1. Silver foil inscription from Sambas, West Kalimantan

In the annual report of the Archaeological Service in the Dutch East Indies (Oudheidkundig Verslag, OV) for the year 1948, in §6 dealing with “Discoveries”, we find an entry on Sambas (West Borneo, now West Kalimantan province), which reads (p. 13):Footnote ¹⁶

Subsequently, the September 1949 issue of the Journal of the Malayan Branch of the Royal Asiatic Society (vol. 22, pt. 4) was dedicated entirely to the Sambas finds, with papers contributed by Roland Bradell, Nilakanta Sastri, H.G. Quaritch Wales, Tom Harrisson and Tan Yeok Seong. No mention was made in this publication of any inscription as part of these finds. The collection was acquired by the British Museum in 1956, and the technical staff of the Museum then discovered that a rolled-up silver foil had been hidden in the base of the largest sculpture of the Sambas hoard (object 1, plates 1 and 2 in JMBRAS 22/4, 1949). At some moment after being unrolled, this foil broke into two fragments, currently kept at the British Museum under registration number 1956,0725.8.b.Footnote ¹⁹ Putting the two fragments close together, we can determine that the original would have been about 16 cm in length and 2.35 cm in height.

As far as I know, the inscription on this silver foil, covering both sides, has only rarely been mentioned in the scholarly literature and remains unpublished.Footnote ²⁰ I will first present my reading of this document, then I will analyse its contents and discuss the importance of this artefact for the history of Buddhism and Buddhist art in Indonesia.

Text

The foil is extremely hard to capture in its entirety photographically. After repeated provisional attempts based on different photographs offered by Michael Willis, and a tracing of the letters as observed by his research assistant Beatriz Cifuentes (without knowledge of the script), I was able to determine the reading from the original at the British Museum on 20 June 2012, using a microscope. As we will see on some other specimens of mantra/dhāraṇī inscriptions included in this article, the writing is sloppy, so that it often takes some goodwill to read a given akṣara in a way that makes sense. I subjoin photos of the original both with and without the mentioned tracing (Figures 2 and 3). While it does not seem necessary to point out all scribal errors, my notes anticipate some of the textual identifications made after the decipherment is presented.Footnote ²¹

Figure 2.

The Sambas foil, obverse and reverse.

Figure 3.

The Sambas foil, obverse and reverse, with tracing of the akṣaras by Beatriz Cifuentes.

Obverse:

(1) // na[ma] (°ār)yyā(na)[x n]tabhādrāFootnote ²² // yāvada kāci daśaddiśi loke sarvvatriyādhvaganara narasi(h)hāFootnote ²³ tāhu(2)tān ahnaFootnote ²⁴ vanda[mi] (sa)rvva °aśeṣāṁ (k)āyatu vācamanena prasannaḥ // yāś ca °imaṁ pariṇāmanarā(3)narāda(m·) x tvaFootnote ²⁵ (sa)kj janayed adhimukti (vo)dhivaram anuprārthayamāno °agru viśiṣṭa bha(4)[ve]d i[mu] p[u]ṇya(m·) //Footnote ²⁶ rakṣoghnā paramā hy eṣā pavitrā pāpanāsaśanī śrīka(r)ī dhī-

Reverse:

(1)Footnote ²⁷karī cāṭiFootnote ²⁸ sarvvaguṇavivarddhanī // ye dharmmā hyetoprabhavā hetun teṣā(2)ñ ca yo [n]irodho hy evam·vādī mahāśramanaḥ // ye māṁ (rū)peṇa °adrākṣu (3) ye māṁ [gho](ṣe)ṇa °anvayuḥ mathyyāprahāṇaprastā na mā drakṣyanti te janāḥ ||| (4) dra yaṁ bhadra[pra]nidhānaFootnote ²⁹ °ini candravarmma x (na)ta (p)una °ā(va)tu CCi x (na)ta(rañca) hi xFootnote ³⁰

2. Stone inscription of Batu Bedil in Lampung

Let us move now to that great bastion of Buddhism in Indonesia, the island of Sumatra. In the far south of this island, in the province of Lampung, there is a complex of megalithic sites now known as Batu Bedil (Rifle Stone), after the most eye-catching megalith. It seems to have been the German ethnographer Friedrich W. Funke who first discovered this complex in 1953, and found there a large inscribed stone (Batu Surat) at about 40 m from a field marked by twenty-seven menhirs arranged in a square (originally probably 8 × 8 stones, arranged regularly over 40 × 40 m).Footnote ⁴⁸ In Reference Funke1958, he published an extensive report of his findings (see especially pp. 20–22), including a ground plan of the field of menhirs in relation to the Batu Surat (pl. I), with a good photo of the inscription (fig. 8 on plate III). He assumed that the position of the inscribed stone in relation to the field of menhirs indicated a significant correlation; he also speculated (p. 21 n. 3) that the text of the inscription would be in Old Sundanese. Both of these speculations were rightly rejected by van der Hoop in a very critical review of Funke's book (v.d. Straaten and v.d. Hoop Reference Straaten1960: 285).

The site was visited again, in 1954, by a team of Indonesian and foreign archaeologists during a voyage of exploration that has become famous thanks to the full reports published by the leader of the team, Soekmono, on two occasions (1955 and Reference Soekmono1962), both with the inscription illustrated in the form of the OD photo 19576.Footnote ⁴⁹ In the 1955 report (pp. 49–50), we find much more reliable information on the inscription:

More or less the same facts were mentioned by Damais (Reference Damais1962: 283–4, n. 5), who had himself participated in the voyage. He announced that a publication by J.G. de Casparis was forthcoming, but this has never materialized.Footnote ⁵¹ I visited the site myself in September 2011, and was able to make the estampage of which a photograph is included here (Figure 4). Although I still find three of the ten lines totally unreadable, I can publish a greater part of the inscription here because I am able to confirm the suspicion voiced by Soekmono that it is a known dhāraṇī, and have identified its source: the Aparimitāyuḥsūtra. With this source text in hand, I am able to read a number of akṣaras (here printed in italic) that it would otherwise probably have been impossible to decipher.

Figure 4.

EFEO estampage n. 2043 of the Batu Bedil inscription.

(1) || namo bhagavate

(2)°aparimitāyurjñā-

(3)nasuviniścitate-

(4)jorājāya tathā-

(5) gatāyārhate sa-

(6)...

(7) ...

(8) ...

(9)svabhāvapariśuddhe ma-

(10) hānayaparivāre svāhā

These elements are sufficient to exclude the possibility that we are dealing with any other known dhāraṇī. We may initially compare the reading with the text of the dhāraṇī in the Aparimitāyuḥsūtra, edited on the basis of Nepalese manuscripts by Konow Reference Konow and Rudolf Hoernle1916: 301:

But comparison with the version of the dhāraṇī in the Khotanese manuscript (dating to about the middle of the tenth century ce)Footnote ⁵² that was edited in parallel by Konow, on the facing page of his edition, immediately shows that the Nepalese manuscripts, which are approximately 700 years more recent, give an expanded text:Footnote ⁵³

That this shorter text is older, and that the Nepalese manuscripts transmit an expanded version, is suggested by the fact that all three versions included in the Taishō edition of the Chinese canon show variants of the same short version.Footnote ⁵⁴ Although our inscription is totally worn in the middle of the dhāraṇī, we may deduce from the fact that the preserved lines show an average of seven akṣaras per line that there would certainly not have been enough space for the sequence tadyathā o puṇyamahāpuṇya aparimitāpuṇya aparimitāyupuṇya jñānasambhāropacite.Footnote ⁵⁵ In other words, our inscription was engraved with a variant of the original shorter version of the dhāraṇī, and it must be noted that compared to the Khotanese and Chinese versions, ours may have been even shorter, for the number of syllables on lines 6–8 will have been about 21, whereas the number of syllables required to obtain the text of those other versions is 28. In this connection, it is a striking fact that the Sūtra itself repeatedly states that the dhāraṇī consists of 108 syllables, which it only does in the longer Nepalese recension. In the present context, I am unable to throw any light on this numerical problem.

In his recent article dedicated to the text, Richard K. Payne has drawn attention to the fact that the Aparimitāyuḥsūtra and dependent texts seem to have been among “the most popular bodies of literature in Nepal, in Dunhuang, and elsewhere throughout the Buddhist cosmopolis” (Reference Payne2007: 277).Footnote ⁵⁶ The entire scripture consists of little more than the advice that one should copy, or have others copy, the sūtra, or the dhāraṇī, and promises diverse benefits for those who do, using i.a. the phrase (Konow Reference Konow and Rudolf Hoernle1916: 315–7):

Translation adapted from Silk (Reference Silk and Lopez2004: 427):

It seems reasonable to assume that the ancient local Buddhists who were involved in the manufacture of our inscription did so with precisely the motivation of founding a “true shrine”. More specifically, Soekmono's observation (cited above) that the text is provided with a lotus cushion, just like a Buddha image would be, is an unmistakeable expression of the intent behind engraving this text, namely that its presence as dharmakāya would be equivalent to the presence of the Buddha, and would sacralize the site.Footnote ⁵⁸

It should be noted that there is considerable overlap in the terminology used by the Aparimitāyuḥsūtra to describe, on the one hand, the ways in which the text (or its dhāraṇī) can be used as well as the benefits that will result from such use and, on the other, the way we see use and benefits described in other scriptures that will be discussed further on in this article, notably the Amoghapāśah daya (section 10).Footnote ⁵⁹

Finally, regarding the date of our inscription (which Soekmono situated around 900 ce, see above), it should be noted that Damais (Reference Damais1967: 687 n. 1) was inclined to date it about a century older, without specifying the more archaic traits underlying this estimate. The text is in Kawi script that, frankly, seems less archaic to me than the script of the Sambas inscription (section 1).Footnote ⁶⁰

3. Gold foil from sector Segaran IIA at Batujaya, West Java

During excavations in 2005 at the Buddhist complex of Batujaya, home to the oldest Buddhist monuments identified in Indonesia so far, the team led by Pierre-Yves Manguin discovered in the sector called Segaran IIA a piece of gold foil, folded up into a small squarish lump. When it was unfolded, it turned out to bear an inscription.Footnote ⁶¹ The existence of this inscription, besides that of other gold foils bearing the ajñānāc cīyate formula found at the same site, was mentioned by Hasan Djafar (2010: 93).Footnote ⁶² But the inscription so far remains unpublished. Thanks to excellent visual materials furnished by Manguin (Figure 5), I am now able to propose the following reading:

The reading is rather tentative, for the word eṭana makes no sense,Footnote ⁶³ while the reading yā is also far from secure. I translate:

Despite the problem posed by eṭana, it is clear enough that the dhāraṇī is an adaptation of some of the numerous variants of the final line of the Gāyatrī mantra, reading tan no nom pracodayāt, found in brahmanical literature (see Bloomfield Reference Bloomfield1906: 50–51). The word taking the place of nom is in each instance the nominative singular case form of the name of a deity. I am unable to trace this formula in any Buddhist source, but the idea that it is Buddhist finds support in the monumental context where the foil was discovered.

Figure 5.

The foil found at Batujaya, Segaran IIA. Photo EFEO/P.-Y. Manguin.

The palaeographic appearance of the inscription is inconsistent, and comprises both some archaic features of Southern Brāhmī (i.e. “Pallava”) and some more Kawi-like characteristics. The stratigraphic context in which the artefact was found is somewhat difficult to interpret, the only thing that is certain being that it must date from before 800 ce. There is a greater archaeological likelihood that it belongs to a level dateable to 330–550 than to the subsequent level of 550–700/750.Footnote ⁶⁴ From the epigraphist's point of view, however, this early dating bracket seems hard to accept. The other gold foil inscriptions from the same site give a more archaic impression that would more easily be compatible with such an early dating, although I would still be reluctant even in their case to consider a date before the sixth century.

4. Gold foil from the Nan Han period “Cirebon” shipwreck, coast of West Java

A sunken ship was discovered off the coast of West Java near the city of Cirebon in 2003. After the cargo had been salvaged, it could be determined that the shipwreck must date from about the last quarter of the tenth century, or later. Alongside large quantities of ceramics and a substantial number of bronze cult objects, the divers also recovered a piece of gold foil which is inscribed with a dhāraṇī of a total of six lines. It is very difficult to produce legible photographs of this small, irregularly shaped artefact (Figure 6), but Horst Liebner, who is writing his doctoral thesis about this shipwreck and has had extensive access to the cargo, has taken considerable effort to document the inscription. He helped me execute the tracing of the akṣaras that is shown here in Figure 6.Footnote ⁶⁵

Figure 6.

The foil recovered from the “Cirebon” shipwreck, obverse and reverse. Photo Horst Liebner, with tracing of the akṣaras by the author.

Obverse:

(1) namo ratnatrayāya nama °āryyāva(lo)-

(2) kiteśvarāya bodhisatvāya mahā(sa)-

(3) tvāya mahākā(r)uṇikāya tadya(thā)

(4) °o mārutākuha x kāmāvrā x (mālya)

Reverse:

(1) (prabhā)devī (°ada)lān driyase nam(u)

(2) (dha)man de(v)ī

Despite the fact that the writing is very faint, and that the photographs available do not show all parts of the texts with equal focus, it is possible to read more than half with certainty, due to the fact that the opening (up to the beginning of line 4 on the obverse) is highly formulaic, being found in a number of Buddhist scriptures. But I am unfortunately unable to identify the mantra part that begins with o , and am (therefore) unable to reconstruct the parts that are not formulaic. Since we find here the words māruta, meaning “wind” and (if my reading is correct) driyase, meaning “you are borne”, it may be suspected that the dhāraṇī was intended to protect against the dangers of voyage at high sea, and that the object on which it is inscribed was worn as an amulet, while the text was addressed to the goddess personifying the spell.

For the structure of this dhāraṇī, I refer to the Ekādaśamukha – a text of the same genre as the Aparimitāyuḥsūtra and the Amoghapāśah daya referred to in sections 2 and 10 of this article – of which a copy was found among the Gilgit manuscripts (seventh century). Here we find a sequence of six repetitively structured mantric sequences beginning with namo ratnatrayāya | namo vairocanāya tathāgatāya | nama āryāvalokiteśvarāya bodhisattvāya mahāsattvāya mahākāruṇikāya | nama atītānāgatapratyutpannebhyaḥ sarvatathāgatebhyo ’rhadbhyaḥ samyaksaṁbuddhebhyaḥ | oṁ diri diri dhuru dhuru | iṭṭe viṭṭe | cale cale | pracale pracale | kusume kusumavare | ili mili ciṭi svāhā | and ending with namo ratnatrayāya nama āryāvalokiteśvarāya bodhisattvāya mahāsattvāya mahākāruṇikāya | tadyathā piṭi piṭi tiṭi tiṭi | viṭi viṭi gaccha gaccha bhagavan āryāvalokiteśvara svabhavanam svabhavanaṁ svāhā | […] āryāvalokiteśvara gaccha svabhavanam ||.Footnote ⁶⁶

The Sādhanamālā (eleventh century) furthermore contains three citations of a Siṁhanādadhāraṇī “Lion's Roar dhāraṇī”, that again shows the same structure (Bhattacharya Reference Bhattacharya1968: 48, 52, 54, 63): namo ratnatrayāya nama āryāvalokiteśvarāya bodhisattvāya mahāsattvāya mahākāruṇikāya tadyathā oṁ akaṭe vikaṭe nikaṭe kaṭaṁkaṭe karoṭavīrye svāhā.

On the basis of these comparisons, we expect the word svāhā somewhere in the part that I have not been able to decipher. In any case, it is clear that we have here a variation on dhāraṇī material of the type seen in the Ekādaśamukha and the Sādhanamālā. The palaeographic appearance of the script is consistent with a date in the ninth or tenth century.

5. Gold and silver foils from Candi Plaosan Lor, Central Java

In his Prasasti Indonesia II (1956: 170–2), J.G. de Casparis included as item no. V the edition of a “plate of gold leaf” which had been discovered during excavations at Candi Plaosan Lor in 1947. The discovery itself had been the subject of a brief discussion in OV 1948 (p. 30), where it was also mentioned that a little silver plate had been found near the plate of gold foil, bearing the same inscription. It seems to be this silver foil inscription that is captured on the photograph OD 17094. Unfortunately, the current whereabouts of neither of the two foil inscriptions appears to be known, and no archaeological photograph seems to have been taken of the gold foil inscription at the time of its discovery.Footnote ⁶⁷ From their archaeological context, the inscriptions, written in Kawi script, may be assigned to the ninth century. The published reading is as follows:

De Casparis indicated no substantial difficulty in reading the text of the gold foil, and there is no reason to doubt that his decipherment is correct. Indeed, several elements of this reading can also be recognized on the OD 17094 photo of a silver foil inscription. De Casparis correctly identified this text as a dhāraṇī, and he translated:

This scholar was unable to identify a precise scriptural source for the dhāraṇī, but noted its affinity with the Sarvadurgatipariśodhanadhāraṇī. He knew the latter only from two manuscripts kept at the Bibliothèque nationale at Paris, and from the manner that he cites them it seems that he was not aware of the further contents of these manuscripts.Footnote ⁶⁹ The scriptural source for this dhāraṇī is the eponymous Tantra, from which I cite (Skorupski Reference Skorupski1983: 127–8):

This mantra was very widespread in Buddhist Asia, and existed in many variants. A version with only slightly different readings is one of two dhāraṇīs included in Taishō 1395 Baji kunan tuoluoni jing 拔濟苦難陀羅尼經 “Sūtra of the Dhāraṇī for Salvation from Tribulations” (21: 912c), a short dhāraṇī-sūtra about which Rolf Giebel (Reference Giebel1993) has published an article, where he cited all the versions he had managed to identify at the time (see p. 152 for the Taishō version). Now the Javanese dhāraṇī seems to comprise a combination, not known to me at this time from any other Buddhist tradition, of the Sarvadurgatipariśodhanadhāraṇī with another known dhāraṇī, pointed out to me by Rolf Giebel.Footnote ⁷⁰ It appears in another short dhāraṇī-sūtra, translated by Faxian in 1001: Taishō 1399 Miechu wu nizui da tuoluoni jing 滅除五逆罪大陀羅尼經 “Sūtra of the Great Dhāraṇī for Extinguishing the Five Heinous Sins” (pañcānantaryāṇi)” (21: 915c–916a):

As Giebel points out to me, in this Chinese version ajñāle is uncertain, and jaloharaṇe is probably an error for rajoharaṇe, as we read on the foils from Candi Plaosan. Although a precise scriptural source for the Javanese dhāraṇī may of course still be discovered, it seems adequate to say that it is built substantially on the same text as seen in Taishō 1399, with insertion of elements that are all known from the Sarvadurgatipariśodhanadhāraṇī, except the epithets gaganaviśodhani Footnote ⁷¹ and cittaviśodhani.Footnote ⁷² It has been reported that the same dhāraṇī has also been found engraved on silver and gold foils from Candi Sojiwan, which have not, to my knowledge, ever been published, but which were mentioned by Boechari (1976: 19 = 2012: 172, n. 21).

Subsequent restoration work at the site of Candi Plaosan Lor in 1993 has brought to light other gold, silver and bronze artefacts bearing Buddhist inscriptions of a similar type, but now in Siddhamātkā script closely resembling that of the Kalasan, Kelurak, Plaosan and Ratu Baka stone inscriptions, so datable to the period between the late eighth and mid-ninth centuries.Footnote ⁷³ A provisional reading of two gold foils, without diacritics, was included in the report on these restorations (Gutomo and Niken Wirasanti Reference Hidas1998: 54 with photo no. 27 on p. 100) and later reproduced verbatim in Miksic et al. Reference Miksic, Nayati, Klokke and van Kooij2001: 323–4. Besides an evident ye dharmāḥ formula at the head of the second inscription cited by Miksic et al., it is hard to make any sense of these two inscriptions in this provisional reading. However, the Bureau of Conservation of Archaeological Remains (BPPP) for Central Java province obligingly assented in August 2012 to my request to put photographs of these artefacts at my disposal. Besides the two gold foils, the photos I received from the BPPP also included a third item, that appeared to be in silver, and of which I found no mention in the report by Gutomo and Niken Wirasanti, or the publication by Miksic et al. This third photograph showed an inscription in two long lines of Siddhamātkā script identical to that on the two gold foils, but the photograph was not sufficiently focused for me to be able to read the whole. I was subsequently allowed to come and inspect the three foils and photograph them myself.Footnote ⁷⁴ On this basis, I am now able to publish the following readings and source identifications.

The taller gold foil was found in association with the reliquary in the SE corner below the floor level of the central chamber of the northern main temple at Candi Plaosan Lor.Footnote ⁷⁵ It measures 5.2–5.8 × 25.5 cm, weighing 7.9 g (Figure 7).

This inscription gives the text of the Bodhigarbhālaṁkāralakṣadhāraṇī, a text which has drawn quite a bit of attention in Buddhist studies in the last three decades, and which was thus far not known to be attested in Indonesia. I refer for the most recent publication on this dhāraṇī to Strauch Reference Strauch, Mevissen and Banerji2009. Admitting the emendations proposed above, the version from Candi Plaosan Lor may be translated as follows:

The reason why this dhāraṇī was deposited in the foundations of the monument is very clear from the framing Sanskrit text in which it is embedded in several other versions from India and Tibet. Let me quote here only Schopen's translation of the relevant passage (Reference Schopen1989a [Reference Schopen2005]: 329):

Another rolled piece of gold foil was found west of the SE reliquary. It measures 39 × 3.7–4.0 cm, weighing 15.2 g. Its reading is as follows (Figure 8):

We have here the ye dharmāḥ formula, followed by a dhāraṇī called the Saptasaptatisamyaksaṁbuddhakoṭibhiruktā “Proclaimed by seventy-seven crores of completely awakened ones”, and two further dhāraṇīs of which the first is attested in transliteration in the Chinese canon (Taishō 1383, 21: 904b),Footnote ⁹³ while the last remains unidentified. As for the Saptasaptatisamyaksaṁbuddhakoṭibhiruktā, I take this title from the single South Asian Sanskrit text where I have found the dhāraṇī in this full form: it stands at the start of section 2.7 of the Kāraṇḍavyūhasūtra, in the most commonly used edition, straight after the section where the Buddha's pupil has been taught the famous six-syllabled formula oṁ maṇi padme hūṁ (Vaidya Reference Vaidya1961: 301):

I have made only the most evidently necessary emendations, in consultation of the facsimile edition of a Nepalese manuscript published by Lokesh Chandra (1981),Footnote ⁹⁶ but even then, it is hard to escape the impression that the passage is incoherent, and therefore probably corrupt. The following translation makes this clear:

The passage is not only hard to understand by itself, but also seems to lack coherence with its context in the text as edited from Nepalese manuscripts. Moreover, it is unattested in the significantly older Gilgit manuscript, and this absence does not seem to be an accident that can be attributed to the very fragmentary condition of this witness (see Mette Reference Mette1997: 122–3). One may therefore suspect that the passage is an interpolation in the Nepalese tradition of this text, but if this suspicion is justified, it begs the question from which source the passage might have been borrowed. Beyond noting that a variant dhāraṇī namaḥ reading saptānāṁ samyaksaṁbuddhakoṭīnāṁ, namaḥ cale cunde namaḥ is attested in Taishō 1169 (20:678c); that another variant namaḥ saptānāṁ samyaksaṁbuddhakoṭināṁ [sic], oṁ cale cule cunde is attested in the same text (20:686b); that the Javanese version of the dhāraṇī is found in precisely the same form among the Sanskrit texts brought to Japan by Kūkai in the early eighth century; and that this same collection of texts also includes a Cundāstotra which alludes to the same idea of the dhāraṇī being uttered by seven (instead of seventy-seven) crores of Buddhas,Footnote ⁹⁷ I must leave these textual problems open here, and return to the inscription that was my starting point. In the case of this apparently composite second inscription, the reason for its deposit is unfortunately not provided by the identification of the source texts.

Figure 7.

Gold foil recovered from the North shrine at Candi Plaosan Lor. Photo Arlo Griffiths.

Figure 8.

Gold foil recovered from the North shrine at Candi Plaosan Lor. Photo Arlo Griffiths.

Nothing seems be known with certainty about the exact provenance of the third foil, mentioned above, but we may safely assume that it too was found during the restoration process of the same monument. Its dimensions are 56.5 × 3.6 cm, its weight 15.2 g. I read as follows (Figure 9):

Figure 9.

(Colour online) Silver foil recovered from the North shrine at Candi Plaosan Lor. Photos Arlo Griffiths.

Here again we find the ye dharmāḥ formula followed by the Saptasaptatisamyaksaṁbuddhakoṭibhiruktā. Then follows a variant of the mantra that, as we have seen above, forms part of the expanded recension of the dhāraṇī of the Aparimitāyuḥsūtra: tadyathā o puṇyamahāpuṇya aparimitāpuṇya aparimitāyupuṇya jñānasambhāropacite. I am unable to trace this variant with akṣa akṣa akṣaya° in any other text. Then follow the words namo bhagavate ratnaśikhine tathāgatāyārhate samyaksaṁvuddhāya which we saw previously on the second gold foil, but it is possible that they occur here with a termination that is slightly different from that we saw there (tadyathā oṁ ratna ratna ratnasaṁbhave svāhā). The silver foil is unfortunately damaged at the point of possible difference. As noted above (n. 93), a variant of this dhāraṇī addressed to the Buddha Ratnaśikhin is attested in the Chinese canon.

6. Two maṇḍala stakes (kīla) from Central Java

A recent publication from the Bureau of Conservation of Archaeological Remains (BPPP) of the special region of Yogyakarta includes, among numerous other items of interest, an inscribed object in tuff stone, resembling in shape some of the numerous inscribed boundary marking stones whose inscriptions form a distinct category of Javanese epigraphy.Footnote ⁹⁹ The object is stated to have been found in 1975, but this cannot be correct. The very fact that an excellent set of photographs is preserved at Leiden University proves that the discovery must have taken place earlier, because photographs stopped being supplied to Leiden by 1959. According to the Leiden website (see n. 49), the creation of the prints of these OD photos 21151–21160 kept in Leiden dates to 1958. The provenance there cited is Kali Tirto, Berbah, Sleman; this disagrees slightly with the information in the recent publication, which states Jragung, Berbah, Sleman.Footnote ¹⁰⁰ The stone is currently held at the site museum in the Prambanan complex (Taman Wisata Candi Prambanan) and bears the inventory number BG 37. One of the OD photos is reproduced here as Figure 10. The shape of the akṣaras points to a date in the ninth or tenth century. The published reading, virtually without diacritical marks, bears only a vague resemblance to the text that is actually engraved, which I read as follows:

I have presented the text in two paragraphs, for it appears that it is composed of two parts, identifiable in distinct Sanskrit texts preserved in manuscripts from Nepal. The first paragraph finds a remarkable parallel in the Sarvadurgatipariśodhanatantra. In the context of the ritual eradication by the tantric adept (yogin) of obstructions to his practice, we read a section intermingling ritual instruction with virtually the same mantras that we find on the stone. This section must be quoted here in full along with the editor's translation (Skorupski Reference Skorupski1983: 134–6; trans.: 12–3).Footnote ¹⁰⁴ The mantras corresponding to those in the first paragraph of our inscription are indicated in bold type:

tato vajrānalena mudrāsahitena vighnadahanādikaṁ kuryāt | oṁ vajrānala hana daha paca matha bhañja raṇa hūṁ phaṭ ity udīrayet | abhyantaravajrabandhe ’ṅgulijvālāgarbhe ’ṅguṣṭhavajram utthitam iyaṁ vajrānalamudrā | tad anu | vajranetri bandha sarvavighnān iti | mudrāyuktyā sarvavighnabandhaṁ kuryāt | vajrabandhaṁ baddhvāṅguṣṭhadvayaṁ prasārya samaṁ dhārayet | vajranetrīmudrāFootnote ¹⁰⁵ | prasāritavajrabandhaṁ bhūmyāṁ pratiṣṭhāpyādhobandhaṁ kuryāt | oṁ vajra dḍho me bhava rakṣa sarvān svāhā | vajrabhairavanetreṇa mudrāsahitenordhvabandhaṁ kuryāt | oṁ hulu hulu hūṁ phaṭ | iti | vajramuṣṭidvayaṁ baddhvālātacakraṁ bhrāmayitvā śirasopari tarjanyaṅkuśākāreṇa dhārayet | vajrabhairavanetramudrā | tasyādhastād vajrayakṣeṇa mudrāsahitena punar bandhaṁ kuryāt | oṁ vajrayakṣa hūṁ iti | vajrāñjaler aṅguṣṭhadvayaṁ prasāritaṁ tarjanīdvayaṁ daṃṣṭrā | vajrayakṣamudrā | vajroṣṇīṣeṇa mudrāyuktena pūrvāṁ diśaṁ bandhayet | oṁ druṁ bandha haṁ iti | druṁ iti vā | vajramuṣṭidvayaṁ kanyasāśṅkhalābandhena tarjanīdvayasūcīmukhaṁ parivartoṣṇīṣe sthāpayet | vajroṣṇīṣamudrā | punar vajrapāśena tām eva bandhayet | hūṁ vajrapāśe hrīḥ iti | vajramuṣṭidvayena bāhugranthiṁ kuryāt | vajrapāśāmudrāFootnote ¹⁰⁶ | vajrapatākayā paścimāṁ diśaṁ bandhet | oṁ vajrapatāke pataṁgini raṭeti | vajrabandhe ’ṅguṣṭhasattvaparyaṅkasūcīṁ ktvāgrāsamānāmāvidāritāntyapaṭāgrī | vajrapatākāyāḥ | digvidikṣv adha ūrdhvañ ca vighnanikntanaṁ kuryāt | vajrakālyottarāṁ diśaṁ bandhet | hrīḥ vajrakāli ruṭ maṭ | iti | vajrayakṣamudrām eva mukhe dḍhīktya vajrakālyāḥ | vajraśikharayā dakṣiṇāṁ diśaṁ bandhet | oṁ vajraśikhare ruṭ maṭ | iti | vajramuṣṭidvayena parvatotkarṣaṇābhinayākāraṁ vajraśikharāyāḥ | vajrakarmaṇā maṇḍalabandhaṁ ktvā prākāraṁ dadyāt | hūṁ vajrakarmeti | punar abhyantaraprākāraṁ vajrahūṁkāreṇa hūṁ iti | vajramuṣṭidvayaṁ baddhvā bāhuvajraṁ samādhāya kaniṣṭhāṅkuśabandhitā trilokavijayanāmatarjanīdvayaṁ tarjanī | vajrahūṁkārasya | iyam eva madhyāgradvayaṁ vajraṁ vajrakarmaṇaḥ

Next he should burn the obstruction by applying the gesture of Vajrānala. He should exclaim: OṁVajrānala kill, blaze, consume, convulse, break, battle hūṁ phaṭ. The gesture of Vajrānala is the vajra-thumb raised in the middle of blazing fingers comprised inside the vajra-bond.

Figure 10.

Photo OD 21151. Courtesy of Leiden University Library.

Next he says: OṁVajranetrī bind all the obstructions. He applies the gesture and binds all the obstructions. He makes the vajra-bond, stretches the thumbs and holds them evenly. This is the gesture of Vajranetrī. Placing the outstretched vajra-bond on the ground he should bind beneath. He says: OṁVajra be firm for me, protect all svāhā.

Applying the gesture of Vajrabhairavanetra he should bind above. He says: Oṁ hulu hulu hūṁ phaṭ. Binding the vajra-fists and waving them like a firebrand he should hold them above the head with the forefingers made into hooks. This is the gesture of Vajrabhairavanetra.

Once more by means of the gesture of Vajrayakṣa he should make a new bond beneath, saying: OṁVajrayakṣa hūṁ. The gesture of Vajrayakṣa is a form of the vajra-añjali with the thumbs outstretched and the forefingers projecting like tusks.

With the gesture of Vajroṣṇīṣa he should bind the eastern quarter. He says either Oṁ druṁ bind haṁ or just druṁ. He should place the vajra-fists on the crown of his bowed head with the little fingers linked together like a chain and the forefingers made into a point. This is the gesture of Vajroṣṇīṣa.

Once more he should bind the same quarter by applying the gesture of Vajrapāśa. He says: HuṁVajrapāśa hrīḥ. He should make a knot with the arms by means of the vajra-fists. This is the gesture of Vajrapāśa.Footnote ¹⁰⁷

He should bind the western quarter by means of Vajrapatākā. He says: Oṁ, flyingVajrapatākā flutter. The gesture of Vajrapatākā is the vajra-bond in which the thumbs are crossed. The forefingers are put together and then parted, and the little fingers are made like banners.

He should destroy the obstruction below and above, in the quarters and in the intermediate quarters.

He should bind the northern quarter by means of Vajrakālī. He says: OṁVajrakālī ruṭ maṭ. The gesture of Vajrakālī is the gesture of Vajrayakṣa firmly placed at the heart.

With Vajraśikharā he should bind the southern quarter. He says: OṁVajraśikharā ruṭ maṭ. The gesture of Vajraśikharā is made with the vajra-fists shaped like an arched hillock. He binds the maṇḍala by means of the gesture of Vajrakarma. He makes thus the enclosing wall. He says: HuṁVajrakarma.

The inner enclosure is made by means of Vajrahuṁkāra, saying: Huṁ. He binds the vajra-fists and forms a vajra with his arms; the little fingers he makes into hooks and raises the forefingers into a point known as Trilokyavijaya (Victor over the Threefold World). This is the gesture of Vajrahuṁkāra. Footnote ¹⁰⁸

It should be pointed out that this portion is found only in what Skorupski (Reference Skorupski1983: xvii) has called “version B” of the Tantra, and finds no correspondent in “version A”. Skorupski considered the Tibetan translation of “version A” to be as old as the eighth century, while the oldest evidence for the existence of “version B” would be a Tibetan translation from the thirteenth century. It is at first sight surprising to see our ninth–tenth-century Javanese inscription agreeing with the ostensibly “later” version, but in his review of Skorupski's work, van der Kuijp (Reference van der Kuijp1992) has pointed out that the very ancient date attached to “version A” cannot be taken at face value, and that the transmission of this version is not at all reliable; conversely, Skorupski has himself pointed out that the parts of “version B” not represented in “version A” “certainly represent very early materials” (p. xxix). That the set of deities involved in this portion had been received in Bali was already known from their occurrence in a Buddhist hymn transmitted on that island in manuscripts.Footnote ¹⁰⁹ None of these, however, is earlier than the seventeenth century, although the presence of the set of deities must logically precede the oldest manuscript by some amount of time. It seems to me that our inscription now offers fairly strong evidence of the availability to Buddhist priests in ninth–tenth-century Java of a textual/ritual tradition very closely related to that preserved in the thirteenth-century Tibetan translation and still much later Nepalese manuscripts of the Sarvadurgatipariśodhanatantra.Footnote ¹¹⁰ It would hence seem that we can confirm with entirely independent evidence the conclusions reached earlier by Nihom, using the Balinese manuscript material, “that a tradition reflecting the form and content of [chapter 1 of “version B” of this Tantra] was available at some time in classical Indonesia”.Footnote ¹¹¹ I will return to this issue below.

The second paragraph of the inscription is itself composed of three parts. The first runs from oṁ vajrakīla to phaṭ. Then we find tenfold repetition of huṁ (i.e. the bīja hū). I will return to these first two parts below. The third part comprises the remainder of the inscription. This last part is found almost verbatim in a number of texts. I have found it in the ritual manuals Piṇḍīkramasādhana Footnote ¹¹² and Sādhanamālā,Footnote ¹¹³ but also in the scripture Guhyasamājatantra, at 14.59 (Matsunaga Reference Matsunaga1978):

However, in none of the mentioned texts is this mantra combined with o vajrakīla kīlaya hū phaṭ. It is only in another ritual manual, the Sarvavajrodaya composed by the ninth-century author Ānandagarbha (and the later Kriyāsaṁgrahapañjikā which seems to be based on it in this section), that I find this combination. In the edition based on a single eleventh-century Nepalese manuscript established by a group of young Japanese scholars, we read as paragraph 37:Footnote ¹¹⁴

Returning now to the first two parts of the second paragraph of our inscription, I presume that the tenfold repetition of the bīja hū, for which I do not find a precise textual parallel, indicates destruction of obstacles in the ten directions (the four cardinal and intermediate quarters of space, above and below);Footnote ¹¹⁷ it may be assumed that the Sarvavajrodaya’s instruction hūṁkāram udīrayan indicates a very similar liturgical element, although the number of the stakes there seems to be only five. As for the first part, a slightly more extensive variant of this occurs in another Javanese inscription of the ninth or tenth century, engraved on a liṅga-shaped stone whose precise provenance is unknown, held in the National Museum at Jakarta, under number D. 140.Footnote ¹¹⁸ This inscription was published first by Crucq (Reference Crucq1929: 274), and subsequently drew attention again from other scholars, all misguided by the idea that the language of the inscription was Old Javanese. However, some decades later Boechari (1964: 123 = 2012: 66 n. 5) observed that the text is in Sanskrit and must undoubtedly be read as follows:

This is (with just two minor variants: the repetition of vandha and the displacement of oṁ to the end) identical to the text of the kīlanah daya “heart-mantra for placing a stake” found first in the Sarvatathāgatatattvasaṁgraha, and hence repeated in several later Buddhist tantric texts, including the Sarvavajrodaya and the Kriyāsaṁgrahapañjikā, but not the Sarvadurgatipariśodhanatantra. The passage is as follows (Horiuchi Reference Horiuchi1983/I, p. 526, sections 1264–5):

It is now time to draw conclusions from these data. With regard to the textual correspondences between chapter 1 of “version B” of the Sarvadurgatipariśodhanatantra, on the one hand, and Balinese Buddhist liturgical materials on the other, adduced by Nihom (Reference Nihom1994: 78–80) in favour of his conclusion that this Tantra was disseminated to the Indonesian archipelago, we have to keep in mind that all of the same correspondences can be traced in the Kriyāsaṁgrahapañjikā. This text, dating from the eleventh century, would be old enough to have served as the source for the Balinese material, but is too late to have served as source for the two inscriptions under discussion here. However, all of the relevant textual material is found in the Devatāyoga part of the sixth chapter of the text, and this is precisely the part which is considered to have been adapted from the significantly older Sarvavajrodaya.Footnote ¹¹⁹ This last text, possibly composed in the ninth century, would probably be old enough to have served as source for our inscriptions.Footnote ¹²⁰ Unfortunately, the part corresponding to Kriyāsaṁgrahapañjikā 6 Devatāyoga is not preserved in the sole Sanskrit manuscript known to exist for this text. Because I do not have access myself to the Tibetan version of the Sarvavajrodaya, I limit myself to pointing out the correspondences between the elements that drew Nihom's attention in the Sarvadurgatipariśodhanatantra and the corresponding parts of the Kriyāsaṁgrahapañjikā (see Table 1).

Table 1.

Correspondences between Balinese Buddhist liturgy, Sarvadurgatipariśodhanatantra “version B” and the Kriyāsaṁgrahapañjikā

Among these elements, it must first be recalled that items 3 and 4 concern precisely the passages of the Sarvadurgatipariśodhanatantra (and Kriyāsaṁgrahapañjikā) which I have cited as a possible source for the first paragraph of the first inscription. If we may assume, then, that these and all the other elements, whose place in the Kriyāsaṁgrahapañjikā is indicated in the table, do indeed find their correspondent in the (Tibetan version of the) Sarvavajrodaya, this means that the first paragraph of our first inscription may have been quoted from that text rather than the Sarvadurgatipariśodhanatantra. If we keep in mind, furthermore, that the Tibetan version of the Sarvavajrodaya presumably offers parallels for all the segments of this inscription, and that this text is the only one which does so while also being old enough to have been known in (late) ninth-century Java, whereas the Sarvadurgatipariśodhanatantra does not provide a source for all elements of the first inscription, and does not offer anything corresponding to the part that is also represented in the second inscription, it seems that we must draw the conclusion that it is more likely that the Sarvavajrodaya (or a text closely related to it) than that the Sarvadurgatipariśodhanatantra inspired the rituals during which or with a view to whose execution these inscriptions were produced.Footnote ¹²¹

Although none of the ritual texts cited here prescribes the use of “stakes” (kīla) made of stone,Footnote ¹²² it nevertheless seems clear that these two stones served as stakes – possibly the central ones – during the ritual opening of the ground plan (maṇḍala) of Buddhist shrines. In this connection, given the physical resemblance of these stones to the rather numerous wholly or partly liṅga-shaped boundary markers preserved from Central Java, called “sīma stones” (vatu sīma) in contemporary inscriptions (cf. n. 99), it is significant that certain Indian inscriptions use precisely the word kīlaka to denote cadastral boundary markers.Footnote ¹²³ Obviously, the spheres of a ritual demarcation of agricultural land and the ritual demarcation of maṇḍalas were felt to be connected in ancient India and Indonesia, as is expressed by the use in these spheres of a partly overlapping set of terminological and material culture.

7. Inscriptions with the mantra ṭaki hūṁ jaḥ

Another case where the Sarvavajrodaya could be pertinent to Javanese data is furnished by a group of inscriptions containing mantras with the sequence ṭaki/ī hūṁ jaḥ. In an important study published in 2003, Jeff Sundberg drew attention to and elucidated a unique gold artefact excavated in the Ratu Baka complex near Prambanan, bearing the text oṁ ṭakī hūṁ jaḥ svāhā. He has done his best to find a scriptural source for this mantra. Unable to find a precise match, he pinned much of his argument on the attestation of admittedly very similar, although not identical, mantras in the Sarvatathāgatatattvasaṁgraha.Footnote ¹²⁴ However, oṁ ṭakī hūṁ jaḥ is attested as such in the Guhyasamājatantra (14.22), which Sundberg does not mention;Footnote ¹²⁵ perhaps more significantly, it is also found in the Sarvavajrodaya, whose ritual system is based on that of the Sarvatathāgatatattvasaṁgraha.Footnote ¹²⁶

Sundberg has pointed out that there are a few other, similar and obviously related stone inscriptions from Central Java, but in my opinion he was wrong to maintain, with his predecessors, that these inscriptions read paki instead of ṭaki, on which basis he concluded that “their mantra is not that of the Ratu Baka gold plate”.Footnote ¹²⁷ It is safe to assume that ṭaki is intended throughout,Footnote ¹²⁸ and the evidence suggests that the formula ṭaki hūṁ jaḥ at the beginning of three of these four short stone inscriptions recording the erection of boundary markers served the purpose of removing obstacles. For at Sarvavajrodaya section 56, we read (Mikkyō Seiten Kenkyūkai 1986–87: 267):

From the evidence of Guhyasamājatantra 14.22, where the mantra oṁ ṭakki hūṁ jaḥ is explicitly called sarvatathāgataṭakkirājamahākrodha “the great wrath of all Tathāgatas who is [named] Ṭakkirāja”, it seems that ṭakkirājenāk ṣya can be interpreted as “having adducted by reciting the mantra named Ṭakkirāja”, and that the intended mantra is the one that concerns us here.Footnote ¹³¹ The position of the mantra in the three stone inscriptions corresponds precisely to that of the more familiar expression avighnam astu “let there be no obstacles!” found in several Javanese inscriptions of the ninth century onwards; one of these even opens more elaborately with oṁ avighnam astu gaṇapataye namaḥ “Oṁ, let there be no obstacles! Homage to Gaṇapati (the god who removes obstacles)”.Footnote ¹³²

The fourth stone inscription with ṭaki jaḥ hūṁ comprises only a mantra, and has been read paki hūṁ jaḥ jaḥ hūṁ waho paki hūṁ jaḥ jaḥ hūṁ waho (that is, twice paki hūṁ jaḥ jaḥ hūṁ waho) by Damais, from a plaster cast kept in Leiden, the whereabouts of the original inscription being unknown. In the Kriyāsaṁgrahapañjikā (6.2.9, Inui Reference Inui1992: 139), in a section on the consecration of the Tantric master, I find oṁ ṭakki jaḥ hūṁ vaṁ hoḥ, which is virtually the same mantra and allows us to predict how Damais’ reading may be improved.Footnote ¹³³ I have not yet found this in any scriptural source, and alas we have in this case neither context nor original artefact to help us understand the function of this inscription. Finally, I must add to this record the mantric sequence ṭaki hūṁ jaḥ daḥ hūṁ kiṭa in the lead-bronze inscription from Borobudur already mentioned a few times above.Footnote ¹³⁴

8. Bronze image of Vairocana at the Metropolitan Museum of Art

A small bronze image of Vairocana in vajraparyaṅkāsana and displaying bodhyagrīmudrā is held at the Metropolitan Museum of Art in New York, under number 1987.142.171 (Figure 11). To my knowledge, it is so far unpublished,Footnote ¹³⁵ no doubt because the iconography is very well known from other, better preserved specimens. But the inscription engraved on the figure's base lends it importance from perspectives other than art history. The script is that of Central Java, i.e. Kawi, in the ninth or tenth century. Excellent photos are viewable online through the Museum's website, and allow the following reading:

Figure 11.

Bronze sculpture under inventory number 1987.142.171. Photo courtesy of the Metropolitan Museum of Art.

This is a surprisingly corrupt anuṣṭubh stanza, that I hesitantly restore as follows:Footnote ¹³⁶

If the emendations are correct, then we may translate as follows:

This is a very basic expression of the core Buddhist doctrine of dependent origination (pratītyasamutpāda). Although I am unable to trace this verse in any preserved scriptural text, it could very well have been put in the Buddha's mouth in some Buddhist scripture now lost (or no longer available in Sanskrit), for a gāthā much like this one, and likewise untraceable in known scriptures, is found attributed to the Buddha in Candrakīrti's Prasannapadā on kārikā 4.19 (de la Vallée Poussin Reference De la Vallée Poussin1903–13: 505):

I quote the translation by Jacques May (Reference May1959: 240):

It seems reasonable to assume that this gāthā, engraved on the base of the image of Vairocana, served as dharma-relic, just like the stanzas engraved on a silver foil inserted into the base of the Sambas sculpture discussed above (section 1) or the texts on gold foil or clay tablets inserted into relic caskets and stūpas (sections 5, 9 and 10). And in this sense, this inscription is comparable to the other material presented in this paper.

9. Clay sealings from Bali

In Machi Suhadi's article on “Buddhist mantras in the ASEAN states”, the author's “mantra formula II” (1989: 117), a text originally published by Stutterheim (Reference Stutterheim1929: 38), can be identified with a widespread Buddhist dhāraṇī. The text is found stamped in Siddhamātkā script on small clay sealings found at Pejeng and (not far from Pejeng) at Blahbatuh,Footnote ¹³⁷ inserted into miniature stūpas of unbaked clay. The reading published by Stutterheim was as follows:Footnote ¹³⁸

This text can now be identified as the Vimaloṣṇīṣadhāraṇī. The following is the version of that dhāraṇī in a Dunhuang manuscript dating to about the ninth/tenth century published by Cristina Scherrer-Schaub (Reference Scherrer-Schaub and Kvaerne1994: 712):

There are some differences between this version and the Balinese sealings, but it is in any case clear from Schopen's (Reference Schopen1985) study of this text that the transmission of this dhāraṇī throughout Buddhist Asia was not very stable.Footnote ¹³⁹ Moreover, it is only natural that Stutterheim's ignorance of the identity of the text, combined with the challenges of reading a text stamped in miniature size on this kind of material, has caused several misreadings. Luckily, a fairly good photograph showing two specimens of clay sealings stamped with the same matrix of this text was published by Putu Budiastra and Wayan Widia (Reference Budiastra and Widia1980–81: photo 9), which I reproduce here as Figure 12. On this basis, I am able to propose the following synthetic edition of the text as it was transmitted to Bali:

Figure 12.

Two sealings from Bali. Reproduced from Putu Budiastra and Wayan Widia Reference Budiastra and Widia1980–81, photo 9.

Finding myself somewhat in disagreement with the way Scherrer-Schaub has translated her reconstituted text for the Dunhuang manuscript (Reference Scherrer-Schaub and Kvaerne1994: 713), I propose here the following translation, starting from the Balinese text, which is not essentially different:

It is clear from the rich evidence adduced by Schopen (Reference Schopen1985) that the practice of stamping this dhāraṇī on clay and installing it in a stūpa is precisely in agreement with the way this text was used in other parts of the Buddhist world. The dating of these specimens is very hard to estimate on the basis of palaeography, but a date between 800 and 1000 ce seems likely.

More exceptional is the case of “mantra formula III” on the other clay object mentioned by Machi Suhadi (Reference Suhadi1989: 117–8). It was originally read as follows by Stutterheim (Reference Stutterheim1935: 88):

I do not have access to a photo of any sealing bearing this text, so I cannot verify Stutterheim's decipherment, but the text must be very similar if not identical to one of the many variants of the list of ten epithets of the Buddha.Footnote ¹⁴³ A particularly close (if not identical) parallel is the phrase ity api sa bhagavāṁs tathāgato ’rhan samyaksaṁbuddho vidyācaraṇasaṁpannaḥ sugato lokavid anuttaraḥ puruṣadamyasārathiḥ śāstā devamanuṣyāṇāṁ buddho bhagavān iti, found in the Bodhisattvabhūmi 1.7 (Wogihara Reference Wogihara1930–36: 93). It means something like this: “Thus also the Lord is [known as] Tathāgata, fully awakened, Arhant, endowed with knowledge and [good] conduct, the Sugata, Supreme knower of the world, Coachman of men who need to be tamed, Teacher of gods and men, the Buddha, the Lord”. This formula is in any case not a mantra.

10. A gold foil from Pura Pagulingan on Bali

Finally, the last section of Machi Suhadi's article (Reference Suhadi1989: 118–9) was dedicated to a number of inscribed artefacts recovered during the dismantling of the temple Pura Pagulingan in preparation for its reconstruction. It was discovered then that the foundations showed an octagonal groundplan. It was at this level that two pieces of gold foil were discovered, beside a terracotta tablet (bearing the ye dharmāḥ formula) and a silver foil bearing a few akṣaras.Footnote ¹⁴⁴ The first gold foil, rectangular in shape, was inscribed with a repeated ye dharmāḥ formula. Machi Suhadi also gives the text of the second gold foil, oval in shape, with three lines of writing in circles, referring in n. 31 to an unpublished report by Boechari (1985b). Having recently prepared the publication of this report from the author's own typescript (Boechari 2012: 526–8), I am able now to cite the reading precisely as Boechari saw it, without the copying mistakes in Machi Suhadi's publication:

This comprises five mantras, each introduced by o , of which the fourth and most extensive is clearly identifiable as that which is called the upah dayamantra of the Amoghapāśah dayadhāraṇī in the edition by Kimura (Reference Kimura1979: 245 [12]), where this mantra does not appear to hold a particularly important position. This edition, however, is based only on late Nepalese manuscripts, and is altogether much less useful for historical research than the two painstaking publications by Meisezahl (Reference Meisezahl1962, Reference Meisezahl1965), which present witnesses conserving the text in a considerably more archaic state. Several witnesses – a Dunhuang manuscript containing an old Tibetan recension (possibly dating to the ninth century), a twelfth-century Sanskrit manuscript held at Cambridge University Library, and a Siddhamātkā manuscript preserved in Japan, apparently representing an old transmission of the Sanskrit text independent from that in Nepal – do not show the labels homamantra, h dayamantra, upah dayamantra, etc., identifying the ritual use of various constituent parts of the dhāraṇī. Furthermore, in both the Dunhuang and the Japanese manuscript, we find the mantra corresponding to this Balinese inscription precisely at the end of the dhāraṇī, without any of the added mantras that all (?) other manuscript witnesses show.Footnote ¹⁴⁵ In other words, in the oldest versions of the text, the mantra that we see in Bali was the climax of the mantra portion, and hence it is not surprising that it would be chosen to stand for the whole. Also, the various manuscript witnesses show substantial variants among themselves for this sequence of mantras at the end of the dhāraṇī, so it seems unproblematic that the Balinese version is not identical to any other known version. Anyhow, now that the identity of this mantra has been determined, it must be kept in mind that the readings of this gold foil inscription from Bali are liable to be improved if the original could be inspected or good reproductions be made available.

Even though we do not know the precise archaeological context in which this gold foil was found, the function of this artefact may again rather clearly be inferred from the frame text of the dhāraṇī (Meisezahl Reference Meisezahl1962: 313):

This is the translation given in Schopen (Reference Schopen1976 [Reference Schopen2005]: 36):

We may hence assume that our inscription represents the epitome of the dhāraṇī, and its production must have been motivated by a similar intent to the inscription of Batu Bedil presented in section 2. In contrast to the Aparimitāyuḥsūtra, which was the source text there, it must however be pointed out that there is not, in the Amoghapāśah daya, any emphasis on “writing down” the text.