maspéro, Royaume de Champa, p. 105. 74 Boisselier, Statuaire du Champa, p. 62.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

The names of gods, kings, and mythical lineage ancestors show clearly that Pāõóuraïga

did not represent an emigration of royalty from North to South. The records indicate the

emergence of a local elite – probably, as Southworth describes, because

the collapse of the early Tang trade and the demise of Guangzhou during the mid-eighth century CE also destroyed the commercial dominance of the Thu Bồn valley system, and when the South China Sea trade was transferred to the ports of the Red river delta during the late eighth century CE, an entirely new pattern of trade emerged… [There was] a major shift in the displacement of trade wealth to south to central Vietnam, in particular to the ports of Nha Trang and Phanrang, on the sea-route from Java to northern Vietnam’… [and] the independent states of Kauñhāra and Pāõóuraïga … flourished here during the late eighth and early ninth centuries.75

From Pāõóuraïga back to the North

An example of the unnecessary problems caused by the assumption of a unitary Champa

is seen in the Maspéro-Finot controversy over the beginning of what Maspéro called the ‘sixth

dynasty’: that of Đồng Dương, again in the Thu Bồn but not at Mỹ Sơn, at the end of the ninth

century following the ‘hegemony of Pāõóuraïga’. For Maspéro this dynasty began with a king

Indravarman mentioned in the so-called ‘first inscription’ of Đồng Dương (C66 dated 875) as

King of Champa. He became king because the ‘childless Vikrāntavarman [last king of the

southern dynasty in Pāõóuraïga, last date C14/854] would himself have chosen him as

successor’.76 There is no record of what children Vikrāntavarman did or did not have; Maspéro

was here indulging in pure historical fiction. The inscription in question (C66), however, says

that Indravarman was son and grandson of Bhadravarman and Rudravarman, both called ‘king’,

although according to Maspéro they did not reign. His unexpressed reason for that belief was that

in their time the king of Champa was still in Pāõóuraïga. Indravarman’s ancestry possibly also

went back to still another king named Dharmarāja.

Finot, on the other hand, argued that they did reign, and that for Bhadravarman this belief

was supported by the inscription of An-Thái dated 902 – from the reign of Indravarman, but

referring to Bhadravarman as a previous king. In Finot’s review of Maspéro, however, there is a

75 Southworth, ‘Origins of Campā’, p. 318. 76 Maspéro, Royaume de Champa, pp. 109-11. ‘First’ has been the traditional designation of this inscription, also discussed above, although Schweyer, ‘Chronologie des inscriptions’ precedes it with two newly discovered and undated texts from that period (C205 and 206) which refer to Rudravarman, a predecessor of Indravarman; she has published these texts in ‘Vaisselle en argent’.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

curious statement relating to Maspéro’s suggestion that Indravarman had granted his ancestors

fictitious titles. Finot said, ‘For supposing that a Cham ruler was able to bestow the title of king

on his ancestors…he could certainly not call them kings of Champa, which everyone knew they

had not been’. In the view of Finot, like Maspéro, the real kings of Champa in the time of

Bhadravarman and Rudravarman were in Pāõóuraïga. 77 Of course, between the last date of

Vikrāntavarman of Pāõóuraïga (854) and the first date of Indravarman (875) there is room for

two ancestors of Indravarman in the North, and it should be assumed that he had been preceded

in the Thu Bồn valley region by at least two named ancestors who were contemporary with the

last kings known in Pāõóuraïga.

The controversy could have been avoided if it had been recognized that the ‘Champa’ of

the region of Đồng Dương was a quite separate polity from Pāõóuraïga, and that the dynasty

represented by Indravarman could well have begun with his father, grandfather and great-

grandfather, even though no inscriptions from their reigns have been found.78 That is, pace Finot,

Maspéro and Schweyer, there was no transferal of the capital from North to South in the eighth

century, nor again from South to North in the ninth; rather, a new political economic ascendance

of the North took place, probably related to changes in the international maritime routes. Like the

rise of Pāõóuraïga and the apparent decline of the Thu Bồn area in the eighth to ninth centuries,

the reemergence of the North was undoubtedly linked to another change in the international trade

routes from China to Indonesia and India.

Maspéro began the events of the reign of the first Indravarman of Đồng Dương with a

piece of total fiction based on a misunderstanding of Cambodian history which was in vogue

until undermined in 1927 by Philippe Stern and given the coup de grâce by Coedès in 1928,

namely the elief that the monuments and some of the associated inscriptions now known to date

from Jayavarman VII (1181-1220?) belonged to the ninth century. Thus, relying on Aymonier’s

erroneous interpretation of the Banteay Chhmar inscription, Maspéro garbled an account of a

Cambodian intervention in Champa in what he thought was the late ninth or early tenth centuries,

but which actually belongs in the twelfth century, as demonstrated later by Coedès.79

77 Finot, review of Maspéro, p. 290. 78 The two inscriptions (C205, C206) on silver objects published by Schweyer (see footnote 75) and attributed to Đồng Dương, naming a Rudravarman whom she accepts as grandfather of Indravarman, are small movable objects in a private collection, and the attribution can only be hypothetical. 79 Maspéro, Royaume de Champa, p. 113; Philippe Stern, Le Bayon d’Angkor Thom et l’évolution de l’art khmer (Paris: Librairie Orientale Paul Geuthner; George Coedès, ‘La date du Bayon’,


ARI Working Paper No. 37 Asia Research Institute ● Singapore

From the mid-tenth century reign of the king whom Maspéro named Indravarman III but

Schweyer Indravarman II, the Chinese again provide names of Champa rulers, none of which

agrees with any -varma name in their inscriptions.80 An interesting aspect of this period is that

this so-called ‘Đồng Dương, or Indrapura, dynasty’ expanded its area northward as far as Quảng

Bình, as evidenced by the fact that nearly half of their inscriptions were set up north of Huế – a

subject not touched upon by Maspéro but given importance in the special studies of this period

by Schweyer.81 As Boisselier noted, moreover, in two of these extreme northern locations there

were large sanctuaries reminiscent of Đồng Dương itself: at Đại Hữu and, in particular, Mỹ Đức.

In Boisselier’s words, ‘this fairly large complex [Mỹ Đức] was made up of a front section of

three sanctuaries, coming after a tower with three doors and a southern building, inside a wall

which closed off a large room…’ These dispositions were reminiscent of Đại Hữu and Đồng

Dương and, as Huber remarked, ‘the large stela of Lạc-Thành’ (farther south in Quảng Nam),

which he described as ‘along with the great inscription of Hà-trung in Quảng Trị, the most

beautiful epigraphical monument left behind by this dynasty’. The Đà Nẵng museum catalogue

also lists several objects from another site in Quảng Trị named Đa Nghi.82

This is important to note because in the 980s the first war with Vietnam broke out, ending

with a Champa defeat, a change of dynasty and – in the standard interpretation – a new

southward move of ‘the capital’. The story has been told based almost entirely on the Chinese

records, with names of Champa kings from epigraphy fitted in haphazardly; the result, I shall

argue, is serious misunderstandings in the literature to date. Let us start with an inscription of a

king Jaya Indravarman at Pô Nagar in Nha Trang dated 965 (C38D2), the only inscription so far

ascribed to the Đồng Dương dynasty in the South. This would be Schweyer’s Indravarman II,

who reigned from about 916 to 972 and was the author of an inscription (C148) in Lai Trung

(Thừa Thiên), not far from Huế. He was probably also responsible for another inscription at Pô

Nagar (C39) which can only be dated to sometime within that century. Maspéro also mentioned BEFEO, 28 (1928): 81-103; Coedès, ‘Nouvelles données chronologiques et généalogiques sur la dynastie de Mahīdharapura’, BEFEO, 29 (1929): 297-330, including the inscription of Banteay Chhmar (pp. 309-315). 80 Schweyer is quite right to say that ‘it seems impossible to me to use continuous numbering for the kings of Campā as Maspéro does…for [they] belong to different dynasties, each of which must have had its own internal numbering system’ (‘Dynastie d’Indrapura’, p. 205, n.1). This statement, however, implicitly negates her acceptance elsewhere of a single Champa with kings shifting capitals between very distant locations. 81 Ibid. and Schweyer, ‘Vaisselle en argent’. 82 Boisselier, Statuaire du Champa, pp. 133,136; Huber, ‘Épigraphie de la dynastie’, pp. 285 and 298-9 (in Jacques ed., Études épigraphiques, pp. 259 and 272-3); Association Française des Amis de l’Orient, Musée de sculpture, p. 179. Note that in the latter source Hà Trung is erroneously placed in Quảng Nam.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

such a Jaya Indravarman, but as successor of his Indravarman III (Schweyer’s II), apparently

because of a slight change in the Chinese designation; this perhaps led Maspéro, who took every

name in the Chinese accounts literally, to invent a non-existent king. Schweyer seems to have

recognized this, attributing it to confusion in the annals at the time of dynastic change in China.83

According to Maspéro, based entirely on Chinese and Vietnamese sources, the Cham

offended the Vietnamese by seizing their ambassador. Vietnam invaded in 982, and the king

(who had reigned since 972) – let us say King A, because he is not mentioned in an inscription

and the Chinese transcription is not intelligible – was killed. The Vietnamese sacked the Cham

capital (for Maspéro Indrapura/Đồng Dương) and the next Cham king – based on Chinese

transcriptions, King B – fled, ‘probably to Phan Rang’ – a supposition for which there is no

justification either in the Chinese and Vietnamese sources or in extant inscriptions. Maspéro

admitted that for the events of this period, ‘these struggles…are not overly clear’ and that in 978,

when his King A would still have been alive, the Chinese records note a Champa king ‘Ji-nan-

da-zhi’, different from either A or B, who sent tribute.84 This is not a problem so long as it is

recognized that there was no single Champa.

It is interesting that the chronicle Tt merely says that the Vietnamese attacked ‘Chiêm

Thành’ (the Cham capital) and seized much booty; there is no indication of where the capital was.

Then, under the date 983, it says that ‘before, when the king went to attack Chiêm Thành, Lưu

Kế Tông ran away to that country and someone was sent to seize him’. The next sentence is, ‘the

new canal on the sea route has been completed’, and the reason for this is in the continuing entry:

‘When the king went to attack Chiêm Thành, via the Đồng Cổ mountain and the Bà Hoa river,

the mountain route was dangerous and difficult’. In footnotes to the text, those two locations are

said to be in Thanh Hoá, which might suggest that the Chiêm Thành concerned was in the far

north of Champa – not surprising given the northward expansion of the Cham rulers of the


Maspéro, based on Chinese sources, says that Lưu Kế Tông (whose name he incorrectly

transcribes as ‘Lưu Ký Tông’) did not disappear until 988. Under this ‘usurper’ many people fled

to Hainan, while others ‘gathered around a national claimant [to the throne] and took him off to 83 Schweyer, ‘Dynastie d’Indrapura’, p. 208. 84 Maspéro, Royaume de Champa, pp. 122-4. Maspéro (p. 121, note 1) wished to read the Chinese transcriptions (variously Bo-mei-shui, Xi-li-tuo-ban-yin-cha, Bo-mei-shui-he-yin-cha and Bo-mei-shui-yang-bu-yin-cha) as ‘Parameśvaravarman’, and this has been accepted by Schweyer, ‘Dynastie d’Indrapura’, p. 208, but this requires a large input of imagination. 85 TT, Bản kỷ, 1: 16a-b, v. 1, p. 222 and notes 2, 3.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Vijaya in 988’, where he took the name Harivarman II. However, in a long footnote Maspéro

wrote that ‘at least this is what seems to come out of the rather vague information provided by

the chronicles.He added that according to the Chinese source, in 988 a certain Bǎng Vuong La of

Champa set himself up in Phật Thành (literally, ‘Buddha city’) and Maspéro wished to interpret

the king’s titles via Chinese (Ju-shi-li-he-shen-pai-ma-luo, Viet. Cu-thi-lợi-hà-thân-bài-mà-la) as

indicating the ruler he had identified as Indravarman IV.86 Not only is there no epigraphic record

for such a person, but the identification from the Chinese source is not acceptable.

Now, where was the ‘Buddha city’ whither a Champa king moved after defeat by

Vietnamese at a place which the sources do not name clearly? The most likely place to be called

‘Buddha city’ at the time was Đồng Dương, famous from its remains as a very large Mahayana

site. For the traditionalists, however, that construal was impossible because they believed that

Đồng Dương was the capital of what Maspéro called the ‘sixth dynasty’ (875- 991), which the

Vietnamese attacked and plundered in 982. In the standard treatment since Maspéro, that Phật

thành or Phật thệ thành (Foshicheng) has been considered to be Vijaya, modern Quy Nhơn. The

interpretation of Phật thành as Vijaya results in part from another entry in the Vietnamese annals

Đại Việt sử ký tiền biên in 988, citing ‘old [Chinese? Vietnamese?] annals’ as saying that in 1000

a Champa king with the titles ‘yaï po ku vijaya’ retreated to Phật thệ thành, ‘700 li away from

the former capital’.87

Within Southeast Asian toponymy it has become accepted since Coedès’ study of Śrī

Vijaya that Chinese ‘Fo-shi’ (Viet. Phật thệ) may designate ‘vijaya’ – although this was not the

accepted construal in the time of Pelliot, who transcribed it as ‘bhoja’ – and Vijaya has so far

been the interpretation of Phật thệ thành in Champa at that time. In this case, however, it should

be noted that the titles of the king concerned, known from epigraphy as yāï po ku vijaya śrī, were

rendered comprehensibly in the Sino-Vietnamese transcriptions. Following characters which

may reasonably be construed as ‘yāï po ku’, the next term in transcription is bì-sà-xà, not

unreasonably vijaya. But why, then, did the Vietnamese and Chinese historians not transcribe the

name of the city in a similar manner as well? Given that they had a reasonably phonetic

transcription for the Sanskrit term vijaya, there is nothing known about the Champa polity of

86 Maspéro, Royaume de Champa, pp. 125-6 and 126 note 3; emphasis added. 87 Viện Nghiên Cứu Hán Nôm, Đại Việt sử ký tiền biên (Hanoi: Khoa học Xã hội, 1997), bản ký, 1:25a, p. 172.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Vijaya that should have obliged them to transcribe its name in a way also denoting ‘Buddha


The Vijaya problem

Before continuing with the historical narrative, it is necessary to open a discussion of

Vijaya, which appears – or has been interpreted as appearing – in later entries of the Chinese and

Vietnamese chronicles and in the eleventh- to twelfth-century inscriptions. It will be argued here

that ‘Vijaya’ has been misunderstood as both a name and a location, leading to erroneous

interpretations in the historical narrative. The name ‘Vijaya’ in all modern literature on Champa

is conventionally understood as the old Champa center in Bình Ðịnh near the modern city of Quy

Nhơn. The old generations of historians, starting already with Finot and continuing emphatically

with Maspéro, assumed that this was the capital of Champa, briefly from around 1000 CE and

definitively from at least 1044. As Southworth notes, for Finot ‘Vijaya’ was one in a ‘list of

provinces’ of the kingdom of Champa which ‘has remained essentially unchallenged, being cited

for all periods from the second to fifteenth centuries’.89 Even one of the new generation, Pierre-

Bernard Lafont, referred to ‘Campā, whose capital was Vijaya’; ‘the north of the country

[apparently in that context everything north of Phan Rang], which had Vijaya as its capital’; and

‘Vijaya, center of Cam power and civilization until its annexation by the Vietnamese in 1471’.90

This situation of Vijaya was not, however, the immediate interpretation of the first

Chamists. Aymonier at first believed that Vijaya ‘must have been a small buffer state located in

the present-day plain of Phantiet [i.e., Phan Thiết]’; In 1903 Finot said that ‘Vijaya has not bee

located’ in reference to inscription C17 with dates from 1147 to 1160 CE, in which Jaya

Harivarman, ‘having been attacked…by the Cambodian army together with that of Vijaya,

defeated them at Caklyang’.91 A year later, however, Finot compared inscriptions C101, C17,

C30 and C92 (all concerning the twelfth-century wars with Cambodia) and saw that Nagara

88 Maspéro, Royaume de Champa, p. 126, n. 3.This true Chinese rendition of the title vijaya is also seen at that time in the Song huiyao (see Wade translation). 89 Maspéro, Royaume de Champa, p. 25; Finot, ‘Notes d’épigraphie XI’, p. 915 (Jacques ed., Études épigraphiques, p. 97); quotation from Southworth, ‘Notes on the political geograhy’, p. 237. 90Pierre-Bernard Lafont, ‘Études Cam III. Pour une réhabilitation des chroniques rédigées en Caü moderne’, BEFEO, 66 (1979): 107, 109, n. 1, 111. 91 Finot, ‘Stèle de Śambhuvarman’, p. 639 and n. 1. Aymonier’s opinion is in Étienne Aymonier, Le Cambodge III, Le Groupe d’Angkor et l’histoire (Paris: Ernest Leroux, 1904, p. 514), cited in Finot, ‘Notes d’épigraphie XI’, p. 906, n. 1. This inscription in Cham has not been published in its entirety; its contents were summarized in Aymonier, ‘Première étude’, pp. 39-41.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Campā was then Vijaya; ‘at this time’, he said, ‘the capital was at Bình Định’. Thence came the

new certainty that Vijaya/Nagara Campa was at Bình Ðịnh, a certainty which was less absolute a

few pages further in the same article: ‘Vijaya probably corresponds to the province of Bình-định,

and the name of this city was probably Chà-bàn’. 92

Finot did not explain the change of view, but with respect to Chà Bàn, it may have been

influenced by the discussion in Pelliot and by Aymonier’s work on the legendary histories of the

Cham.93 Pelliot indeed showed convincingly on the basis of Chinese histories that in the twelfth

and thirteenth centuries there was a Cham ‘capital’ in Bình Ðịnh, but he did not yet recognize the

name ‘Vijaya’, and thought that its name was Chà Bàn, a name which he said first appeared in

the Vietnamese histories in 1312. He also thought that the Cham capital Foshi/Phật thệ attacked

by the Vietnamese in 1044 and 1069 was at or near Chà Bàn. However, he restored ‘Foshi’ not

as ‘vijaya’ but as ‘bhoja’.94

William Southworth has most strongly – and correctly, I believe – insisted that ‘Champa’

was neither a unitary polity nor even a federation, but rather consisted of several separate entities,

the interrelationships among which varied from time to time (total separation, alliance, peace,

hostility, trade). He has asserted firmly that ‘the earliest surviving reference to Vijaya’ is in an

inscription (C101), probably dating to c. 1153-84, and that ‘its application to earlier periods in

the academic literature should therefore be considered an historical anachronism’.95 Well said –

but then what about the earlier references to places interpreted as located in Bình Ðịnh, the site

later named Vijaya and sometimes referred to as such in the academic literature? Southworth

does not get into this problem.

Before continuing, I wish to emphasize that although the general acceptance of Bình

Ðịnh as the location of the Vijaya named in twelfth to thirteenth century inscriptions is eminently

reasonable, there is no document showing unequivocally that such is true. Champa and

Cambodian inscriptions indicate Angkor hegemony in ‘Vijaya’ (written thus in Sanskrit, Khmer

92 Finot, ‘Notes d’épigraphie XI’, p. 906, n. 1 and p. 915, n. 4 (emphasis added). 93 Étienne Aymonier, ‘Chronique des anciens rois du Cambodge’, Excursions et Reconnaissances, 4, 2 (1880): 149-8. 94 Pelliot, ‘Deux itinéraires’, pp. 202-7. He observed that ‘this name Foshi is identical to that of the country of Foshi…which existed at the time of the Tang and which shows up in Jiadan’s second itinerary. There is more or less general agreement that the name Foshi from the Tang era should be reconstructed as Bhoja; the same solution should probably be adopted for the Cham capital at the end of the tenth century’ (p. 202 note 2). 95 Southworth, ‘Notes on the political geography’, p. 238; see Finot, ‘Notes d’épigraphie XI’, pp. 963-5 for the inscription. Since the inscription contains no dates, it is uncertain how Southworth found those he proposed; C17 may in fact be earlier, but this is of no consequence, for they are both from the same reign.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

and Cham) in the late twelfth and early thirteenth centuries. The seven brick towers and tower

complexes still standing, which are the most prominent evidence for the importance of Bình

Ðịnh-Quy Nhơn at the time, have been described as exhibiting Angkorean architectural influence

(although I find it difficult to recognize), but as architectural wholes they are quite different from

anything in Cambodia.

For Maspéro and those who followed him, the first reference to Vijaya-Bình Ðịnh was in

988, when following a Vietnamese victory the Cham king retreated to Phật Thành or Phật Thệ

Thành, which Maspéro believed was Vijaya in Bình Ðịnh. Again in 1044 the Vietnamese

attacked and ‘arrived in Vijaya’. Here Maspéro refers to the Vsl which first records a Vietnamese

attack on Chiêm Thành, then finally that they reached the city of Phật Thệ.96 Once more, in

1069, the Vietnamese attacked. The Vietnamese histories Tt and Cm record very briefly that their

king went to attack Chiêm Thành, location unspecified, and captured the Cham king Chế Củ,

taking him as hostage until he gave up as ransom the three districts of Ðia Lý, Ma Linh and Bố

Chính, far in the north in Quảng Trị and Quảng Bình. For the account of that operation Maspéro

relied on Vsl which has a much more detailed account in which the Vietnamese arrived ‘at the

port of Śrī Banöy’, which Maspéro accepted as the port of Vijaya-Bình Ðịnh, called Thị-Lợi-Bì-

nại (Ch. Shi-li-pi-nai) in Vsl. Then they pursued the Cham king to Phật Thệ.97

Although Maspéro did not refer to a source, the name ‘Śrī Banöy’ is from the Cham

‘légendes historiques’ published by Aymonier, the literal geographical details of which were

never accepted and which have now received a new interpretation. According to the research of

Po Dharma, followed by Lafont, the first of the legendary capitals – ‘Bal Sri Banay’, presumably

Śrī Banöy – was in Pāõóuraïga. None of the five legendary capitals was placed in Bình Ðịnh. In

other contexts, however, they place ‘Binnai’ – a name which seems to be a fairly accurate

transcription of the names in Vietnamese and Chinese histories of the time, and which Pelliot and

Maspéro called Śrī Banöy – in Quảng Ngãi.98 I shall take this up in discussion of the narrative


96 Maspéro, Royaume de Champa, p. 136; Cm, Chính biên, 3: 7-9, p. 326; VSL, 2: 7b-8a, pp. 86-7. 97 Maspéro, Royaume de Champa, pp. 141-2; Vsl, 2: 13b, p. 104. 98 I have not seen Po Dharma’s thesis in which he discusses the five legendary capitals, and rely for its details on Lafont, ‘Études Cam III’, p. 109, which apparently follows Po Dharma. For the location of Binnai see Po Dharma, Le Pāõóuraïga (Campā) 1802-1835 (Paris: École Française d’Extrême-Orient, 1987), vol. II, first map in ‘Annexe 4’ (following p. 227), entitled ‘Le Campā et les étapes de l’expansion vietnamienne vers le sud’; and the map in Actes du séminaire, p. 1.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

After the war of 1069, according to Maspéro, the Cham rulers returned to

Indrapura/Đồng Dương and there is no further reference to Vijaya-Bình Ðịnh until it really

appears in epigraphy in the twelfth- to thirteenth-century wars with Cambodia, which will be

discussed later on. The term ‘vijaya’, however, occurs in other contexts – not specifically as a

place name but in contexts which may have misled Finot, Maspéro and others, including Coedès,

who in his study of Śrī Vijaya in Sumatra said that ‘it is known for certain from epigraphy that at

this time [the late tenth century and during the tenth century] the Cham capital was in Bình-định

and was called Vijaya’.99 Coedès, as I shall show, was mistaken about the epigraphy, for ‘vijaya’

as the name of a capital does not appear until the twelfth century. Two Cham kings (both named

Harivarmadeva) in two Cham-language incriptions from Mỹ Sơn (C75 and C94, dated 991 and

978-1078 respectively) took the titles yāï po ku vijaya śrī. This may have misled Maspéro and

others concerning ‘vijaya’, resulting in erroneous interpretations of historical events as discussed

below. In these titles the term ‘vijaya’ probably had no relationship with a location, but rather

indicated ‘victory, victorious’, as construed for one context by Bergaigne.100

In addition to its use as a toponym for a particular place (which does not appear in the

epigraphic record until mid- to late twelfth century), the term vijaya (literally, ‘victory’) was

used in the eleventh to twelfth centuries to designate a geographic-administrative division, for

which one would expect the term Vi.saya. Finot gave attention to this and listed the names of 12

‘vijaya’, usually associated with an elite individual who in some way was involved with that

vijaya. Aymonier at first thought it meant that he had been victorious in the region named. This

use of vijaya was very late in Cham epigraphy, the first instance being in C89/1088 and the last

in C83C/1263, with six in a single inscription, C92/1170; all were in Mỹ Sơn. In addition to

those 12 vijaya (perhaps meaning vi.saya) listed by Finot, two more – śilāvandhavijaya (C29B)

and manā .hvijaya (C31) – are in the inscriptions of Pô Nagar in the same period, and a ïauk glauï

vijaya is in the very late inscription of Biên Hoà, C1/1421. This last occurrence has been

erroneously construed as the name-title of a prince. The text, however, refers to king (yāï po ku

śrī) Jaya Siïhavarmadeva, a ‘man (uraï) [of] ïok gloï vijaya’ – clearly a place name, but

translated by Aymonier as ‘person (uraï) on a high (ïok gloï) victory (vijaya)’.101

99 George Coedès, ‘Le royaume de Çrī Vijaya’, BEFEO 18, 6 (1918): 24 (emphasis added). 100 Bergaigne, ‘Ancien royaume’, pp. 282-5, construed ‘vijaya’ in the title ratnabhūmivijaya as ‘victory’. 101 Schweyer, ‘Chronologie des inscriptions’, p. 343; Finot, ‘Inscriptions du Musée’, p. 14 (‘prince ðauk Glauï Vijaya’); Aymonier, ‘Première étude’, pp. 84-5.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

The expected vi.saya, in ‘Caum vi.sayaü ‘, occurs in the Sanskrit of the seventh-century

C96.102 This apparent confusion of ‘vijaya’ and ‘vi.saya’ occurs once in Old Khmer epigraphy,

in the Khmer part of the Sdok Kak Thom inscription (K.235); there it refers to vijaya Indrapura,

recognized by Coedès and Dupont as a confusion with vi .s aya, which appears in the

corresponding Sanskrit section.103 The few inscriptions found in Bình Ðịnh are not of much help.

None employs the name ‘Vijaya’, and when it appears in epigraphy in the twelfth century and

may with confidence be localized near Quy Nhơn, it is never referred to in Chinese and

Vietnamese sources as ‘Foshi/Phật Thệ’. The identification of Foshi/Phật Thệ with Vijaya has

been accepted since Coedès’ study of Śrīvijaya in Sumatra, where the term vijaya was known

from local inscriptions; some of the Chinese writers, beginning with Yijing, called it ‘Foshi’.

Thus we must take a new look at Coedès’ study of the problem.

On re-reading Pelliot and then Coedès on this question, it seems to me that there was

circularity in the argument: the Sumatran evidence was used to justify an explanation of the

Champa evidence, which was in turn used to support the explanation for Sumatra. Besides that,

the reasoning of Coedès was a bit slippery. His argument appeared in his article of 1918, ‘Le

royaume de Çrīvijaya’, and the problem was to reconcile the name ‘Srīvjaya’ – known from

inscriptions and found in Cola records of places they attacked – with the name ‘Shilifoshi’. This

term, found first in Yijing, was believed to represent a highly developed port city in Sumatra,

probably at Palembang, known to Chinese Buddhists as an important center for the study of

Sanskrit and Buddhism and considered by them as an important stage for preliminary study

before proceeding to India. Coedès’ solution was that the Chinese Shilifoshi was a transcription

of Srī vijaya. That transcription was not straightforward and involved discussion of the phonetics

of both vijaya and Foshi. The Chinese ‘Shili’ for śrī was not, and has not since been,


The first European interpretation of ‘Foshi’, seen in Pelliot, was ‘bhoja’, troubling

because it had no relevant meaning, but there was agreement that the character fo佛 (‘Buddha’),

could only represent syllables with vowels /o/ or /u/. There was also a problem with the che in 102 Ibid., p.42; Finot, ‘Notes d’épigraphie XI’, pp. 915-6, 921. 103 George Coedès and Pierre Dupont, ‘Les stèles de Sdòk Kăk Tho .m, Phno .m Sandak et Prá .h Vihãr’, BEFEO, 43 (1943-46): 104-5 and note 9. The opposite confusion – ‘viùaya’ for ‘vijaya’ – occurred in the Tamil inscriptions concerning Śrīvijaya in Sumatra, apparently because of a peculiarity in Tamil script (Coedès, ‘Royaume de Çrī Vijaya’, pp. 4-5). 104 Ibid., p. 23 insisted that it was the name Foshi ‘appeared for the first time’ in Yijing’s text’.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Foshi which, it was thought, could only represent syllables with /i/ or /e/, but not with /a/. Thus

Coedès first rejected ‘bhoja’ but asked, ‘is Foshi = vijaya any better?’ First, it could be shown

that Yijing, the first to write of ‘Shilifoshi’, had in other writings used that same shi character to

represent /je/, /ji/, and also /jai/ and /jaya/ in other Sanskrit terms, which seemed to solve one

problem. However, he continued, ‘the correspondence between fo = vi is less satisfactory. I

think, though, that it is possible.’. The old pronunciation of fo, he said was *pw’iδ. (Now,

according to Geoff Wade it seems to be /but/). ‘For the change from vi, or more exactly bi

[because in the Southeast Asian languages in question an alternation of initial v/b is common] to

bu through labialisation is…possible, and examples do exist, which is enough to justify the use

of fo. In fact, fo-che (foshi) fo-ts’i may represent a form bujai, a corrupted spoken form of vijaya’.

‘Bujai’, let us not forget now, is a completely speculative form, not found in any text.105

The examples offered by Coedès as support for bi > bu were taken from the

representations of original Sanskrit words in modern Khmer, and are not at all good evidence.

Coedès’ three examples were Sanskrit vīja, ‘seed’ > Khmer būj (puč); bimba, ‘picture’ > bum

(pŭm); and bhīmasena, a type of camphor, > bumsen (pŭmsèn). I repeat, these are modern Khmer

forms, and it is not certain they derive directly from Sanskrit. In Angkor Khmer-language

inscriptions, however, we still find vija (seed) in the fourteenth century and bhimasena in the

eleventh century, meaning that the Sanskrit vocalization was still maintained with only

shortening of the vowel and that the ī/i > ŭ/u in Khmer was a later development and thus of no

relevance for the argument about vijaya > hypothetical bujai in seventh-century Śrī Vijaya.

Neither ‘bimba’ nor ‘bum’ occur in published Khmer inscriptions. Moreover, Coedès

acknowledged that Sanskrit vijaya has remained bijai in modern Khmer and Thai.106 Moreover,

even if the development in Khmer had been as Coedès proposed, it would be of no significance

for vijaya > bujai = foshi in Sumatra, where the language was Malay, whose phonetic

development could not be hypothesized on the basis of Khmer.

If, then, the proposal for vijaya/bijaya > bujai as a ‘corrupted spoken form’ is quite weak,

in fact Coedès’ main argument for Foshi = vijaya collapses. What, then, did Yijing mean with

his Foshi, especially when we know that the Chinese had a perfectly transparent transcription for

105 Ibid., pp. 23 (‘Is…any better?’) and 24 (phonetic discussion). Information on the reconstruction of the earlier form is from Geoff Wade, personal communication. 106 Ibid., p. 24, notes 4 and 3 respectively. The Khmer examples are respectively in inscriptions K.470 and K.455. K.470 is in George Coedès, Inscriptions du Cambodge, Vol. 2 (Hanoi: Imprimerie d’Extrême-Orient, 1942), p. 188 and K.455 is in Coedès, Inscriptions du Cambodge, Vol. 7 (Paris: École Française d’Extrême-Orient, 1964), p. 80.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

vijaya (bì-sà-xà) in their reports concerning Champa and that there was a king Harivarma whose

titles in inscriptions were yāï po ku vijaya śrī, with vijaya written exactly as in inscriptions

referring to locations named ‘vijaya’?

Not only did the Chinese transcribe vijaya correctly in Cham royal titles, but in texts

other than Yijing’s they had more accurate transcriptions for Śrī Vijaya in Sumatra. Coedès

already noted in 1918 that other Chinese transcribed the name of what was apparently the same

place in Sumatra as Jin-li-pu-shi, with the character jin a scribal error for the very similar shi,

which in the full Chinese transcription ‘represents very precisely Çrībijaya = Çrīvijaya’. Wade

has made the same point, citing tenth-century Chinese texts referring to ‘Jin-li-pi-jia’ and ‘Jin-li-

pi-shi’, accepted as the same place in Sumatra and with ‘jin’ as scribal error for ‘shi’.107 It would

thus seem that Yijing’s foshi for v(b)ijaya was his own idiosyncratic choice. Why?

Yijing was a good Sanskritist, and he would have known that the country’s official name

was Śrī Vijaya, written that way in their own inscriptions, and the Buddhist intellectual elite with

whom he probably mixed would have pronounced it that way, or at least as /bijaya/. Why would

he have adopted a ‘corrupted spoken form’ which, if it existed (on dubious evidence), would

have been that of the streets? Even if he had assimilated that spoken form, why transcribe it with

precisely the character which means ‘Buddha’? Yijing was not only a Sanskritist who would

have understood vijaya, he was also a devout Buddhist who was in Śrī Vijaya because of its

reputation as a Sanskrit and Buddhist studies center. I propose that his choice was deliberate, to

represent not a popular, non-literate pronunciation of the local name, but what for him was the

important character of the place, its Buddhism. Thus Foshi = but (Buddha) jay(a), ‘Buddha


As supplementary support for his argument, which I have treated as circular above,

Coedès turned to Champa, where, ‘at the end of the tenth century and over the course of the

eleventh, the Chinese and Vietnamese mention the city of Foshi [written just like the Foshi in

Sumatra] as the capital of Champa… We know for certain from the epigraphy that at this time

the Cham capital was in Bình Định and was called Vijaya’.108 Here, as in the evidence

concerning ‘bi > bu’, Coedès slid over some of the details. The epigraphic evidence which he

cited for Vijaya as a capital, allegedly in Bình Ðịnh is not from the tenth to eleventh centuries,

but from the twelfth; and at that time the Vietnamese chronicles do not write of Phật Thệ (Foshi),

107Coedès, ‘Royaume de Çrī Vijaya’, p. 24; Geoff Wade, personal communication, 3 Dec. 2004. 108 Coedès, ‘Royaume de Çrī Vijaya’, p. 24; emphasis added.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

but only of Chiêm Thành, ‘Cham city’. They also have very little to say about the Champa-

Cambodia warfare which accounts for the frequent references to Vijaya in the epigraphy. Even

when the Vietnamese chronicles are concerned with the real Cham center in Bình Ðịnh (possibly

Vijaya), in the fifteenth century before and during the conquest of 1471, they still write only of

Chiêm Thành, Chà/Đồ Bàn, or Thị Nai. That is, the real Vijaya of Champa, in contexts where

contemporary epigraphy shows that it was certainly the site of reference, was never called

Foshi/Phật Thệ by the Chinese and Vietnamese, just as the Cham royal title ‘vijaya’ was never

transcribed as foshi, but with characters truly representing vi(bi)jaya.

This should, I think, engender some doubt about Foshi representing vijaya in Champa.

The first references to ‘Phật thệ’ plus ‘thành’ (‘city’) are in the 980s – the time of the first war

between newly independent Viet Nam and a Champa entity whose ‘capital’ had been established

by 875 in Indrapura (the ruins of which are now known as Đồng Dương), where the dominant

religion was Mahayana Buddhism. The first Vietnamese reference to their victory, moreover,

calls the location not Phật thệ thành, but Phật thành, ‘Buddha city’, an understandable epithet

for Đồng Dương.109

The reason that the identity Phật thành/Phật thệ thành = Đồng Dương /Indrapura was not

understood in the time of Pelliot, Coedès and Maspéro was that the Chinese had reports of the

Cham elite withdrawing 700 li after their defeat, Đồng Dương was considered the northernmost

important Champa city in the ninth to tenth centuries, and from there to Vijaya in Bình Ðịnh was

precisely 700 li. The extension of Champa territory at that time, demonstrated by large-scale

structures and epigraphy throughout Quảng Trị and Quảng Bình, was not yet well-known. Now,

however, it is easy to understand that the Chiêm Thành which the Vietnamese attacked in 982

was somewhere in that northern area, from which the distance to the ‘Buddha city’ Đồng

Dương/Indrapura was also around 700 li. This could explain the Chinese note of 989 in the Song

Shi about Foshiguo (‘The country of Foshi) being ‘newly established’ as meaning simply that a

new Champa ruler re-established his capital at ‘Buddha city’ Đồng Dương, where it had been

located – except for the expansions northward – since at least 875. A later Chinese report (also

from the Song Shi) of a Cham envoy in 1007 saying that ‘my country was formerly subject to

Jiaozhou, then we fled to Foshi, 700 li south of our former location’, fits well with a Viet attack

109 Pelliot, ‘Deux itinéraires’, p. 204, n. 2, insisted that this was an error, and should be corrected to Phật thệ thành, the Sino-Vietnamese equivalent of Foshicheng found in the Chinese chronicles, as though the Vietnamese did not know the name of their conquest.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

in 982 in Quảng Bình and a Cham retreat to Đồng Dương. The insistance of the Vietnamese

chronicle Cương Mục that Foshi was Huế, if not quite accurate, is not as confusing as Pelliot

thought, given that Huế is close to the latitude of Đồng Dương.110

Maspéro himself was forced to admit, after an unconvincing story of a retreat southward

even as far as Phan Rang, that the Cham rulers soon returned to Indrapura, the city where the

Đồng Dương temple complex was located. That is, when his fiction about the two decades or so

after 982 is bracketed out, it is easy to see that there was no move of Cham kings or capital south

of the Indrapura - Huế region.111 At the date of that first Champa-Vietnam war, however, there

was a real king Harivarman, known from his late tenth-century inscriptions (C75/991 and

probably C78) as yān po ku vijaya śrī, which undoubtedly influenced Maspéro’s belief that he

had retreated to Vijaya. The inscriptions are from Mỹ Sơn, however, and thus do not justify an

argument that he had moved to the Vijaya in Bình Ðịnh. He is one of only two kings known from

inscriptions with these titles – the other appearing nearly a century later and neither of them

having any connection to the city of Vijaya – and it was probably to him that the Vietnamese

referred in their chronicle entry for 1000.

Moreover, to add to the confusion, Maspéro says that in 990 – that is, after Harivarman

had supposedly moved to Vijaya – the Vietnamese attacked again and ‘devastated’ ‘the citadel’

of Địa Lý. The latter, as Maspéro himself notes, was in Quảng Bình, far north of the 700 li of

territory believed lost in the alleged withdrawal from Indrapura to Vijaya and thus of no

relevance to a Champa center supposedly located in Vijaya. Nevertheless, Maspéro says that

‘Harivarman’ – in the Chinese text just ‘a new king’ whose title in Chinese transcription seemed

to him to be ‘Indravarman’ of Foshi, which he believed to be Vijaya – complained to the Chinese.

A Cham king making such a complaint was more likely in Indrapura or somewhere in the North,

not in Vijaya-Bình Ðịnh, and was probably the Harivarman who in 991 left an inscription in Mỹ

Sơn. Two years later the Vietnamese released 360 prisoners whom they had taken in that battle,

and the Tt refers to Địa Lý as cựu thành (Ch. jiucheng), ‘old city’ – perhaps, in this context, even

‘old capital’, given that ‘Chiêm Thành’ was the conventional Vietnamese reference for the

Champa capital. The Champa inscriptions, together with the Vietnamese and Chinese sources

110 Maspéro, Royaume de Champa, p. 132, has the quotation from the Song Shi; Pelliot, ‘Deux itinéraires’, p. 203, for the Cm citation locating Foshi at Hué. The Cm (Chính biên, 3:9, v. 1, p. 327), citing the Chinese gazetteer Qing yitongzhi, places it in what was then Hương Thủy district of Thừa Thiên province. 111 Maspéro, Royaume de Champa, pp. 126-30.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

used by Maspéro, still imply that the area of Champa-Vietnamese contact was from Quảng Nam

northward to Quảng Bình.112

Some slightly different details appear in Geoff Wade’s new translation of Song Huiyao.

First, in 990 a new Champa king named in Chinese Yang-Tuo-pai ‘claimed that he had been

born (自稱所生) in the country of ‘Vijaya (Foshiguo 佛逝國)’; the latter is thus designated as a

‘guo’(country) instead of the usual ‘cheng’ (城), meaning ‘city’. A problem here, however, is

that the Song huiyao says that in 990 the new king only claimed to have been born in Foshiguo,

not that he had newly re-established it, as written in the Song Shi. Obviously one or the other

Chinese text is corrupt. Then in 1006 the Cham king sent an envoy to China, and he ‘advised that

their country had formerly been subject to Jiao-zhou… but later it had given its allegiance [or

‘fled to’] to Foshi (佛逝) (Vijaya)’. Wade’s treatment of this is peculiar, writing in his main text

that Champa ‘had given its allegiance to Fo-shi’, but in a note saying ‘literally: “fled to”’,

apparently like the Song Shi. Obviously ‘giving allegiance’ and ‘fleeing to’ imply quite different

historical circ*mstances. This seems to be a case in which the Chinese scribes, writing long after

the event and at second hand, did not understand the situation. In this second entry, of 1006, the

Champa envoy said that ‘to the north they lost 700 li of their former territory’; as noted above,

this could correspond to the territory north of Indrapura as far as Quảng Bình, which had been

occupied with impressive monuments by the Indrapura/Đồng Dương ‘dynasty’, and in which the

‘old city’ of Địa Lý was located.

In conclusion, concerning Vijaya, the identification of Foshi/Phật thệ in Champa with

vijaya must be rejected.

Narrative history continued

Champa activity in the far North resumed in 995 and 997 with attacks on Hoan and Ái

again, beyond Quảng Bình northward in Nghệ An and Hà Tĩnh – hardly attributable to a Cham

polity in Vijaya-Bình Ðịnh which had lost 700 li of territory between Bình Ðịnh and Indrapura to

the Vietnamese. The Vietnamese emperor Lê Hoàn is said to have commented to the Chinese

that his borders were contiguous with Champa. Perhaps that is why Maspéro wrote that

Harivarman, ‘although he had been proclaimed king in Vijaya, had re-established the court in

112 Ibid.; TT, Bản kỷ, 1:20a, v. 1, p. 227.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Indrapura [Đồng Dương]’.113 Neither of these claims is justified by anything in the texts, but it

seems that Maspéro finally saw that the defeat of 982 must have been farther north and that

Đồng Dương/Indrapura – or at least the Thu Bồn region – became the main northern Cham

center again after that war. There is no evidence that yāï po ku vijaya śrī Harivarman, Maspéro’s

‘national claimant to the throne’, ever moved to Bình Ðịnh, and there is thus no need to

hypothesize a subsequent return to Indrapura. The northern Champa rulers of the time had

always been in that region, and they were probably responsible for the earlier attacks on Hoan

and Ái in 803, noted above. Here we may understand the significance of what initially appear to

be contradictory contexts in the the Song huiyao. The introduction says that it was two days by

sea from Champa to the Red River delta, but by 1076 the same trip is recorded as having taken

17 days; Wade interprets the difference as resulting from a loss of Champa territory after the

‘presumably ... earlier time’.114 If the two-day sail is not a textual error, it can only refer to

Champa’s farthest northern extent (northern Quảng Bình) and an attack on Hoan and Ái; the

territory lost by 1076 would have been those far northern provinces, not the territory from

Indrapura to Bình Ðịnh.

In Maspéro’s interpretation, Harivarman’s successor – known only by his titles, yaï po ku

vijaya śrī, as transcribed in Chinese – moved definitely to Vijaya in 1000; there is no source for

this assertion, which is probably based only on his title. As noted above, this was probably still

the same Harivarman. Maspéro’s story continues from the Chinese, with yaï po ku vijaya śrī

replaced in 1010 by another ‘Harivarman’ (inferred from a Chinese transcription but probably

the same Harivarman). In 1018 the Chinese transcription changed to Shi-mei-pai-mo-die and

Maspéro interpreted it unconvincingly as ‘Parameśvaravarman’, a new king. In fact, following

yāï po ku vijaya śrī Harivarman, the Chinese transcriptions of Cham royal names up to 1044

changed four times in ways that give little support to Maspéro’s interpretations.115

In the 1020s and 1030s there was more warfare in Quảng Bình, odd if the Champa center

was in Vijaya; then in 1044, the Vietnamese decided on a major invasion, which has always been

interpreted as a conquest of Vijaya in Bình Ðịnh. The campaign began in January, and the

fighting (described in the same detail in both Maspéro and Tt) took place north of Huế and

around Đà Nẵng. The Vietnamese moved by boat; having reached a branch of the Thu Bồn River, 113 Maspéro, Royaume de Champa, p. 129; in note 3 he quotes the Vietnamese source An Nam Chí Lược, 11:7b, which quotes Lê Hoàn’s remarks to the Song. 114 Wade, Song huiyao. 115 Maspéro, Royaume de Champa, pp. 133-6.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

they left the boats and landed, defeating the Champa army there. It is impossible from the

description to determine how long they took to reach that region, but the next statement,

immediately after the description of the battle, is that in the seventh month ‘the king took his

soldiers and entered the city of Phật Thệ’. This implies a site not far from the battlefield, that is

the ‘Buddha city’ (Đồng Dương), not Vijaya as has conventionally been construed in modern

academic literature. A march from the Thu Bồn to a Vijaya in Bình Ðịnh would have taken

nearly another month, probably with more fighting on the way, or at least at their destination; by

sea it would have taken almost as long, but in the eighth month they had already returned to

Nghệ An. A campaign from the Thu Bồn to Bình Ðịnh and return to Nghệ An could not have

been compressed into the specified time period.116 Thus, it should be accepted that the so-called

invasion of Vijaya in 1044 was in fact an attack on the Thu Bồn valley region and that this was a

final Vietnamese attempt to defeat the successors of the so-called ‘Đồng Dương dynasty’ who

had taken over the old Linyi policy of pushing northward and had established important centers

as far North as Quảng Bình.

Champa in the eleventh century

The period from the war of 1044 to the war of 1069, which Maspéro – on the basis of one

Vietnamese source but not the most official chronicles – called another attack on Vijaya, is

perhaps the most fictionalized segment of his history. (Peculiarly, it should be noted, the Song

Huiyao makes no reference to the Vietnamese attacks on Champa in 1044 and 1069). Following

the events of 1044, a new ruler, ‘whose ancestors were merely warriors, Īśvaras, vassals of the

preceding rulers…took power and had himself crowned under the name of Jaya

Parameśvaravarman’.117 How do the inscriptions contribute to Maspéro’s interpretation? First,

they provide no information at all about Vijaya until the late twelfth century, when that name

appears for the first time in connection with Champa-Cambodia warfare. After the 991

inscription of yāï po ku vijaya śrī Harivarman in Mỹ Sơn there are no more inscriptions until

1050, and throughout the eleventh century all inscriptions were set up either in Mỹ Sơn (or at

least in Quảng Nam) or in the South in Nha Trang and Phan Rang. In particular, the two kings,

116 Stein, Linyi, p. 129, cites a Chinese text (the Qianhanshu) which says that an army on the march covered only 30 li per day, whereas the distance from Đồng Dương to Bình Ðịnh was considered to be 700 li, a nearly 50-day round-trip trek, not counting time in Vijaya-Bình Ðịnh. By sea, the distance from the Thu Bồn to Bình Ðịnh is about half the distance from the Thu Bồn back to Nghệ An, thus more or less a month. 117 Maspéro, Royaume de Champa, p. 137.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

both named Harivarman and holding the title yaï po ku vijaya śrī, showed no interest in the site

of the later Vijaya (Bình Ðịnh) in their inscriptions, and that term in their titles was probably

intended as ‘victorious’, without any territorial connotation.

Following the war of 1044, between 1050 and 1064, two new kings, Parameśvaravarman

and Rudravarman, left inscriptions at Pô Nagar in Nha Trang; the second of these rulers is

mentioned in credible transcriptions in Chinese. The first, who between 1050 and 1055 left two

inscriptions in Pô Nagar of Nha Trang or its surrounding province, four in Phan Rang and one in

Quảng Nam, seemed particularly concerned with troubles in Pāõóuraïga, not warfare with

Vietnam. Maspéro abusively said he ‘had quite a job to rebuild the ruins that the Vietnamese

armies and civil wars had left behind…. In fact, neither in the inscription cited nor elsewhere did

this king refer to Vietnamese; he mentioned only ‘this Kali age where Conflict holds sway over

the world’, a standard formula for ‘modern’ times in contrast to better days in the past.118 He

made no reference to previous kings, least of all to say that his family had been their vassals.

Then in 1064 King Rudravarman left a Sanskrit inscription in Pô Nagar, saying that he ‘belonged

to the noble family of’Īśvaras [not ‘mere warriors…vassls of the preceding rulers], of Śrī

Parameśvara [perhaps Parameśvaravarman]’. He also claimed that he was the younger brother of

Śrī Bhadravarman, perhaps the immediate successor (briefly) of Parameśvaravarman, as

reasonably interpreted by Maspéro – who added without reason, however, that Rudravarman

‘may have had some connection with the death of his older brother’.

In the meantime, in C95/1056, a certain Śrī Yuvarāja Mahāsenāpati set up an inscription

in Mỹ Sơn, in which he boasted of having conquered the city of Śaübhupura in Cambodia,

destroying its sanctuaries and taking Khmer prisoners whom he offered to the sanctuaries of

Śrīśānabhadreśvara, the main temple of Mỹ Sơn, thus following in the traditions of former kings

of Champa who honored this site. There is no indication of his family, and no reason to connect

him with the family of Parameśvaravarman-Bhadravarman-Rudravarman who were ruling in

Nha Trang-Phan Rang between 1050-64, at least. If he may be connected with any previous

rulers, it would be with the family of the Harivarman of 991, and later. It must be accepted that

there were still ruling chiefs in the Thu Bồn valley concurrently with the kings in the South, both

before and after the war of 1044.119

118 Ibid., p.139 and notes 6-7. 119 Inscription C95, in Finot, ‘Notes d’épigraphie XI’, pp. 943-6. Note that at the time of their publication the numerical dates in the Champa inscriptions were often not properly understood. Finot (p. 946) read its date as 789


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Finot believed this Śrī Yuvarāja Mahāsenāpati was the Yuvarāja Mahāsenāpati Prince Pāï

(younger brother of another Harivarman), who became King Paramabodhisatva. He is mentioned

with those qualities in the dated inscription C89/1088, as Yuvarāja in the undated (tenth c. ś.

[AD 978-1078]) inscriptions C93 and C94, and as Paramabodhisatva in C30A1. This still makes

a link with the Harivarman of C75/991, because the ruler of that name who was the elder brother

of the Yuvarāja (later Paramabodhisatva) also had the titles yāï po ku vijaya śrī in C94. Thus

from 991 until 1088 there are records of a line of related kings leaving their inscriptions at Mỹ

Sơn concurrently – from 1050 to 1064 – with Parameśvaravarman and Rudravarman in Nha

Trang and Phan Rang.

In the time of Rudravarman there seems to have been contact between his part of Champa

and China, for the Chinese on occasion recorded a credible phonetic transcription of his name. It

is not certain, however, that all the envoys recorded as coming to China from ‘Champa’ were

from the same region. In Maspéro’s interpretation of the Chinese sources, the Cham were

preparing to attack Vietnam, although the interpretation requires a certain element of imagination.

Thus, at a date which he does not make clear, a Chinese source says that because of attacks by

Giao Chỉ, Champa was preparing its forces to resist. Then the Champa king ‘continued his

preparations’, as evidenced by a request to buy mules from China; and finally ‘he launched

hostilities along the border’ – the provocation for a Vietnamese invasion in 1069. Maspéro’s

citation is from Tt, which in fact makes light of the incident, saying that in 1068 Chiêm Thành

offered a white elephant and then ‘caused trouble along the border’.120

Maspéro’s citation of Tt here is peculiar, since for the full campaign (which does not

appear in Tt), he uses Vsl, the only one of the Vietnamese histories which contains it. The other

histories say only that the Vietnamese king attacked Chiêm Thành, captured the Cham king Chế

Củ, and held him for ransom. Moreover, the border was far in the north in Quảng Bình-Quảng

Trị, as we shall see from the outcome of the war, while the inscriptions show two Champa

centers, one in the Thu Bồn and one in Nha Trang-Phan Rang; the latter could hardly have been

involved in squabbles along the northern border. śaka/867, rather than 978/1056, and thought it must have referred to a previous reign. He also said that it ‘appears to be a continuation of the preceding inscription’ (C94), which seems correct. I attribute the undated C94 to the time of Harivarman of Mỹ Sơn, whose other inscriptions are C75/991 and C93/tenth c. śaka, i.e. 978-1078 CE. For the corrections of the numerical dates in general see Finot, ‘Les inscriptions du Musée’; on the date for C95 see Finot, ‘Errata et addenda’, BEFEO, 15, 2 (1915): 191: ‘IV, 944...and 946… “798”: read “978” and eliminate the note on p. 946’. Maspéro, Royaume de Champa, describes the event (p. 145) but provides no date, situating it in his narrative between other events in 1076 and 1080; the same is true for Coedès, Indianized states, p. 152. 120 Maspéro, Royaume de Champa, pp. 140-41, citing the Song Shi and the Tt.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

The so-called campaign against Vijaya in 1069, based only on Vsl, proceeded as follows.

The Vietnamese fleet set out in the third month and reached Thị Lợi Bi Nai where the troops

disembarked and fought with the Cham, killing many. The Cham king, Ðệ Củ (Chế Củ in the

other Vietnamese histories) – for Maspéro Rudravarman of Phan Rang – fled to Phật thệ but was

pursued and captured on the border of Cambodia in the fourth month. The Vietnamese remained

about one month in Phật thệ and started their return to their own capital in the fifth month. Thus

the journey from Thăng Long (modern-day Hanoi) to Phật thệ took less than one month and the

return nearly two months.

Pelliot commented on the same passage of Vsl and was convinced that the object of the

invasion was Vijaya-Bình Ðịnh. He identified Maspéro’s Ðà Dùng, one of the ports en route

(reached on the 28th day of the third month), as Tư Dùng, the southern entrance to the lagoon of

Huế, and said that the voyage from there to ‘Vijaya’ (that is Thị Lợi Bi Nai, reached on the third

day of the fourth month) was only six days. This seems a very short time, given that the distance

from Huế to Quy Nhơn is more than half the total distance of the campaign from the Vietnamese

capital which Vsl says took two months.121 This is enough to show that the Thị Lợi Bi Nai and

Phật thệ in this account were not in Bình Ðịnh. In any case, it should be noted for comment

below on subsequent events, that the story found in Vsl has the Vietnamese fleet and troops

proceeding directly to their goal, and after success returning directly home.

I maintain that this story is incoherent and results from Maspéro’s insistance on a single

Champa and confusion about supposed references to ‘Vijaya’. As we have seen, all the

inscriptions from the end of the tenth century to the 1060s, and later, were in Mỹ Sơn, Nha Trang

and Phan Rang. None shows any interest in Vijaya-Bình Ðịnh, but some of them do show

concern with Pāõóuraïga. The border problems which supposedly provoked the Vietnamese

attack had been far in the North in Quảng Bình, as were the provinces which Vietnam obtained

as a king’s ransom following the war. Cham chiefs in Nha Trang and Phan Rang would have had

little interest or military power in that area, and those who left inscriptions in Mỹ Sơn were more

preoccupied with the far South. The Cham king there was Rudravarman in Maspéro’s version,

but the Cham king in the Vietnamese histories was Chế Củ or Ðệ Củ. He was captured and taken

prisoner, and in order to gain release he ceded Ðịa Lý, Ma Linh and Bố Chính (far to the north in

121 Pelliot, ‘Deux itinéraires’, pp. 206-7. Ðệ Củ as the name of the Cham king is also from Pelliot’s reading.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Quảng Bình and Quảng Trị) to Vietnam. That was the border where Tt recorded that the Cham

had been causing trouble.

Did the 1069 Vietnamese invasion reach Bình Ðịnh, however? According to Maspéro,

the Vietnamese arrived ‘at the port of Śrī Banöy’, called Shili Pinai in Chinese and Thị Lợ’i Bi

Nai in the Vietnamese Vsl; thus ‘Śrī Banöy’ was his construal of the Chinese and Vietnamese

names. It is the name of one of the Champa capitals in the historical legends collected and

published by Aymonier; the latter had placed it in Quảng Bình, but Pelliot argued that it should

be identified with the port of Chà Bàn in Quy Nhơn.122 Recent work by modern Chamists,

however, has given a completely new and contradictory reading of the Cham historical

chronicles. According to the new interpretations, those chronicles were formerly discredited

because it was believed that they claimed to refer to kings ruling in Vijaya, whereas in fact they

were only intended as chronicles of Pāõóuraïga /Phan Rang. The major revisionist work in this

area is the thesis by Po Dharma, which I have not seen, but maps in his Pāõóuraïga work and in

the Copenhagen papers place Binnai in Quảng Ngãi. According to him, the first of the five

legendary capitals – ‘Bal Sri Banay’, presumably Śrī Banöy – was in Pāõóuraïga, and none of

them was located in Bình Ðịnh.123

Quảng Ngãi is the site of a large citadel (in fact the remains of an ancient city) now called

Châu Sa, and another citadel, Cổ Lũy, on the north and south banks of the Trà Khúc river near its

mouth; there is also the ruined temple of Chánh Lộ, dated to the eleventh century, from which

important works of sculpture have been retrieved. The two citadel sites were never explored by

the French, but it is clear that like other river mouths of Champa, that delta was also important. I

thus propose that the Vietnamese attack of 1069 was on Châu Sa – a more credible but still very

rapid six-day sail from Huế – and that Chế Củ was not Rudravarman, but a Cham chief

somewhere in the North.124

122 Ibid., p. 209; 123 See Pierre-Bernard Lafont, ‘Avant propos’ and ‘Les recherches sur le Campā et leur évolution’, in Actes du séminaire, pp. 3-5 and 7-25 respectively. Lafont notes (p. 15) that Po Dharma, Pāõóuraïga ‘has shown that the list of kings given in the chronicles [written] in “modern” Cham does not correspond to those appearing in the epigraphy because the former is a list of kings who ruled in the South after the fifteenth century, while the latter are kings who ruled in the North before the fifteenth century’. 124 Southworth, ‘Origins of Campā’ shows that Châu Sa may have been an important center as early as the seventh to eighth centuries (pp.149, 151, 170). Eventually Pelliot’s identification of places along the route may have to be modified.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

How do subsequent Champa inscriptions fit with Maspéro’s story of the 1069 invasion?

He says that while Chế Củ-Rudravarman was in captivity in Đại Việt, the country of Champa

collapsed in civil war, with more than ten chiefs fighting for the title of king. This assertion is

based on a single Cham inscription, C30A1 from Nha Trang; its translation gave Aymonier some

difficulty, and in his study most of the original Cham text is not published, but only summarized.

Contrary to Maspéro, it says nothing about ‘Annamites’, and the warfare about which it speaks

seems to have involved Phan Rang and Nha Trang. The date of this inscription is 1084 and the

king was named yāï po ku śrī Paramabodhisatva. He is said to have preserved the realm of

Nagara Campā during the disasters of the war. Then there are terms which Aymonier did not

understand, followed by ‘went in to capture yā≡ pu ku Rudravarman’ and expel him from

Nagara Campā (the words ‘expel him’ are reinterpreted by Maspéro as ‘sack the capital’), and

Champa was at war for 16 years with ten men struggling for the throne. One man in Phan Rang

ruled there for 16 years until removed by Paramabodhisatva. After his success Paramabodhisatva

was sole king and he and his son gave offerings to Pô Nagar in Nha Trang.125 In a vague

way, then, this inscription does seem to reflect a war around 1069 (in traditional arithmetic 16

years before 1084) and a bad end for Rudravarman, but not a war with Vietnam. The important

struggle was between Nha Trang and Phan Rang, and we must not lose sight of the Vietnamese

story, which relates only an attack on Thị Lợi Bi Nai and a rapid return to Đại Việt, without any

action farther south.

Another problem with the story of 1069 is the appearance in 1080 of another group of

royalty, led by a Harivarman Prince Thāï, who from the combined evidence of C93, C94,

C90A/1080-1081 and C89 (all in Mỹ Sơn) is seen as the elder brother of Prince Pāï, later King

Paramabodhisatva. This group had established themselves in the Thu Bồn. Their story starts with

a king Prāleyeśvara Dharmarāja who had a son Harivarmadeva, Prince Thāï. It says that the

temple of Śrīśānabhadreśvara had been taken and devastated, something which might have been

connected with a war around 1069. Prince Thāï defeated enemies a dozen times – but not

‘Annamites’, pace Maspéro. In particular, he defeated the Cambodians at a place named

Someśvara and took much booty. Prince Thāï’s son was pu lyaï śrī Rājadvāra. He was raised in

rank to yāï po ku śrī Jaya Indravarmadeva when he was only nine years old, and his father died

in 1081. The story continues in C89 with Prince Pā≡ Paramabodhisatva replacing his nephew, 125 Maspéro, Royaume de Champa, p. 143 note 1, citing Aymonier, ‘Première étude’, pp. 33-4; the words ‘les Annamites’ are inserted by Maspéro in parentheses as the subject of the verbs translated as ‘went in to capture’.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

but the latter became king again with the title Jaya Indravarmadeva ya≡ Devatāmūrtti and was

ruling when the inscription was erected in 1088. During that time they rebuilt the temples of Mỹ

Sơn because, for an unspecified reason, the city of Champa (Nagara Campa) was ruined and


The same story of ruin in Mỹ Sơn is related in C94, with restoration by Harivarman

(apparently Prince Thāï) and his younger brother (elsewhere called Prince Pāï). This inscription

starts as follows: ‘the enemies had entered the kingdom (nagara) of Champa, installed

themselves as masters, and had taken all the royal possessions and all the wealth of the gods’.

Who were the enemies? Since C90A/1080-81, written during the period when Mỹ Sơn was

undergoing restoration, boasts of a successful invasion of Cambodia with much rich booty taken

and presented to the temples of Mỹ Sơn, and makes no reference to Vietnamese, it would appear

that the cause of the devastation had been a war with Cambodia. Moreover, if the Vietnamese

had attacked Bình Ðịnh in 1069 as described by Maspéro, their route would have completely by-

passed the Thu Bồn area. The operations of Paramabodhisatva against Phan Rang related in

C30A1/1084 indicate that the rulers based in Mỹ Sơn were expanding their influence as far south

as Phan Rang, which had been in a state of civil war for 16 years. This could hardly have been

related to a Vietnamese attack on Bình Ðịnh in 1069 as described by Maspéro, relying on Vsl.

The story of a Vietnamese attack on Bình Ðịnh in 1069 is thus discredited. Inscriptions

from separate polities in Thu Bồn and in Phan Rang suggest wars at about that time, but not with

Đại Việt; in the North the enemy was most likely Cambodia, and in the South the conflicts were

local. If the Vietnamese had attacked Bình Ðịnh on the route described in Vsl and Maspéro, they

would have by-passed the Thu Bồn, and there is no suggestion in that account that they were

involved with Phan Rang. If, however, they attacked Binnai-Quảng Ngãi, the warfare could

perhaps have spread to the Thu Bồn, although the later inscriptions show more concern with

Cambodia than with Đại Việt. As mentioned above, Chế Củ, whom the Vietnamese said they

captured and then released for a ransom of three northern provinces, was not the Rudravarman in

the South, but a Cham chief somewhere in the North.

There is now a new full reading and translation of C30A1 by Schweyer, and I offer it

here as information to be eventually confirmed. It still contains some incoherence, but if correct

in its main lines it totally changes the picture, saying that it was King Paramabodhisatva who

126 These inscriptions are found in Jacques ed., Études épigraphiques; see the index on pp. i-iii.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

went to Phan Rang and captured Rudravarman. If so, this is conclusive evidence that

Rudravarman had not been king in whatever place the Vietnamese attacked.127

Some final questions on the alleged withdrawal of the Champa capital to the South and

loss of 700 li of territory in the 980s: If that had been the territory from Indrapura to Vijaya-

Bình Ðịnh, why would the Vietnamese have found it necessary to attack the ‘old city’ of Địa Lý

in Quảng Bình in 990? How was it then possible for the Cham later in the 990s to attack Hoan

and Aí even farther to the North beyond Quảng Bình? And how, supposedly based in Vijaya-

Bình Ðịnh, did they regain those northern provinces in Quảng Bình-Quảng Trị which they were

able to give to Vietnam as ransom for the captured king Chế Củ? These statements only make

sense if the war in the 980s was in Quảng Bình, where the Indrapura kings had established

important centers, one of them important enough to be considered as Chiêm Thành (‘city of the

Cham’) by the Vietnamese and from which the Cham were forced – though only temporarily – to

withdraw 700 li to the original capital of that dynasty in the Thu Bồn valley, Indrapura-Đồng


Champa and Cambodia, twelfth to thirteenth centuries

The next section in Maspéro concerns the wars between Champa and Cambodia in the

second half of the twelfth century, in particular in the reigns of the Cambodian kings

Sūryavarman II (1113?-1145/50?) and Jayavarman VII (1181-1220?). There is less controversy

about the Champa inscriptions concerning these events because they are much more precise and

detailed, but there is still considerable confusion in the interpretation of the events based on

syntheses of the Champa and Cambodian inscriptions and the Chinese sources. The currently

accepted outline, as summarized in Coedès’ Indianized states (which is based literally on

Maspéro for the Champa details), is that both Sūryavarman II and Jayavarman VII attacked

Champa – the first with some short-term success and the second restoring Cambodia after a

Cham invasion of Angkor in 1177 and effecting a real conquest of Champa for over 20 years,

starting in 1190.128

There is in fact nothing about these events in the Cambodian epigraphy from

Sūryavarman’s time, and all that is known comes from Champa inscriptions and Vietnamese

histories. For the second period, there are brief – and vague – references in post-1181

127 See Schweyer, ‘Po Nagar’, deuxième partie. 128 Coedès, Indianized states, pp. 169-70.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

inscriptions of Jayavarman VII concerning a large-scale invasion by Champa sometime before

that date, which in the synthesis of events since Maspéro has been attributed to 1177. That date

for the event is not entirely secure, however, and in fact the details of the invasion and its

preparations come from Chinese sources, which in some cases are wrong. Much detail about this

second period, the reign of Jayavarman VII, is found in the inscriptions of Champa, but –

peculiarly, given their boasts about earlier victories over Cambodia – there is nothing certain

about a major Champa attack around 1177.

To facilitate the argument, we must keep in mind that Maspéro and Coedès held the view

that the Cham were the remnants of an ‘Indonesian’ migration out of China which went through

the mainland and on to the islands, and their maritime skills were not given sufficient attention. It

is also obvious from the epigraphy of both peoples that the Cham and Khmer had been in close

contact from earliest historical times, probably from a distant prehistoric period, and that each

was familiar with all the land and sea routes into the territory of its neighbor. This new period of

bellicose relations between Champa and Cambodia was initiated by a new type of foreign

expansion toward the East and the coast of Champa under Sūryavarman II and Jayavarman VII,

together with attacks on Vietnam and new contacts with China by Sūryavarman, credibly in the

interest of participating in the growing maritime economy. After a break of several reigns,

relations with China were renewed by Sūryavarman II, with missions sent from Angkor in 1116

and 1120, and in 1128 the conferral of special dignity on the Cambodian king.129

Sūryavarman also attacked Đại Việt, but without success; on the other hand, he

succeeded in subjugating part of Champa for a short time. Both areas were important for their

coastal access. Unfortunately the Angkor epigraphic record is quite unhelpful in terms of details

about this aspect of state activity, and it is in fact only an assumption that the attacks on Champa

in the 1140s were by Sūryavarman (see below).130 The seaward expansionism of Angkor at this

time fits precisely into the terms of the relationship between Chinese commercial policy and the

rise, decline and transformation of Southeast Asian states. Sūryavarman’s reign coincided almost

exactly with the first years of the Southern Song, whose dependence on the sea after land routes

westward had been cut impelled them to open trade with Southeast Asia beyond the level

allowed by previous dynasties.

129 Lawrence Palmer Briggs, The ancient Khmer Empire (Bangkok: White Lotus reprint, 1999), p. 189; Coedès, Indianized states, pp. 159, 161-2. 130 Ibid., pp. 159-60.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

If the various treatments of these events in Champa and Cambodia are compared, we see

that near-total confusion reigns. First, let us take the details in Coedès. Very briefly, he wrote

that the Champa king Jaya Indravarman IV (who in Maspéro’s interpretation had usurped power

in 1166-7) attacked Cambodia twice, the first time in 1170 by land – ‘“moving his army on

carts”, says one inscription, but the fighting remained stalemated’. So, seven years later (in 1177),

guided by a shipwrecked Chinese, Jaya Indravarman went up the Mekong and Tonle Sap,

surprised the Cambodians and pillaged the city. The source of these details is inscription K.485

from the Phimeanakas, written some time after 1181 when Jayavarman had reconquered his

country, but apparently no earlier than 1191 as it records offerings by the queen to the Preah

Khan. The inscription records, without date, one apparently land-based (‘moving his army on

carts’) Cham invasion. Why, if we follow the interpretation of Coedès in that context, did

Jayavarman refer only to a first – unsuccessful – Cham invasion in 1170, and what is the source

for that event? In another context he implied that the inscription did refer to the final battle,

referring to ‘the heavy task [faced by Jayavarman VII] of pulling Cambodia “out of the sea of

misfortune into which it had been plunged” by the Cham invasion of 1177’, the latter date being

inserted by Coedès. He also said that Jayavarman ‘fought a series of battles against the Cham, in

particular a naval battle…which liberated the country once and for all’. Thus, for Coedès there

were three major battles – 1170, 1177 and a naval battle – before Jayavarman became king.131

The confusion of Coedès is understandable, for he was following – very succinctly – the

interpretation of Maspéro (also based on confusing sources), so let us look there. The usual

reference to Maspéro is to his Royaume, first written in 1911 and then republished in 1928. There

he also occasionally refers to his treatment of events in Empire khmer, in particular concerning

the alleged attack of 1170. Jayavarman VII, he wrote in Empire khmer, ‘had to repel several

Cham incursions led by Jaya Harivarman. The one which took place in 1170 (1092 ś.) was

particularly disastrous for the Khmer.’ The reason for Maspéro’s interpretation here is not clear.

Chinese sources record warfare in 1170, but it was not ‘disastrous’; they say the two sides were

equal and the result a stalemate. Then, again in Maspéro, ‘in 1175…Jaya Harivarman’s successor

Sri Jaya Indravarman II carried out his own invasion of Cambodia and caused considerable

131 Ibid., pp. 299-300, 309-10. In fact Coedès’ statements on these events were anything but straightforward. In his 1929 article, he considered that the Cham invasion with their troops on chariots was in 1177 and resulted in the seizure of the city (‘Nouvelles données, p. 324), but in 1964 his treatment permitted the inference that the land invasion had been in 1170 and that the final successful invasion was upriver in 1177. The Phimeanakas inscription is translated in Coedès, Inscriptions du Cambodge, v. 2 (Hanoi : Imprimerie d’Extrême-Orient, 1942), pp. 161-81.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

damage there. He reached the very gates of the capital, which Jayavarman had to flee in haste.’

After this, Jayavarman reorganized his army for revenge, ensured the neutrality of the

Vietnamese, and invaded Champa in 1190. Here there was no mention of a Cham occupation of

Angkor, nor of any event in 1177, but rather of two earlier attacks.132 Of course, at that time

Maspéro was mistaken about the dates of Jayavarman’s reign, which in his text began in 1162,

but this seems to be where the attack of 1170 had its origin in modern scholarship.

Before continuing with a critique of the various interpretations, it may be good to review

the evidence on which they are based. There is the Phimeanakas inscription, described above,

which refers to a single land invasion from Champa – undated, but before 1181 (when

Jayavarman dated the beginning of his reign) and after 1166 (the date of the so-called usurping

King Tribhūvanādityavarman, overthrown by that invasion. This inscription also records that

Jayavarman had been in Vijaya at some time before the invasion. There is also the still

incompletely understood inscription K.227 of Banteay Chhmar, recording a battle between

Khmer and Cham somewhere in Champa in which the Khmer were defeated and their leader

nearly killed. Coedès believed that it was the story of a campaign by a son of Jayavarman VII in

Champa before Jayavarman became king, thus sometime in the 1160s. Other, less direct,

evidence is in the bas-reliefs of the Bayon and Banteay Chhmar. They depict scenes of combat in

boats between two sides distinguished by different headgear – interpreted without epigraphic

evidence as Cham and Khmer. They also show land warfare with infantry and elephant troops,

again distinguished by headgear but mixed, as though some of the warriors believed to be Cham

were fighting on the side of the Khmer.

Coedès summarized the main points of the epigraphic records as follows. In K.288, one

of the Chrung inscriptions, Jayavarman VII gave an outline of the reign sequence preceding his

own: ‘Dharanīndravarman [I] was despoiled by Sūryavarman [II], Yaśovarman [II] despoiled by

Tribhūvanādityavarman, despoiled by the king of the Cham named Jaya Indravarman’. There is

no controversy about the first event, and since the last date of Sūryavarman has been

hypothesised – without epigraphic confirmation – as between 1145 and 1150, the time of

Yaśovarman and Tribhūvanādityavarman would have been between the end of the reign of

Sūryavarman and the victorious Cham invasion, which for Maspéro (followed by Coedès) was

1177. Tribhūvanādityavarman is dated by one inscription (K.418) to 1166, and there are no other

132 Maspéro, Empire khmer, p. 45.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

inscriptions after 1145 (the last dated inscription to name Sūryavarman II during his lifetime)

until Jayavarman VII. 133 Except for lack of mention of that or any other date except

Jayavarman’s claim to enthronement in 1181, this seems to agree with the passage in K.485,

Phimeanakas: ‘Yaśovarman [II] having been …[effaced] by a servant with ambitions to obtain

royal power [presumably Tribhūvanādityavarman], the king [Jayavarman VII] returned from

Vijaya … to come to the aid of the ruler [Yaśovarman]’. Yaśovarman had already been

overthrown, however, and Jayavarman VII waited; thus he claimed to have been in Vijaya at

some time before he became king in 1181.

Other relevant passages in this badly damaged inscription are of interest. ‘Śrī Jaya

Indravarman, king of the Cāmpas…transporting his army on carts, went to fight the country of

Kambu…’; and a following badly damaged passage which Coedès interpreted as ‘he (Jaya

Indravarman) killed the usurper (Tribhūvanādityavarman)’. Finally, ‘having…defeated this (king

of the Chams) in battle…he (Jayavarman VII) had possession of…through the conquest of

Vijaya and the other countries’; and in one more context a reference to ‘the Vijaya expedition’.

As will be clear below in the discussion of the Champa inscriptions, Coedès (following

Maspéro) was mistaken in amalgamating two Jaya Indravarmans of Champa – one denoted as of

Grāmapuravijaya and the other as Oï Vatuv – into a fictional Jaya Indravarman IV. It was

undoubtedly the first who led the campaign(s) against Cambodia which overthrew

Tribhūvanādityavarman and attacked Angkor, but his last dates in the Champa records are 1163,

1164, 1165, 1167, 1168, 1170 and possibly 1183.134 The Phimeanakas inscription thus implies

that Jayavarman VII already occupied Vijaya in the 1180s. Thus, during some period of the

1160s when there were internal conflicts in both Cambodia and Champa, the future king

Jayavarman VII of Angkor had been in Champa (Vijaya). Although he said that he tried to return

to help Yaśovarman, he was too late; he did not challenge Tribhūvanādityavarman, but waited

until the latter had been overthrown by the Cham.135

133 In fact, the nature of the the inscription is peculiar and its provenance is uncertain. K.418 is really two inscriptions on two trays (‘plateaux’) (thus easily movable) found at Phnom Svām in the far south of Vietnam. The date is on one and on the other the name-title Kamrate≡ añ śrī Tribhuvanādityavarmadeva. 134 The referent of the last date, in inscription C30A3, is uncertain; only the date 1167 refers with certainty to Jaya Indravarman Grāmapuravijaya 135 Coedès, Indianized states, p. 169 considered that he was on an expedition to Vijaya, which seems to fit the available evidence. Vittorio Roveda, Khmer mythology, secrets of Angkor (Bangkok: River Books, 2000) proposed that ‘like Rama, he had been unjustly exiled’ (p. 33), but there is no evidence for this.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

The question which stares one in the face is: what were the relations between the future

Jayavarman VII in Vijaya and the contending Cham factions of the time, in particular the Jaya

Indravarman who allegedly invaded Cambodia? The Champa inscriptions discussed below,

although giving much detail about relations between Cham princes and Jayavarman VII, have

nothing about an invasion in 1177 or a Champa occupation of Angkor at any date. This date and

that of 1170 for a first attack were extrapolated from Chinese sources, which are hardly credible,

as we shall see. The Vietnamese recorded very little about these events, but one of their records

casts doubt on the likelihood of a great invasion in 1177.

Bernard-Philippe Groslier attempted an original interpretation of the events of the time

through a detailed discussion of the Bayon bas-reliefs. For him, Jayavarman VII ‘was in

Vjiaya around 1165…He presumably was making war as G. Coedès has interpreted stela K 485

from the Phimeanakas’; ‘like Coedès’, Groslier added, ‘I would lean towards [the idea of] a

military expedition in Vijaya around 1165’. He also hypothesized that Jayavarman returned from

Vijaya in 1165-6, and from then until 1177 resided at the Preah Khan of Kompong Svay. This

last is of course pure speculation. 136 Groslier’s discussion here is concerned with the

interpretation of an impressive Cham victory over the Khmer depicted on the East side of the

northern outer gallery of the Bayon – but what event was it? The Cham assult against…

Tribhuvanadityavarman ... between 1170-71 and 1177 quite naturally comes to mind. Indeed,

Jaya Indravarman IV of Champa, after having ensured the neutrality of Đại Việt, attempted a

first attack on Cambodia around this date. It came by land…[with] carts’ (K. 485). ‘This attack

by land having been inconclusive, however, the Cham, with the help of a Chinese pilot, took to

the sea and then went up the Mekong in 1177, taking Angkor by surprise, burning it, and killing

the usurper’.137

Still referring to the scene at the Bayon, however, Groslier wondered whether it was

really the Cham victory over Tribhūvanādityavarman: ‘If they wanted to emphasize this decisive

moment, then why choose the first assault by land, which remained inconclusive, rather than the

naval invasion which dealt the final blow? Groslier’s conclusion was that this scene of a great

Cham victory represented the story in the Banteay Chhmar inscription, but with the future

Jayavarman VII as the endangered Cambodian leader rather than his son, as Coedès had

136 Bernard Philippe Groslier, ‘Inscriptions du Bayon’, in Jacques Dumarçay, Le Bayon histoire architecturale du temple (Paris: École Française d’Extrême-Orient, 1973), pp. 149-52. 137 Ibid., p. 156.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

proposed.138 The obvious negative response to this is that Jayavarman VII would hardly have

taken the story of an embarrassing defeat as the subject of a major scene in his central temple.

The same question may be asked about the Phimeanakas inscription, in which Jayavarman

recorded his final victory after a Cham invasion. Why refer only to the supposedly failed land

assault (speculatively dated by Coedès, Maspéro and Groslier to 1170) instead of the final Cham

victory in 1177, from which Jayavarman wished, some time after 1181, to claim that he saved

the country?

Groslier continued, still in the context of an explanation of the Bayon reliefs: ‘We know

that a new Cham attack occurred after Jayavarman had come to power and after the pacification

of Malyang and before the conquest of Vijaya around 1190…all reasons for dating this battle

around 1182-83 and [locating it] on the very site of the future Preah Khan of Angkor’. (K.908

says it was the site of a victory). This explanation is based on Maspéro’s remark that ‘in 1112 ś.

(1190) King Jaya Indravarman on Vatuv revolted against the King of Cambodia’ (Mỹ Sơn

inscription 92B); Groslier impermissibly adjusted the 1190 date, which is explicit in the Champa

inscriptions.139 In contrast to other versions of the history of this period in which the Cham

attack (or revolt) was in 1190 and was soon followed by Jayavarman’s conquest of Champa,

Groslier, having put the event in 1182, says that ‘the naval expedition against Champa which

ensued must have been around 1183-84’; his interpretation was based on the apparent naval

battle scenes at the Bayon.140 For him, the final victory over the invading Cham was this naval

battle in the 1180s, not the conquest of Champa which occurred in 1190.

Groslier’s reconstructions of events are much too speculative to be acceptable, in

particular with respect to his new (hypothesized) dates, sometimes in conflict with inscriptions,

and his illegitimate multiplication of battles. Because of the lack of information from Cambodia,

and the uncertainty about the period between Sūryavarman II and Jayavarman VII as well as the

latter’s background, the Champa records are also of importance for reconstructing the history of

Cambodia at that time. There are four inscriptions of what was apparently a new royal faction:

C17 (Phan Rang), C30A2 (Nha Trang), and C100 and C101 (Mỹ Sơn). They tell of a king Jaya

Harivarmadeva (Prince Śivananda), who was the son of Jaya Rudravarmadeva (C17, C30) –

posthumous name (Parama) Brahmaloka (C100, C101) – and the queen Nai Jiññyaï (C100), also

138 Ibid., p. 156. 139 Ibid., p. 164; Maspéro, Royaume de Champa, p. 164 n. 9. 140 Groslier, ‘Inscriptions du Bayon’, p. 164.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

called Paramasundarīdevī (C101). This king was the grandson of Rudraloka and of the lineage of

Paramabodhisatva (C101), who was himself a person of Ratnabhumivijaya.141

In 1147 Jaya Harivarman followed his father Jaya Rudravarman to Pāõóuraïga, where the

latter died and the people asked the son to reign. In that year troops of Cambodia and Vijaya

attacked Pāõóuraïga (C17); in 1148 Jaya Harivarman defeated them at battles in Caklyaï and

Vīrapura (C17). When he went to the South, the king of Cambodia (presumably Sūryavarman II)

sent a general (senāpati) Śaïkara to attack in the plain of Rājapura. (The precise locations of

these three toponyms are unknown, but the context suggests that they were somewhere between

Quy Nhơn and Phan Rang.) The Cambodians were defeated; the king then sent a much stronger

force to fight in the plain of Vīrapura, and it was also defeated (C101). (These events are not

dated in C101, but they agree with the dates in C17). The Cambodian ruler had installed his

brother-in-law Harideva, accompanied by troops, as king in Vijaya (C101); in 1149 Harideva

was captured by Jaya Harivarmadeva (C17). There were more victories over the Cambodians: in

Cambodia, presumably in the time of Yaśovarman II; in Vijaya in 1151, 1155 (C17), 1157 (C100)

and 1160 (C17, C30); and over the Vietnamese (C30). During this time Jaya Harivarman had

been victorious in the North as far as Amarāvatī (Quảng Nam) and in the South down to

Pāõóuraïga (C30); he defeated Cambodian and Cham opponents and reigned supreme after that,

presumably in both Vijaya and the South (C101). The future Jayavarman VII must have been

involved in some of the last of these campaigns.

It is in fact not certain which Cambodian king was reigning at the time of the attacks

between 1147 and 1160, recorded in the Champa inscriptions. The last mention of Sūryavarman

II is in 1145 and there are no more Cambodian inscriptions until the time of Jayavarman VII in

the 1180s, except for one recording Tribhuvanādityavarman in 1166. It has been assumed that

the Cambodian conquests in Champa in the 1140s and the attacks on Đại Việt recorded in the

Vietnamese sources were by Sūryavarman because of his reputation and the assumption that his

successors were weak; thus Coedès was willing to extend his reign to 1150 because the

Vietnamese histories record an attack on Nghệ An by Cambodia at that date.142 But then why not

allow Sūryavarman to have lived until the invasions of Champa in the 1150s? Because they all

141 I would propose that the pu po nei yāï cei Dav Veõi Lakùmī Sinyān of C91 was a princess who later became queen Nai Jiññyaï, with nei and nai representing Cham vinai ‘woman’ and the lakùmī certainly indicating a female. It is quite aberrant for Schweyer (‘Vaisselle en argent’, p. 335) to insert ‘[Jaya Indravarman]’ as part of her name-title. 142 Coedès, Indianized states, p. 160; see Tt, Bản kỷ, 1:7a, vol. 1, p. 318.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

ended in defeats? We must realize that quite arbitrary choices have been made by Maspéro and


Then there was war with the ‘Kirāña’, hitherto interpreted as the non-Cham peoples in

the mountains and forests to the West of the Champa coast; the fighting was in ‘the plain and

forest in the South’, and Jaya Harivarman defeated them (C101). After that, the text of C101 is

confusing, probably because Finot was unable to translate the Cham completely. As it reads, the

chiefs of the Kirāña proclaimed Jaya Harivarman’s brother-in-law Vaïśarāja as king in the city of

Madhyamagrāma. Jaya Harivarman defeated him, and the Cham text says that he killed (vunuþ)

him; Finot’s translation, however, only says that they fought.143 Then, the translation continues,

‘the king of the Yavana [Vietnamese] proclaimed as king “a man of Champa” named Vaïśarāja

and gave him several Yavana senāpati and troops.’ Jaya Harivarman deployed the troops of

Vijaya and defeated these opponents, afterward giving much booty to various temples, including

a temple to his father Paramabrahmaloka and another to his mother Jiñjyaï (C101) (presumably

the Nai Jiññyaï cited above from C100).144

Finot’s version, in addition to translation problems, is obviously corrupt here. First he

makes Vaïśarāja a Cham prince chosen as king by the Kirāña, then after his defeat says that ‘a

man of Champa’ named Vaïśarāja was chosen by the Vietnamese, with the added incongruity

that they proclaimed him king of Champa because they had learned that the king of Cambodia

was raising obstacles to Jaya Harivarman. Maspéro ‘solved’ the problem by concluding that

there was a single Vaïśarāja who, after his first defeat by Jaya Harivarman, fled to Vietnam,

where he obtained support. Of course, if the Vaïśarāja of the first context was killed, he could

not have escaped to Vietnam, nor been appointed as king by the Vietnamese. Perhaps, for a

different speculation, ‘vaïśarāja’ should be construed literally as ‘lineage chief’, rather than as a

name. At least, if the inscription has been read at all accurately, it means that there were two

individuals whom the Cham called vaïśarāja.

As usual, the Vietnamese sources are hard to use, because of the completely different

names for the individuals concerned. They do say that in 1152 the Vietnamese king appointed a

Cham named Ung Minh Ta Ðiệp as king of ‘Chiêm Thành’ (‘Champa city’), but he was killed by

143 The text is mçsuþ vunuþof which the first term was translated consistently by Aymonier and Finot as ‘combattre’. Finot and the glossary in Jacques ed., Études épigraphiques translate vunuþ as ‘combattre’ as well, which in the given context is redundant, and as cognate with Malay bunuh it must be taken as ‘kill’. I suggest that here Finot was forcing a translation into a misunderstood context. 144 Finot’s translation is in Jacques ed., Études épigraphiques, pp. 145-8; Maspéro, Royaume de Champa, pp. 158-9.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

the Champa king Chế Bì La Bút.145 Chế, as in all Vietnamese records concerning Champa, is

probably the Cham princely title cei/ciy, but ‘Bì La Bút’ cannot be restored as anything related to

Jaya Harivarman. Even if this may coincide with the Cham inscription concerning Vaïśarāja

appointed by the Vietnamese, it does not justify Maspéro’s supposition about his escape to

Vietnam after a first defeat.

The time period of these events is from near the end of the reign of Sūryavarman II in

Cambodia to the time (presumably) of Yaśovarman, and reflects the initial victory of the

Cambodians in Vijaya in the time of Sūryavarman. It would seem that the move of Jaya

Rudravarman and his son Jaya Harivarman to Pāõóuraïga might have resulted from that first

Cambodian victory in Vijaya; the story of Jaya Harivarman would then reflect their eventual

defeat and a greater unification of Champa than before: at least Pāõóuraïga with Vijaya,

implicitly including Nha Trang (site of C30B4) and possibly extending to Amarāvatī in the North.

It has been accepted since the writings of Maspéro and Coedès that some 10-20 years

after the above events, the Cham invaded Angkor, after which in 1181 Jayavarman VII became

king in Cambodia and, starting in 1190, reconquered Champa. Before going on to the relevant

Champa inscriptions, let us summarize their version of the history of Champa in that period.

Maspéro starts the story in 1167 with a new king in Champa whom he calls Jaya Indravarman IV,

a ‘usurper’, an amalgamation of two persons named in the inscriptions as Jaya Indravarman oï

Vatuv and Jaya Indravarman of Grāmapuravijaya. This introduces the confusion found in his

recounting of events because, those two individuals were not the same person, as Finot had

pointed out. Maspéro’s Jaya Indravarman IV, preserved in Coedès, is thus a fiction.146 Based on

Chinese reports, Maspéro then says that this Jaya Indravarman attacked Cambodia in 1170, but

the two sides were equal and the struggle inconclusive. Then a shipwrecked Chinese officer

showed him how to maneuver cavalry and shoot arrows from horseback. Thinking that this

would give him an advantage over the Cambodians, Jaya Indravarman tried to buy horses from

Hainan but was refused, the Chinese emperor saying that it was forbidden to export horses from


Part of this, of course, is utter nonsense. The bas-reliefs both of Angkor Wat and of the

Bayon show that the Cambodians were perfectly familiar with the use of cavalry, and if so, the 145 TT, Bản kỷ, 4:10a-b, v. 1, p. 320. 146 Maspéro, Royaume de Champa, p.161 and n. 5, where he rejected Finot’s correction (‘Inscriptions du Musée’, p. 250 n. 2). 147 Ibid., pp. 163-4.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Cham must have had equal familiarity. Given the divisions within Champa and the long relations

between the two countries, the Cham could conceivably have obtained horses from Cambodia, if

they had not already known them as early as the Cambodians – probably the true situation.

Wade’s new translation of the Song Huiyao shows that the Cham were very familiar with horses

and had been receiving them from China since at least the tenth century. Pictorial proof of Cham

horsemanship is in sculpted scenes of Cham polo players and horsem*n, and their general

familiarity with horses is demonstrated in scenes on the pedestal of the Vihāra of Đồng


Failing to get horses, Jaya Indravarman, according to Maspéro, decided on a naval attack;

guided by (another?) shipwrecked Chinese person, in 1177 his fleet went down the coast then up

the river (Mekong-Tonle Sap), surprised and pillaged the Cambodian capital and returned with

enormous booty. Note that in this version the Cham did not remain in occupation of Angkor, as

some later interpretations would have it.149 Here is more nonsense. First of all, it expressly

contradicted by the passage of the Phimeanakas inscription quoted above that ‘Śrī Jaya

Indravarman, king of the Cāmpas… transporting his army on carts, went to fight the country of

Kambu…’ Moreover, after centuries of close relations with Cambodia, both amicable and

bellicose, including several invasions of their neighbor’s territory, the Cham knew well all the

routes into Cambodia and had no need of a shipwrecked Chinese to show them the way.

Moreover, whether by land or by river, the campaign would have taken weeks, and they could

not possibly have taken Angkor by surprise. This story was only credible at the time when it was

believed that the Cham were remnants of an overland migration by ‘Indonesians’ and their own

seafaring abilities were ignored, and when the French scholars studying Indochina considered

that everything written in Chinese should be taken literally as holy writ. The date 1177, which

has become a solid ‘factoid’ in everything written subsequently, has its origin here, but in this

case even Maspéro recognized that its source was wrong on another date; this, together with the

tales of shipwrecked Chinese helping the Cham, casts doubt on its other details.150

So far, then, there is a problem as to whether the main Cham attack was by land or sea. If

by sea, why did Jayavarman VII in his inscription – written after defeating the Cham – portray 148 Geoff Wade 2004; Association Française des Amis de l’Orient, Musée de sculpture, respectively figures 124, 126, 38, 44. In the new section of the Đà Nẵng Museum, and not illustrated in the catalogue, is a small (82 cm high) sculpture (number DN 19, tenth century) of a pair of horses being ridden and guided by a single rider straddling the backs of both – just the sort of scene common among people familiar with raising and using horses. 149 Maspéro, Royaume de Champa, p.164; Népote, ‘Champa propositions’, part 2, p. 98. 150 Maspéro, Royaume de Champa, p. 164, n. 6, 8.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

the latter as having arrived on chariots? Before going further let us take a look at the Chinese

sources (as interpreted by Maspéro) for the date 1177, and the real primary sources, the

Cambodian and Champa inscriptions.

First, let us look at the date 1170. Chinese histories record an unsuccessful Champa

attack on Cambodia at that date; as will be noted below, a Champa inscription allows the

inference that 1170 was the last of a series of incursions beginning in the 1160s. This inscription

was not used by Maspéro, however, and it seems that his evocation of 1170 was based on an

imaginative construction of the records of Champa-Vietnamese diplomatic relations, in addition

to the Chinese histories. Thus he said that Jaya Indravarman, having secured the neutrality of Đại

Viet by a mission in that year, tried to attack Cambodia by land. Indeed, the standard Vietnamese

chronicles say that in that year ‘Chiêm-thành’ came to cống (offer tribute). They have the same

entry in 1152, 1153, 1160, 1164, and so on. 151 This was part of normal Champa-Vietnam

relations, and it is impossible to give a special significance to any such record.

Then for the date 1177, Maspéro relied on late Chinese sources – which he considered to

be in error, however, because they situated Jayavarman’s conquest of Champa in 1195-1201

rather than the accepted date of 1190.152 Based on that, and on the clear fiction of the Cham

depending on a Chinese pilot to find the way into Cambodia, it seems to be possible to say that

the date 1177 was also in error. Nevertheless, the Song Huiyao says that in 1177 – without a

shipwrecked Chinese guide – ‘Champa used a boat-borne force to attack Cambodia... [the

Cambodians] sued for peace but were ignored and massacred’. This source apparently does not

claim that the Cham invaded and occupied Angkor, nor is it clear where the attack on the

Cambodians took place. The text continues that in 1198-9 ‘Cambodia launched a massive attack

on Champa ... Champa sent a missive indicating their surrender, but the Cambodians

exterminated everyone and established a Cambodian as ruler of the area’. The date here and the

‘extermination’ are contradicted by Champa epigraphy; this casts some doubt on the

interpretation of 1177 as referring to a Cham conquest of Angkor, and even on the date itself.153

The problem of 1177 is made more complex by the Vietnamese histories, which record

that in that year the Cham attacked Nghệ An. If, as I have argued, there was no unified Champa,

the northernmost Champa entity could have attacked Nghệ An while another was invading 151 See Tt and Vsl under the entries for these years. 152 Maspéro, Royaume de Champa, p.164 and n. 4, citing the Wenxian tongkao (referred to by French scholars as Méridionaux) and the Song Shi. 153 Extract from the Song Huiyao, personal communication from Geoff Wade.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Cambodia. Then, however, if only one region/chiefdom of Champa attacked Cambodia, the

invasion might well have been less violent than has been interpreted. Maspéro, interestingly,

recognized the problem presented by the Vietnamese histories but slyly pushed it aside, writing

that as in 1170, Champa sent a mission to Đại Việt in 1184. ‘This’, he commented, ‘did not stop

the Chams from the northern provinces from occasionally raiding Nghệ An’; he failed to mention,

however, that such an event only took place in 1177.154 Maspéro was here really cheating in

refusing to report that the Vietnamese sources were in conflict with the hypothesis of a major

Champa attack on Angkor in 1177. As noted above, the records of Champa envoys to Đại Việt

are frequent, differ in no way one from another, and may not be exaggerated as special attempts

to gain Vietnamese neutrality.

After the Cham attack, Jayavarman VII – again according to a Chinese source – vowed to

take a terrible revenge’; in 1190, ‘profiting from an act of aggression by the Cham’, he attacked

and eventually conquered Champa. 155 Coedès summarized the same story, differing from

Maspéro only in recognizing that the father of Jayavarman VII, Dharanīndravarman II, followed

Sūryavarman II, and was in turn succeeded by two more kings, Yaśovarman II and

Tribhūvanādityavarman; the last was displaced by the Cham. Maspéro had believed that

Jayavarman VII directly succeeded his father.156

In addition to a rejection of the Chinese tales of Chinese guides, adopted by Maspéro and

Coedès, the next Champa inscriptions force more nuances of interpretation in the story.

Following the inscriptions related to the intervention in Champa of Sūryavarman II, there are two

short inscriptions (C53, C54) in Bình Định which name a king Jaya Indravarmadeva; Bergaigne,

dates them only to the end of the eleventh century śaka (1178-1278 CE) apparently on

palaeographic grounds. They are difficult to relate to anything else because of the uncertain date

and lack of concrete information, but they may be relevant evidence for the growing importance

of Vijaya in the twelfth century. Because C53 records a donation to yāï pu nagara, usually

considered to be the goddess of Pô Nagar in Nha Trang, that may have been the center of Jaya

Indravarman’s territory.

Perhaps this Jaya Indravarman was the Jaya Indravarman of Grāmapura pradeśa/vijaya

mentioned in C85/1163 and C92A in Mỹ Sơn and C30A3 in Nha Trang. In the first he is only

154 Maspéro, Royaume de Champa, p. 163, n. 5. 155 Ibid., p. 164, note 6, citing Wenxian tongkao. 156 Coedès, Indianized states, pp. 164, 168-9; Maspéro, Royaume de Champa, p. 164 note 7.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

named, with no other information, but the second contains a list of his donations to various

temples/gods in the years 1163, 1164, 1165, 1167, 1168 and 1170. These included

Śrīśānabadreśvara, the main temple of Mỹ Sơn; a Buddha Lokeśvara, which suggests the site of

Indrapura/Ðông Dương; and two Bhagavatī, of which one must have been the temple in Nha

Trang. This type of donation suggests the results of a victorious campaign, or campaigns; it

would seem that at the dates implied – between 1160 and 1170 – this can only mean, successful

warfare against Cambodia.

In this connection it is interesting to note that in spite of their great detail concerning the

struggles with Cambodia in the reigns of Sūryavarman II and Jayavarman VII, the Champa

inscriptions are silent about the supposed great victory over Angkor in ‘1177’. This stands in

contrast to the earlier boasts of the exploits (apparently in southern Cambodia) of a famous

warrior, Senāpati Pār (C19/seventh c. ś., C37/813, C31/817, in Phan Rang and Nha Trang, see

above), and later about pillaging Śambhupura (C95/1056, see above). There is only the

fragmentary C30A3 from Nha Trang containing two dates, 1167 and 1183; lists of donations by

Jaya Indravarman of Grāmapuravijaya to a Bhagavatī Kauthāreśvarī, certainly in Nha Trang but

not one of those mentioned in C92A; and a truncated phrase about ‘going to conquer Cambodia’,

which Aymonier construed as ‘lorsqu’il va’ (‘when he went’), but which Maspéro preferred to

render as ‘étant allé’ (‘having gone’). In addition to Aymonier, Maspéro cited Bergaigne; the

latter, however, did not recognize the phrase in question at all, but only the word ‘Kamvujadeśa’

with a footnoted query, ‘still another victory over the Cambodians?’157

In Schweyer’s new and allegedly complete reading of this inscription, the phrase in

question is ‘niy … nau [‘go’] mak [‘take’] Kamvujadeśa’. Schweyer has preferred to follow

Maspéro and translate ‘après être allé’ (‘after going/having gone’), but that is an impossible

rendition of niy, a simple demonstrative which is closer to Aymonier’s ‘lorsque’ (when).158 The

problem is determining which of the two dates relates to ‘go take Cambodia’. Those who

translate ‘after going’ assume that the date of the campaign must have been 1177 and that the

date at the end of the text (1183) refers back to it. Even if a retrospective interpretation may be

possible, the translation ‘after having gone’ is not. The earlier date, 1167, relates just as well –

even better, I would say – to the incomplete phrase, and it fits better with the series of important

donations listed between 1163 and 1170 in C92A.

157 Ibid., p. 164, n.6; Aymonier, ‘Première étude’, pp. 44-5; Bergaigne, ‘Ancien royaume’, p. 87. 158 Schweyer, ‘Po Nagar’, deuxième partie.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

The combined evidence of these two inscriptions, however, does not suggest a great

conquest in 1177 as implied by the the versions of Maspéro and Coedès, which were based on

admittedly poor Chinese sources, some compiled long after the event and obviously at second

hand. These inscriptions suggest rather a series of successful raids in the 1160s and 1170s, when

Cambodia was in political turmoil and the future Jayavarman VII was himself in Champa,

according to his own inscriptions in Vijaya. Coedès, moreover, embellished the story with a

naval battle, presumably suggested by the Bayon and Banteay Chhmar reliefs which show two

different forces, believed to be Cham and Khmer, fighting in boats which look like very large

pirogues or canoes. He claimed that this was recorded in Verse LXX of the Phimeanakas

inscription, but that text is so vague and allegorical that were it not for the reliefs, it would not

have been interpreted as recording a naval battle.159

The next inscriptions are C92B and C92C in Mỹ Sơn, with dates 1182, 1190, 1192, 1193

and 1194 – that is, after the supposed Cham conquest of Angkor, their expulsion by Jayavarman

VII, his coronation in 1181 and then during the period beginning in 1190 when the traditional

academic treatment says that he effected his conquest of Champa. These inscriptions relate the

story of a Cham king Sūryavarmadeva (Prince Vidyānanda of Tumpraukvijaya) who went to

Cambodia in 1182 and was taken into the service of the king of Angkor.160 In that capacity he

led troops to put down a rebellion in Malyang, believed to be in western Cambodia, and he was

then made Yuvarāja by the king of Cambodia. Then in 1190 there was a rebellion, apparently

within Champa, by a king Śrī Jaya Indravarman oï Vatuv; Maspéro, forcing the story into his

preconception of events, called it a ‘Cham aggression’, while Coedès referred to ‘a new attack by

the Cham king Jaya Indravarman ong Vatuv’ – who, in fact, does not appear in any of the earlier

records. This provoked the invasion of Champa by Jayavarman VII.161 Against this rebellion the

king of Cambodia sent the prince (Sūryavarmadeva) with Cambodian troops to take Vijaya and

capture Jaya Indravarman oï Vatuv; he succeeded, and the latter was sent to Cambodia.

159 The text as rendered by Coedès (Inscriptions du Cambodge, v. 2, p. 177) is ‘Having through his patience in the midst of misfortune [or with vessels whose strength…] defeated in combat this (king of the Cham) whose warriors were like an endless ocean, after having received royal consecration, he possessed, through the conquest of Vijaya and other countries…’. It takes real imagination to see evidence of a naval battle here, and the reference to vessels seems to have been only a secondary possibility in translating the complex Sanskrit. 160 In Cham royal titles of the time names of men entitled ‘king’ (pu po tana raya) are often followed by their (presumably earlier) princely (cei, ciy, ‘prince’) title, as in this case. 161 Maspéro, Royaume de Champa, p. 164; Coedès, Indianized states, p.170. As noted above, Maspéro erroneously assimilated Jaya Indravarman oï Vatuv to Jaya Indravarman of Grāmapuravijaya, and Coedès was following that misinterpretation; Finot was certainly correct to object to this.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

It should be noted here that if the event of 1190 was a rebellion within Champa but an act

hostile to Cambodia, it implies that Champa – or at least the part under Jaya Indravarman oï

Vatuv (Vijaya, it would seem, where the future Jayavarman VII resided sometime in the 1160s

or 1170s) – had already been conquered by Cambodia, as implied in the Phimeanakas inscription.

In the same operation in 1190 a certain Śrī Sūryajayavarmadeva Prince In, brother-in-law of the

king of Cambodia, was established as ruler in Vijaya, following which the prince

[Sūryavarmadeva] went to reign in Rājapura, Phan Rang. Prince In was expelled by a Prince

Raùupati/Raghupati, who then ruled in Vijaya with the title Śrī Jaya Indravarmadeva (a third

prince with that name-title).162

In 1192 the king of Angkor sent Cambodian generals with Jaya Indravarman oï Vatuv

(who had implicitly switched sides and joined the Cambodians); he came to meet the prince

(Sūryavarmadeva) in Rājapura. They led the Cambodian troops, took Vijaya and captured and

killed Jaya Indravarman cei (‘prince’) Raù(gh)upati; the prince [Sūryavarmadeva] then reigned

in Vijaya. Later in the same year Jaya Indravarman oï Vatuv deserted the Cambodians and went

to Amarāvatī (Thu Bồn valley, Mỹ Sơn). He rebelled, raised troops from several locations and

went to take Vijaya. The prince [Sūryavarmadeva] pursued and killed him and ‘then reigned

without opposition’. The following sentence of the inscription is damaged, but apparently there

was a break between the prince (Sūryavarmadeva) and the king of Cambodia, for in 1194 ‘the

prince fought and defeated the Cambodians’. Two years later, the king of Cambodia sent more

troops, who were also defeated, following which the prince [Sūryavarmadeva] went to

Amarāvatī and made offerings to Śrīśānabhadreśvara.

A view of those last events from another point of view is in C90D, undated but easily

placed in the same temporal context. It is by a Yuvarāja oï Dhanapatigrāma, whose career at first

paralleled that of Sūryavarmadeva, for he also went to Cambodia and led troops to suppress the

rebellion in Malyang. The inscription continues immediately, ‘King Sūryavarmadeva rebelled

against the king of Cambodia’, who ordered Dhanapatigrāma to take Cambodian troops and

capture him. The results of the campaign are not mentioned, but it was apparently successful, and

was followed by another rebellion by King Ājñā Ku whom Dhanapatigrāma captured and sent to

Cambodia. The last sentences in the inscription are unclear. Maspéro, relying on Vietnamese

162 Finot read the name as Raùupati, but now Schweyer (‘Chronologie des inscriptions’, p. 337) and Claude Jacques (Études épigraphiques, p. xvi note 10) prefer the reading ‘Raghupati’. This is a question which I leave to the Sanskritists, but I think that this new reading, which originated with Jacques, is probably correct.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

histories, says that Sūryavarmadeva finally sought refuge in Vietnam, but because of differences

in names between the Champa inscriptions and these texts, that conjecture may not for the

moment be considered reliable. He also, on the basis of uncertain interpretations of the

Vietnamese sources, says that Dhanapatigrāma was paternal uncle of Sūryavarmadeva. Coedès

here follows Maspéro.163

All of these details from the Champa inscriptions, contrary to the standard histories,

imply that Champa – at least the center and South, but perhaps not the Thu Bồn area (Amarāvatī)

– was subordinate to Cambodia since sometime before 1190 and that the victories of Jayavarman

VII both at home and in Champa had depended to a large extent on Cham supporters, but that

once given authority in their homeland, the latter were unreliable. These Champa inscriptions of

the end of the twelfth century were the work of Champa chieftains who had at times been allies

of Jayavarman VII, and they reflect some degree of sympathy with his activities.

There is then a long break in the epigraphy until the 1220s, when three inscriptions

(C4/1220, 1227, Phan Rang, C30B4/1226, Nha Trang, and C86/1230, 1234, Mỹ Sơn) refer

retrospectively to a 32-year war with Cambodia and show a quite different attitude toward their

neighbour and its relations with Champa. The first two are available in the usual summaries by

Aymonier, which make control of their contents impossible. In the third (C86) a King Jaya

Parameśvaravarmadeva oï Aïśarāja of Turaiyvijaya claims to have been sole ruler in Champa

during that time, thus leaving the fate of Dhanapatigrāma uncertain. Taken together they say that

in 1190 Jayavarmadeva of Cambodia conquered the earth and took nagara Champa or that in

1201 he came, appointed a Yuvarāja and left. There was a 32-year war until 1220 (which would

thus have begun in 1188), when the Cambodians went to Vraþ Nagar (Angkor) and the Cham

went to Vijaya. In 1226 King Jaya Parameśvaravarman was enthroned.164

Maspéro embellished the story without sources, saying that Oï Aïśarāja of Turaiyvijaya

‘was raised at the court of Jayavarman VII…receiving in 1201 the title of yuvarāja and

permission to rejoin the governor Dhanapatigrāma in Champa’, all of which could be true, but

without textual evidence is historical fiction. The fate of Dhanapatigrāma is unknown, although

Maspéro assimilates him to a Yuvarāja Mnagaþïa oï Dhanapati who in a very brief inscription

(C92C, following the story of Sūryavarmadeva) dated 1244 claimed to govern (uncertain 163 Maspéro, Royaume de Champa, p.166; Coedès, Indianized states, p.171. 164Although this last detail is from Aymonier, ‘Première étude’, pp. 50-52, the Cham text in this case is supplied and does indeed seem to say this. Note that a full treatment of inscriptions C4 and C30B4 is in Schweyer, ‘Po Nagar’, deuxième partie.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

translation) Champa. This is in conflict with King Jaya Parameśvaravarman’s inscription C6 in

Phan Rang in the same year and implies that the two were rivals controlling different Champas,

one in the North and one in the South – a situation which was not unusual, as has been

emphasized here. Jaya Parameśvaravarman was certainly an important figure. He is named in six

inscriptions dated between 1220 and 1244 and four others without dates; all but two of these are

in Nha Trang or Phan Rang, C86 is in Mỹ Sơn and C52 in Bình Ðịnh.

Maspéro, mentions Cham warfare with Viet Nam, but says that ‘the documents are not in

agreement either on the dates or results’.165 The documents are a Cham inscription (C4) from

Phan Rang and the Vietnamese histories. The inscription is one of those which Aymonier only

summarized without a complete transcription and translation, and which seems to say that there

was a joint Cambodian-Champa campaign to seize some Yvan, presumably Vietnamese, in śaka

1123/CE 1201 (date corrected by Claude Jacques in marginal notes to my copy of Aymonier

from Aymonier’s and Maspero’s 1129/1207), and that both Yvan and Khmer suffered losses.

However, much is unclear in the text presented by Aymonier, and so far no attempt has been

made to restudy it, except for Jacques’ informal correction of the dates. It does not say where the

battles took place. Moreover, the text continues with the date 1142/1220, saying – this time

clearly – that ‘the Khmer went to Vraþ Nagar [undoubtedly Angkor] and the people of Champa

came to Vijaya’ (kvir nau Vrah Nagar uraï Campa marai Vijaya). The last date in the text is

1148/1226 (corrected by Jacques from Aymonier's 1149/1227), when the Champa ruler was

crowned and built palaces, temples, gods, etc in Śrī Vijaya. Given the great timespread indicated

by these readings and the fragmentary character of the text as read by Aymonier, it is possible

that the first date of 1201 is incorrect. Aymonier and Maspéro believed that the Cham chief in

question was Jaya Parameśvaravarman, but his own inscriptions do not begin until 1220.

The Vietnamese histories say that in 1216 and 1218 Cambodia and Champa jointly

attacked Nghe An but were easily defeated. Perhaps Aymonier's date of śaka 1129 (1123 for

Jacques) should be re-read as 1138/1216, given the known confusions between 9 and 8, 2 and 3.

These vague sources, like the Champa inscriptions of the 1190s, seem to suggest that the

situation was still as it had been preceding the domination of Jayavarman VII in Cambodia, when,

in the words of Claude Jacques, ‘there were Khmer and Cham against different groups of Khmer

165 Maspéro, Royaume de Champa, p. 168.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

and Cham’ (as is clear in the Bayon reliefs). ‘There were several kingdoms in the area of

Champa and different alliances were formed with diverse Khmer factions’.166

All of this suggests that the traditional academic history of the time needs revision. There

is no good evidence of a great Cham conquest of Angkor in 1177, certainly not with the details

supplied by the Chinese. During the time when Cambodia was in turmoil in the 1160s and 1170s,

there may have been more or less successful raids from Champa while the future Jayavarman VII

was in Vijaya and, we may assume, was part of the Champa political scene. We may accept his

own statements that he saw a King Jaya Indravarman (presumably the one from

Grāmapuravijaya) as a rival, but I would suggest that the real conquest of Angkor was by

Jayavarman VII and his Cham allies – probably in the 1170s, at least before 1181 – and that the

subordination of central and southern Champa to him dated from that time. This is because one

Mỹ Sơn inscription called the event of 1190 a ‘rebellion’ against him in Champa, apparently in

Vijaya – not a Cham aggression which offered him an excuse to invade and conquer Champa, as

stated by Maspéro and Coedès. The other reference to that date (C30B4) has indeed been

interpreted as saying that he conquered the ‘earth’ (sarvvadvīpa, literally ‘all continents’) and

took nagara Champa, but it is from Nha Trang and may only mean that he then took that

particular nagara Champa.167

One of the interesting, and very large, bas-relief panels at the Bayon (north outer gallery

east side) is the only panel showing a victorious army pursuing – indeed routing – their enemies

who are fleeing in panic; it depicts the forces who have always been interpreted as the Cham

victors. Coedès did not discuss it in his article on the subject, but Groslier interpreted it as the

story of a Cambodian defeat before the reign of Jayavarman recorded in the Banteay Chhmar

inscription.168 Why would Jayavarman VII have this scene given such importance in his central

temple unless that Cham army was his own? In the opinion of Claude Jacques, the interpretations

to date of the bas-reliefs

166 Michael Freeman and Claude Jacques, Ancient Angkor (London: Thames and Hudson, 1999), p. 85. The Vietnamese reference is in TT, bản ký, 4:29b, vol. 1, p. 337. On the confusion of numbers see Finot, ‘Inscriptions du Jaya Parameśvaravarman’, in Jacques ed., Études épigraphiques, pp. 187-200, especially pp. 51/199 on the period in question here. 167 Both in Aymonier’s summary (‘Première étude’, pp. 47-8) and in Schweyer’s newer unpublished reading (‘Po Nagar’, deuxième partie), the word for ‘conquer’ is conjectural, and another word preceding ‘come take Nagara Campa’ could not be translated. 168 Coedès, ‘Nouvelles données’ and Coedès, ‘Quelques suggestions sur la méthode à suivre pour interpréter les bas-reliefs de Bantāy Chmàr et de la galérie extérieure du Bàyon’, BEFEO, 32 (1932): 71-81; Groslier, ‘Inscriptions du Bayon’, discussed above.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

remain conjectural in spite of everything…Generally speaking, the interpretation of the scenes on the second storey as a Cham defeat of the Khmer seems improbable: the Khmers were scarcely in the habit of talking about their defeats, and I have difficulty believing that they would have depicted them on their State temple…I must admit, however, that I have no alternative solution.169

Here I have proposed an ‘alternative solution’ in which all the known details fit together.

Besides that scene, other battle scenes on the east outer gallery north side show Cham

military (including what are apparently officers on elephants) among the Khmer troops fighting

other Cham, as well as a group of what seem to be Chinese military mixed in with the Khmer

forces. At the very least, these reliefs in the Bayon galleries do not support a story of strictly

Khmer versus Cham warfare. We may recall in this connection that after becoming king,

Jayavarman VII broke with certain traditions. After nearly 300 years of the increasing use of

Khmer language in the epigraphy, all of his important inscriptions are in Sanskrit, which could

be seen as an international elite language serving both countries, and he adopted as his state

religion Mahāyāna Buddhism, which had always been more important in Champa than in

Cambodia. Perhaps it was his Champa associations rather than religion which sparked the so-

called Hinduist reaction against his creations – allegedly in the thirteenth century, a date which is

completely hypothetical. Although it was natural for Europeans in the nineteenth and early

twentieth centuries to assume hostility between different religions, the entire history of pre-

Angkorean and Angkorean Cambodia shows religious co-existence.

The conquest, sacking and occupation of Angkor by the Cham in ‘1177’ must be

accounted a bit of academic folklore, based on an uncritical synthesis of poor quality sources. In

fact, according to the Vietnamese Tt, in that year ‘Chiêm Thành raided [khấu 寇] the châu [Ch.

zhou, an administrative division] of Nghệ An’.170 Of course, once more we do not know which

Champa was involved, but it was almost certainly not the Champa which was either at war with

Cambodia or under Cambodian occupation. This reinforces the intepretation of the Song Huiyao

story of 1177 as relating not to a great invasion of Angkor, but to some lesser battle betwen

Cham and Cambodians.

169 Claude Jacques, ‘Le Bayon et l’épigraphie’, in First symposium on the Bayon, Siem Reap, 31 August 1996, final report, vol. I, p. 82. 170 TT, Bản kỷ, 4:18a, v. 1, p. 327.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Following the epigraphy of Jaya Parameśvaravarman, whose last record is from the mid-

thirteenth century, there are 23 more inscriptions: 14 from Nha Trang or the Phan Rang region;

four in Bình Ðịnh, which contain no useful information but at least show that the Cham were

maintaining an interest there; two in Mỹ Sơn; two in the highlands west of Nha Trang and Qui

Nhơn; and one dated perhaps to 1421 in Biên Hoà, just north of Saigon. By this time the political

center of Champa was indeed moving southward under Vietnamese pressure. These inscriptions

of the thirteenth to fifteenth centuries do not together provide much more than names of chiefs,

and I shall not try to pursue the story of Champa history further. All of the important events –

war with the Mongols (1278-88), the near conquest of Vietnam by the Cham warrior Chế Bồng

Nga (1360-90) (for both of which periods there are no inscriptions) and the further struggles

leading to the Vietnamese conquest of Vijaya in 1471 – must be pieced together from

Vietnamese and Chinese histories and competently restudied.

There is, however, an interesting story of new friendly and then again hostile relations

between the two neighbours after the failed Mongol interventions in both Vietnam and in

Champa. Trần Nhân Tông (r. 1278-93), the old Vietnamese emperor who had led the country

against the Mongols, abdicated and went to Champa for nine months to visit the holy sites –

undoubtedly, since he was a Buddhist, the Mahayana temples in Quảng Bình, Quảng Trị and

Đong Dương. His visit was thus probably restricted to northern Champa. While there, he and a

Champa ruler whom the Vietnamese sources call Chế Mân agreed on the marriage of a

Vietnamese princess with the latter in exchange for the northern Champa provinces of Ô and Lý.

According to Tt, the princess was sent to Champa in the sixth month of 1306. The following year

the Cham king Chế Mân died, and the Trần court sent envoys to rescue the princess who

otherwise, according to the Vietnamese sources, would have been cremated with the corpse of

her husband.171

In Maspéro’s synthesis one of the late inscriptions – that of Po Sah (C22) near Phan Rang,

with dates 1274, 1298, 1300, 1301 and 1306 – has been assimilated awkwardly to the story of

royal marriage and exchange of northern provinces found in Vietnamese histories. This

inscription is of intrinsic interest because its dates cover the period of war with the Mongols,

though without reference to it, and because of its apparent record of Champa relations with Java.

When carefully analyzed, it shows that at the time of the marriage of a Vietnamese princess with

171 See the account in ibid., Bản kỷ, 6: 21a-23a, v. 2, pp. 90-92.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

a Champa king and ensuing conflicts, the South of Champa (Nha Trang and Pāõóuraïga) was a

separate entity from the Thu Bồn area in the North. The full treatment of this inscription, which I

summarize here, has been published elsewhere.172

The inscription records two queens – one apparently from Java – of a King Jaya

Siühavarman, the third of that name in Maspéro’s list of kings. The identity of the other queen

was uncertain (at least in the literal readings of the inscription), but Aymonier, Finot and Coedès

eventually agreed that she must have been the Trần princess who the Vietnamese histories say

was given to a Champa king entitled Chế Mân. As Coedès concluded, after the discussions of

Aymonier and Finot, Jaya Siühavarman III ‘married a Javanese princess, the queen Tapasī’ and

in 1306 married a sister of the reigning Vietnamese emperor (Trần Anh Tông, son of Nhân Tông)

who received the title parameśvarī.173 A problem with this interpretation is that the Po Sah

inscription, which mentions a princess in the beginning who is well established as a parameśvarī

(literally, ‘top queen’), ends in 1306 – the year in which Cham-Vietnamese negotiations over the

marriage, as related in the Vietnamese histories, were just being concluded and the princess

newly sent to Champa.

Another unresolved question is the identity of ‘Champa’. Aymonier, Finot, Maspéro and

Coedès all assumed that there was a single unified Cham ‘kingdom’ and that it was possible to

identify kings named in Cham inscriptions with the quite different names of contemporary Cham

kings in the Vietnamese histories. The Po Sah inscription provides evidence for one of the main

points in the present discussion: that there was no unified Champa kingdom, that at least the

North (Quảng Nam, Thu Bồn and beyond) was always quite separate from the South (Nha Trang

and Phan Rang), and that inscriptions from other parts of Champa (especially the South) may not

be related to Vietnamese or Chinese records of the North. The Po Sah inscription was found near

Phan Rang in the far South, and none of the other three inscriptions of that king (Jaya

Siühavarman III in Maspéro’s numbering) was farther north than Nha Trang, while the

Vietnamese histories seem to indicate dealings with a Champa located in the area between Huế

and Quy Nhơn, and the negotiations concerned the districts of Ô and Lý, north Huế. Thus I

would propose that there is no connection between the Po Sah inscription and the Vietnamese

princess given to a northern Cham king in 1306.

172 Vickery, ‘Cambodia and its neighbors’. 173 Coedès, Indianized states, p. 217; Maspéro, Royaume de Champa, pp. 189-90.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

From another angle, all of Jaya Simhavarman’s inscriptions show devotion to Śiva,

whereas the former Vietnamese emperor was Buddhist. In a visit to Champa to see holy sites he

would only have been interested in those which were Buddhist, all of which were in the North.

The events of 1306 and Jaya Simhavarman’s inscriptions show a clear separation between

Śivaitic Pāõóuraïga and those Buddhist centers.

The ‘Chế Mân’ – Jaya Simhavarman connection is weak on linguistic grounds as well.

Maspéro said of the king whom the Vietnamese histories call Chế Mân, ‘Chế is the Annamite

transcription of Śrī. Man represents the final sound of the name Jaya Siïhavarman’. He was

wrong on both points. The term rendered as ‘chế’ in Vietnamese (cei/ciy in Cham inscriptions) is

‘prince’, and as L.-C. Damais explained, ‘varman’ is impossible: ‘there is no reason to restore a

form which, even in Sanskrit, is only theoretical [varman], for only the forms in -warm(m)a

actually existed’.174 Moreover, this is the only case in which a Cham royal ‘varma(n)’ name is

rendered in the Vietnamese histories by a term ending in ‘man’. This was a noteworthy example

of special pleading by Maspéro.

For the rest of the fourteenth century (1306-1401) there are no inscriptions, and historical

reconstruction depends on Vietnamese and Chinese sources, in particular the former. They relate

frequent conflict – interesting in that the Cham, under princes entitled ‘Chế’ by the Vietnamese,

were attacking and trying to recover the northern provinces of Ô and Lý in Quảng Trị. The

‘capital’ of the Champa leaders is not certain, perhaps Vijaya but certainly not Pāõóuraïga.

Another interesting twist in the story of Champa-Vietnam relations follows the death of Chế

Bồng Nga in 1390 and the end of the threat to Đại Việt from Champa after thirty years of

warfare, a period in which there are no Champa inscriptions against which to check the story in

Chinese and Vietnamese sources. According to Maspéro, following Vietnamese sources, two

sons of Chế Bồng Nga joined the Vietnamese after their father’s death. Although they received

no material support from the Trần, after the latter’s overthrow by Hồ Quý Ly in 1400, he

appointed one of the sons to govern a northern Cham province ceded (again) after 1390; this

Cham prince died fighting off other Cham trying to reconquer it.175 This, of course, is typical of

the splits in Champa noted throughout its history and the total absence of any ‘national

consciousness’. 174 Ibid., p. 188, n. 2; Louis-Charles Damais, review of Riwajat Indonesia by Poeratjaraka, BEFEO, 48, 2 (1957): 608, n. 2. 175 Maspéro, Royaume de Champa, pp. 219-24. According to Maspéro the province in question was Indrapura, only then finally ceded to Vietnam after having been ‘conquered’ more than once in the past.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Until that point in time the traditional view in modern historical literature (first Western,

then taken up by modern Vietnamese historians) that Viet Nam was an aggressive nation pushing

constantly against its southern neighbors, Champa and Cambodia, is certainly inaccurate.176 If

Linyi was early Champa, the Chinese records describe it as constantly menacing the Chinese-

dominated provinces to its north, and the Chinese were still concerned by attacks of Linyi’s

immediate successor, Huanwang, on the districts of Hoan and Ái far north in Nghệ An and Hà

Tĩnh.177 If that view of Linyi is rejected, still the first Champa-Viet warfare was brought about

by attempted Cham intervention into Vietnamese internal politics near the end of the tenth

century. Thereafter both sides alternated as aggressors, with the Cham nearly conquering Viet

Nam at the end of the fourteenth century.

Only from the beginning of the fifteenth century is the traditional conception of a

continuous push southward (Nam tiến) by the Vietnamese at all accurate.178 This is another

period for which there is only the sparse record of the Vietnamese histories, which have never

been given the critical study they require with respect to Champa. Conflict continued, and in

1402 the Cham gave up large parts of Quảng Nam and Quảng Ngãi to Hồ Quý Ly. They were

plagued by their traditional disunity in the face of Vietnamese pressure, which eventually led to

the decisive event – the conquest of Vijaya-Qui Nhơn in 1471, after which only the southern

provinces of Kauñhāra (Nha Trang) and Pāõóuraïga were left as Champa.

Although there were no more inscriptions and little solid architecture – Pô Rômê near

Phan Rang attributed to the sixteenth century and perhaps other structures and vestiges in the

central highlands, there was still a Champa which did not completely disappear until the early

nineteenth century.179 Its history has been reconstructed by Po Dharma in his Le Pāõóuraïga and

summarized in his ‘Survol de l’histoire du Campā’ in the catalogue of the Cham museum in Đà

176 This is not an impression which comes forth directly from an objective reading of the primary sources, and it no doubt developed to serve the needs of colonialists searching for a benevolent impulse in the conquest of a Vietnam constantly menacing its neighbors. Even Paul Mus was mesmerized by this view of a malignant Vietnam; see his Viêt-nam: sociologie, p. 17, and the last page of Mus, L’Inde vu de l’Est: Cultes indiens et indigènes au Champa (Hanoi: Imprimerie d’Extrême-Orient, 1934). Then Vietnamese nationalist writers adopted it as evidence for the greatness of their country in the past. 177 Maspéro, Royaume de Champa, pp. 102,105. 178 Indeed, a new generation of scholars of Vietnam reject entirely the concept of Nam tiến. As Li Tana puts it, it was a ‘series of different episodes responding to particular events or opportunities’; Li, Nguyên Cochinchina, pp. 19, 21, 28. Keith Taylor, one of the leading historians of Vietnam, also sees it as episodic, writing that ‘I do not believe that such an event [Nam tiến] ever took place’ (‘Surface orientations’, pp. 951, 960). 179 On this later architecture see Doanh, Chămpa ancient towers, ch. 20 (including notes).


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Nẵng.180 By the end of the next century the Cham had regained territory as far north as Cù Mông,

just south of Quy Nhơn. Pushing farther north into Quảng Nam, however, they provoked a

Vietnamese reaction which in 1611 took all the territory as far as Cape Varella south of the

mouth of the Đà Rằng river almost as far as Nha Trang. In 1653 the Cham attempted another

push northward and lost everything but Pāõóuraïga, in 1692 the Vietnamese court stopped using

the name ‘Chiêm Thành’; ‘Pāõóuraïga, the last vestige of Campā, ceased to exists as an

independent country’, although it still had a special status. Revolts nevertheless continued

throughout the eighteenth century until in 1832 its territory was divided between two Vietnamese

districts and, ‘Campā definitively ceased to exist’.181

At some time during the last centuries (there is no evidence of it in epigraphy or

architecture) many Cham, including kings mentioned in their chronicles, adopted Islam, which is

the religion of all but a few small communities of Cham in Cambodia.182 The largest community

of Cham today in Vietnam, however, around Phan Rang, are still not Muslim.

Toward a new history of Champa

First, the history of Linyi must be separated from that of early Champa. Then it must be

realized that there was never a unified Champa, and that events recorded for one part of Champa,

whatever the source, may not be extrapolated to the rest. Students of the subject must also be

aware that the Vietnamese and Chinese histories are not always the best sources and that their

data may not automatically be assimilated to what is in the Champa inscriptions – which, where

they exist, must be treated as the primary sources. The Chinese sources, in particular, require

new study by competent Sinologists. With respect to events, Champa must be seen as consisting,

from the time of the earliest sources, of at least three equally important regions: (1) the North,

consisting of the Thu Bồn valley (Mỹ Sơn, Trà Kiệu, Ðông Dương/Indrapura, and extending at

times into Quảng Trị and Quảng Bình; (2) Nha Trang; and (3) Phan Rang/ Pāõóuraïga. Each of

these regions could be called nagara campa, in the Vietnamese records ‘Chiêm Thành’ (‘Cham

city’). Vijaya, now Quy Nhơn, became important later and was also then a ‘Chiêm Thành’ and

nagara champa.

180 Association Française des Amis de l’Orient, Musée de sculpture, pp. 39-56. 181 Po Dharma, ‘Survol’, pp. 51, 53. See also Po Dharma, ‘Les frontières du Campā (Dernier état des recherches)’, in Les frontières du Vietnam: Histoire des frontières de la péninsule indochinoise, ed. P. B. Lafont (Paris: L’Harmattan, 1989), pp. 128-35. 182 Pierre-Yves Manguin, ‘L’introduction de l’Islam au Campā’, BEFEO, 66 (1979): 255-87.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Champa was founded by Austronesians who arrived by sea and established themselves in

river port areas, and the rise and decline of one or another region depended on the vicissitudes of

the international maritime trade networks from China through what is now northern Vietnam,

along the Champa coasts and on to southern Cambodia, Nusantara and India. The shifts in

predominance of one or another region were neither the result nor the cause of royalty moving

from one center to another. Major shifts in dominance were from the Thu Bồn region of the

North to Pāõóuraïga in mid-eighth century and then, a century later, the reemergence of the North

in another location when the so-called Ðông Dương/Indrapura dynasty appeared as the dominant

entity. A new feature of this new northern dominance was expansion well into Quảng Trị and

Quảng Bình, as the earlier Linyi had done, and one of the important centers was probably there.

In the eleventh century Nha Trang and Pāõóuraïga show increasing importance in the epigraphic

record, while in the twelfth to thirteenth centuries Vijaya in Bình Ðịnh midway between the Thu

Bồn valley and the South may have been the most important economically and strategically.

Detailed study of Champa’s international trade relationships, which could not be

undertaken here, will be of major importance in further work. The importance of maritime

activity for Champa has been recognized but not studied in detail for the pre-fifteenth century

period. K.R. Hall’s treatment is too speculative and not based on solid sources, although he is

quite correct in saying that ‘the authority of a Cham monarch was concentrated within his own

river-mouth plain’; Anthony Reid’s discussion focuses on a later period and the information

about Champa in early European sources. A beginning in the type of study required now is an

article by Momoki Shiro in which he cites the Chinese sources for the products exported from or

traded by Champa, some of which are not native to Champa and thus indicate involvement in

entrepot trading. More detail on such matters may help to understand the development and

decline of the different Champa ports and river-valley hinterlands in accordance with changes in

demand for different products in China.183

Struggles between north Champa and Vietnam began after the latter’s independence in

the late tenth century, but it was not, as conventionally believed, a constant push southward by

the Vietnamese. The first war, and others later, was provoked by the Cham. Real Champa 183 Kenneth R. Hall, Maritime trade and state development in early Southeast Asia (Honolulu: University of Hawaii, 1985), pp. 253-4; Anthony Reid, Changing the shape of early modern Southeast Asia (Chiang Mai: Silkworm, 1999); Momoki Shiro, ‘Chămpa chỉ là một thể chế biển?’ [Was Champa only [a] maritime polity?], Nghiên Cứu Đông Nam Á (Southeast Asian Studies), 4 (37) (1999): 43-8.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

weakness with respect to Đại Việt began only after the 30 years of involvement with Cambodia

in the twelfth to thirteenth centuries, which may have caused more political and economic

damage to Champa than any previous intervention from Vietnam.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Bibliography L’Association française des amis de l’orient (AFAO), 1997 Le Musée de sculpture Ca∝ de Đà Nẵng, Editions de l’AFAO, Paris. Aymonier, Étienne 1880 “Chronique des anciens rois du Cambodge”. Excursions et Reconnaissances, IV, 2 (1880), pp. 149-80. 1891 “Première étude sur les inscriptions tchames”, Journal Asiatique, janvier-février 1981, pp. 5-86. 1901, 1904 Le Cambodge. Vol I: Le royaume actuel. Paris: Ernest Leroux. Editeur, 1900. Vol. II: Les provinces siamoises. Paris: Ernest Leroux, Editeur, 1901. Vol III: Le Groupe d’Angkor et l’histoire. Paris: Ernest Leroux, Editeur, 1904. 1911 “L’inscription came de Po Sah”, Bulletin de la Commission Archéologique de l’Indochne, pp. 13-19. Aymonier, Étienne et Antoine Cabaton 1906 Dictionnaire Čam-Français, Paris, Imprimerie Nationale, Ernest Leroux, Éditeur. Bellwood, Peter 1985 Prehistory of the Indo-Malaysian Archpelago, Orlando, Academic Press. 1992 “Southeast Asia Before History”, The Cambridge History of Southeast Asia Volume I, pp. 53-136. 1993 "Cultural and Biological Differentiation in Peninsular Malaysia: The

Last 10,000 Years", Asian Perspectives Vol. 32, No. 2 (Fall 1993), pp. 37-59.

Bergaigne, Abel 1888 “L’Ancien royaume de Campā dans l’Indochine, d’après les inscriptions”, Extrait du Journal Asiatique, Paris, Imprimerie Nationale. 1893 Inscriptions sanscrites de Campa , Inscriptions sanscrites du Cambodge. Notices et extraits des manuscrits de la Bibliothèque Nationale, Tome 27, (1ere partie), 2e fascicule, Paris, Imprimerie Nationale. Blust, Robert A. 1988 "The Austronesian Homeland: A Linguistic Perspective", Asian Perspectives 26, pp. 45-67. 1992 “The Austronesian Settlement of Mainland Southeast Asia”, in Papers From the Second Annual Meeting of the southeast Asian Linguistics Society 1992, Edited by Karen L. Adams and Thomas John Hudak Arizona Sate University, Program for Southeast Asian Studies, Tempe, Arizona, 1994.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Coedès, George 1918 “Le royaume de Çrī Vijaya", BEFEO 18, 6 (1918), pp. 1-28 1928 “La date du Bayon”, BEFEO 28, pp. 81-103 1929 “Nouvelles données chronologiques et généalogiques sur la dynastie de Mahīdharapura”, BEFEO 29, pp. 297-330, including the inscription of Banteay Chhmar, pp. 309-315. 1932 “Quelques sugestions sur la méthode à suivre pour interpreter les bas-reliefs de Bantãy Chmàr et de la galérie extérieure du Bàyon”, BEFEO 32, pp. 71-81. 1939 “La plus ancienne inscription en langue chame”, in Eastern and Indian Studies in honour of F.W. Thomas, C.I.E., pp. 46, ff. 1942. Inscriptions du Cambodge Vol. 2. École Française d’Extrême-Orient. Collection de textes et documents sur l’Indochine III. Hanoi: Imprimerie d’Extrême-Orient. 1944 Histoire ancienne des états hindouisés d’extrême-orient.

Hanoi. Imprimerie d’extrême-orient. 1964a Les états hindouisés d’Indochine et d’Indonésie. Nouvelle édition revue et mise à jour. Paris: Éditions E. de Boccard. 1964b Inscriptions du Cambodge Vol. 7. École Française d’Extrême-Orient. Collection de textes et documents sur l’Indochine III. Paris: École Française d’Extrême-Orient. 1966 Inscriptions du Cambodge Vol. 8. Paris, École Française d’Extrême-Orient. 1968 The Indianized States of Southeast Asia, edited by Walter F. Vella, translated by Sue Brown Cowing. Honolulu, University of Hawaii Press. Coedès, George et Pierre Dupont 1943-46 "Les stèles de Sdòk Ka k Tho .m, Phno .m Sandak et Prá .h Vihãr.” BEFEO 43:56-154. Coedès, George et Henri Parmentier 1923 Listes générales des inscriptions et des monuments du Champa et du Cambodge, Hanoi, Ecole Française d’Extrême-Orient, 1923. Copenhagen 1988 Actes du séminaire sur le Campā, organisé à l’Université de Copenhague le 23 mai 1987, Paris, Travaux du centre d’histoire et civilisations de la péninsule indochinoise. 1988 Damais, L[ouis]-C[harles] 1957 Review of Prof. Dr. Poerbatjaraka. Riwajat Indonesia, djilid 1 (Histoire de l’Indonésie, vol. 1). Djakarta. Jajasan Pembangunan. Bulletin de L’École Française d’Extrême- Orient 48 (1957), pt. 2, pp. 607-49.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Demi ville, Paul 1951 “R.A. Stein, ‘Le Lin-yi, sa localisation, sa contribution à la formation du Champa et ses liens avec la Chine’”, T’oung Pao 40:336-351. Diffloth, Gérard 1992 “Vietnamese as a Mon-Khmer Language”, in Papers from the 1st Annual Meeting of the South to East Asisan Linguistivcs Society. M. Ratliff and E. Schiller, eds., Tucson, Arizona State University Press, pp. 125-139. 1995 Paper presented in Hanoi, cited in Nguyên Hu’u Hoành, Tiêng Katu, 21-22, 219. ÉÉPC 1995 Études Épigraphiques sur le Pays Cham, réunies par Claude Jacques. Paris: Presses de l’École Française d’Extrême- Orient, Réimpression No 7 Filliozat, Jean 1969 ‘L’inscription dite de ‘Vỏ [sic! Võ]-Cạnh’”, Bulletin de L’École Française d’Extrême-Orient 55 (1969), pp. 107-116. Finot, Louis 1902 “Deux nouvelles inscriptions de Bhadravarman Ier”, Bulletin de L’École Française d’Extrême-Orient II, pp. 185-191.

1903 “Stèle de Śambhuvarman à Mi-so’n”, Bulletin de L’École Française d’Extrême-Orient III (1903), pp. 206-211

1904a “Notes d’épigraphie VI: Inscriptions de Quảng Nam”, Bulletin de L’École Française d’Extrême-Orient IV (1904), pp. 83-115 1904b “Notes d’épigraphie XI: Les inscriptions de Mi-so’n”, Bulletin de L’École Française d’Extrême-Orient IV (1904), pp. 897-977 1915 "Les inscriptions du Musée de Hanoi, BEFEO XV (1915), pp. 1-19 1915 b ‘Les inscriptions du Jaya Parameśvaravarman I roi du Champa’,

Bulletin de L’École Française d’Extrême-Orient, 15/2, pp. 39-43. 1915c ‘Errata et Addenda’, Bulletin de L’École Française d’Extrême-Orient,

15/2, pp. 185-192. 1928 Review of Maspéro, Le royaume de Champa, BEFEO XXVIII, pp. 285-292. Reprinted in Maspéro 1928. Glover, Ian 1997 "The Excavations of J.-Y. Claeys at Trà Kiệu , Central Vietnam, 1927- 28: from unpublished archives of the EFEO, Paris, and records in the possession of the Claeys family", Journal of the Siam Society, Volume 85, Parts 1&2, pp. 173-186.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Groslier, Bernard Philippe 1973 Inscriptions du Bayon, in Jacques Dumarçay, Le Bayon histoire architecturale du temple, Publications de l’Ecole Française d’Extrême- Orient, Mémoires Archéologiques III-2, Paris, l’Ecole Française d’Extrême-Orient, 1973, pp. 83-306. Hall, Kenneth R. 1985 Maritime Trade and State Development in Early Southeast Asia. Honolulu: University of Hawaii Press, 1985 Huber, Edouard 1905 "Le clan de l’aréquier", Bulletin de l’École Française d’Extrême-Orient V,1905, pp. 170-5. 1911 “L’épigraphie de la dynastie de Đồng Dương”, BEFEO 11, pp. 268- 311. Jacques, Claude 1969 “Notes sur la stèle de ‘Vỏ [sic! Võ]-Cạnh’”, Bulletin de L’École Française d’Extrême-Orient 55 (1969), pp. 117-124. 1995 Études Épigraphiques sur le Pays Cham, réunies par Claude Jacques. Paris: Presses de l’École Française d’Extrême- Orient, Réimpression No 7. 1996 "Le Bayon et l’épigraphie", in First Symposium on the Bayon, Final Report – Volume I, pp. 78-86. Lafont, Pierre-Bernard 1979 “Études Cam III. Pour une réhabilitation des chroniques rédigées en Ca∝ moderne”, BEFEO 1979, pp. 105-111. 1988 “Avant Propos”, Actes du séminaire sur le Campā, organisé à l’Université de Copenhague le 23 mai 1987, Paris, Travaux du centre d’histoire et civilisations de la péninsule Li Tana 1998 Nguyên Cochinchina, Southern Vietnam in the Seventeenth and Eighteenth Centuries, Ithaca: Cornell Southeast Asia Program Publications. Pierre-Yves Manguin 1979 "L’Introduction de l’Islam au Campā", BEFEO, LXVI, 1979, pp. 255- 287. Maspéro, Georges 1928 [1988] Le royaume de Champa, réimpression de l’École Française d’Extrême- Orient, Paris. Moussay, Gerard 1971 Dictionnaire Căm-Vietnamien-Français, Phanrang, Centre Culturel Căm.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Népote, Jacques 1993 (1-2) “Champa, propositions pour une histoire de temps long”, Péninsule, Nouvelle série, 26-27. Pelliot, Paul 1903 “Le Fou-Nan”, Bulletin de l’École Française d’Extrême-Orient III, 1903, Hanoi, pp. 248-303. 1904 “Deux itinéraires de Chine en Inde à la fin du VIIIe siècle”, Bulletin de l’École Française d’Extrême-Orient IV, Hanoi, pp. 131-385. Po Dharma 1987 (I-II) Le P ⎞∂ura≡ga (Camp ) 1802-1835, tomes I-II. Paris, Publications de l’École Française d’Extrême-Orient. 1988 "Etat des dernières recherches sur la date de l’absorption du Campā par le Vietnam", Actes du séminaire sur le Campā, organisé à l’Université de Copenhague le 23 mai 1987, Paris, Travaux du centre d’histoire et civilisations de la péninsule indochinoise, pp. 58-69. 1999 Quatre lexiques malais-can anciens rédigés au Campā, Presses de l’École française d’Extrême-Orient, Paris. Reid, Anthony

1999 Changing the Shape of Early Modern Southeast Asia, Chiang Mai, Silkworm Books 1999 Roveda, Vittorio 2000 Khmer Mythology, Secrets of Angkor. Bangkok, River Books. Schweyer, Anne-Valérie 1998 “Le dynastie d’Indrapura (Quảng Nam, Viet Nam)”, Southeast Asian Archaeology 1998, Proceedings of the 7th International Conference of the Europezan Association of Southeast Asian Archaeologists, Berlin, 31 August – 4 September 1998, Wibke Lobo and Stefanie Reimann Editors, Centre for South to East Asian Studies, University of Hull, Special Issue & Ethnologisches Museum, Staatliche Museen zu Berlin, Stiftung Preussischer Kulturbesitz, pp. 205-217. 1999a “Chronologie des inscriptions publiées de Camp , Études d’épigraphie cam-1”, Bulletin de l’École Française d’Extrême-Orient 86, pp. 321-344. 1999b "La vaisselle en argent de la dynastie d’Indrapura (Quảng Nam, Việt Nam, Études d’épigraphie cam - II", Bulletin de l’École Française d’Extrême-Orient 86, pp. 345-355. 2004 ‘Po Nagar de Nha Trang’, Aséanie, 14, pp. 109-40. 2005 ‘Po Nagar de Nha Trang’, deuxième partie, Aséanie, 15, forthcoming.


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Southworth, William Aelred 2001 “The Origins of Campā in Central Vietnam, A Preliminary Review”, Ph.D. thesis, Archaeology, SOAS, University of London, 2001 Stern, Philippe 1927 Le Bayon d’Angkor Thom et l’évolution de l’art khmer. Paris: Annales du Musée Guimet, Bibliothèque de vulgarisation, no. 47. Librairie orientale Paul Geuthner. 1942 Philippe Stern, L’Art du Champa (Ancien Annam) et son Evolution, Toulouse. Stein, Rolf 1947 Le Lin-Yi, sa localisation, sa contribution à la formation du Champa et ses liens avec la Chine. In Han-Hiue, Bulletin du Taylor, Keith W. 1983 The Birth of Vietnam, Berkeley, University of California Press. 1998 “Surface Orientations in Vietnam: Beyond Histories of Nation and Region”, Journal of Asian Studies, vol. 57, no. 4, 1998, pp. 949–978. Thurgood, Graham 1999 From Ancient Cham to Modern Dialects, Two Thousand Years of Language Contact and Change. Honolulu, University of Hawaii. Vickery, Michael 1998 Society, Economics and Politics in Pre-Angkor Cambodia: The 7th to eighth Centuries. Tokyo.The Centre for East Asian Cultural Studies for Unesco, The Toyo Bunko. 2003 “Funan Reviewed: Deconstructing the Ancients”, publication pending in BEFEO. 2004a “Two Historical Records of the Kingdom of Vientiane”, in Christopher E. Goscha and Sören Ivarsson (eds.), Contesting Visions of the Lao Past Lao Historiography at the Crossroads, Richmond: Curzon Press. 2004b Cambodia and its Neighbors in the 15th Century, Asia Research Institute Working Paper Series No. 27, Singapore. www.ari.nus.edu.sg/pub/wps.htm Wade, Geoff

2005 Champa in the Song hui-yao: A Draft Translation, Asia Research Institute Online Working Paper Series, No. 53, Singapore.

http://www.ari.nus.edu.sg/docs/wps/wps05_053.pdf * * *


ARI Working Paper No. 37 Asia Research Institute ● Singapore

Vietnamese histories. Cm Khâm Ðịnh Việt Sử Tông Giám Cương Mục Tt Ðại Việt Sử Ký Toàn Thư Vsl Việt Sử Lược


WPS 37 Champa Revised - [PDF Document] (2024)


What is the purpose of WPS in welding? ›

A Welding Procedure Specification, or WPS, is a formal written document describing standard welding procedures. Like a recipe for welders, these procedures provide clear direction for your welders to make quality products that are up to code and industry standards.

What is the welding procedure specification? ›

The Welding Procedure Specification (WPS) is a document that outlines the specific welding procedures that are to be used for a particular welding job. The Procedure Qualification Record (PQR) is a document that certifies that a particular welding procedure has been qualified to meet the required codes and standards.

Does a WPS need a PQR? ›

A WPS and PQR are required documents in the fabrication process when a weld must comply with a certain standard. In simple terms, a WPS is a set of welding instructions and the PQR is a checklist to ensure those instructions were satisfactorily met. The PQR document is created first and is needed to create the WPS.

What are the two types of WPS? ›

WPS is of two types- Prequalified WPS(pWPS) & qualified WPS.

Who can prepare WPS? ›

A welding engineer or any knowledgeable and competent person in the welding metallurgy with appropriate experience can write a WPS.

What is G in welding? ›

Welds with a 1 are flat position, 2 is horizontal, 3 is vertical and 4 is overhead. F stands for fillet weld, while G is a groove weld.

What are the purposes of WPS? ›

The WPS requires employers to obtain written consent from employees before making any deductions from their salaries, such as for loans or other expenses. This protects employees from being exploited by employers who may take advantage of their financial vulnerability.

What is the function of WPS? ›

Wi-Fi® Protected Setup (WPS) is a built-in feature of many routers that make it easier to connect Wi-Fi enabled devices to a secure wireless network. This information is provided to help connect your TV, Blu-ray Disc™ player, or other supported home video products to a wireless network using WPS.

Why would you use WPS? ›

WPS stands for WiFi Protected Setup. It is a wireless network security standard that allows connections between a router and wireless devices, such as security cameras, faster and easier.

What is the importance of WPS? ›

If you are in the Welding business you clearly know that a Welding Procedure Specification (WPS) is a document to guide the effective creation of a weld that meets all applicable code requirements and production standards. The WPS contains details that are necessary to create the desired weld.

Top Articles
Latest Posts
Article information

Author: Ray Christiansen

Last Updated:

Views: 5323

Rating: 4.9 / 5 (49 voted)

Reviews: 80% of readers found this page helpful

Author information

Name: Ray Christiansen

Birthday: 1998-05-04

Address: Apt. 814 34339 Sauer Islands, Hirtheville, GA 02446-8771

Phone: +337636892828

Job: Lead Hospitality Designer

Hobby: Urban exploration, Tai chi, Lockpicking, Fashion, Gunsmithing, Pottery, Geocaching

Introduction: My name is Ray Christiansen, I am a fair, good, cute, gentle, vast, glamorous, excited person who loves writing and wants to share my knowledge and understanding with you.