The Mañjuśrīvāstuvidyaśāstra is the most complete Mahāyāna Buddhist architectural treatise known to have survived into modern times. The first three chapters describe various rituals and procedures pertaining to the founding of Buddhist cities, towns, monasteries and temples, while the remaining chapters ― collectively identified as the Citrakarmaśāstra in the colophons ― are primarily devoted to the subject of Buddhist iconography. The text’s oral transmission is attributed to Mañjuśrī, which is perfectly understandable given the bodhisattva’s close association with wisdom in general, as well as with astronomy and the other sciences in particular. Of specific interest to the subject at hand is the text’s description of the planning of the image house, which presents the ground plans and locations of the Buddha, the bodhisattvas and the devīs within the temple’s inner sanctum.

Written in a Sinhalese script, this Sanskrit text is highly relevant with respect to the study of Central Javanese architecture. The influence of Sinhalese monks on Javanese Buddhism is well known from the Ratu Boko inscription, which describes the founding of a branch on Java of Anurādhapura’s famous Abhayagiri monastery on the island of Sri Lanka.[30]

“Among the wide distribution of śilpa texts, the Mānasāra, the Mayāmāta and the Kāśyapaśila are Śaivāgamic texts which evidently had their origin in Dravidian India. Considering the great similarities that exist between such Hindu texts” and the Mañjuśrīvāstuvidyaśāstra “it is feasible to deduce that the latter too is of South Indian origin. Yet, unlike in the case of the Hindu treatises, the contents of which are corroborated by extant physical remains, no such parallel monastic remains are found in India that would fit into the monasteries (ārāma) as described in the Buddhist treatise of Mañjuśrī.... On the other hand...some of the Sri Lankan monastic complexes seem to show the prevalence of a wider knowledge of architectural concepts advocated in the text of Mañjuśrī, thus indicating the availability of the text material to Sri Lankan architects from early times.”[31]

Of particular interest to our present study, the Mañjuśrīvāstuvidyaśāstra describes the same lineup of cosmograms collectively known under the rubric vāstupuruṣa-maṇḍala, but without specifically embracing the Śaiva-Siddhāntist term. Instead, the Mañjuśrīvāstuvidyaśāstra employs the word caṇḍita, and to which the text assigns a nearly identical list of Hindu vāstudevatā as those alloted to the squares (koṣṭha) of the vāstupuruṣa-maṇḍala by texts such as the Mayāmātam.

The Mañjuśrīvāstuvidyaśāstra presents caṇḍita as its all-embracing term for referring to temples of varying size and plan. This strongly suggests to me that it had served as the basis for the Javanese word caṇḍi, which likewise is used generically when referring to the island’s ancient temples. As Chandra has noted, the word caṇḍi is used for the first time on Java in the Ngabean Inscription of AD 882, which speaks of the deities in the sanctum (devatā in pacaṇdyan). The word also appears in the Paguyangan Inscription from the beginning of the eleventh century: sang hyang caṇḍi i Burwan.[32]

Chandra has suggested that the word caṇḍi is a derivation of the Sanskrit word sandhī, “from sam + the root dhī ‘to reflect, think’ but more within the context of meaning ‘meditation’. Zoetmulder investigated the [Javanese] poet’s vision and his divine strivings in the introductory mangala stanzas of the kakawins wherein he found many examples of the word caṇḍi used as a technical term in yogic practices. Soekmono (1995:89) calls it poet’s special yoga, and ‘the poet’s goal in composing a kakawin was to mystically unite himself with the divinity worshiped.’ Both Zoetmulder and Soekmono have come very close to the original semantics of sandhī as ‘contemplation’.” Chandra also concluded that the transition of sandhī to the Old Javanese caṇḍi (via the Tamil canti) had been influenced by the architectural term caṇḍita. However, here Chandra inaccurately defines the term, at least with respect to its use in the Mañjuśrīvāstuvidyaśāstra, by claiming it refers to one of seven kinds of nine-storeyed buildings.[33] In no instance does this text actually employ the word caṇḍita to refer to vertical structures of any kind. Instead the term is repeatedly applied generically when referring to specific grid plans, whether pīṭha (9 squares), upapīṭha (25), maṇḍūka (64), paramaśāyikin (81), or triyuta (256). For example, “the garbhagṛha [divided into] sixteen parts on each side [to form] two hundred and fifty-six [squares or pada] is said to belong to the caṇḍita plan known as triyuta.”[34]

The fact that the Mānasāra calls the maṇḍūka of 64 squares a caṇḍita shows that the latter term was likewise known to the text’s presumed Tamil-speaking, South Indian composer. Employing Occam’s Razor ― all other things being equal, the simplest solution is the best ― the term caṇḍi appears to have arisen on Java by way of the transmission of the architectural term caṇḍita, either from an early version of the Mañjuśrīvāstuvidyaśāstra or from the arrival of another suitably-dated architectural text of similar disposition.[35] Perhaps the Javanese shortening of caṇḍita to caṇḍi had been influenced by the Sanskrit word sandhī, rather than the other way around as Chandra had suggested.[36]

As Marasinghe has observed, the Mañjuśrīvāstuvidyaśāstra is the only surviving example of what may have once been a vast collection of Buddhist śilpa texts in circulation within the ancient kingdom of Anurādhapura. “But the long-drawn rivalry between the Māhāvihāra...and the Abhayagiri-vihāra...and the Cola invasions of the 10th and 11th centuries resulted in the destruction of many valuable works belonging to both schools, and the Mahāyānists suffered the worse damage.” A similar destructive surge occurred “in the 16th century when the Portuguese razed Buddhist monasteries to the ground, set fire to libraries and wantonly destroyed the books they contained.  In the same century, the Sinhalese king Rājasimha I who renounced the Buddhist faith and embraced Śaivism threw into fire a large number of religious texts that belonged to Buddhist temple libraries. The survival of the present work even in a single manuscript may, therefore, be described as truly portentous.”[37]

Given the well known connection between the Abhayagiri in the ancient kingdom of Anurādhapura and the Śailendra monarch who had sponsored the founding of a branch of the famous Sri Lankan vihāra on Java in the late-eighth century, it is possible that the term caṇḍita had made its way to Central Java at that time, where it was subsequently shortened to caṇḍi.[38] Jeffrey Sundberg noticed that “the spellings employed in the Vāstuvidyaśāstra exactly parallel the strange Javanese spellings of Sanskrit words: the ‘dharmma’ (double ‘m’s) from the Garung shrines at Plaosan are exactly matched by the same spelling in the Vāstuvidyaśāstra.”[39]

“Perhaps, it is possible that the Abhayagiri-vihāra, Anurādhapura, which encouraged the pursuance of Mahāyāna Buddhist teachings from very early times, had played a major role in the absorption and committing to writing of such treatises.... The Abhayagiri-vihāra was in constant communication with South Indian Buddhists following Mahāyāna teachings. Hence, the treatise, Mañjuśrīvāstuśāstra, which possibly originated in this part of India, could have been made available to the Abhayagiri monks, who applied the treatise in their building activities, as seen in the case of the pabbatavihāras.”[40]

The Mañjuśrīvāstuvidyaśāstra “judging by the Sinhala script written with cursive forms, could not be assigned a date anterior to the 14-15th century period. Yet, the contents of the first three chapters themselves would indicate that the subject matter should go back to a much earlier period....” In particular, “the pabbatavihāras of ancient Sri Lanka draw a parallel particularly to the Hastayaramana type discussed in the text of Mañjuśrī. These pabbatavihāras are datable to a period between the 3rd and 9th centuries. Moreover, the description of the image-house in its developed form seems to correspond to the well formed image-houses of Sri Lanka of the 5th century A.D.... The sabhā, on the other hand, which is the central member of the pañcāvāsa system of the Mañjuśrī, does not seem to appear until about the 8th-9th century...” The sabhā’s presence in the VVS “indicates the prevalence of the pañcāvāsa system described in the Mañjuśrīvāstuśāstra as far down as the end of the first millennium A.D. The Mañjuśrīvāstu-śāstra, in its developed form, may be attributed to the 11th-12th century period. The present manuscript in Sinhala script, which we would attribute to the 14th-15th century period, indicates the continuity of earlier compilations of technical texts, though with much lacunae and in corrupt form.”[41]

Certain iconography descriptions that appear in the Citrakarmaśāstra chapters of the text are not found on Java. For example, the description of the Bodhisattva Maitreya, which places a tiny figure of the Tathāgata Akṣobhya on his crown, is at variance with how this bodhisattva was presented in Central Java.[42] Still, the content of this text may reasonably be considered, at least in part, as indicative of certain Buddhist architectural practices that had been in vogue on Sri Lanka during the eighth and ninth centuries. Furthermore, the text amply demonstrates that Mahāyāna Buddhist architects had likewise followed many of the same concepts and practices honored by the Śaiva-Siddhāntist temple-builders of South India.

In Chapter 1 of the Mañjuśrīvāstuvidyaśāstra, the Buddha initially mentions the use of the caṇḍita plans of 64 and 81 squares, the maṇḍūka and the paramaśāyikin, and then delineates the 45 deities that occupy the building site in a manner that is consistent with the data presented in South Indian texts such as the Mayāmātam. This is followed by a list of the Buddhist monastic buildings that qualify for placement within specific squares of the caṇḍita plan.[43] By contrast, Chapter 2 details the pīṭha and upapīṭha plans consisting of nine and 25 squares, respectively, as the preferred layouts for placing specific buildings within the confines of a monastery and presents lists of the nine and 25 deities that occupy the squares in the two plans.

The Citrakarmaśāstra (7.1) subsequently brings the paramaśāyikin once again to the forefront by presenting it as one of the plans for laying out the image house, the inner sanctum where the main image or images are installed. Of particular interest is the statement that image-house, “can be square, rectangular, octagonal, sixteen sided, circular and oval.” Thus the design of the summit area at Borobudur is entirely in conformance with a paramaśāyikin caṇḍita plan of 81 squares, despite the unusual design. The abrupt transition that takes place when one enters the summit area is one good indication that the worshiper has moved from the outer world into the inner core of the temple. The absence of any carved representations of worldly figures and events within the summit area also suggests a locale that is anterior to physical manifestation. Although the Mañjuśrīvāstuvidyaśāstra does not specifically describe the use of the paramaśāyikin for the layout of an entire temple, it makes us realize that Mahāyāna Buddhists had been aware of its parameters and associated ritual details, and thus potentially well disposed to its use in a manner consistent with what is expressed in the Mayāmātam, the Mānasāra and other South Indian texts.

Fig. 11: The paramaśāyikin caṇḍita (left); and the application of the sthaṇḍila plan (right) at Caṇḍi Sewu.[44]

       
The Citrakarmaśāstra (7.1) names the outermost tier of 32 squares (fig. 11) along the periphery of the paramaśāyikin as the quarter of goblins and ghosts (paiśācapāda), the second tier of 24 squares as the quarter of human beings (mānuṣapāda), the third tier of 16 squares as the quarter of divine beings (daivikapāda) and the inner core of nine squares as the Supreme Principle (brāhmikapāda). In South Indian architectural texts, these same four divisions are assigned to a 7x7 grid of squares considered to reflect a separate system.[45]

When we apply the paramaśāyikin caṇḍita exclusively to the summit of Borobudur, we can see that this particular organization suggests some most interesting possibilities with respect to interpreting the rationale behind certain architectural components. For example, the Citrakarmaśāstra generally calls for no images to be placed in the nine central squares of the diagram that collectively form the residence of Brahmā, thus adding another facet to the running debate as to whether the main stūpa at the apex of the pyramid had ever housed an image or images. Yet enough room remains to allow worshipers to enter the brāhmikapāda when circumambulating the main stūpa (fig. 12). The 16 stūpas with square-shaped openings would then presumably represent the daivikapāda, while the remaining two tiers of 24 and 32 stūpas with diamond-shaped openings would represent the mānuṣapāda and paiśācapāda, respectively.

[29] Jayasuriya, M.H.F.; Prematilleke, Leelananda; and Silva, Roland. Mañjuśrī Vāstuvidyaśāstra.  Sri Lanka: Bibliotheca Zeylanica Series I. Archaeological Survey of Sri Lanka (1995): 17.
[30] See Chandra, Lokesh. Cultural horizons of India. New Delhi: Aditya Prakashan (1995): 10-21 [vol. IV].
[31] Ibid., p. 18.
[32] Chandra, Lokesh ed. Society and culture of Southeast Asia: continuities and changes. New Delhi: Aditya Prakashan (2000): 133.

[44] The partitioning of the four quarters of the South Indian sthaṇḍila plan for the image house at Caṇḍi Sewu is comparable to the design of the Bṛhadeśvara temple at Tanjore. See Klostermaier, Klaus K. A survey of Hinduism. Albany: SUNY Press (1989): 319-320.

Fig. 15: The vāstupuruṣa-maṇḍala for the cella and vestibule.

As previously seen in Chapter I, the sub-basement at Caṇḍi Mendut is surrounded by a running band that contains a total of 51 relief panels (A1 through A51), except for the space occupied by the staircase itself. These can be categorized into two distinct groups: 24 which contain flower decorations and 27 more that either portray anthropomorphic male figures or animals. The 24 floral patterns perhaps symbolize the 24 waxing and waning periods of the annual lunar cycle, and the 27 anthropomorphic or animal figures the 27 ‘mansions’ that the moon and sun periodically inhabit during their respective daily, monthly and yearly cycles. The makara-adorned staircase, which perhaps with respect to placement was intended to coincide with the zodiac constellation Makara (Capricorn), would complete the astral series by representing Abhijit ― the 28^th nakṣatra. And the outermost border at Caṇḍi Mendut (fig. 15) may have been intended to provide much the same function as the outermost tier of the vāstupuruṣa-maṇḍala, albeit while reflecting a different numerical total.

However, Caṇḍi Mendut’s divergence from the typical outer border of 32 squares potentially can be explained by referring to the Sarvatobhadra-cakra long used by Indian astrologers for casting horoscopes (fig. 16). Sarvata means ‘everywhere’ while Bhadra means ‘auspicious’. The diagram is therefore intended to show all-round well-being as well as all kinds of auspicious (and inauspicious) results. Unlike other Indian astrological (Jyotiṣ) charts, the Sarvatobhadra-cakra incorporates the 28^th nakṣatra; by contrast, Abhijit does not figure in the astrological systems set forth in the Bṛhat Jātaka and the Sārāvalī. Some astrologers consider this chart to be an archaic remnant of Vedic astrology precisely because it includes Abhijit.

 
Fig. 16: The Sarvatobhadra-cakra of Indian Astrology (‘vowels’ are magenta; ‘consonants’ are blue).

The Sarvatobhadra-cakra is based on the now familiar plan of 9 x 9 squares,[61] with the 28 asterisms and with 12 zodiac constellations occupying much the same positions that their Vedic equivalents possess in the vāstupuruṣa-maṇḍala. As is also the case at Caṇḍi Mendut, the solstices and equinoxes are not explicitly represented along the outer band of the Sarvatobhadra-cakra. Seven planets corresponding with the seven days of the week are assigned to various squares at the center of the diagram; each of the squares without a celestial occupant either displays a Sanskrit vowel or vowel/consonant combination from the alphabet. The layout of the vāstupuruṣa-maṇḍala at Caṇḍi Mendut otherwise corresponds in all respects to the caṇḍita of 81 squares [62] with two tiers of 24 and 16 squares, respectively, surrounding the central nine squares of the plan.

Initially drafted on palm leaves in all probability, the diagram would have subsequently been drawn upon the cleared and flattened ground of the site on which the actual building was eventually constructed. At Mendut, the nine Buddhist divinities originally installed within the main cella ― of which only three are still in their original places today ― may have been considered to occupy these nine squares as an expression of the Supreme Principle (Brahmasthāna) of the vāstupuruṣa-maṇḍala, which symbolizes the eternally effulgent dark source of all the light sources arrayed around it.

At Caṇḍi Mendut, a total of 81 deities were originally portrayed on the temple’s exterior walls, thus symbolically filling up all 81 squares of the diagram (fig. 17):

·         9 images within the cella’s interior out of which only 3 have been preserved;

·         8 bodhisattvas in B34; B35; B37; B38; B40; B41; B43; B44;

·         17 deities in B36 (5); B39 (5); and B42 (7, including the two nāga attendants);

·         2 Hārīti (C5) and her yakṣa spouse (C6) inside the vestibule;

·         8 flying devas in fields C7 and C8 within the vestibule;

·         8 deities on the vestibule’s exterior walls in B32, B33, B45, and B46;

·         27 nakṣatra panels on the basement;

·         2 Indra and Airāvata on the left staircase wing.

Furthermore, the core maṇḍala resolves down to 45 Buddhist deities, not including attendants:

·         9 images within the cella’s interior;

·         8 bodhisattvas in B34; B35; B37; B38; B40; B41; B43; B44;

·         9 deities in B36 (3), B39 (3); and B42 (3);

·         2 Hārīti (C5) and her yakṣa spouse (C6) inside the vestibule;

·         8 flying devas in fields C7 and C8 within the vestibule;

·         8 deities on the vestibule’s exterior walls in B32, B33, B45, and B46;

·         1 image of Indra on the left staircase wing.

Fig. 17: A representative distribution of the deities at Caṇḍi Mendut.

Twenty-three nakṣatra ‘squares’ on the basement either depict seated or kneeling gandharva. Here their well-known passion for the erotic [63] is turned to more pious pursuits. This is not only reflected in their faces but also evident from the overall arrangement of their panels on the basement. “By means of both direction and attitude, they continually draw our attention to the middle of the structure. Moreover, their devotional activities make it seem as if they are expressing on their faces a recollection of what is taking place at the center. In addition, the sembah gestures of the middle figures on the three sub-basement sides (A13, A26 and A39) suggest that these three figures discern the great divide that exists between them and the leading characters that are enthroned on the great walls above; they not only express their loving adoration but also convey an anticipation of still greater things to come.”[64]

Though subordinate to the Buddhist gods and goddesses on the exterior walls of the cella above, the serial of 23 gandharva figures on the cella’s sub-basement nonetheless exhibit a marked correspondence with the raised ambulatory [65] path that surrounds the main cella. It is consistent with the Vedic role of gandharvas as the protectors of the celestial station where “the doers of good attain the sweet-juiced” (soma-amṛta) upon their deaths. Possessed of water, they “go clothed in the liquid water, to the great celestial abode to (take) the sacrifice.”[66]

At Caṇḍi Mendut, the gandharvas on the sub-basement are the dedicated soma-priests who protect the fabled elixir of immortality. However, we also must keep in mind that the primordial Gandharva is Soma, the Lord of the moon who is said to cohabit with each of his 27 wives on successive nights. The regions of the gandharvas, who live in the sky, are the air and the heavenly waters and thus they are associated with both clouds and moisture. They also regulate the course of the nakṣatras; hence 27 are mentioned.[67]

The temple’s front-projecting porch, which “extends outward by six times the width of any of the other three sub-basement projections,” deviates “in both size and presentation from the 23 remaining fields portraying figures, of which 20 assume a kneeling posture and three are in a seated posture. This projection corresponds — as does the overall architectural plan — with the higher structure of the temple porch above.” The dissidence of this state of affairs “makes it seem as if it were loose and on its own from the rest of the main building.”[68]

By contrast, two animals are carved on the sides of the temple’s front-projecting porch — the monkey and the mongoose — that may plausibly be explained as astrological symbols. Each nakṣatra represents a unique quality of human consciousness. In this respect, each animal is intended to evoke the personality and behavioral pattern of the nakṣatra with which it is associated based in the belief that, as Carl Jung has framed it, “whatever is born or done this moment of time has the qualities of this moment of time.”