CANDI BADUT AND THE TRAJECTORY OF HINDU-JAVANESE TEMPLE ARCHITECTURE

ROY E. JORDAAN

CANDI BADUT AND THE TRAJECTORY OF HINDU-JAVANESE TEMPLE ARCHITECTURE

INTRODUCTION

In a suburb of Malang, a city in East Java, lie the ruins of a small temple complex known of old as Candi Badut, named after the village long since annexed by the sprawling city. Soon after its discovery in 1923, B. de Haan, a Dutch architect working for the Archaeological Service of the Netherlands East Indies, visited the site and reported that “all the fragments are pure central Javanese in character, [thereby] demonstrating that the ruins are very old.”1 Two years later the site of the Badut temple was cleared of trees and other natural overgrowth and systematically excavated, which led to a plan to reconstruct the main temple, albeit partially. This partial reconstruction, finished in 1926, can be seen in a photograph taken by the Archaeological Service (fig.1). The reconstruction drawing in the final archaeological report on Candi Badut gives a fair picture of the temple in its original state (fig.2).

In this article, I am less interested in the reconstruction of the temple than in the problem of explaining the presence of this Central Javanese–style temple so far away in eastern Java. Obviously, this question can only be answered if we know when, albeit approximately, Candi Badut was built. The archaeological literature shows that whereas there is no disagreement among archaeologists and art historians about Candi Badut’s Central Javanese character, the consensus on the founding date took more time to develop. Following the opinion of its talented restorer, de Haan, most archaeologists assigned the temple to an early phase of Central Javanese temple architecture, in the second half of the eighth century. Others have agreed with this dating, but ventured the suggestion that the temple was structurally altered in the thirteenth century. Only one art historian, E.B. Vogler, has objected to the early dating, contending that Candi Badut was built towards the end of the Central Javanese period, around 900 CE.2 But Vogler’s ideas were rejected by R. Soekmono, who restated the hypothesis of Candi Badut’s eighth-century origins, a theory which is now widely accepted. In my opinion, however, the current archaeological consensus is ill-founded. With the aid of hitherto neglected evidence I shall demonstrate that Vogler’s arguments were prematurely rejected. As for the relevance of this exercise, it may suffice to cite Vogler’s remark: “It goes without saying that it is of utmost importance for our knowledge of Javanese art to establish what views on this matter should be considered correct.”3

1 B. de Haan, Oudheidkundig Verslag 1923 (Weltevreden/’s-Hage: Koninklijk Bataviaasch Genootschap van Kunsten en Wetenschappen [hereafter KBG], 1924), 87; cf. Oudheidkundig Verslag 1925 (Weltevreden/’s-Hage: KBG, 1926), 9. 2 E. B. Vogler, “De stichtingstijd van de tjandi’s Gunung Wukir en Badut,” Bijdragen tot de Taal-, Land- en Volkenkunde 108 (1952): 313–46. 3 Ibid., 313.

REVIEW OF THE DISCUSSION ON CANDI BADUT’S FOUNDING DATE

After mentioning a number of architectural and stylistic similarities between Candi Badut and several Central Javanese temples, de Haan raised the question of Candi Badut’s antiquity. Recalling the find of the Dinoyo inscription in a nearby village, he queried whether Candi Badut could be linked with the sanctuary mentioned in that inscription. He admitted that there is a problem in doing so, as the inscription concerns the consecration of a statue of the Hindu sage Agastya, whereas the central object found in situ in the main chamber of Candi Badut is a stone lingga, a phallic symbol of ·iva. Furthermore, Candi Badut was not the only ancient monument found in the area: the remains of Candi Besuki lie less than a kilometer away. Still, de Haan argued, while Candi Badut cannot as a matter of course be connected with the edifice alluded to in the Dinoyo inscription, “the year 682 ·aka [760 CE] mentioned therein is not without significance for determining the founding date of this temple.”4 Referring to the analysis of the Dinoyo inscription by F.D.K.Bosch,5 who had suggested that the inscription might stem from the same dynasty which some thirty years earlier had issued the charter of Canggal in Kedu in Central Java, de Haan claimed that Candi Badut cannot therefore be much older than 760. In his opinion, this dating is supported by the style of the monument, “the language of which cannot be misunderstood.”6 He conjectured that descendants of this old Hindu-Javanese settlement in eastern Java returned to the central part of Java about a century later, where they founded the temples of Prambanan and other Central Javanese monuments. “If this hypothesis is correct,” de Haan continued, “the founding date of Candi Badut can then be fixed between 670 [and] 785 ·aka, which would allow us to rank Badut not only among the oldest monuments of eastern Java, but even of the whole of Java.”7

As said, de Haan’s views were accepted by most of his contemporaries, including the leading trio in colonial Dutch East Indies’ archaeology: N.J.Krom, F.D.K.Bosch, and W.F.Stutterheim. In his earlier analysis of the Dinoyo inscription, Bosch had classed Candi Badut among the very oldest specimens of Hindu-Javanese art “in view of the remarkable similarities between the makara, lion-ante fixes, and other ornamental elements, and the art of Diëng, Kalasan, and Borobudur.”8 For his part, Krom speculated about a possible relationship between the information provided by the Canggal and Dinoyo inscriptions, and Chinese records reporting the Javanese capital’s eastward transfer between 742 and 755, which led him to propose the hypothesis that the Dinoyo inscription had been issued by

4 B. de Haan, Oudheidkundig Verslag 1929 (Weltevreden: KBG, 1930), 247.

5 F.D.K.Bosch, “De Sanskrit-inscriptie op den steen van Dinaja (682 Çaka),” Tijdschrift voor Indische Taal-, Land- en Volkenkunde57(1916): 410–44; “Het lingga-heiligdom van Dinaja,” Tijdschrift voor Indische Taal-, Land- en Volkenkunde 64 (1924): 227–86.

6 De Haan, Oudheidkundig Verslag 1929, 248.

7 Ibid., 248.

8 Bosch, “Het lingga-heiligdom van Dinaja,” 285. Considering the extent of the territory covered, this grouping by Bosch of Badut, Diëng, Kalasan, and Borobudur may surprise, but should be understood in the context of the time when the chronology of Hindu-Buddhist monuments was uncertain. The early dating of these monuments on the basis of similarities between their ornamental elements was led into error by the linkage of the Kalasan inscription of 778 with the present Kalasan temple building instead of with an earlier structure that was later found to be encased in it (see the main text).

a group of settlers related to those of the Canggal inscription, but who had moved to eastern Java for political reasons, such as the arrival of the Buddhist ·ailendra dynasty.9 In this connection, he also referred to de Haan’s surmise that Candi Badut was built by the same community that had issued the Dinoyo inscription. Without taking a clear stance in this matter himself, Krom did note that “it was very important in connection with the later evolution of East Javanese art and culture, as distinct from that of central Java, to know whether monuments such as Badut and Besuki originated from the same source as Dinoyo, which would allow us to assume that the beginning of Javanese art history was grounded in the same Hindu-Javanese civilization.”10 Stutterheim, finally, considered Candi Badut to be one of the oldest Javanese buildings, along with Candi Gunung Wukir, which he referred to as “monuments dating from 732.”11

In 1952, E.B.Vogler challenged all the above arguments in an article specifically devoted to the founding date of the Gunung Wukir and Badut temples.12 As to the architectural and stylistic elements which de Haan and Bosch had mentioned in favor of their early dating, Vogler observed that none of these elements offers a reliable indicator of Candi Badut’s position in the evolution of HinduJavanese monumental art. As for identifying the excavated temple of Gunung Wukir as the sanctuary of the Canggal inscription, Vogler noted that Stutterheim had not adduced any argument other than the fact that this inscription had been found about one hundred and fifty meters south of the excavated complex. Reminding us of what had happened with the Kalasan inscription of 778, which had for a long time been linked with the temple structure we see today (Kalasan III) rather than with the first of two older buildings encased in it, Vogler cautioned against a rash identification of the Canggal inscription’s sanctuary as the ruins of the Gunung Wukir temple.13 The remainder of Vogler’s article concerns his art-historical research on the evolution of the so-called kâla makara ornament framing the entrances and niches of Javanese temples.14 The aim of this narrowly circumscribed yet complicated research was to determine the chronology of the stylistic changes in the kâla makaraornament, on the basis of which he concluded that both Candi Badut and Candi Gunung Wukir had to be assigned to the later phase of Central Javanese temple architecture.

It is relevant to note that the same year in which Vogler’s article appeared in print, the Old Java scholar R.Ng. Poerbatjaraka published his lecture notes in a stencilled paper with the title Riwajat Indonesia (History of Indonesia), in which he elaborated on Krom’s idea of associating Dinoyo with the Chinese reports on the transfer of the Javanese capital between 742 and 755.15 Poerbatjaraka spec-

9 N.J. Krom, Hindoe-Javaansche Geschiedenis, 2nd ed. (The Hague: Nijhoff, 1931), 146–48. 10 Ibid., 161.

11 W.F. Stutterheim, Oudheidkundig Verslag 1940 (Batavia: KBG, 1941), 12.

12 Vogler, “De stichtingstijd van de tjandi’s Gunung Wukir en Badut.” 13 Ibid., 315.

14 See also E.B.Vogler, De Monsterkop in de Hindoe-Javaanse Bouwkunst (Leiden: Brill, 1949), and “Ontwikkeling van de gewijde bouwkunst in het Hindoeïstische Midden-Java,” Bijdragen tot de Taal-, Land- en Volkenkunde 109 (1953):

249–72. 15 R.Ng.Poerbatjaraka, Riwajat Indonesia-djilid 1 (Djakarta: Fakultas Sastra Universitas Indonesia, 1952); cf. Agastya in den Archipel (Leiden: Brill, 1926), 109–10; “Çrîvijaya, de Çailendra- en de Sañjayawamça,” Bijdragen tot de Taal-, Landen Volkenkunde 114 (1958): 254–64; G. Coedès, The Indianized States of Southeast Asia (Honolulu: University of Hawai‘i Press, 1968), 90.

ulated that a certain Chi-yen, who in Chinese sources was held responsible for moving the Central Javanese capital “to the East,” was the same as Gajayâna, the ruler who in the Dinoyo inscription of 760was said to have moved his residence seemingly in response to an internal political conflict. According to Poerbatjaraka, the conflict arose from the alleged conversion to Mahâyâna Buddhism of Rakai Panaraban, alias Rakai Panangkaran, a Central Javanese ruler. Resenting this conversion, fanatical

·aivas supposedly migrated to eastern Java. These refugees founded the eastern Javanese kingdom of Kañjuruhan, near the present-day city of Malang. It was Gajayâna who issued the Dinoyo inscription of 760 commemorating the replacement of a wooden statue of Agastya with one made of black marble. Poerbatjaraka presumes that Gajayâna may also have been the patron-builder of Candi Badut and Candi Besuki.

The arguments put forward by Vogler for dating Candi Badut to the very end of the Central Javanese period instead of the early beginnings, remained unanswered until 1979 when they were briefly discussed by R. Soekmono in his influential article on the archaeology of Central Java before 800 CE.16 Unfortunately, Soekmono’s review is partly beside the point, as is evident from his remark about Vogler’s reminder to be careful in linking the Canggal charter to the structural remains found at the same site. His comment was that “Vogler’s warning against Stutterheim’s identification of Candi Kalasan I with the Kalasan Charter is not quite clear,” a statement followed by detailed information on the location where the Canggal inscription had been found, to demonstrate that the inscription hailed from the very place where the temple had been built, something which Vogler had not questioned. What Vogler had warned against, though, was precisely what Soekmono did, namely to conclude from this “that we now have good reason to assume that Candi Gunung Wukir was founded in 732.”17 However, as was precisely borne out by Kalasan, the inscription’s location does not suffice for this conclusion. In order to qualify for this, the information concerning the site should have been supplemented by art-historical arguments demonstrating that the present building of Candi Gunung Wukir (or the little that is left of it) is indeed the same as the one described in the Canggal inscription. Aside from this, Soekmono could have tried to refute Vogler’s theory by pointing out invalidating flaws, if any, in the art-historical analysis of Candi Gunung Wukir’s and Candi Badut’s kâla makara ornaments, which had led Vogler to assign a late date to the construction of these temples in the first place.18 Indeed, Vogler had claimed that the two temples in question were younger than Candi Prambanan.19

16 R. Soekmono, “The Archaeology of Central Java before 800 A.D.,” in Early South East Asia: Essays in Archaeology, History, and Historical Geography,ed. R.B. Smith and W.Watson (New York and Kuala Lumpur: Oxford University Press, 1979), 457–73. In this review, I have ignored appraisals of Vogler’s views which are not supported by arguments or evidence, for instance those by Louis Frédéric, who opined, “Bien que certains auteurs (M.E.B.Vogler) pensent qu’il est postérieur au Lara Djonggrang, je serais porté à croire que, en régard à sa décoration, sa fondation pourrait au contraire dater aux débuts du IXe siècle,” in Sud-Est asiatique: Ses temples, ses sculptures (Paris: Arts et Métiers Graphiques, 1964), 158–59.

17 Ibid., 462.

18 It is not altogether clear why Soekmono did not review Vogler’s article on the founding dates of Candi Badut and Candi Gunung Wukir, except for his remark that “the dating of a monument on the basis of only one single ornamental design has too many weak points”; Soekmono, “The Archaeology of Central Java,”

460. However, he fails to Soekmono also seems to have underestimated the problems in linking the Dinoyo inscription to Candi Badut. Although he admitted that the site of the Dinoyo inscription so far from the temple is not quite convincing and that other temples in the vicinity, such as Candi Besuki, could also lay claim to the Dinoyo charter, Soekmono stated, “In spite of all that, the purely square plan of Candi Badut, and its entirely plain base, are not comparable with any other specific features than those found at Candi Kalasan I and Candi Gunung Wukir. Therefore, if the year 760 is not applicable to the installation of the Agastya statue of Candi Badut, the middle of the eighth century can certainly be assumed as the period which witnessed the foundation of the monument.”20 However, whereas the square plan and plain base are perhaps reliable indicators of Candi Badut’s Central Javanese character, they do not justify dating the foundation to the middle of the eighth century. Indeed, according to Daigoro Chihara, “it is wrong to infer that a structure will always be of an early date just because it has a flat-type podium.”21Vogler’s dissenting opinion has made it clear that the early dating of Candi Gunung Wukir is far from certain.22 Thus what remains for comparison is Candi Kalasan, and this sole example simply offers too slender a basis for deciding this important issue.

This brings me to the point that Soekmono neglected to support his early founding date of Candi Badut with other architectural and art-historical evidence. Though he had twice regretted the fact that the encased remnants of Kalasan I and II, and also Candi Gunung Wukir, lack their superstructures so that no more architectural elements could be included in his study, he failed to notice that the remaining walls at Candi Badut do offer ample opportunities for an art-historical comparison of its ornamentation with other Central Javanese temples. Such an analysis certainly would have enabled him to test his conclusions. I will return to this matter shortly.

In his contribution to the catalogue of a 1990 exhibition of Indonesian sculpture held at the National Gallery of Art in Washington, D.C., Soekmono repeated his opinion that “[i]n all probability this temple is connected with an inscription datable to the year 760.” However, he now observed do justice to the fact that the kâla makara ornament was analyzed as a composite item comprising distinct features (called kenmotieven by Vogler). Soekmono’s decision not to follow Vogler’s hypothesis, which is justified as far as Candi

Kalasan III’s dating to 778 CE is concerned, cannot simply be extended to Vogler’s later dating of Candi Badut and Candi Gunung Wukir. In his separate article on these temples, Vogler had in the meantime adjusted his chronology in light of the more recent findings, both with respect to Kalasan and the political division of Central Java during the reign of the ·ailendras. Therefore Vogler’s revised chronology should be re-evaluated, but that is beyond the scope of this article.

19 Vogler, “Ontwikkeling van de gewijde bouwkunst in het Hindoeïstische Midden-Java,” 256–57. It should be noted that at the time of his writing, Candi Prambanan was generally dated to the first half of the tenth century.

20 Soekmono, “The Archaeology of Central Java,” 462–63.

21 Daigoro Chihara, Hindu-Buddhist Architecture in Southeast Asia (Leiden: Brill, 1996), 108. Chihara mentions Candi Pringapus and Candi Sambisari as counterexamples of Soekmono’s thesis that a flat-type podium invariably implies an early date of construction. Moreover, he is merely prepared to provisionally endorse Soekmono’s dating of Candi Badut “in view of the fact that at the present point in time we know almost nothing about the history of eastern Java during the Central Javanese period owing to a dearth of material other than the Dinoyo inscription and some Chinese accounts.” Ibid, 108–9.

22 See also N.J.Krom, Inleiding tot de Hindoe-Javaansche Kunst (’s-Gravenhage: Martinus Nijhoff, 1923), 167, who allows for the possibility that the ruins belong to a later renovation of the sanctuary.

that “[t]he carved floral decoration on the walls of the body of the temple and the kâla-makara framing the niches and the doorway display a stylistic development more advanced than what we see at Diëng and Gedong Songo.” This led him to assign a “slightly later date” to Candi Badut.23 Conceivably, Soekmono’s adjustment was inspired by Joanna Williams’s essay on the date of Boro budur in relation to other Central Javanese monuments, which appeared in a book on Borobudur with his foreword. Even though Candi Badut was excluded from Williams’s analysis, she did note that “[t]ypologically, Badut belongs close to Kalasan’s final phase and to Pringapus, for it has round moldings, finials transitional between 2 and 3, and kâla heads without lower jaws, but with incipient hands, and apparently no dentils.”24 She also remarked that the temples of Badut and Songgoriti “raise questions of the degree of correspondence between East and Central Java.”25

Other archaeologists and art historians, however, had no such questions and presented Soekmono’s early dating of Candi Badut as an established fact. For instance, John Miksic reports that Candi Badut is much older than most of the other east Javanese temples: “This temple, dedicated to Siva, was built in 760, during the earliest period of stone architecture in Java, but was drastically altered during the 13th century.”26 He does not mention his source for these alterations, but presumably it was The Temples of Java by Jacques Dumarçay.27

Perusal of this popular booklet and several scholarly publications by the latter author makes clear that his hypothesis about the late-eighth-century founding date of Candi Badut is mainly based on the received view that the temple was built by the same dynasty that had issued the Dinoyo inscription of 760.28 See, for instance, his statements that “the construction must have taken place in about 760, the date of the Dinoyo inscription,”29 and that “[a]t the end of the eighth century, several tem-

23 R. Soekmono, “Indonesian Architecture of the Classical Period: A Brief Survey,” in Jan Fontein, The Sculpture of Indonesia, with essays by R. Soekmono and Edi Sedyawati (Washington D.C.: The National Gallery of Art, and New York: Abrams, 1990), 69. This “slightly later date” is not specified but, in line with Soekmono’s 1979 analysis that dates the Diëng and Gedong Songo temples to c.730–c.800 CE, we must assume that the founding of Candi Badut should be dated to around 750 CE to be able to connect it with the Dinoyo inscription of 760. To my knowledge, Soediman was the first to cast doubt on Soekmono’s chronological classification of Javanese temples in a paper on Candi Sambisari presented at an Indonesian archaeological conference; Soediman, “Candi Sambisari dan masalah-masalahnya,” Pertemuan Ilmiah Arkeologi di Cibulan, 21–25 February 1977.

24 Joanna Williams, “The Date of Barabudur in Relation to Other Central Javanese Monuments,” in Barabu∂ur: History and Significance of a Buddhist Monument, ed. Luis O. Gómez and Hiram W. Woodward Jr. (Berkeley: The Regents of the University of California, 1981), 45. Going by the positions of Kalasan and Pringapus in her time chart, which summarizes the chronology of the monuments of central Java (fig.4, p.39), this would seem to imply a dating for Badut to around 850 CE. This is a little earlier than Loro Jonggrang, which she dates to 850–900 CE.

25 Ibid., 45

26 John Miksic, “Ancient Sites in the Brantas River Basin,” in Java, ed. Eric Oey (Singapore: Periplus, 1991), 326.

27 Jacques Dumarçay, The Temples of Java (Singapore: Oxford University Press, 1986), 19.

28 See also Jacques Dumarçay, “Gubahan arsitektur di Jawa Timur,” Amerta 9 (1984): 7–13; Le savoir des ma∆tres d’œuvres Javanais aux XIIIe et XIVe Siècles (Paris: Publications de l’École Française d’Extr◊me-Orient, 1986), 5–8; Histoire de l’architecture de Java (Paris: Publications de l’École Française d’Extr◊me-Orient, 1993), 60; Cultural Sites of Malaysia, Singapore, and Indonesia (Kuala Lumpur:

Oxford University Press, 1998), 74; “Le Candi Badut,” Archipel 63 (2002):

7–14; Architecture and Its Models in South-East Asia (Bangkok: Orchid Press, 2003), 68–72.

29 Dumarçay, Histoire de l’Architecture de Java, 60.

ples were built in the region of Malang (dated to 760, after the so-called Dinoyo inscription), contemporary with the monuments of the Dieng and Gedong Songo….”30 Regarding the evidence for the thirteenth-century structural alterations, which actually falls outside the scope of this article, Dumarçay has so far adduced some evidence of partial reconstructions of the staircase and the base, but provided nothing to support his claim that “this shrine must have been considerably altered much later, in the thirteenth century, and this no doubt caused it to collapse.”31 Whether these partial reconstructions of the staircase and the base can really be assigned to the thirteenth century must be doubted. Verification, however, is well-nigh impossible because of the fact that Dumarçay does not support his claims with datable evidence such as inscriptions, coins, shards of Chinese ceramics, and the like.32 Earlier, de Haan had also noted that part of the relatively wide base is a later extension (een later aanbouwsel, “a later annex”), but lacking further clues, prudently did not offer any

explanation regarding the date.33 Unless datable evidence shows otherwise, the extension of the base could have been carried out at the end of the ninth or beginning of the tenth century, shortly after the completion of the structure. Perhaps this reconstructive work was prompted by the builders’ discovery of some architectural flaw or was carried out after an earthquake. An important consideration is that no Central Javanese inscriptions more recent than 928 – nor other signs of Central Javanese royal involvement in eastern Javanese affairs – were found.34 As A.M. Barrett Jones confirms, “After 928 A.D. the inscriptions and monuments found in Java without exception come from East Java and not Central Java.”35This implies that whatever there is to be seen at the Badut temple site was put in place before that date. Those conversant with the literature know that the most plausible explanation for the sudden socio-cultural break between Central and East Java is the devastating eruption of Mount Merapi around 928.36Whatever caused this discontinuity, there is little to suggest that the modifications perceived by Dumarçay really date from the thirteenth century and that they concern structural adaptations connected with fundamental changes in ritual, as the architect would have us believe. Dumarçay’s casual remarks about “ritual changes,” a new “wave of Hindu influence,” and “refashioning” cannot be accepted with-

30 Dumarçay, “Le Candi Badut,” 7.

31 Dumarçay, The Temples of Java, 19.

32 This problematic aspect of his work has also been detected with respect to other Hindu and Buddhist temples in Java. For instance, Soekmono noted that his “technical studies are not always in conformity with, and are sometimes in contradiction to, the results from historical and archaeological research.” See Jacques Dumarçay, Candi Sewu dan Arsi tektur Bangunan Agama Buddha di Jawa Tengah/Candi Sewu and Buddhist Architecture of Central Java (Jakarta: KPG, 2007), 13 and 175. For specific examples, see Jeffrey Roger Sundberg, “Considerations on the Dating of the Barabudur Stûpa,” Bijdragen tot de Taal-, Land- en Volkenkunde

162, 1 (2006): 105–6; Roy E. Jordaan, “Be˘lahan and the Division of Airlangga’s Realm,” Bijdragen tot de Taal-, Land- en Volkenkunde 163, 2/3 (2007): 340–1.

33 De Haan, Oudheidkundig Verslag 1929, 248–49.

34 Krom, Hindoe-Javaansche Geschiedenis, 196.

35 A.M.Barrett Jones, Early Tenth Century Java from the Inscriptions, Verhandelingen van het Koninklijk Instituut voor Taal-, Land- en Volkenkunde 107

(Dordrecht: Foris, 1984), 1.

36 M. Boechari, “Some Considerations of the Problem of the Shift of Mataram’s Center of Government from Central to East Java in the 10th Century A.D.,” in Early South East Asia, ed. Smith and Watson, 473–93; cf. Roy E. Jordaan, “Ecological Factors in the Transfer of the Seat of Government from Central to East Java in the Tenth Century,” Indo nesian Environmental History Newsletter 9 (1997): 1–8. See also Caesar Voƒte and Mark Long, Borobudur: Pyramid of the Cosmic Buddha (New Delhi: D.K. Printworld, 2008), especially 50–60.

out further specification and pertinent religious-historical evidence.37 In any event the alleged thirteenth-century structural changes of pediment, false storeys, and niches do not tally with the art-historical evidence of the decoration and ornamentation on the temple’s outer walls (discussed further below). Moreover, Dumarçay does not demonstrate that the alleged thirteenth-century addition of two miniature buildings to the three on each side of the pseudo-storey was meant “to bring their total number to five, in accordance with the current practice of that epoch.”38

More recently, art historian Marijke Klokke has included Candi Badut and Candi Songgoriti among a few temples in East Java that clearly date from the Central Javanese period. She states, “Their style is so very similar to the Central Javanese style that they may be called Central Javanese in style.”39 She does not mention to what phase of the Central Javanese period the temple should be assigned – the early phase or the final one. Elsewhere, she even suggests that Candi Badut was “by chance” located in east Java.40

Lastly, the present East Javanese provincial government has included Candi Badut in its online list of important tourist sites and objects and connected the temple with the ancient kingdom of Kañjuruhan mentioned in the Dinoyo inscription. Assigned a founding date in the late eighth century, Candi Badut is said to have been constructed “at the time the ·ailendra kings ruled in Central Java,”41 which is in line with Poerbatjaraka’s historical reconstruction mentioned above.

NEW DATING ARGUMENTS

There are three reasons calling for a reconsideration of the current consensus on the founding date of Candi Badut. The first has to do with W.J. van der Meulen’s neglected reinterpretation of the Dinoyo inscription of 760, which includes a well-argued rejection of the theory that the inscription was linked with the Central Javanese capital’s eastward transfer between 742 and 755, as presumed on the basis of ancient Chinese reports.42 After a close reading of the Old Javanese text, van der Meulen plausibly

37 See, for instance, Dumarçay, Le savoir des ma∆tres d’œuvre Javanais, 6; Cultural Sites of Malaysia, Singapore, and Indonesia, 74; architecture and Its Models in South-East Asia, 72.

38 Dumarçay, “Gubahan arsitektur di Jawa Timur,” 8. As one anonymous reviewer, who corrected my interpretation of this question, noted, “Whether Dumarçay was right about this part of a thirteenth-century makeover of Candi Badut, we may not know because Plate 1 shows that the twentieth-century partial reconstruction did not proceed above the decorated base of the roof structure, while the reconstruction drawing (Plate 2) shows only three bangunan miniatur [miniature buildings] on each pseudo-storey.”

39 Marijke J. Klokke, “Candi Gunung Gangsir, a Unique Temple in East Java,” in Fruits of Inspiration: Studies in Honour of Prof. J.G. de Casparis, ed. Marijke J. Klokke and Karel R. van Kooij (Groningen: Forsten, 2001), 221, n.8.

40 Marijke J. Klokke, “Ornamental Motifs: The Stylistic Method Applied to Ancient Javanese Temple Art,” in Southeast Asian Archaeology 1998, ed. Wibke Lobo and Stefanie Reimann (Hull: Centre for South-East Asian Studies, The University of Hull, 2000), 96.

41 Online at www.eastjava.com/books/glorious/html/badut/html (accessed January 2003).

42 W.J.van der Meulen, “The puri Pûtikeªvarapâvita and the pura Kâñjuruhan,” Bijdragen tot de Taal-, Land- en Vol ken kunde 132 (1976): 445–62. Van der Meulen’s new reading was facilitated by de Casparis’s epigraphical re-examination of the inscription; J.G.de Casparis, “Nogmaals de Sanskrit-inscriptie op den steen van Dinojo,” Tijdschrift voor In dische Taal-, Land en Volkenkunde 81 (1941): 499–513.

argues that the religious conflict hinted at in the Dinoyo inscription was one between Viß∑uism and ·aivism, not between Buddhism and ·aivism as most authors (myself included) had assumed. Thus he undermined Poerbatjaraka’s hypothesis about a connection with Rakai Panangkaran’s alleged conversion to Buddhism or with the arrival of the ·ailendra dynasty. Subsequently, and following the opinion of the linguists R. Kern and L.-Ch. Damais (who both had separately argued that the “Chi-yen” in the Chinese reports was an incomplete transcription of the Javanese title [[[Ra]]]kryan), van der Meulen proposed identifying Chi-yen as Sañjaya, the rakai (rakryan) of Mataram. He argued that the contention of Poerbatjaraka, Krom, and others that Chi-yen had to flee because of a religious controversy or the inimical pressures by the Buddhist kingdom of ·rîvijaya, is “simply preposterous. The whole setting of both the Javanese and Chinese evidence is one of expansion and increasing power. There is, moreover, no intimation in the Chinese communication that the move was precipitous or forced or that the new kraton-city [capital] was in any way inferior in stature to the kraton-city of Jawa. It was probably therefore a strategic move to a more central site, or, perhaps,

the planning of a new start….”43 In other words, the transfer of the capital “to the East” could just have been a move to another place in Central Java itself, perhaps resembling the later transfer of royal palaces of the principalities of Surakarta and Yogyakarta. In any case, the severance of the hypothetical link between the Dinoyo inscription and the move of the Central Javanese capital to eastern Java implies that a mid-eighthcentury community of Central Javanese settlers in the area can no longer be taken for granted and first needs to be substantiated. To admit that Candi Badut is not linked to the Dinoyo inscription, while maintaining the assumption that the temple was built in the eighth century, is not acceptable either, because it conveys the impression that the temple was built by an East Javanese ruler without explaining how he had recruited the Central Javanese architects and artisans.44 Consequently, the dating of Candi Badut to the second half of the eighth century is no longer tenable, and its construction should be assigned to a later century after all.

Obviously, Candi Badut’s location in East Java cannot sensibly be related to mere “chance.” It should, in my opinion, be explained by the eastward expansion of Mataram at the end of the ninth century and the beginning of the tenth. The expansion can be inferred from the fact that several royal edicts from this period come from the area of Ka∂iri, in eastern Java, yet also include references to Central Javanese dignitaries and to the old capital Meˇ∂ang in Central Java.45 Krom concluded that several contemporary Javanese rulers, such as Balitung, Dakßa, Tulo∂ong, and Si∑∂ok, had ruled over both parts of the island. The last mention of Central Java in Si∑∂ok’s inscription of 929 refers to Meˇ∂ang in the past tense, namely as the seat of his deified predecessors, which is probably related to the destruction of the capital or its abandonment after the eruption of Mount Merapi around 928, mentioned above.

43 W.J.van der Meulen, “King Sañjaya and His Successors,” Indonesia 28 (1979): 19.

44 The question remains as to whether there is some other monument that could date from 760. In my opinion, we should reckon with the possibility that the temple was lost. Considering the inscription’s emphasis on replacing the wooden statue of Agastya by one made of stone, it cannot be excluded that the temple itself was a wooden structure.

45 Krom, Hindoe-Javaansche Geschiedenis, 184–96.

The second reason to challenge the established archaeological opinion regarding Candi Badut is inspired by my visit to the temple in 1995, after attending the Eighth Indonesian Archaeological Conference (PIA-VIII) in Yogyakarta. My personal inspection of this monument offered me a closer look at various ornamental designs that had remained vague and indistinct in the old photographs of the Archaeological Service. They showed that Soekmono’s research had not been complete, focused as it initially was on the temple’s square plan and the plain base. Leaving aside the style of the kâla makara ornament, which had been the object of Vogler’s detailed analysis, I discovered several other ornaments and reliefs that also seemed difficult to square with both de Haan’s and Soekmono’s early dating, and Dumarçay’s thirteenth-century architectural alterations. Perhaps the most convincing example is provided by the so-called behangsel motief (wall-paper motif), also known as the “triªûla cakra ornament” or “full floral relief.”46 Reliefs with this motif adorn the outer walls of Candi Badut, as in the panels on either side of the niche with the statue of Durgâ Mahißâsuramardinî (fig.3).

The presence of this motif on Hindu and Buddhist monuments (figs.4, 5) is, in my opinion, difficult to reconcile with de Haan’s original theory about Candi Badut having been founded by Hindu refugees who had migrated to eastern Java to withdraw from ·ailendra-Buddhist influences. If the structure was really built by “fanatical ·aivas” who resented the conversion of their homeland to Buddhism, as Poerbatjaraka suggested, they probably would not have used elements reminiscent of Buddhist monumental art. A more likely explanation is that no antagonism existed between Hinduism and Buddhism during the Central Javanese period, as is also suggested by the close proximity of, and the numerous architectural and stylistic resemblances between, Hindu and Buddhist temples in the Prambanan plain, such as Candi Prambanan (also known as Candi Loro Jonggrang), Candi Sewu, and Candi Plaosan. Indeed, the architectural and stylistic resemblances are so conspicuous as to cast doubt on the validity of the conflict theory of de Casparis and Dumarçay, which has dominated the archaeo- logical discourse since the early 1960s. In my opinion, the resemblances can be more satisfactorily explained in terms of peaceful religious co-existence and cooperation.47

As art historian Mary-Louise Totton points out in her doctoral thesis on the Prambanan temple complex, the triªûla cakra ornaments or full floral reliefs are found on the interior walls of the main chamber of the ·iva temple of Candi Prambanan and on the outer walls of Candi Sewu, Candi Plaosan Lor, Candi Sambisari, Candi Mendut, and Candi Bagong, which were “all built or revetted during the mid-ninth century.”48In a subsequent article, Totton takes the ornamental designs for representations of luxury cloth and adds the names of Candi Badut and Candi Barong to the list of temples which have

46 F.D.K.Bosch, De Gouden Kiem: Inleiding in de Indische Symboliek (Amsterdam: Elsevier, 1948), pl. 57; The Golden Germ (The Hague: Mouton, 1960), pl.60.

47 The theory on the peaceful co-existence of the Indian religions and the mutual exchange of stylistic elements – defend ed in Roy E. Jordaan, Imagine Buddha in Prambanan: Reconsidering the Buddhist Background of the Loro Jonggrang Temple Complex (Leiden: Vakgroep Talen en Culturen van Zuidoost-Azië en Oceanië, 1993) – has recently been discussed by art historian Bo-Kyung Kim in “Indefinite Boundaries: Reconsidering the Relationship between Borobudur and Loro Jonggrang in Central Java” (Ph.D. diss., University of California, Los Angeles, 2007).

48 Mary-Louise Totton, “Weaving Flesh and Blood into Sacred Architecture: Ornamental Stories of Candi Loro Jonggrang” (Ph.D. diss., University of Michigan, 2002), 282, n.26.

Fig. 1 Candi Badut, northwest corner after the reconstruction. Photo: Oudheidkundige Dienst in Nederlandsch-Indië, 1926 (O.D. 8324).

Fig. 2 Reconstruction drawing of Candi Badut’s main temple building. The specified ornamental elements are among the temple’s most conspicuous Central Javanese features, apart from the tiered roof as a whole. Adapted from B. de Haan, Oudheidkundig Verslag 1929 (Weltevreden: KBG, 1930), pl. 6.

Fig. 3 Candi Badut, northern side with the damaged statue of Durgâ Mahißâsuramardinî. Photo by the author, 1995.

Fig. 4 Triªûla cakra ornaments. Candi Badut inserted. After F. D. K. Bosch, De Gouden Kiem: Inleiding in de Indische Symboliek (Amsterdam: Elsevier, 1948), pl. 57; and F.D.K. Bosch, The Golden Germ (The Hague: Mouton, 1960), pl. 60.

5a

Fig. 5a Details of the triªûla cakra ornament depicted on Candi Kedulan’s exterior wall. Photo taken by the author at the excavation site, 2004. Fig. 5b Details of the triªûla cakra ornament depicted on Candi Losari’s exterior wall. Photo courtesy Dwi Pradnyawan. Fig. 6 Pantheon of the simple, noncomposite temple. After B. de Haan, Oudheidkundig Verslag 1927 (Weltevreden: KBG, 1928), 15. Fig. 7 Plan of the Badut temple site. After B. de Haan, Oudheidkundig Verslag 1929 (Weltevreden: KBG, 1930), pl. 5. Fig. 8 Pantheon formed by the temples in the central courtyard of Candi Prambanan. Adapted from B. de Haan, Oudheidkundig Verslag 1927 (Welte vreden: KBG, 1928), 16. Fig. 9 Map of Java showing the locations of the Indo-Javanese temples mentioned in this article. Drawing by Hans Borkent, 2008.

Fig. 9

such textiles depicted on their exterior walls.49 As for the ·aiva temples and for the Buddhist temple complex of Candi Ploasan, I believe this dating could be correct. But I have yet to be convinced of the revetment during the mid-ninth century of Candi Sewu and Candi Mendut and whether it included the floral reliefs, for which Totton provides no evidence.50 Be this as it may, the presence of similar floral reliefs on the upper part of the second gallery of Borobudur indicates that the triªûla cakra ornament probably has to be dated earlier as far as the Buddhist monuments of Central Java are concerned.51 Figure 5 offers two other specimens of the ornament. The first floral relief adorns the outer walls of Candi Kedulan, a small ·aivite temple complex in central Java presently being excavated. Significantly, two stone inscriptions found at the site are both dated to 856.52 The second triªûla cakra relief is from Candi Losari, a slender shrine recently discovered in Losari, a village located on the YogyakartaMagelang highway, near Mendut and Borobudur.53 Other ornamental elements, such as the antefixes and the so-called garland motifs of small birds, would also seem to support my later dating of Candi Badut. However, I must leave to others the delicate stylistic research on their exact sequential position in the evolution of these ornaments. This type of research, also known as seriation research method, is often presented as offering a breakthrough in

49 Mary-Louise Totton, “Cosmopolitan Tastes and Indigenous Design: Virtual Cloth in a Javanese Candi,” in Textiles in Indian Ocean Societies, ed. R. Barnes (London: RoutledgeCurzon, 2005), 129, n. 27. Totton reiterates her opinion that all the temples concerned were built or revetted during the mid-ninth century. Apparently, she was not aware that her dating does not accord with the current dating of Candi Badut nor with Dumarçay’s hypothetical refashioning of the temple in the thirteenth century. For other textile motifs, see Hiram W. Woodward, “A Chinese Silk Depicted at Candi Sèwu,” in Economic Exchange and Social Interaction in Southeast Asia: Perspectives from Prehistory, History, and Ethnography, ed. Karl L. Hutterer (Ann Arbor: The University of Michigan, 1977), 233–43. 50 During the first partial reconstruction of Mendut, begun by Brandes and continued by van Erp, it became apparent that the present temple encased an older and somewhat smaller building. Interestingly, the inner core of the base was made of red bricks, which indicates an early time for the original construction. See Krom, Inleiding tot de Hindoe-Javaansche Kunst, 313–14. According to Dumarçay, it seems possible “that Candi Mendut was transformed by the addition of a vestibule, built with characteristic Hindu construction techniques which were introduced no earlier than 830 A.D.” See Jacques Dumarçay, “Buddhism and Architectural Change,” in Ancient History, ed. John Miksic (Singapore: Editions Didier Millet, 1989), 71. However, it needs reminding that the dating of these construction techniques, notably the in-fill method, is uncertain. Thanks are due to Mark Long for this information.

51 See N.J.Krom and T.van Erp, Beschrijving van Barabudur, Archaeologisch onderzoek in Nederlandsch Indië 3 (’s-Gravenhage: Nijhoff, 1920–31), Portfolio III.4: Lichtdrukken, gaanderijen, pl.129 and pl.132. As pointed out to me by Dwi Pradnyawan, another fine specimen of a triªûla cakra ornament is depicted on the side of a chariot in the narrative bas-reliefs; see A.J.Bernet Kempers, Ancient Indonesian Art (Amsterdam: Van der Peet, 1959), pl.82: “Bara- budur. Queen Mâyâ on her way to the Lumbinî grove.”

52 Candi Kedulan, in the plain of Prambanan, eight kilometers west of Candi Prambanan and seven kilometers north of Kalasan, was discovered in the closing years of the twentieth century, underneath layers of mud and earth (as was Candi Sambisari, located about nine kilometers to the southwest). Regrettably, nothing has yet been published on Candi Kedulan except for short abstracts of oral presentations at various international conferences by the Indonesian archaeo- logist Timbul Haryono. The inscriptions are presently being analyzed by Andrea Acri and Arlo Griffiths; hopefully their translations of these and other inscriptions will soon be available.

53 Candi Losari, discovered in 2006 and visited by the author in July 2008, still awaits excavation and ultimately, perhaps, reconstruction. Also noteworthy, in addition to the triªûla cakra motif, are the beautiful kâla with pronounced fangs adorning Candi Losari, the pointed antefixes, and the garlands with flowers.

Javanese art-historical research, but it is still in its infancy and as yet unable to meet the high expectations.54 Nevertheless, in the future the seriation research method will hopefully yield a more precise dating of Candi Badut in relation to other late Central Javanese temples, both in Central Java itself and in East Java (such as Candi Gurah, Candi Besuki, and Candi Songgoriti). The third reason to reject Badut’s early dating has to do with the architectural similarity between the layout of the Badut temple complex and those of other

·aiva temples in Central Java, such as Candi Ijo, Candi Meˇrak, and Candi Prambanan. Ironically, some of the similarities in plan and layout also attracted the notice of de Haan,55 the same man who two years later was to assign Candi Badut to the early phase of Javanese monumental art. In a short note published in Oudheidkundig Verslag 1927, de Haan demonstrated that the typical layout of ·aiva temples in Java reflects a regular system that encompassed simple temples as well as the most complicated ·aiva temple complex in Java, Candi Pramba nan. Taking the simple, non-composite temple as nucleus and point of departure, he proposes the arrangement depicted in figure 6.56 The second type of sanctum is the smaller temple complex exemplified by Candi Badut, Candi Meˇrak, Candi Ijo, and others. All of these temple complexes consist of a main temple and three smaller subsidiary temples. Some of the smaller subsidiary temples displayed the following arrangements:

LEFT CENTER RIGHT

Candi Badut lingga Nandi57 ?

Candi Batu Gono ? Nandi Mahâyogi (?)58

Candi Ijo lingga Nandi ?

54 See, for instance, Klokke, “Ornamental Motifs,” 85–98.

55 De Haan, Oudheidkundig Verslag 1927 (Weltevreden: KBG, 1928),14–7. More recently, Ono Kunihiko of Waseda University has also called attention to the off-center positioning which Candi Badut has in common with other HinduJavanese temples such as Candi Sambisari, Candi Prambanan, Candi Ijo, Candi Gunung Wukir, Candi Gunung Sari, Candi Gurah, Candi Kidal, Candi Jawi, and Candi Randuagung. Ono Kunihiko, “The Symbolism of Temple Sites on Old Javanese Temples: Asymmetrical Temple Sites of Hindu Candi,” Journal of the Society of Architectural Historians of Japan 36 (2001): 2–35 (in Japanese, with an English abstract).

56 For the purpose of this article, I have slightly adapted de Haan’s terminology and plans. For instance, the statue that he referred to as “Guru” is now known to represent the sage Agastya, and the name has therefore been changed accordingly. The different temple arrangements are illustrated in a new and hopefully more comprehensible way.

57 Here, for the bull placed in front of a ·iva temple, de Haan uses the designation current among art historians. This usage was challenged by Gouriswar Bhattacharya in a paper presented at the 29th Conference for German Orientalists. He argues that the correct designation for that bull is V®ßabha Virabha or V®ßa. Gouriswar Bhattacharya, “Nandin and Virabha,” Zeitschrift der Deutschen Morgenländischen Gesellschaft, Supplementum III, 2 (1977): 1545–67. I wish to thank Professor Arlo Griffiths of EFEO in Jakarta for the reference to this “epoch-making” paper.

58 Both the question mark and the note are de Haan’s. The note reads, “the statue of Meˇrak has not yet been identified definitely. It concerns a seated male figure with the familiar build of the Guru with two kneeling devotees.” De Haan, Oudheidkundig Verslag 1927, 16, n.1. Unfortunately, I found no additional information on the Mahâyogi of Candi Batu Gono (for which, I assume, de Haan’s shorthand “B. Gono” stands) and of Candi Meˇrak. Perusal of back volumes of Oudheidkundig Verslag and other colonial archaeological reports has not yielded any useful information on the statues in question. Combining these data, the arrangement for the smaller temple complex as a whole reflects the plan of the Candi Badut site, where five of the seven enumerated objects were preserved (fig.7).

De Haan then turns to Candi Prambanan, the most elaborate ·aiva sanctuary in Java. Schematically, his sketch of its layout looks like the image in figure 8. As is known, de Haan used the above data to argue against the use of the name candi vâhana for the three subsidiary temples opposite the Trimûrti temples, replacing it by the neutral, albeit unimaginative names Candi B, Candi Nandi, and Candi A.59De Haan observed that Candi Prambanan “comprises the same elements in exactly the same arrangement as the smaller types elsewhere. Here, too, the simple temple constitutes the nucleus and key element of the plan, while in combination with the three eastern temples it displays the arrangement of the smaller complexes.”60 As was already noted elsewhere, the elements are the same but their internal arrangement is not, as is clear from the reversed positioning of Durgâ and Agastya vis-à-vis the central deity and Ga∑eªa. This fact can be accounted for by the different orientation of the Pramba- nan temple complex: it faces east instead of west.61 In both cases, however, the statue of Agastya is facing south as it should, and that of Durgâ faces north.

Considering the strictly systematized character of the design, one could argue that as a consequence of this, Candi A may have contained a lingga, and Candi B a statue of Mahâyogi, if at least the latter identification can be maintained.62 Candi B, it may be recalled, contained a heavily damaged statue of ·iva, but not in the shape of an ascetic. IJzerman, who discovered the broken statue underneath the rubble of what he referred to as a “chapel,” reported, “The statue undoubtedly was one of the finest of Prambanan. The neatly worked maku†a had the skull and the crescent moon, the forehead [had] the third eye; the face was beardless. [This is] a new statue of Mahâdeva, the latest recently found at Prambanan, for the southern temple opposite the Brahmâ temple was empty.”63 Krom added, “In the temple opposite the Viß∑u temple there was a four-armed statue of ·iva in a quite damaged condition, but with the pedestal and the body still present. The head, which is a very finely carved piece

59 Regrettably, the use of the vâhana misnomer still persists. See, for instance, Fredrick W. Bunce, The Iconography of Architectural Plans: A Study of the Influence of Buddhism and Hinduism on Plans of South and Southeast Asia (New Delhi: Printworld, 2002), 152; and Mary-Louise Totton, “Narrating Animals on the Screen of the World,” The Art Bulletin 85, 1(March 2003): 6–24. In this article as well as in her Ph.D. thesis, Totton returns to the former names of Candi A and Candi B, viz. “Candi Garuda” and “Candi Haµsa,” respectively. She does this in spite of de Haan’s early reminder that IJzerman (Beschrijving, p.57) had found a mutilated statue of ·iva in the northern subsidiary temple and that the old name “Candi Garu∂a” was based on the erroneous assumption that the Garu∂a statue, installed there by van Erp, originated from this subsidiary temple. The Garu∂a statue was later removed, but the wrong name of the subsidiary temple lingers on. Candi B was found empty, and there is no valid reason to assume that it contained a Haµsa statue (or Angsa, according to Bunce) that served as a mount for Brahmâ.

60 De Haan, Oudheidkundig Verslag 1929, 16.

61 J.W.IJzerman, Beschrijving der oudheden nabij de grens der residentie’s Soerakarta en Djogdjakarta (Batavia: Landsdrukke rij, 1891), 50. As pointed out in Poerbatjaraka, Agastya in den Archipel, 103, Agastya’s positioning is not fixed in relation to the presiding deity, ·iva, but in accordance to his own appropriate direction: south.

62 As stated in note 58 above, this identification was tentative. When de Haan published his note in 1927, he had no knowledge of Poerbatjaraka’s doctoral study of 1926 on Agastya in the archipelago, which would stop Old Java scholars confusing this pot-bellied sage with ·iva as Bha†âra Guru (“Divine Teacher”) and Mahâyogi (“Great Ascetic”).

63 IJzerman, Beschrijving, 57.

with high head-dress, skull, crescent moon, and third eye, is now in Batavia [[[Wikipedia:Jakarta|Jakarta]]].”64 Given the differing identifications, the safest conclusion seems to be that both at Candi Batu Gono and Candi Prambanan, one of the subsidiary temples contained an anthropomorphic representation of ·iva. The question arises whether the three subsidiary temples of Candi Prambanan and of the smaller ·aiva temple complexes, rather than housing the mounts of the Trimûrti, did not in fact each contain a distinct representation of ·iva, namely a phallic form (lingga), a theriomorphic form (“Nandi” or rather V®ßabha), and an anthropomorphic form. Perhaps the choice of the anthropomorphic form was not fixed, provided it was a representation of ·iva that fitted into the iconographic plan of the temple complex concerned.65 This could help to explain the discrepant identifications of Mahâyogi and Mahâdeva. The question marks that de Haan inserted in the northern and southern candi apit have to be maintained as we still have no clue as to their function. What is interesting to note though is that the positions of the two stone cubes (wadas) at either side of the main shrine of the Badut temple complex correspond to the candi apit in the central courtyard of the Prambanan temple complex. Perhaps they had a similar function.66 Crucial to my argument is the question of how the correspondences between Candi Badut and the ·aiva temple complexes in central Java should be explained. As is clear from his remarks quoted above, de Haan tried to explain these correspondences by assuming a later return of the Dinoyo refugee settlers to Central Java, allegedly resulting in an architectural boom, but this idea has never found recognition among his peers.67 In my opinion, it was just the other way around: Candi Badut followed Cen-

64 Krom, Inleiding, 1:486.

65 In line with this hypothesis, I would venture the suggestion that the Brahmâ-like statue at Candi Gurah represents Maheªamûrti, a four-headed anthropomorphic manifestation of ·iva (the fifth head supposedly being invisible). See Gopinatha T.A. Rao, Elements of Hindu Iconography (Madras: Law, 1914–16), II–2: 379–82. In my opinion, Soekmono’s alternative identification of this unique statue as Agastya is untenable, as the sage is not a divine person but a human being. Consequently, he is represented with two arms and one head, and not seated on a throne or seated/standing on a lotus cushion; cf. Poerbatjara, Agastya in den Archipel, 87. R. Soekmono, “Gurah: The Link between the Central and East-Javanese Arts,” Berita Lembaga Purbakala dan Peninggalan Nasional/Bulletin of the Archaeological Institute of the Republic of Indonesia6(Jakarta: Jajasan Purbakala, 1969), 18. According to Jan Fontein, the statue may represent another form of ·iva, “perhaps Bhattâra Guru, the divine teacher, or the deified Agastya”; Fontein, The Sculpture of Indonesia, 152–53.

66 As for the location and function of two large flat stones, de Haan provides the following information: “Both typical and extremely rare was the discovery of two large square blocks of stone [wadas] to the north and south of the main temple. These stone cubes are about 60 cm. and have a deep square hole in the top side. The position of the blocks was such that their upper surfaces were in line with the original ground level, which was at least how they were found. The purpose of the blocks is not quite clear; it is not impossible that they served as pedestals for banners raised during festivities, as is still the case now in Bali.” Casting doubt on the hypothetical similar function is the report on the find of a small Durgâ statue, now lost, in or near the northern candi apit. IJzerman, Beschrijving, 58.

67 Except, perhaps, Dumarçay, who mentions a change in Badut’s ground plan, “probably around 850,” adding that “[t]he same ground plan was used when Candi Canggal [= Candi Gunung Wukir] and Candi Sambisari were de signed”; Dumarçay, Candi Sewu, 53, n.3 and 215, n.3. It remains unclear, however, whence Dumarçay got this idea, and why he did not support it with art-historical arguments and/or datable evidence. Against this, I hold that Badut’s ground plan follows Central Javanese prototypes such as Candi Gunung Wukir, Candi Sambisari, Candi Prambanan, and Candi Kedulan. tral Javanese prototypes.68 Historically, the construction of Candi Badut and other Central Javanese temples in eastern Java far better fits the late-ninth- to early-tenth-century eastward expansion of Mataram. The distance to the center of the kingdom (where the leading architects and master artisans were permanently engaged in more prestigious building projects such as Candi Prambanan) offers the most plausible explanation, in my view, of why Candi Badut was kept so small and sober, with a plain base and relatively simplified floral reliefs.

CONCLUSION AND ART-HISTORICAL IMPLICATIONS

Reconsideration of the current theory assigning Candi Badut to the early phase of Central Javanese temple architecture, in the second half of the eighth century, has revealed three invalidating flaws. The first concerns the presumed connection between Candi Badut and the community that had issued the Dinoyo inscription of 760, the second relates to the incongruity of the art-historical arguments with the extant art-historical and architectural evidence, and the third concerns the unsatisfactory explanation for the correspondences between the plan of Candi Badut and other ·aiva temple complexes in Java. New arguments have been put forward in support of Vogler’s hypothesis about a founding date in the final phase of Central Javanese monumental art, in the second half of the ninth century or at the beginning of the tenth. This revised founding date of Candi Badut fits in with the eastward expansion of the Kingdom of Mataram during this period. If the new dating is correct, it would show that the development of Hindu-Buddhist monumental art followed a less erratic trajectory than in de Haan’s theory.

An interesting question in this context is whether the transition of Hindu-Buddhist temple art from central to eastern Java went on smoothly and without a clear break, or whether the differences between the Central Javanese and East Javanese temples are in fact so numerous and conspicuous as to validate the use of the term discontinuity.69 One of the implications of my art-historical research on Candi Badut’s founding date (and of Candi Prambanan, previously) is that the hiatus and difference between Central and East Javanese temple art are more pronounced than is generally acknowledged, and in need of explanation.70 The explanation reiterated here is that during the reign of the ·ailendra

68 The relative sequence of the Central Javanese prototypes deserves further investigation. Joanna Williams believed “Canggal” [= Candi Gunung Wukir] to date from the early eighth century, and its arrangement to “prefigure” the Prambanan temple compex, “which must be understood as deliberately reviving a number of such earlier elements”; Williams, “The Date of Barabudur,” 30. It should be noted, however, that at the time of writing, Candi Sambisari was still being excavated and little was known about its layout and the design (ibid., 45), to say nothing of Candi Kedulan and Candi Losari, which were discovered only recently. Personally, I feel inclined to think of Candi Badut and Candi Besuki as offshoots or satellites from Central Javanese prototypes, including Candi Gunung Wukir, in which Candi Prambanan came to function as a kind of “mother” temple.

69 Roy E. Jordaan, “Continuity or Discontinuity in Hindu-Buddhist Temple Art in Ancient Java?” Dialogue [Astha Bharati] 5.1 (2003): 93–100.

70 See, for instance, Fontein, The Sculpture of Indonesia, 48; and Marijke J. Klokke, “Hinduism and Buddhism in Indonesia,” in Ann R. Kinney (with Marijke J. Klokke and Lydia Kieven), Worshiping Siva and Buddha: The Temple Art of East Java (Honolulu: University of Hawai‘i Press, 2003), 21. dynasty (c.775–850 CE), central Java was exposed to more direct and vigorous cultural influences from the Indian subcontinent than in the later East Javanese period (c.928–1500). When, in the wake of the huge volcanic eruption of Mount Merapi, Javanese royalty decided to abandon their temple domain of Central Java and permanently settle in East Java, they initiated a more Javanese form of Hindu-Buddhist art.71 Following J. Brandes, I think that Candi Gunung Gangsir represents the “missing link” between the monumental art of Central Java and that of East Java.72 Here, perhaps, lies the explanation for the use of the molds in the production of Candi Gunung Gangsir’s ornamental terra-cotta stones and plaques. This practice could be seen as an attempt to revive Hindu-Javanese temple architecture, which had ground to a sudden halt after the eruption of Mount Merapi and the shift of the Javanese capital to East Java. Seen in this perspective, the molds could provide evidence, despite Jan Fontein, of “a temporary eclipse of a great artistic tradition.”73 Further stylistic research is needed into the motifs of these molds so that we may trace their origins and establish which Central Javanese temples served as models. Such information would in turn help us to reconstruct the precise trajectory of Hindu-Buddhist monumental art in Java.

ACKNOWLEDGEMENTS

A shorter version of this paper, entitled “The Foundation Date of Candi Badut Reconsidered,” was presented at the International Conference on Indonesian Art, hosted by the Indira Gandhi National Centre for the Arts, New Delhi, 4–6 March 2003. Regrettably, the publication of the conference proceedings is seriously delayed and may never materialize. I wish to thank IGNCA and coordinator Dr. Bachchan Kumar for the hospitality extended during the conference. Dwi Pradnyawan of the Universitas Gajah Mada, in Yogyakarta, has been helpful in facilitating my recent field trips to Candi Kedulan and Candi Losari. I am indebted to Hans Borkent and Siebolt Kok for their help with the illustrations, and to Jeff van Exel for editing the English of this article. Finally, I wish to thank the two anonymous reviewers for their comments and useful editorial suggestions.

71 Cf. Lydia Kieven, “The Architecture and Art of Ancient East Java,” in Kinney, Worshiping Siva and Buddha, 40–41; cf. Krom, Hindoe-Javaansche Geschiedenis, 179.

72 J. L. A. Brandes, “Driemaandelijks rapport, over April, Mei en Juni 1903,” Rapporten van de Commissie in NederlanschIndië voor Oudheidkundig Onderzoek op Java en Madoera (Batavia/’s-Gravenhage: Bataviaasch Genootschap van Kunsten en Wetenschappen, 1903), 38–39. Not Candi Gurah, as claimed by Soekmono, which looks as much a Central Javanese temple complex as Candi Badut, as far as its architectural arrangement is concerned. The statuary of Candi Gurah is believed to date from the eleventh or twelfth century.

73 Fontein, The Sculpture of Indonesia, 48.