Maritime Links Between China, Sumatra, the Malay Peninsula, and Buddhist Monasteries in India (c. 11th–12th Centuries) in the Light of Two Fragmentary Inscribed Strips of Copper from Muara Jambi

Abstract

This article explores the maritime connections relating to Buddhism and diplomacy between polities in Sumatra, the Malay Peninsula, the Indian subcontinent, and China from the beginning of the 11th century up to the 12th century CE. It focuses on new epigraphic evidence from Muara Jambi in the form of two inscribed strips of copper mentioning the Cūḍāmaṇivarmavihāra, a monastery funded by the king of Śrīvijaya in Nagapattinam (South India), and the Bālādityavihāra, probably located in Nālandā (Northeastern India). These new findings are compared to archaeological and textual materials from elsewhere in the Buddhist world that cast light on the web of transregional connections between Nusantara, China, and India in the early centuries of the second millennium.

1. Introduction

Maritime connections relating to Buddhism—or, more generally, “temple diplomacy”1—between polities in Sumatra and the Malay Peninsula, South India, and China from the beginning of the 11th century up to the 12th century CE have been discussed by scholars since the early 20th century, mainly on the basis of epigraphic and textual sources in Sanskrit, Tamil, Tibetan, Chinese, and Arabic. Given the dearth of primary sources from the Buddhist site of Muara Jambi (Sumatra, Indonesia) and the Malay Peninsula in the period under discussion,2 any new local sources are welcome, all the more so if they mention the names of persons or institutions known so far only from non-local sources. This article focuses on one such new source, namely a pair of fragmentary inscriptions from Muara Jambi, which are here translated for the first time and interpreted as two complementary fragments of distinct copies of a single text. Our discussion presents this new source in the light of contemporary ones from other parts of the Buddhist world with which Muara Jambi had close connections.

The two inscription fragments take the form of small strips of copper and were reportedly recovered from the bed of the Batanghari River near the Muara Jambi temple complex. They are now kept, respectively, at the Culture Conservation Bureau3 and at Yayasan Rumah Menapo,4 both in the provincial capital of Jambi. Both artifacts have been cursorily described and photographically reproduced for the first time in a catalogue dealing mainly with inscribed tin foils from Sumatra by Tejowasono et al. (2019, pp. 13, 49), while the lines of the Old Malay text engraved on them have been transliterated in a publication on inscriptions from the Śrīvijaya polity (Andhifani and Tedjowasono 2021).5 If these lines of text are confirmed to be genuine, then they will constitute a new piece of historical evidence on the late-10th–early 11th-century Śailendra King Cūḍāmaṇivarman, who is known from inscriptions from the Cōḻa domains in South India mentioning him as father of King Māravijayottuṅgavarman (as well as in Tibetan and Chinese sources), and two Buddhist monasteries, which are situated, respectively, at Nagapattinam (Nākappaṭṭiṉam) in South India and Nālandā in Northeast India, namely the eponymous Cūḍāmaṇivarmavihāra and the Bālādityavihāra, named after Gupta King Bālāditya.

Since past historiography has tended to associate Cūḍāmaṇivarman with either Kedah in the Malay Peninsula (present-day Malaysia) or Palembang (present-day Indonesia), him being mentioned in source material recovered from Jambi not only adds to the scant epigraphical evidence of this king, but it also throws new light on the historical and geographical context of the polity in those areas (which have been labeled as “Śrīvijaya” by previous scholarship as well as non-indigenous sources, though we still lack any local source using that name in this period) and its links with South India in the framework of temple diplomacy. This complements the evidence available so far of (Buddhist) transregional diplomatic networks linking South India, polities in the Malay Peninsula and Sumatra, and China. However, since the new source material was not derived from scientific excavations and there is no information other than hearsay about the circumstances of its recovery, concerns about authenticity must be raised. This topic is revisited in the conclusion, where relevant parallels with pointillistic cold-working and the possibility of forgery are explored further. On the basis of the palaeographic and linguistic features of the two inscription fragments, as well as of our visual inspection of the artifacts, we conclude that the documents must be authentic.

2. Description of the Artifacts

On 10 and 11 June 2024, a team constituted by Andrea Acri, Hedwi Prihatmoko, Mathilde Mechling, Gregory Sattler, and Wahyu Rizky Andhifani visually inspected the two inscription fragments (Figure 1 and Figure 2), kept respectively at the BPK V (A) and Yayasan Rumah Menopo (B) in Jambi, and took full photographic documentation. In the following subsections we describe the results of our examination of the physical objects bearing the inscriptions.6

2.1. Strip Kept at BPK V Jambi

The first strip (A) is catalogued as “010, 01/BPCBJB/Cu/VIII/2019” in Tejowasono et al. (2019, p. 13). It is reproduced in recto and verso and described as being written in Old Malay language and Old Sumatran script, and has having been recovered from the Batanghari River. No classification of its contents is given the catalogue, and no portions of the text are transcribed. No information on the archaeological context and circumstances of its discovery are given either, probably because it was not recovered during an official archaeological excavation but was purchased from a private individual. The artifact, most likely made of fairly pure copper, as is the second artifact (B) presented below,7 is a rectangular, narrow-shape strip (measuring 10.8 cm in length × 2.3 cm in height × an estimated thickness of 0.1 cm)8 with its left end narrower and pierced with a round hole. This may have served for attaching the strip to another artifact. The right end of the strip shows traces of tear and has visibly been cut off. Given the thinness of the strip, this could have been achieved by repeatedly folding the artifact in half one way and then the other, possibly after its discovery, as there would have been no reason for the strip to be folded during its time of use.9 The inscription it bears is incomplete due to the loss of a part of the support. While the length of the original artifact is unknown, it seems likely that it would have been roughly twice its current dimensions.

The recto of the strip displays two lines of engraved text, the first one of which opens with a spiral symbol (Figure 1, top). Instead of forming a continuous engraved line, as is most common for inscriptions on metal in Southeast Asia, the characters (akṣaras) have been traced by small dots. Upon closer inspection with a magnifying glass, the “dots” were observed to be V-shaped. The “V” shape of the tool marks corresponds to the shape of a burin, which left slightly raised edges along the engraved marks (Figure 3). The deep and sharp V-shaped tool marks suggest cold working in the metal after the manufacturing of the strip. The engraved lines are straight, with regular intervals between the characters (Figure 4). The tool marks have a consistent shape and depth.

The single line engraved on the verso contrasts remarkably with the two lines on the recto. The tool marks are irregular, the characters are not regularly spaced, and the inscribed line is not straight (Figure 5). Some tool marks are very deep, while others are not (Figure 5 and Figure 6). The surface on this face is not smooth, presenting a great number of small marks and scratches. The central part displays ripples—a feature that none of our photos captures clearly—which might suggest that the artifact was cast, perhaps in an open, flat mold (the ripples could be traces of the liquid metal as it cooled down and solidified). However, it is also interesting that the recto surface is very smooth, with rounded edges (Figure 3 and Figure 4), while the verso surface is rougher, and the edges are raised (Figure 5). The verso also displays a line incised along the edge, but it does not seem to have been engraved like the characters. The curled edge and the line running parallel to it are characteristics observed in hammered gold sheets cut to the desired shape with scissors or fine shears (Craddock 2009, p. 375). Our copper strip was probably cut to the desired shape using a similar technique, perhaps after hammering the copper into a thin sheet.

2.2. Strip Kept at Yayasan Rumah Menapo, Jambi

The second strip (B) is catalogued as “045; 02/PADMA/Cu/VIII/2019” in Tejowasono et al. (2019, p. 49). It is reproduced in recto and verso and described as being made of copper,10 written in Old Malay language and Old Sumatran script, and recovered from the Batanghari River. The measurements of this artifact are in the same order of magnitude as the first (11 cm in length × 2.2 cm in height × an estimated thickness of 1 mm). The content is described as “mantra”, and the following transcription is given of the first few and last characters:

… nmu ga sakṣa da mcu oṁ ra …

… ddhan m kara (ṅa) la …

As with the first strip, no record about the archaeological context of its finding exists for this one either. However, while in the village of Muara Jambi, we were able to speak to the person—his identity remained undisclosed—who claims (accidentally) to have recovered this inscription from the bed of the Batanghari River, right in front of the main settlement, while extracting sand. According to him, it was found in an earthenware jar with various objects in tin and copper, including ancient coins. This is in harmony with the green corrosion marks described below, which could be the result of corrosion due to contact between the strip and such objects.

The recto of the strip displays two lines of text, whose characters have been traced by means of small, engraved dots (Figure 2). These dots are V-shaped and have visibly been made by means of a burin, as is also the case for strip A. Both inscribed lines on this face are irregular. The characters are not set with regular spaces in between each (Figure 7). The tool marks do not appear to be regular either, but it must be noted that observations on this artifact were more difficult for several reasons. The surface displays deep horizontal marks, which suggest that the strip has undergone treatment of corrosion with an abrasive tool. Furthermore, dust remains encrusted in the tool marks, and chalk powder had previously been rubbed on the surface to facilitate reading of the inscription. We removed the chalk with a soft, non-abrasive cotton cloth, as well as by blowing on the surface, but we could not remove all chalk and dust with this method. Although this prevents a precise examination of the engraving, our general impression is one of less regularity than the recto of inscription A (Figure 1).

On the verso, there is one line of text that ends with a spiral symbol, probably marking the end of the text in the same way that this symbol marks the opening of the text on fragment A (Figure 8). However, near the broken lower left side there are also three dots/tool marks, which could suggest the presence of a second line of text in the missing portion of the strip. The surface on the verso is not smooth and shows slightly regular bumps that could suggest that this artifact was shaped through the hammering technique. This remains uncertain, as the surface is patinated, shows spots of green corrosion (likely where the artifact touched another metal object), and shows scratches resulting from cleaning with an abrasive tool. However, similarly to fragment A, but this time on the recto, fragment B displays curled edges and a line running parallel to the edges, indicating that it was cut to shape with scissors or shears, likely after being hammered (Figure 2, recto). The left edge is also noteworthy, as it shows traces of tear where the artifact has broken off, perhaps by being folded, while the lower left corner is cleanly cut off at an angle, in a straight line, as if cut with a sharp tool.

3. Text and Provisional Translation

Below, we offer a diplomatic transliteration of the text engraved on the two strips, with the one facing the other. This transliteration is virtually identical to the one published by Wahyu Rizki Andhifani and Hedwi Prihatmoko in Andhifani and Tedjowasono (2021, pp. 64–66), except for the division of a few words. The + marks correspond to the spiral-shaped opening/closing symbols found at the start of the text on A and at its end on B.

A	B
Ar1. + kathora kāṅśa kaṇḍa 1 | barisi valānda … Ar2. ndenan· barkna pamūja ka bharāla śrībā...	Br1. …(lā)nda tāsak· damcu khaṇḍa 5 brātña kā 5 su 1 Br2. … (śrī)bālādityabihāra | dṅan· ka bharāla
Av1. …lendracūḍāmaṇivarmmabihāra …	Bv1. … (bi)hāra +

The parallel arrangement above allows us to note that the last words preserved on each of the three lines on strip A can be matched with the initial words on the corresponding lines of strip B, so much so that words preserved incompletely on the one can be confidently reconstructed by means of the other, as follows:

valānda—[valā]nda

śrī-bā[lāditya]—śrī-bālāditya

bihāra—(bi)hāra

The two strips seem to have slightly different dimensions and bear slightly overlapping texts. In combination with the technical differences observed above, these observations prove that we are not dealing with two halves of one original strip. Initially, we suspected that either of the two strips could be a forgery that was created to match a genuine one. However, the most plausible interpretation seems to be that we are dealing respectively with authentic artifacts, one being the left and the other the right half of two distinct copper strips of similar size that bear a similar text. While it would be possible to imagine a scenario whereby the two strips bore related but not identical texts, nothing stands in the way of assuming that both strips originally contained exactly the same text. Although it is admittedly a surprising coincidence that two complementary halves of distinct strips engraved with an identical text would have been recovered in entirely uncontrolled archaeological circumstances, this is the hypothesis that underlies the reconstructed textual layout for both copies of the text that we now present in the Noto Sans Kawi font that closely resembles the original script:

The characters printed in red color are entirely lost only in the respective copy, and those printed in blue are lost in both copies, meaning that we assume the two parts not to be fully complementary, because we need to restore two akṣaras at the line transitions Ar1/Ar2, Ar2/Av1, Br1/Br2, and Br2/Bv1. These restored characters are represented in square brackets in the conventional epigraphic edition of the combined texts that now follows, along with a translation, while still leaving some problematic and untranslatable terms in italics.

Reconstruction

kathora kāṅśa kaṇḍa 1 | barisi valānda tāsak· damcu khaṇḍa 5 brātña kā 5 su 1 [valā]ndenan· barkna pamūja ka bharāla śrī-bālāditya-bihāra | dṅan· ka bharāla [śrī-śai]lendra-cūḍāmaṇivarmma-bihāra.

Provisional Translation

1 kathora of kaṅśa, filled with 5 valanda tāsak damcu, weighing 5 kāṭi, 1 suvarṇa. These valanda are for worship to the Lord of the Śrī-Bālāditya monastery and to the Lord of the Śrī-Śailendra-Cūḍāmaṇivarman monastery.

4. Linguistic and Philological Commentary

Even a cursory glance at our edition of the text is sufficient to confirm the provisional identification of the language by Ninny Susanti Tejowasono et al. (2019) as Old Malay (OM). Affixed forms like bar-isi and brāt-ña are distinctive among the epigraphically attested early languages of the Indonesian archipelago. On the corpus of OM documents known so far (consisting of about 35 inscriptions and one manuscript), we refer to Griffiths (2018); for further discussion, as well as a map showing the distribution of the other OM inscriptions known so far, we refer to Clavé and Griffiths (2022). For studies of the linguistic features of the forms of Malay found in this small corpus, we refer to the same publications, as well as Mahdi (2005, 2015).

OM documents from diverse parts of the Malay world, covering a period from the 7th through the 14th centuries, reveal an orthographic pattern whereby vowels spelled long are assumed to reflect stress on penultimate vowels in pronunciation (because vowel length as such was not a phonemic feature of the language). This is observed in the present text in the immediately transparent word brāt-ña (Modern Malay berat-nya) “its/their weight” and in the more challenging keyword valānda. The latter figures once before the demonstrative inan to which it is joined in sandhi so that it appears as valāndenan.11 An instance of sandhi that also touches on word formation is the form tāsak, which is not immediately recognizable from the perspective of Classical or Modern Malay (C/MM), that we interpret as ta-asak with a variant of the coincidental tar- prefix that normally corresponds to C/MM ter-.12

After those general remarks on spelling and grammar, we turn to a discussion of the new lexical material roughly in the order of the words’ or word combinations’ appearance in the text.

kathora kāṅśa—The second word, the precursor of both gangsa and kangsa in C/MM and of Indian origin,13 is also attested in the Tanjung Tanah manuscript where we read (p. 22, lines 4–5): jaka bahu:taṁ mas· pirak· riti ra:ncuṁ kaṁśa tambaga. Kozok (2015, p. 77) translates the whole passage as “if one owes a debt of gold, silver, brass, rancung, bronze, copper”, interpreting kaṁśa to mean bronze. By contrast, Mahdi (2015, p. 203) translates the list as “gold, silver, bronze, brass, copper”, taking riti-rancung as a single expression meaning bronze and assuming kaṁśa to mean brass. The fact that this list of terms is translated differently on different pages of the same book dedicated to the manuscript illustrates the uncertainty inherent in interpreting premodern terms designating metals and alloys. The problem of which Malay term corresponds to which copper-based alloy has been analyzed in Manguin (1976, pp. 251–53), and he considers that C/MM kangsa means “bronze”.14 This has been observed likewise for the use of kaṁsa in Cambodian inscriptions of the same general period as our inscription.15 Although dictionaries of Malay tend to gloss gangsa and kangsa as “bell-metal”, we interpret kāṅśa in our inscription to mean “bronze”.

The word kathora, though apparently likewise of Indian origin,16 is not known to us in this precise form in any Indian source, but it is found in an Old Khmer inscription of 924 Śaka (1002/3 CE) to denote an object of value offered to a deity, though little more can be inferred from the passage.17 The corresponding entry in Jenner (2009) cites related words in Modern Khmer and in Thai—where the meaning is “spittoon, chamber pot, bedpan”—and glosses the Old Khmer word as “A vessel of unidentified form and function, perhaps a cuspidor”. A word that expresses an apparently similar meaning and has the closely similar form kaṇṭora is found in only one Old Javanese (OJ) text, namely the Tantu Paṅgəlaran, on which basis the corresponding entry in Zoetmulder (1982) is formulated as “(Skt kaṭora) a kind of cup or vessel”, although the textual evidence here suggests a vessel significantly larger than a cup.18 A rare epigraphic occurrence of the Sanskrit word cited by Zoetmulder, or more precisely the diminutive form kaṭorikā, is found in a 6th-century inscription from East Bengal, in a list of utensils donated to a community of ascetics; the context allows the translation “small cup”.19 In our context, the meaning of kathora seems to be related to the meaning of kaṇṭora in the Tantu Paṅgəlaran; whatever the exact type of container designated by kathora, it must have been large enough to contain five smaller containers and, if our interpretation of damcu is correct, nearly 4 kg of cinnabar.

Although other scenarios cannot be excluded, we are inclined to suppose that the context is one of transshipment of goods from China to India. Metalwares were commonly exported from China to the Indian Ocean region beginning as late as the 9th century and continuing until recent times; thus, the appearance of such a sizable bronze vessel in our text corresponds with much of the textual and material record. Such items were perhaps the most valuable type of commodity on board ships departing from China beginning in the 11th century.20 In terms of large metal vessels, we find examples in the form of cauldrons recovered from several shipwrecks. For instance, a variety of cast iron cauldrons were recovered from the early 9th-century Belitung wreck, as well as two copper alloy handles that were once riveted to a large cauldron (Flecker 2017, pp. 30–31). Thirty-two cauldron handles were also recovered from the early to mid 10th-century Intan wreck, with some of those vessels believed to have been made of copper alloy (Flecker 2002, p. 98). Iron cauldrons were also discovered at the sites of the late 10th-century Cirebon wreck (Liebner 2014, p. 206) and the 12th–13th-century Java Sea wreck (Flecker 2002, pp. 26, 87, 98, 138), while the export of such items to Java and the Philippines is likewise attested to in Song-period texts (Wheatley 1959, p. 117). Most or all of these items found in shipwrecks would appear to have been Chinese manufactures. Perhaps the best example of a large bronze vessel would be a fully preserved bronze cauldron that Claudine Salmon describes as “a straight-handled deep belly cauldron that was recovered in 1994 near Candi Kedaton (in Muara Jambi)” (Salmon 2003, p. 94). Scholars believe that this item was likewise imported from China. These examples suggest to us that the kathora in question was a similar commodity of high value that additionally functioned as a storage container as it was shipped overseas.

khaṇḍa—Our text uses first kaṇḍa, then khaṇḍa, as a numeral classifier, a usage attested neither in C/MM nor in any OM document. The latter spelling seems to be the original one, allowing us to link this usage to Sanskrit khaṇḍa “portion”. The same tatsama word is used as a numeral classifier in the new Indo-Aryan language Oriya (Neukom and Patnaik 2003, sct. 3.4.3), while the corresponding tadbhava form khān is used in Bengali (Rácová 2007). On the use of vatu “stone” as a numeral classifier in an OM inscription from Java, see Griffiths (2018, p. 279). On the use of numeral classifiers in the Tanjung Tanah manuscript, see Mahdi (2015, pp. 207–8).

valānda—This term, with a long ā on the penultimate syllable in accordance with the OM spelling habits outlined above, is found in the exact same spelling also in the Laguna copper-plate inscription from Luzon, the Philippines (see below), and corresponds to valanda or valaṇḍa in some sources written in OJ, both from Java itself and from Bali. All these terms may be antecedents of belanja in C/MM and blanja Modern Javanese (see our n. 22 for the meanings of these words), although the correspondence of /d/ in OM and OJ to /j/ in C/MM and Modern Javanese is linguistically irregular and, hence, requires an ad hoc explanation. In a recent publication (Clavé and Griffiths 2022, p. 193), one of the present authors argued against this idea, as it had been proposed by Antoon Postma (1991, p. 170).21 The main argument against it was the theory that belanja is a loanword from Sinhalese-Pali valañja,22 because “there would be no reason for the OM loan to show nd if valānda too were based on the same Sinhalese-Pali word with its ñj, i.e., /nʤ/”. The connection with C/MM belanja thus having been rejected, it was argued that valānda “must be analyzed as vala + nda, with typical OM marking of vowel length on the penultimate vowel”. Arlo Griffiths now wishes to withdraw this argument because it failed to give account of the complexity inherent in the argument relating C/MM belanja to a Sinhalese-Pali term and of the occurrence of valaṇḍa/valanda in OJ sources.

As for the hypothesis connecting Malay belanja to a Sinhalese/Pali term, the complexity is manifold. First, the early (epigraphic) Sinhalese form of the word in fact has an /nd/ cluster, and there is reason to believe that the Pali word valañja with an /nj/ cluster is historically derived from the vernacular word with /nd/.23 Second, the semantics are also complicated in the Sinhalese data, with the Modern Sinhalese $vala\breve{n}da$ (singular direct case form of valan, which is simultaneously stem and plural form) not only meaning “sign, mark” but also “earthenware vessel, pot” and the verb $vala\breve{n}danav$ā “to use, to eat, to enjoy” apparently being derived from the latter sense of the noun.24 Geiger (1941, p. 158, nos. 2356 and 2357) mentions both noun and verb, though without citing the meaning “vessel, pot” for the former, and indicates that the historical spelling, attested in inscriptions as old as the 10th century, was with ḷ instead of l.25 In the Dictionary of Sinhala Epigraphical Words (Ranawella 2007), one finds vaḷan to mean “pot, vessel” and numerous related entries all with /nd/ and meanings derived from the notions of “enjoying, using”. In short, whether or not Malay belanja is ultimately connected with them, these Sinhalese data are potentially relevant to the interpretation of valānda in OM and valanda/valaṇḍa in OJ. Let us now first present the OJ data.

In redacting the entry for his Old Javanese–English Dictionary, Zoetmulder (1982) was aware only of literary, and not of any epigraphic, data, and he was unable to draw a clear conclusion as to the word’s basic meaning; however, for the two derived forms amalandani and pamalandan he proposed the meanings “to place or keep st. in a container” and “that in which st. (valuable) is kept or offered?” These forms are exceedingly rare, with each being citable only from a single text, but the texts in question are relatively old (ca. 12th c. CE), and the contexts leave little doubt about the involvement of some kind of container, though not necessarily a “box”, as published translations presume.26

Turning to epigraphic evidence, there is only one occurrence in an inscription from Java proper, viz. in the Telang I charter (825 Śaka, 904 CE), line 1v13: vineh pirak dhā 1 vḍihan yu 1, valandaniran ulih pirak mā 8 sovaṁ “they were given 1 dhāraṇa of silver, 1 pair of vḍihan (cloth); their valanda when going home was 8 māṣa of silver per person”. The word valanda here could mean “payment, remuneration”, although we cannot exclude that it designates a container for valuables. There are several occurrences in OJ inscriptions from Bali, dating to the 11th and 12th centuries, where the word valanda also appears in the insignificantly variant spelling valaṇḍa.27 In the Campetan charter (no. 556b, 1071 Śaka, 1149 CE), line 3r2, we read: maśraha ya valaṇḍa mā 4 saput thāyu i hyaṅ bapa “they should deliver 4 māṣas of valaṇḍa each to God Bapa”. In this context, “payment” or “tribute” might be a suitable meaning, though again we cannot exclude a kind of container. Otherwise, we have in the epigraphic corpus of Bali three inscriptions containing one or multiple instances of the phrase taṅgapana riṅ pirak lumaku valanda or its equivalent taṅgapana riṅ pirak lumari valanda with lumari taking the place of lumaku. There are also some cases where the words manahurānusuna “will repay [and] pile up” or tahilakna “will pay as tribute” take the place of lumaku/lumari, and there is a number of cases where valantən—apparently a krama-like derivation, cf. Griffiths (2020, pp. 124–25)—takes the place of valanda/valaṇḍa.28 Very few specialists of Balinese epigraphy have so far attempted to translate any of these variant phrases.29 For our part, we have the impression that the one with taṅgapana means something like “they should be hired with silver currency (if lumaku/lumari, literally ‘moving’, can have this meaning) as valanda”.

In all these epigraphic passages, the earliest dating to 904 CE, it is evident that valanda/valaṇḍa (or valantən) appears in contexts where payments are being made and values are expressed with the weight measure māṣa (abbreviated mā), which represented one-sixteenth of the unit suvarṇa (abbr. su) that we see in our text as well as in the OM Laguna copperplate inscription, dating to 900 CE, line 5: anu diparlappas· hutaṁda valānda kā 1 su 8.30 In this context, valānda was interpreted by Clavé and Griffiths (2022) as related to the Sanskrit word bala “(military) force)”, so that the phrase in question was translated to “what was resolved was his military debt (amounting to) 1 kāṭi and 8 suvarṇa”. We would now rather interpret the phrase as meaning either “what was resolved was his valānda debt of 1 kāṭi and 8 suvarṇa” or “what was resolved was his debt, a valānda of 1 kāṭi and 8 suvarṇa”.

Perhaps we may reconcile the usage related to payments in the epigraphic contexts with that related to the storage of valuables observed in the two OJ literary texts referred to above by assuming that the term was able to express not only such senses as “outlay; disbursement; expenditure; budget” but also the container in which accumulated wealth was stored, much like the English words “treasury” or “bursary” (and French “bourse”). It is curious that our text furnishes the only epigraphic occurrence of valanda that clearly designates a container rather than a payment.31 Given the fact that five of these containers need to fit together into one bronze vessel, we are inclined to think that these valāndas may have been pliable containers like bags rather than boxes.

In conclusion, we have the impression that the OM and OJ evidence presented above is not incompatible with the hypothesis that C/MM belanja and Modern Javanese blanja derive from valānda/valanda, although the modern terms can no longer be used to designate a container. It also seems possible that the premodern words are ultimately connected with the Sinhalese word $vala\breve{n}da$, although there are aspects of this question that we lack the competence to judge. Regarding the irregular sound correspondence between C/MM belanja and OM valānda, we have suggested above that an ad hoc solution is necessary, and we are inclined to seek it in the phenomenon of krama-forms already observed above in the local Balinese usage of valantən besides valanda. While the creation of polite-register (krama) words besides inherited vocabulary is generally considered to be characteristic of Javanese, the phenomenon is known to have impacted Malay as well.32 It is therefore imaginable that belanja is in origin a krama-transformation based on valānda. We can cite no precisely analogous case of a krama form, but we can point out the existence of some other unexplained cases of correspondence of /d/ to /j/ in the languages of Indonesia.33

tāsak—We have considered identifying this as C/MM tasik/tasek “sea, lake” or as Old Javanese tasak “ripe, cooked” (cf. C/MM masak), but neither option seems particularly suitable. Hence, we tentatively interpret the form as ta-asak derived from asak “to pack, to fill, to stuff” (cf. C/MM sesak, originally sa-asak). Cf. Mahdi (2015, pp. 194–95) on ta- = ter- in the Tanjung Tanah ms., with examples like taamit and taulih from the Tanjung Tanah ms. We only need to assume that sandhi has taken place to obtain tāsak.

damcu—To interpret this word, which can be matched neither with any word known to us from Malay or related languages nor with Sanskrit, we are forced to explore the possibility of borrowing from other languages. We have considered but rejected the possibility of borrowing from Tamil தமிசு tamicu “East Indian kino” (which would have been perceived by a Malay speaker as something like /tamisu/). From Chinese, we have considered 淡酒 (Mandarin danjiu, Hokkien tamchiu) “light wine”, whose Hokkien pronunciation would have been perceived in terms of Malay phonology as /tamciu/, and 銅子 “copper coin” (Mandarin tongzi, Hokkien tangchi or tangchu), probably perceived in Malay phonology as /tangci/ or /tangcu/, but we have rejected these too as likely sources of damcu for semantic and phonological reasons and, in the second case, also because of the low likelihood of the term having been in use in the 11th and 12th centuries.34 Although the phonological match with any Chinese pronunciations likely to have been in use in Southeast Asia in the 11th and 12th centuries is also not perfect, we believe the best combination of plausible sense and similar sound is with the Chinese term for cinnabar, written 丹朱 (Mandarin danzhu, Hokkien tanchu) and pronounced in Hokkien in such a way that a Malay speaker might have perceived it as /tancu/ or /dancu/.35 Why it was actually spelled damcu, reflecting perception of the Malay sound and /m/, is a problem, though one that we deem to be relatively small, since the range of spellings that might have been adopted for transcribing 丹朱 in OM is broad, given how many relevant variables are unknown to us.

Cinnabar was a commodity commonly imported to and exported from China. Considering that it was shipped from China to markets across much of East and Southeast Asia, it fits well into the historical context of our inscribed strips. It was used in the production of medicine, handicrafts, cosmetics, paint (for structures, paintings, and statues), and in funerary rites across much of the world (Gliozzo 2021). The two essential components that derived from cinnabar were vermilion, which was used as a coloring agent, and mercury, which was used to make medicine and in the gilding of metals. China has always possessed large cinnabar deposits, as has Vietnam (Schafer 1967, p. 157; Zhu fan zhi, juan 1, Jiaozhi section, ed. R. Zhao n.d.). During the Tang dynasty (618–907), cinnabar was shipped out of ports in the Zhedong region (present-day Zhejiang province) (Schafer 1963, p. 153) and from Guangzhou. In the latter port city, it was freely used as a common medium of exchange along with gold and silver (Schafer 1967, p. 153), which is noteworthy because Guangzhou was the main port for overseas trade with states in Southeast Asia and the Persian Gulf in that period.

In addition to its varied uses, the popularity of cinnabar as a trade commodity can be attributed to the fact that sizable deposits of the mineral did not exist or were not known to exist in many areas that actively participated in long-distance sea trade. Such areas included Japan, Java, India, and possibly Sumatra (Gliozzo 2021, p. 7).36 The import of Chinese cinnabar in Japan is well-documented, as it appears on a list of 60 drugs presented to the Tōdaiji Buddhist temple in 756 (Schafer 1963, p. 180), in a record of gifts given from a Chinese merchant to a Japanese courtier in 1028 (von Verschuer 1999, p. 6), and on a list of “Chinese goods” dated to 1052 (Batten 2006, p. 115).37 Cinnabar is also listed as an item of export from China to Java in the early 13th-century work Zhu fan zhi (juan 1, Shepo section). This text further records that, in one of Java’s vassal states, the women used vermilion for makeup, nail polish, and to dye their clothes and that the product was in high demand (Wheatley 1959, p. 91; Zhu fan zhi, juan 1, Sujidan section). In India, around the time of the Song dynasty, minium, a lead-based substance very similar to cinnabar (and often mistaken for it), was known as cīnapiṣṭa, or “China flour” (Schafer 1956, p. 425). This is a strong indication that such substances were commonly imported there from China. The use of mercury compounds is also prevalent in Indian medicinal texts (Gliozzo 2021, p. 8). As for archaeological evidence, cinnabar was discovered on the Quanzhou shipwreck, which sank after returning to China from Southeast Asia in the late 13th century (Museum of Overseas Communications History 1997, p. 49). Although most of the cargo consisted of commodities purchased in Southeast Asia, a certain portion also consisted of unsold Chinese wares, and cinnabar might have belonged in that category. Another example is the Nanhai wreck, which sank around 1183 on its way to Southeast Asia after departing from Guangzhou. Cinnabar was among a wide range of items discovered in the ship’s hold, along with precious metals and ceramics (Zhang et al. 2024, p. 267).

Looking beyond the historical context of cinnabar as a commodity and toward its religious significance, we may point to the lump of cinnabar discovered among precious stones, minerals, gold objects, a ritual deposit box, and other goods in a shaft underneath temple ruins at the Santubong site in Sarawak, North Borneo (Harrisson and O’Connor 1967, p. 213). It is believed that these items were used in a Tantric Buddhist ritual, much like a similar collection of goods—some containing traces of mercury on them—that were found at a temple site in the Bujang Valley area of Kedah (Treloar 1967; Treloar 1968). What is certain is that vermilion was widely used across much of the world to paint religious structures, statues, and paintings. Mercury, extracted from cinnabar, was also commonly used as a component in the gilding of copper-alloy Buddhist statues, one likely example being the gilt bronze sculpture kept in the Victoria and Albert Museum (London), representative of the Nagapattinam style.38

Furthermore, the usage in Old Javanese literary contexts of the term valanda to indicate the storage of wealth in the form of precious metals and jewelry likewise parallels the known instances of cinnabar being included along with gold and silver as precious commodities transported by sea across long distances, and of it being used as a means of exchange along with gold and silver in port cities. Cinnabar is commonly discovered in the geological deposits of gold and silver, which possibly strengthened such an association between the mineral and precious metals in common perceptions (Gliozzo 2021, p. 7). Finally, as Old Khmer (Vincent 2012, pp. 182, 184, 185) and Old Malay do not have any known native terms for cinnabar or vermilion, the use of a foreign loanword adopted from a region with longstanding connections with those cultures would render 丹朱 a strong candidate for the original term from which damcu derived. A notable comparison would be the Indonesian term for vermilion and lipstick, gincu, which derives from a Hokkien term for vermilion, ginchu 銀朱 (Mandarin yinzhu, literally ‘silver vermilion’ or ‘silver cinnabar’).39

brat—The word, spelled here with a long ā presumably because, before suffix -ña, it stands in penultimate position, obviously corresponds to berat in C/MM. While no occurrence of the word is found among OM documents known so far, there are quite a few OJ inscriptions where brat (or vrat) occurs in similar contexts. See Darmosoetopo (1993) and Wisseman Christie (2015) for several examples. As in these OJ instances, and as in the OM Laguna copper-plate inscription (Clavé and Griffiths 2022, pp. 199–200), the weight is expressed using units from a system that seems to have been in use throughout maritime Southeast Asia. In this system, a consistent set of abbreviations is used to express units, kā (for kāṭi) and su (for suvarṇa) being among them. According to Wisseman Christie (2004, pp. 92–93), 1 kāṭi = ca. 760 g, while 1 suvarṇa = 1/20th of that (38 g), so that the total weight indicated here is 3838 g. In OJ inscriptions, the unit dhāraṇa (abbreviated dhā) is used as an equivalent for suvarṇa, specifically for weights in silver, and the next lower unit was māṣa (1/16th of a suvarṇa/dhāraṇa). To illustrate these comparisons, we include in the next section (5) a photograph of such an Old Javanese inscription on a silver plate.

barkna—This form has not so far been encountered in OM, although maṅənakan, another form derived from the same base kəna, is attested in the Tanjung Tanah manuscript. The C/MM morphological equivalent to OM barkna would have to be ber-kena, but this form does not exist (while ber-kena-an, with a further -an suffix, does exist and means “connected, applicable”). From the context, it is clear that barkna must mean “is for” (a meaning that could be expressed in C/MM with bagi, buat, guna, or untuk, not one of which seems to be inherited from Old Malay) or “has the purpose of”. Although there is not a precise morphological match, it seems that this OM usage must be connected somehow with the existence in OJ of pakəna “function, purpose” and the derived form mapakəna “to be of use for, intended for, proper to serve as”.

bharāla—This is at present the oldest known attestation in OM of the ancestor of C/MM berhala, which is ultimately related to the Indic word bha(ṭ)ṭāra that was widely used as such both in Sanskrit and in vernacular sources in various parts of Southeast Asia, including the Malay and Javanese spheres.40 For discussion of the linguistic background, see Griffiths (2014, pp. 228, 242–43) and Hoogervorst (2017, pp. 388–89). For examples of bhaṭāra in OM and OJ contexts, see the inscriptional comparisons given below (Section 5).

dṅan—This word, the ancestor of both dan “and” and dengan “with” in C/MM, expressed both meanings in OM. See Mahdi (2005, p. 192).

5. Final Translation and Comparisons

Although the reconstruction proposed in Section 3 assumes complete identity between the two texts of which fragments have been found, nothing stands in the way of small differences, notably for the quantities, expressions, and weights. Keeping this inherent uncertainty in mind, and with reference to the commentary on the vocabulary items offered in the previous section, we can at last propose the following interpretation of the text’s meaning:

1 bronze cauldron (kathora), filled with 5 bags (valānda) stuffed with cinnabar (damcu), weighing 5 kāṭi, 1 suvarṇa (i.e., about 3838 g). These bags are for worship to the Lord of the Śrī-Bālāditya monastery and to the Lord of the Śrī-Śailendra-Cūḍāmaṇivarman monastery.

It now seems useful to situate the inscription in the broader cultural context of Indonesian epigraphy from the point of view of its textual contents and its palaeography. Above, we mentioned that the report by Tejowasono et al. (2019) identifies the script as “Old Sumatran”. We do not necessarily object to this term, but we wish to emphasize that its pertinence is a question of definition. Depending on the definition used, it would be equally correct to assign a label that is not island-specific, the fundamental problem being that we have no information about script names that were in use in ancient Sumatra41, nor do we have any certainty that the two copper strips were manufactured in Sumatra at all. Not only in language, but also in its palaeographic aspect, we find our text quite similar to the Laguna Copper-plate Inscription (900 CE), manufactured thousands of kilometers away from Sumatra in Northern Luzon, but an equally pertinent comparison can be made with the inscription on the back of the base of a bronze sculpture of the Buddhist deity Lokanātha from Gunung Tua in Northern Sumatra (Figure 9).42

(1)

⚬ svasti śaka-varṣātīta 961 caitra-māsa, tithi tritiya sukla, śukra-vāra, tatkāla juru pā⌈-

(2)

ṇḍai suryya barbvat· bhaṭāra lokanātha ⚬ Imāni kuśala-mulāṇi sarbvasatva-sādharāṇaṁ k$\underset{°}{\mathrm{r}}$-

(3)

tvā Anuttarāyāṁ samyaksambodhau parināmayami (⚬)

Prosperity! Elapsed Śaka year 961, third tithi of the waxing (fortnight) of the month of Caitra, a Friday. At that time, the master smith Sūrya manufactured Lord (bhaṭāra) Lokanātha. [He expressed his intention by citing scripture:] “These roots of what is good do I share with all beings and turn over to the unsurpassed, perfect enlightenment!”.

The date of this inscription, which consists of an Old Malay sentence and a dedicatory sentence in Sanskrit, corresponds to 1039 CE.43 The palaeographic aspect of this Sumatran inscription is perhaps even more similar to our copper strips than the Laguna Copper-plate Inscription, so that this comparison would support dating our copper strips to the 11th century CE. As we see below, their date of manufacture cannot have been before 1005/6 CE.

We see how the Buddhist deity in the Gunung Tua inscription is called bhaṭāra Lokanātha, using the term bhaṭāra from which we have indicated that the term bharāla, in our strips, is derived. An example of use of the same term in an inscription on a metal artifact from Java is given in the next comparison, which also illustrates the use of the word brat and abbreviations dhā, su, and mā for expressing units of weight in silver or in gold (Figure 10).44

(1)

￮ bhaṭāra kakīṅ abhaya brat· dhā 11 (2) bratniṁ mās ya su 1 mā 15

The deity (bhaṭāra) at Abhaya: weight 11 dhāraṇas [in silver]. The weight of the gold, it is 1 suvarṇa, 15 māṣas.

Apparently, one has to understand that the silver plate was the property of the bhaṭāra at Abhaya and that it weighed 11 dhāraṇas in silver, while allusion is also made to a quantity of gold that would perhaps have been placed on the plate. A similar inscription on another silver artifact from Java furnishes further parallels to the text on our two strips (Figure 11).45

(1)

//￮// gavayan· bhaṭāra I ḍi(hya)[ṁ dyun pirak· brat· k]ā 4 rambutnya lpas· //￮//

Work belonging to the deity (bhaṭāra) of Dieng: silver vase weighing 4 kāṭi. Its hairs are loose.

The use of the expression dyun pirak “silver vase” here, combining the name of an artifact type and the metal from which it is manufactured, to label the very artifact on which the text is engraved is exactly analogous to kathora kāṅśa “bronze cauldron” in our strips. We also find the abbreviation kā here, instead of its equivalent dhā, to express the weight in silver. The final phrase, “Its hairs are loose”, is rather mysterious but clearly again aims to add some further detail about the vase in question.

Several of the aspects of the texts engraved on these two pieces of Javanese silverware are analogous to the inscription on our two copper strips, with the notable difference that the Javanese items themselves bear their labeling text, whereas our strips must have served to label other artifacts. Comparable labels attached to cargo found in shipwrecks elsewhere have been found to denote the ownership or intended recipient of certain items and, in some instances, also contained details about the volume of goods and their names. Their purpose was to demarcate commodities and sections of cargo on board a trade ship. See, for instance, the hundreds of wooden strips found in the Sinan and Quanzhou wrecks (Y. Li 2023, pp. 137–43, and Figure 12 below). An explanation for the similarities apparent in both copper plate inscriptions could be that those inscriptions held the same or a similar function of distinguishing cargo on board a ship and that both inscriptions each originally denoted distinct items and quantities of cargo but for the same intended recipient. It should be noted that, in the case of the Sinan wreck, much of the cargo on board was also intended for Buddhist temples, as such temples often held major roles in finance and trade. Additionally, the general shape of the copper strips might also be indicative of a similar function as labels for cargo. Their complete shape, narrower on one end with a pierced hole and pointed on the other end, can be deduced from the shape of each fragmentary strip. Their similarity in shape and textual contents with cargo labels made of wood found in the Sinan and Quanzhou wrecks cannot be a coincidence, as textually quite different land grants inscribed on copper plates from Nusantara are usually also quite different in shape, with significantly larger dimensions. One important difference between the Muara Jambi strips and the Sinan and Quanzhou wrecks strips, however, is their material, a difference about which we prefer not to speculate here.

6. The Inscribed Text in the Context of Transregional Buddhist Networks (c. 11th–12th Centuries)

On the basis of the text preserved on the extant fragments, the reconstructed text lists offerings for the worship to the Lords (bharāla), i.e., deities, of the monasteries called Śailendra-Cūḍāmaṇivarma-vihāra and Bālāditya-vihāra. The first name occurs in virtually the same form in the Sanskrit and Tamil Larger Leiden Copper-plate Inscription of Rājarāja Cōḻa I (r. 985–1014), as well as other nearly contemporary Tamil inscriptions.46 These inscriptions have been discussed in a large number of secondary sources to reconstruct the historical background of diplomatic, religious, and commercial links existing between Sumatra, South India, the Malay peninsula, and China in the period immediately preceding the Cōḻa raids of around 1025 CE, when Rājendra sent a fleet of warships to raid and destroy important ports in Sumatra and the Malay Peninsula, but also in the period following them (see, e.g., Vogel 1919; Krom 1931, pp. 236ff.; Majumdar 1933–1934; Cœdès 1968, pp. 141ff.; Chattopadhyaya 1981, p. 88; Skilling 1997, 2007; Seshadri 2009; Schoterman 2016; Gornall 2014, p. 520, 2020, pp. 59 n. 91, 69–70, etc.).

As related by the charter, whose praśasti portion was added posthumously in 1019, early in the reign of Rājendra I (r. 1014–1032), the Śailendra-Cūḍāmaṇivarmavihāra was built at Nāgīpattane (Sanskrit)/Nākappaṭṭiṉam (Tamil), i.e., Nagapattinam, at the request of Māravijayottuṅgavarman, born in the Śailendra family, lord of Śrīvijaya, king of Kaṭāha (Kedah in the Malay Peninsula),47 who brought to completion—and claimed as his own pious foundation48—the work started by his father Cūḍāmaṇivarman, in whose name the monastery was built.49 This undertaking, which probably took place in the 21st regnal year of Rājarāja I (1005/6), who had granted the village of Āṉaimaṅgalam for the monastery’s maintenance, was supported by his son and successor Rājendra Cōḻa I. A subsequent Cōḻa king, viz. Kulōttuṅga I (r. 1070–1120), continued to support this monastery with endowments (Jacq-Hergoualc’h 2002, p. 275). What is clear is that the Cōḻa dynasty—many of whose kings, like Rājarāja and Rājendra, were staunchly Śaiva—maintained good relations with the rulers of Sumatra and the Malay Peninsula up to the first quarter of the 11th century, and they then challenged them for the control of the commercial maritime routes. However, since Kulōttuṅga I’s charter explicitly records its issue at the request of two ambassadors from the king of kiḍāra (i.e., Kaṭāha), it appears that the relationship between the Cōḻas and Śrīvijaya improved again after the Cōḻa raids.50

The grant by Kulōttuṅga I states that King Rājarāja gave the village “to the Buddha (buddhāya) residing in the surpassingly beautiful Cūḷāmaṇivarma-vihārā of [such] high loftiness [as had] belittled the Kanakagiri (that is, Meru)” (Aiyar 1933–1934b, p. 257; Karashima and Subbarayalu 2009, p. 273). It is striking is that the mention of the Buddha in the monastery, to which the village was gifted, mirrors the mention of the Lord (bharāla) of the Cūḍāmaṇivarman monastery to whom (ka) the goods listed in the Muara Jambi strips were to be offered in part. Even more striking is the mention of a Bālāditya-vihāra next to the Śailendra-Cūḍāmaṇivarmavihāra in the Muara Jambi strips, as both are found to co-occur also in two roughly contemporary historical documents from southern India and Sri Lanka. The first document is the closing verse in the Pali grammatical treatise Rūpasiddhi by scholar–monk Buddhappiya (fl. between ca. 1008 and 1165 CE) (Gornall 2020, p. 54). This monk, also known as Coḷiya Dīpaṅkara, was originally from South India but was active in Sri Lanka. In the aforementioned colophon, he claims that he was a disciple of Ānanda, like a standard for Tambapaṇṇi (Laṅkā), and that he himself was renowned like a lamp in the Damiḷa (i.e., the Tamil) Country, glossed as coladeśaya and soḷīraṭa in a Sinhalese commentary to the text (see below) (Malalasekera 1928, pp. 211, 220; Gornall 2014, p. 520). Indeed, Buddhappiya says that he resided in two (monastic) establishments (vāsa),51 one of which was the Bālāditya.52 The second (slightly later) document is a version of the Rūpasiddhi with interlinear commentary in Sinhalese, which provides the name of the other monastery as cūḍāmaṇikarmavihāraya—obviously a scribal mistake for cūḍāmaṇivarmavihāraya (Paranavitana 1944, p. 20, endorsed by Gornall 2014, p. 520). It is generally assumed that the Bālādityavihāra was situated in South India too (Gornall 2014, p. 520), but there is no direct evidence to prove this, as Paranavitana noted.53 Paranavitana (1944, p. 21) also argued the following:

the Bālādityavihāra was evidently the more important of the two establishments, for Buddhappiya made special mention of it while he left the identity of the other to be explained by his exegetist. Like the Cūḍāmaṇivarma Vihāra and Dhammāsoka Vihāra of Negapatam, Bālādicca Vihāra, too, appears to have been named after an important personage. The most famous Bālāditya of Buddhist history was the king of Magadha who, according to Hieun Tsiang [i.e., Xuanzang], was the founder of the celebrated vihāra of Nalanda but we cannot assert that a monastic establishment in the Coḷa country was named after him.

There being, indeed, no direct evidence that the Bālāditya establishment mentioned in the documents discussed above was located at Nagapattinam, for instance, as an affiliated namesake of the famous Bālādityavihāra at Nālandā,54 our best candidate for it is still the original foundation in Bihar; however, it cannot at present be positively identified with any of the remaining structures.55 Bālāditya developed various construction activities outside of the monastery complex and built a large temple for an image of Śākyamuni Buddha.56 The importance of this (c. 6th century?) Gupta king is highlighted by inscriptional evidence as well as later Buddhist accounts, including Xuanzang’s 玄奘 (c. 602–664) Datang Xiyu ji 大唐西域記 [Record of the Western Regions from the Great Tang (Dynasty)], the Chinese biography of Vasubandhu (Posoupandou fashi zhuan 婆藪槃豆法師傳), and the Sanskrit Mañjuśriyamūlakalpa.57 It is, therefore, not surprising that Buddhappiya only mentioned that monastery, and that the Sinhalese commentary only glossed the lesser known of the two. That he resided in those two monasteries far removed from each other is not surprising either, as they may have been well-connected in the network of Buddhist pilgrimage of the time. More surprising is the curious synchronicity between the Rūpasiddhi commentary and our inscribed strips, which mention both monasteries. To this, the implication of our interpretation must be added, namely that the cargo mentioned in the strips from Muara Jambi was destined for two monasteries at ca. 1800 km distance from each other as the crow flies. This must have posed logistical challenges, but it was not impossible to achieve, especially because the greatest part of the journey would have been by sea, just like the one needed to cover the ca. 2800 km separating Muara Jambi from Nagapattinam.

Contemporary evidence suggests that Buddhappiya’s monastic and scholarly career was cosmopolitan, and that it transcended the borders of the Tamil country and Sri Lanka, indeed. Buddhappiya is acknowledged by other sources from Sri Lanka as a famous Buddhist personality of his time; the Gandhavaṅsa includes him in the list of Sri Lankan masters (laṅkādīpakācariya).58 Buddhappiya had close ties with the Cōḻa court, which, according to Gornall (2020, p. 70), “may have contributed to the split that took place between the monastic orders in South India and Sri Lanka”. An inscription issued by the queen of the Sri Lankan King Vikramabāhu I (1111–1132) found at Polonnaruwa describes Buddhappiya’s teacher Ānanda in similar terms as those found in the colophon of the Rūpasiddhi, and it states that the master had contacts (and possibly spent some time) with the Buddhist community of Tambaraṭṭha,59 which Paranavitana has identified with Tāmbraliṅga, the country around Ligor (Nakhon Si Thammarat, or Nagara Śrī Dharmarāja) in the Malay Peninsula (see Wolters 1958, p. 588). This area was also under the authority of Śailendra kings Cūḍāmaṇivarman and Māravijayottuṅgavarman. Paranavitana goes against another view identifying Tambaraṭṭha with a district in South India, and the matter does not seem to have been settled yet.60 Whatever the case may be, it is likely that Buddhappiya followed in the foot-steps of the propagation of Buddhism in the Cōḻa country (Paranavitana 1944, p. 25). The possible presence of Buddhappiya’s teacher Ānanda in the Śailendra domains of the Malay Peninsula just prior to the foundation of the Cūḍāmaṇivarmavihāra in Nagapattinam further supports the existence of close Buddhist linkages between that region (with Sumatra) and South India, which go beyond the political and diplomatic aspect of the donation of the monastery itself.

Other sources that are often used to reconstruct the aforementioned historical background are Chinese accounts of diplomatic missions and Buddhist texts in the Tibetan canon. The former sources describe the arrangement made between Emperor Zhenzong (968–1022, r. 997–1022) of the Song dynasty (960–1279) and envoys of Śrīvijaya. During their stay in the Song capital of Kaifeng in 1003—just a couple of years before the building of the Cūḍāmaṇivarmavihāra in Nagapattinam—the envoys informed the emperor that their King Cūḍāmaṇivarman had built a Buddhist temple in Śrīvijaya dedicated to praying for Zhenzong’s longevity, and they requested that the emperor provide both a signboard with a temple name, as well as a bell.61 The request was met by Zhenzong, who named the temple Chengtian wanshou 承天萬壽 (“Heaven-Bestowed Longevity”). According to Claudine Salmon and Hans Bielenstein, the Chengtian wanshou temple may have actually been constructed for a Chinese merchant community in Śrīvijaya.62 These merchants may have also advised the Śrīvijaya king on matters of diplomacy, as had happened in Java too, with the merchant Mao Xu advising the Javanese embassy to China one decade prior to Śrīvijaya’s embassy in 1003–1004 (Sattler 2024a, p. 5). Some sources suggest that the Śrīvijayan envoys downplayed the importance of the Cōḻas and even claimed they were under their dominion, which caused a diplomatic reaction by the South Indian dynasty through an embassy to the Chinese Emperor carrying lavish gifts (Sen 2003, pp. 229–34; 2009, p. 62). Later, a Śaiva temple in Cōḻa style was erected in Quanzhou (Guy 1994, 2011, pp. 253–59), and in 1018, the Cōḻa king is recorded as having received large gifts of “Chinese gold” from Śrīvijaya for a temple staffed by Brahmans (Sattler 2024b; Sen 2003, p. 67; Kulke 2009, p. 6). These examples highlight the phenomenon of “temple diplomacy” via maritime links in the region, which might have “originated in and around the Indian Ocean region and circulated with the spread of Buddhism”.63

As for the Tibetan sources, they provide data for reconstructing the history, and transfer, of Buddhism in the region. In particular, they provide some historical details on celebrated masters like Atiśa (a native of northeastern India) and Dharmakīrti of Suvarṇadvīpa (Dharmakīrti II) and their whereabouts in the early- to mid-11th century (see in particular Sinclair 2021; Schoterman 2016; Skilling 1997, 2007). The latter is believed to have lived during Cūḍāmaṇivarman’s reign, as the colophon of one of the Tibetan Kanjur block prints preserving his works, the Durbodhāloka, says that the original text was composed at the request, and during the 10th year of the reign (ca. 1005–8 CE?), of King *Śrī-Cūḍāmaṇivarma-deva (Lha dpal gtsug gi nor bu’i go cha’)64 of *Suvarṇadvīpa (Gser gling), and its place of composition was *Śrīvijayapura (Dpal rnam par rgyal ba’i grong) (see Sinclair 2021, p. 7; Skilling 1997, pp. 188, 194 n. 12). Atiśa traveled to Suvarṇadvīpa to learn from this Buddhist master, who ranked among the most respected and learned personalities of the time. While these Tibetan sources highlight the cosmopolitan nature of the South and Southeast Asian Buddhist world of the time, they are often imprecise; the geographical locations of these political and religious personalities are still problematic. Sinclair (2021, pp. 7–9) and Skilling (1997) identify Suvarṇadvīpa (=Suvarṇabhūmi) with Kedah in the Malay Peninsula. The latter scholar in particular uses information contained in the Tibetan account of Atiśa’s maritime journey to argue that the landing point of the ship on which he was traveling was none other than Kedah. Skilling’s argument that Cūḍāmaṇivarman and his son and successor Māravijayottuṅgavarman reigned in Kedah is derived from his analysis of the Sanskrit and Tamil inscriptions known as the “Leiden Plates”, which state that Māravijayottuṅgavarman was both “lord of Śrīviṣaya” (only in the Sanskrit section) and “lord of Kaṭāha” (in both Sanskrit and Tamil sections). On the other hand, Miksic (2016, p. 262) argues that “for the people coming from the Indian Ocean, the ruler of Kedah was seen as more important than that of Palembang”, which is due almost certainly to Kedah being the first major port that many ships traveled to before entering the Strait of Malacca. This might also explain why the Leiden Plates were mainly concerned with the ruler of Kedah. Miksic does not believe that “references to the ‘Lord of Śrīvijaya’ denote the ruler of Kedah” since “Rulership of Kaḍāra, Kiḍāra, Kataha, Kaḍāram, etc., was probably a glorious title in itself”. To him, the epigraphic and archaeological evidence rather suggests that Śrīvijaya meant Palembang. Still, he thinks that “Malayu was the most important Buddhist center in the Strait of Melaka between 1025 and 1200”, and that “Atīśa may have spent his time in Malayu/Jambi rather than in Kedah” (Miksic 2016, p. 263) for, according to him, the only viable candidate for the location of Buddhist institutions of higher learning as well as royal power that could support the scholarship of Dharmakīrti II and his circle from an archaeological point of view is Muara Jambi. However, the fact that much is still unknown about other sites that could be potential candidates for this royal and religious center, including Kedah itself, makes it difficult to pin so much on Muara Jambi.

While we can only estimate the date of manufacture of the two inscribed strips of this study, we likewise have no certainty as to the location from which the strips were created for use as labels of religious donations. However, an examination of the Chinese textual sources reveals important information that scholars, to date, have overlooked. This information might provide clues to understand the political dynamics in play when the inscribed strips were made. Many scholars have accepted the theory that Palembang served as the capital of Śrīvijaya until sometime after the Cōḻa raids of 1025, and by 1079, the capital was moved to Jambi. This argument is primarily based on O.W. Wolters’ position that the arrival of a tributary embassy from Jambi to the Song capital in 1079 signified that the rulership of Śrīvijaya had shifted from Palembang to Jambi (Wolters 1966; Xu zizhi tongjian changbian, juan 299, ed. T. Li 1881). Following this event, two letters were sent by Jambi leaders to a local Chinese administrator in 1082, though this is the last mention that is made of Jambi in history until later dynasties (Xu zizhi tongjian changbian, juan 330). Wolters believed that, following this period, the frequent references of Śrīvijaya in Chinese sources were, in fact, all references to a Śrīvijaya state with its capital in Jambi, and such a view remains influential to this day. However, there is no additional textual evidence to suggest that this might have been the case at any time prior to the collapse of Śrīvijaya, which Chinese sources situate in the 13th or 14th century. A close examination of the 1082 entry in fact reveals that the Song court considered the leader of Jambi to be a “sovereign” (guozhu 國主), whereas each generation of Śrīvijaya leader was distinguished with a position higher than this as “king” (guowang 國王).65 This was not an arbitrary distinction but rather one that denoted clear positions among the Song court’s diplomatic partners, and a higher title generally entailed greater status and rewards in diplomatic trade. A more nuanced reading of the sources also demonstrates that the Jambi embassy was referred to by Chinese record keepers as “the Śrīvijayan country of Jambi” (Sanfoqi-Zhanbei guo 三佛齊詹卑國), which implies that there were other states within the broader confines of what appears to have been a Śrīvijaya confederacy. This comes as no surprise to scholars who have noted that the Chinese translation of Śrīvijaya (Sanfoqi 三佛齊) directly translates to “Three Vijayas”. Indeed, a range of Chinese sources makes distinctions between Śrīvijaya and Kedah, which is perhaps better interpreted as a distinction between Palembang and Kedah rather than Kedah falling outside the confines of a greater Śrīvijaya confederation.66 In either case, Skilling’s view that Māravijayottuṅgavarman ruled from Kedah is not supported in any way by Chinese sources. Additional distinctions were also made between “Śrīvijaya” and “Jambi”,67 which suggests that “Śrīvijaya” only ever referred to either Palembang or to a large and loosely aligned political entity, in which Palembang was originally the symbolic center.

If there continued to be a primary political center of Śrīvijaya following the Cōḻa raids of 1025, such a location might plausibly have remained in Palembang. Archaeological evidence suggests a very gradual transition in economic prominence from Palembang to Jambi rather than a sudden one, with much overlap in the 11th century (Manguin 1987). It is also understood that trade ships would stop at both Jambi and Palembang in the midst of voyages between China and West or South Asia (Wolters 1967, pp. 207–10). Considering that the copper strips were discovered in or near a riverway used by sea ships that traveled across great distances, the discovery of the strips in the region of Jambi does not guarantee that Jambi was the port where they originated from. Thus, both Palembang and Jambi remain the primary candidates for where the strips were produced until more information is forthcoming.

7. Conclusions

As mentioned at the beginning of the article, the two artifacts discussed here were not recovered from a scientific excavation, but were rather purchased from private individuals, and no written documentation exists about their exact findspots. In the light of these facts, we feel compelled to exert due diligence and weigh the arguments both against and in favor of their authenticity.

As for the arguments in favor of inauthenticity, one may point at some surprising physical features of the strips, the slightly yet noticeably different engraving technique in the verso side of A with respect to the recto side, with more irregularities and a different depth, as well as the rarity of inscriptions engraved by means of a “pointilistic” technique in Nusantara and the wider Indic world.68 It is also important to note the cut in the middle section of the strips, which could be an attempt to disguise inauthentic artifacts into fragmentary antiquities; however, it cannot be excluded that it was deliberately done to have more portions of authentic artifacts for sale. Further, the mention of the [śrī-śail]endra-cūḍāmaṇivarma-vihāra in strip A and śrī-bālāditya-vihāra in strip B, both of which are referred to in a Singhalese interlinear commentary to the Pali treatise Rūpasiddhi by Buddhappiya (as already discussed in Paranavitana 1944), as well as the reference to the donations being for the Lord or idol (bharāla) of the former monastery, which mirrors the same reference in the Greater Leiden Copper-plate inscription by Rājarāja I, looks like a “too good to be true” scenario. Indeed, this scenario might have been brought to the fore by a putative forger in order to concoct the narrative of Muara Jambi as the seat of the political power that supported the funding of the Cūḍāmaṇivarman monastery in Nagapattinam, as well as the donation to the Bālāditya monastery in Nālandā associated with Buddhappiya—a scenario that is reminiscent of the donations made by the Sumatran sovereign Bālaputradeva to build a monastery in Nālandā during Devapāladeva’s reign around 860 CE.69

In spite of this, we believe that their cumulative palaeographic, linguistic, and material features speak against their inauthenticity. Two arguments in particular could be advanced. First, the forger would need to have had a deep and nuanced knowledge of not only the script, but also a level of active command of the OM language that is present not even among the very few scholars involved in this tiny field of study, to have generated unattested but plausible OM forms like tāsak, brātña, and barkna, imagining the use of the loanword khaṇḍa as a numeral classifier, while creating new instances of the challenging word valānda that no previous scholar has adequately understood, deftly inserting it into a context analogous to but not identical with its usage in other early Nusantaran sources, and finally deciding to invent a word that cannot be Indic or Nusantaran, namely damcu. Second, known forgeries tend to be made of gold, be furnished with decoration, reproduce known texts, or if they do not, they tend to be long documents intended to buttress elaborate historical claims.70 This is not the case of our strips. The fact that fragment B, which has been analyzed by XRF (see n. 10), is mainly made of copper, and does not contain any zinc, is also an indicator of its antiquity. More recent metalworks indeed tend to include zinc in the alloy or be of brass (a copper–zinc alloy), but brass was not used on a deliberate basis in South and Southeast Asia before the 16th–17th century (Craddock 2009, pp. 145, 151).

Further, at least one of the aforementioned surprising physical features could be interpreted as not being particularly problematic. For instance, it seems relevant to point out that an inscription (in Chinese) using the pointilistic technique is also found on another metal item also recovered in the Jambi area, namely the bronze cymbal discussed by Salmon (2002, 2003). This technique was known in China as early as the Han dynasty (202 BC–220 AD) (Salmon 2002, p. 108). By the 11th–12th century, at around the time the copper inscriptions were produced, there are records of Chinese merchants learning Southeast Asian writing systems. Thus, one could propose that such writing came from nearby workshops staffed with individuals of different backgrounds, or that the writing could be produced by someone of mixed ethnic heritage or by an Indian or Malay who spent much time in southern China. This could also explain the unusual shape of the strips—narrow on one end with a hole and pointed on the other end—resembling that of artifacts also used as labels for cargo in China, but extant in wood (Figure 12).

Thus, we conclude that the two strips discussed in this article constitute authentic—and exceptional—evidence agreeing with, and supporting, the historical scenario of the existence of two monasteries in South and Northeast India that were linked economically (and otherwise) to Sumatra and China. The strips seem to have been labels recording donations meant for both monasteries to be dispatched to India by ship. One could only speculate about the reasons why the inscriptions never made it to their final destination—for instance, that the boat in which they were transported in was shipwrecked in the Batanghari River, or that part of the cargo fell into the river during its loading onto the ship. Since a great quantity of ancient artifacts, including inscribed tin foils, small bronze statues, coins, and pottery, have been reportedly recovered from the riverbed, either by workers extracting sand or by antique-hunters, this is a scenario that seems to be within the realm of possibility (although, of course, alternative explanations exist, like for instance the fact that the river may have shifted its course and what now lays underwater was once on land). A fragment of a copper plate inscribed in substantially the same script type was also discovered in the tenth-century Intan shipwreck (Flecker 2002, pp. 89–90). Although, in that instance, too little is preserved of the original plate to determine the size/shape, and too few characters are preserved on it to determine the type of text inscribed, at least this discovery shows that the existence of such items on board ships was not without precedent. Thus, an explanation for the existence of the two entirely or nearly identical inscriptions of which the respective remaining fragments have been studied here could be that both inscriptions each originally denoted distinct items of cargo, i.e., bronze containers filled with the same (or a similar) quantity of valuable objects, intended for the same recipients.