Version classiqueVersion mobile

Rites hindous

Gérard Colas
Gilles Tarabout

Débattre des rites, débattre du passé

Reading between the lines

Colliding attitudes towards image worship in Indian religious texts

Phyllis Granoff


Lire entre les lignes : désaccords sur le culte de l’image dans lestextes religieux indiens

Cet article explore le processus par lequel le culte de l’image s’est peu à peu incorporé dans un cadre rituel brahmanique orthodoxe. Est tout d’abord examiné et mis en lumière le fait que le culte des images fut en premier lieu mal accueilli parmi les cercles orthodoxes et devint seulement petit à petit admis. Il apparait que nous pouvons avoir un aperçu des étapes de ce processus en centrant notre attention sur des détails de textes rituels, qui semblent incohérents ou redondants. En nous fondant sur une information issue des textes rituels, en particulier ceux de l’école du Pāñcarātra, des Purāṇa, des inscriptions et hagiographies, l’article illustre la façon dont des rituels ont dû s’adapter et se modifier lors de leur mise en pratique dans un nouveau contexte religieux.

Texte intégral


  • 1 I thank Gérard Colas and Gilles Tarabout for inviting me to participate and for their extensive com (...)

1This paper was originally delivered in a seminar at the École des Hautes Études en Sciences Sociales, June, 2000. It was the last of four lectures.1

2In my presentations to the seminar I tried to examine the issue of rituals and the transfer of rites from a number of perspectives. I began by suggesting that the ritual culture of medieval India was ‘eclectic’; groups self-consciously used rituals that they knew were not their own. They retained their ritual integrity by such strategies as subordinating these alien rituals to their own unique rituals or avowing that other mantras, other rituals are somehow accomplished through the practice of the unique mantras and rituals they themselves are offering. In this ritually eclectic environment rituals did not serve as hard and fast markers of religious affiliation or group identity (Granoff 2000).

3In my second presentation I discussed cases in which rituals could do just that. I focussed on intra-sectarian conflicts in medieval Jainism and suggested that where groups have few other markers to differentiate themselves from their rivals, rituals may be called upon to serve as dividing lines between competing groups (Granoff 2001).

4In my third presentation, “The place of ritual: geographical boundaries and sectarian identity in Medieval Indian religions”, I further suggested that in some cases where rituals clearly divide groups they may be in fact ‘secondary markers’; the primary source of division between groups may even be nonreligious factors such as regional or linguistic identity. Rituals provide these religious groups with a way of talking about their differences in a language that is at least talking about the practice of religion.

5In the present paper I wish to argue that we can eavesdrop on the very process of creating new rituals as a dividing line between competing groups. I would like to suggest that by ‘reading between the lines’ of our texts we can discover some of the intricacies and problems that resulted when a group deliberately sought to construct new rituals or impose its own rituals on a sphere originally not its own. I will be looking at evidence to support an interpretation of early Pāñcarātra texts as the results of efforts to impose on a large scale Brahmanic rituals on an originally non-Brahmanic public cult of images. My research here has only begun and I realize that it is highly speculative.

6Popular and scholarly accounts of medieval Indian religions often emphasize the role played by image worship. There is no question that images and the temples that housed them were at the center of religious practice in India from a relatively early date and in all of the major religions, Hinduism, Buddhism and Jainism. But what is not so often noted is that at the same time there was a strong undercurrent of suspicion towards image worship; indeed occasionally there was even an open hostility towards the cult of images and particularly towards the priests responsible for their rituals. Despite its obvious popularity clearly documented by the sheer numbers of temples and images that remain, the worship of images seems to have been only grudgingly and somewhat belatedly accepted in orthodox priestly circles. The investigation of the reception of image worship and its vicissitudes in many ways offers an ideal opportunity to study how rituals are transformed as they are utilized by new groups and for new purposes.

  • 2 Vaikhānasagṛhyasūtra, p. 62. The passage is discussed in Colas 1996: 22-26. See also Colas 1994: 51 (...)
  • 3 The Purāṇas are notoriously difficult to date. For some information on the dating of various Purāṇa (...)
  • 4 Ed. Anandāśrama Sanskrit Series, vol. 54, 1981, chapter 266.

7The acceptance of image worship would seem at first to have been largely confined to the worship of small scale images in the context of domestic rituals and closely associated with the Śrauta rituals. Thus the Vaikhānasagṛhyasūtra, probably of the 4th c. A. D., in section 4.10 enjoins the worship of Viṣṇu at the conclusion of the morning and evening Agnihotras. It also offers what may be the earliest description of the pratiṣṭhā, the ritual consecration of images, in Brahmanic literature.2 The process by which image worship moved out of the domestic sphere and into the public arena of temples was certainly a complex process, and it is beyond both my abilities and the scope of this presentation for me to trace it. The Matsyapuraṇa, which later authors were so fond of quoting, gives elaborate details for the consecration of images and describes the major participants as twice-born, learned in the mantras of the Vedas.3 Thus the sthāpaka, here the main officiant who performs the consecration ceremony, is to be from that area which had traditionally been the territory of Vedic rites, the area which was the natural home of the kṛṣnasāra deer, and he is to be devoted to the performance of the proper rites (266.3).4 At the same time, however, in a term the exact significance of which puzzles me somewhat, he is also described as “indifferent to or having nothing to do with families of heretics or ‘pāṣaṇḍas”’ (pāṣaṇḍakulaniḥspṛha, 266.3). The mūrtipas, those who carry the image, are explicitly said to be twice-born and of good family (kulīna) (266.5). In a subsequent chapter we are again explicitly warned that the priest conducting the rituals must be a Brahmin who is knowledgeable in the sacred texts (vipreṇa śrutipārageṇa, 267.33). He is also to be a householder and not a ‘false ascetic’. The person who selects as his ritual specialist such a false ascetic or pāṣaṇḍa over learned Brahmins is foreover dishonoured and brings about the destruction of his lineage; the temple is taken over by goblins and is forever without worship. The temple that is consecrated by Brahmins, on the other hand, leads to all good things for the family of the person responsible for its dedication and is worshipped for ages to come. As the text explains,

  • 5 Pāṣaṇdinam yas tu karoti bhaktyā vihāya viprāñ chrutidharmayuktān/guruṃ pratiṣṭhādiṣu tatra nūnaṃ k (...)

The one who with devotion appoints a heretic to carry out rites such as the consecration, instead of Brahmins who know the scriptures and the dharma, will soon find his family in distress and himself in great dishonor. The temple he so consecrates will in short order be taken over by goblins, an object of scorn for all. That which is done by Brahmins brings well-being for the entire family of the patron of the rituals and is worshipped for a long time to come. [267.34-35]5

  • 6 The hypothesis that public image worship was not originally a Brahmin concern is not entirely new. (...)

8It is important to note that the Matsyapurāṇa lumps together in these warnings the ascetic and the pāṣaṇḍa; they are one and the same. Against this heretical false ascetic, the text proposes the proper ācārya, the brāhmaṇa, who is also a householder. In these warnings I would like to see a move on the part of orthodox circles into the ritual sphere of these ‘false ascetics’, the pāṣaṇḍas who had been serving in the capacity of ritual specialists in the temples and whom the author of the Matsyapurāṇa now wants to expel from their ritual roles. My argument in what follows will be that as orthodox circles moved into the public sphere of temple rituals they were not just displacing their competitors; they were also moving into a ritual world that had its own distinctive ideology which was not always or not entirely compatible with orthodox theology and ritual practice.6

  • 7 I have written on the topic in several papers. See Granoff 1998 and Granoff 2003. See also Davis 20 (...)
  • 8 The evidence for medieval Buddhism in India is less abundant, although some early stories do clearl (...)

9There is evidence from a variety of sources that the very concept of image worship was not entirely acceptable in orthodox circles. It is beyond the scope of this paper to trace the complicated process by which image worship entered mainstream Indian religions; here I would like to concentrate mainly on attitudes towards priestly service of images.7 I will draw my material mainly from medieval Hindu texts, although I believe that discomfort with the cult of images in general was also a marked feature of medieval Jainism and Buddhism.8

  • 9 Much of this material has been well studied. See Appadurai 1983: 43-56; StietenCron 1971: 126-138; (...)
  • 10 Śivapurāṣa, koṭirudrasaṃhitā, chapter 13.
  • 11 Seethe Cintrā praśasti of Saraṅgadeva, Epigraphia Indica, 1: 271 ff., verse 48. Bühler, who edited (...)

10The evidence for Hindu hesitation about the professional practice of image worship is of several different kinds. There is the explicit evidence of the Purāṇas and the law books, with their vituperative statements against the ‘fallen Brahmin priests’ or devalakas, who looked after the worship of images.9 There is also the implicit evidence of some of the Purāṇic stories; even when the Purāṇic texts champion image worship they leave behind in their stories hints of this negative appraisal of priests who tend images. Thus the Śivapurāṇa includes a story about the origin of the priests who worship the liṅga. They are called baṭukas and their lineage is traced back to a Brahmin who had violated the rules of ritual purity by having sexual intercourse and not bathing before he worshipped Śiva on the night of Śivarātri. Eventually Gauri takes pity on him and adopts him as her son. Siva decrees that the descendants of this fallen Brahmin are to worship the liṅga.10 In fact in a 13th-century inscription at Somanātha the term baṭuka is used for attendants who cleanse the temple and bathe the god and in return receive money and the leavings of the offerings.11

  • 12 For a parallel story in the Vaikhānasa tradition see Colas 1996: 140.
  • 13 I have also discussed this in Granoff 1998.

11A similar account, disparaging the Brahmin who takes care of a temple, may be found in the Nāgarakāṇḍa, one of the later sections of the Skandapurāṇa. The sage Durvāsas had asked the Brahmin community for land to build a Siva temple, but the Brahmins were too busy with their Vedic study to take much notice of his request. He curses them and one Brahmin attempts to placate the sage, offering him the land he desired. In turn this Brahmin, ‘Suśīla’, “of irreproachable conduct” is cursed by the others to be called ‘Duḥśīla’, “of wicked conduct”. He must move outside the city and he is not permitted to take part in any of the ritual activities of the other Brahmins.12 Again from the Skandapurāṇa is the story of a temple to Siva that the Moon is to dedicate in order to cure himself of his wasting disease that was caused by a curse from his father-in-law Dakṣa. The story is the origin story for the famous temple at Prabhāsa. It is worth looking at in some detail.13

12The Moon is told by Siva to build a temple and perpetuate the worship of Siva’s liṅga. The text provides us with an elaborate description of the rituals that Brahmin priests conduct as part of the building ceremonies. The first steps in the consecration of the liṅga and the building and dedication of its temple are told in great detail (chapter 23). In preparation for the building of the temple several sacrificial enclosures must be made and a great sacrifice must be held. Thus the first structures that the divine architect Viśvakarman builds are not temples at all, but sacrificial maṇḍapas. There is a main maṇḍapa which is adorned with flags. In it is a marvellous throne from which hang garlands of pearls; it was like heaven itself with divine garlands of sandalwood and archways made from the heavenly wishing trees; everywhere it exuded a divine fragrance (37-39). There are separate structures for the various groups of priests who will take part in the sacrifice (100). Eventually the sacrifice is concluded and the temple is built. The Moon turns his mind to the next problem: finding priests to oversee the worship of the temple.

13In fact the Moon founds a city near the main temple where he settles the Brahmins so that they may then go about their normal business:

  • 14 Niveśya nagaraṃ somaḥ śrīsomeśvarasamnidhau / svakarmaṇāṃ pracārārtham athābhyarthayata dvijān // [ (...)

And having settled this city near the god Someśvara, the Moon requested the Brahmins to dwell there and carry out their appointed tasks. [68]14

14He tells the Brahmins that their homes have been made ready; they are filled with everything the Brahmins will need, including rare and precious substances like aloe woods, camphor and sandalwoods, brought from distant lands. The merchants have even been instructed to make over to the Brahmins a portion of their earnings (74). But the Moon wants a special service from the Brahmins in return for giving them a place to live and providing them with subsistence and even luxury goods; the Moon asks these Brahmins to take care of the liṅga he has consecrated in the temple that he has built. The reply the Brahmins give is instructive. They speak through their leader Kauśika:

  • 15 Uvāca Kauśikas teṣu gotrāṇaṃ prathamo dvijaḥ / sādhūpadiṣṭam asmākaṃ dvijarājena sarvathā // [84] s (...)

Kauśika, the foremost Brahmin among those Brahmin clans, spoke. “In every respect we applaud what the moon, king of the twice-born, has instructed us. We will do everything requested of us, but first we beg you to hear this small point we would like to make. If we were to worship in a professional capacity, making use of the leavings of the offerings to Śiva, then we would lose our caste status, a state that is abhorred by the orthodox śruti and smṛti texts. And as śruti and smṛti are the two great commands of the god Rudra himself, what fool would transgress them, even if he were threatened with death? We bring joy to the entire world by performing sacrifices according to the rules of the śruti into the fire, which is the mouth of the gods, and which is one of the eight forms of that god who has eight forms, namely Siva.” [84-88]15

15The meaning of Kauśika’s speech could not be more clear. Orthodox Brahmin priests must worship by performing sacrifices according to the rules of the Vedic texts; professional worship of images leads to a loss of caste status for a Brahmin priest. One has only to think of passages like Manu III. 152, which bars those who tend images from offerings to the gods and pitṛs, two important ritual functions for orthodox priests. Kauśika unambiguously rejects for the orthodox Brahmins a role in caring for the liṅga in the Moon’s new temple. Instead he tells the Moon about some people who might be suitable for the task of overseeing the worship in the temple. Kauśika proceeds to describe a group of people, born in the families of brahmarṣis, “sages who are Brahmins by birth”. These people were originally demons, daityas, enemies of the gods. They had worshipped Śiva and obtained boons from him, but Viṣṇu had nonetheless slain them when they had gotten out of control and oppressed the gods. They had then been reborn in the families of brahmarṣis and were still devoted to worshipping the liṅga (121). It is these people whom the Moon should appoint as priests.

  • 16 It is not only temple priests who were the object of distrust in the Purāṇic accounts. There are pa (...)

16One could not want a more explicit statement about the orthodox Brahmanical attitude towards the temple and the object of worship in the temple. While the orthodox Brahmins were perfectly content to serve as officiating priests in the rituals for building the temple and even for consecrating the liṅga, they refused to serve in the daily worship of the liṅga. Orthodox Brahmins in this story will not be responsible for the worship of the liṅga in the temple, and as the story of the demonic ancestry of the priests who are appointed to that task shows, the orthodox Brahmins remain suspicious of the status of those who deal with temple worship.16 The story also suggests that we need to be somewhat cautious in our interpretation of the significance we attribute to the fact that rituals associated with the building of temples and consecration of images begin to make their appearance in orthodox texts; as this story from the Prabhāsakāṇḍa suggests, Brahmanic participation in these rituals did not necessarily imply a general acceptance of professional participation in image worship and temple rituals. Stories like these in the Purāṇas and the explicit denunciations of the priests who tend images that we find in the law books leave little doubt that at least some of the orthodox Brahmin priests continued to regard the priests who tended images with disgust and disapproval.

  • 17 Epigraphia Indica, XXVI, pp. 212-217.

17I would argue further that this Purāṇic story finds a parallel in a puzzling early medieval inscription. In a 9th-century A. D. grant of a ruler Rāṇaka, from a subordinate branch of the Saindhava lineage, we learn that the revenue of a village was donated in two parts. The first half is to go to the deities, Visnu, Surya, Ganeṣa and the mātṛkas, set up on the outskirts of the village by a certain Brahmin who is named by name in the grant as a Ṛgvedic Brahmin Śivarudra. The second half of the revenue is to go to “the head of the maṭha, whoever he may be” (kasmaicana maṭhapataye). This division puzzled the editor of the grant, who saw in the head of the maṭha someone entirely unconnected with the deities set up by Śivarudra and connected with an entirely different set of temples, also here unspecified. This seems to me unlikely; what follows is a standard formula warning that no one is to hinder the grantee from exercising his rights. The grantee here is said to have rights over the entire village and to use the funds for repairs of the temple, the maṭha and its associated structures. There seems to be in the grant only one person who oversees the granted revenue. I think it is possible that this grant reveals the same bifurcation that was clear from the Purāṇic story: a Brahmin, here Śivarudra, has consecrated the images, but someone else, not even sufficiently important to be named by name, is responsible for their daily worship and upkeep and thus for the actual management of the grant.17

  • 18 Āgamaprāmaṇya [75-76]: na ca tāvatā teṣāṃ takṣnāṃ caikajātitvasaṃśayaḥ, evam ihāpi bhagavacchāstras (...)

18This negative attitude persisted throughout medieval times and in a variety of contexts, as is clear from a text like Yāmunācārya’s Agamaprāmāṇya of the 10th or 11 th century. There Yamuna argues vigorously that the Pāñcarātra priests who worship images are still true Brahmins. His arguments are interesting. At one point in the text Yamuna suggests that there are actually two groups of priests who are at work in the temple. One group performs somewhat menial tasks like cleaning the temple. They are called prāsādapālakas, or ‘custodians of the temple’, and indeed they are of low caste status. According to Yamuna, it is this group to whom statements in the law books excoriating temple worship are directed. For Yamuna the other group, the Pāñcarātrins, Sāttvatas or Bhāgavatas, are true Brahmins. They are duly consecrated and worship according to special rituals laid down in the Pāñcarātra texts. He makes the comparison between the relationship of the prāsādapālakas to the true Bhāgavata priests on the one hand, and the carpenters who carve the sacrificial implements to the priests who officiate at the Vedic sacrifice on the other hand. The carpenters are of low caste while the officiating priests are Brahmins; just because both are involved in the same rituals does not mean they must share the same caste status. Similarly, he concludes, the temple rituals include two different groups of priests. One group is of low caste and tends the image and does things incidentally that many Purāṇas praise, namely sweeping the temple. The other group performs specific rituals enjoined in the Pāñcarātra texts and are Brahmins.18

19I find this comparison particularly important; first of all it makes clear that the audience Yamuna is trying to convince of the appropriateness of Brahmins worshipping images is the circle of Brahmins involved in the religious world of the Vedic sacrifice. This is clear from other examples he uses. One of the main issues for Yamuna, as it had been for Manu, is the rejection of Brahmin status for temple priests because they make a living from their worship. This was the problem of the Brahmin Kauśika in our story as well. In the Āgamaprāmaṇya, Yamuna argues that priests in the Vedic ritual also received remuneration in the form of the sacrificial gift or dakṣiṇā. But even more important than the question of the audience to whom Yamuna is addressing his arguments is the information he gives us about the different ritual specialists involved in the cult of images. He tells us explicitly that there are two fundamentally different groups involved in image worship and temples: the Bhāgavata Brahmins, whose cause Yamuna espouses, and another group, the prāsādapālakas, whose role he disparages. I shall try to establish in this paper that there is evidence from a variety of sources that there were indeed different groups involved in image worship; I even want to go so far as to suggest that Brahmin priestly interest in the public cult of the temples was late and entailed significant problems.

  • 19 For some information on the text see Kane 1962: 868-872. Kane emphasizes Ballālasena’s dislike of T (...)
  • 20 On the Tantric dīkṣā see Gupta, Hoens & Goudriaan 1979: 71-89.

20I believe that orthodox Brahmin resistance to some forms of image worship is also suggested by another source, the Dānasāgara of Ballālasena, written in 1169- 1170. The Dānasāgara can be seen as an effort to codify the correct ritual behavior and define orthodoxy. For this text, the most important ritual is dāna or gift-giving, a term which can encompass gifts to Brahmins and donations to a deity.19 That Ballālasena is using the performance of certain rituals and the rejection of others as the dividing line between orthodox and heterodox emerges from a reading of some of his comments in his introduction to the text. The Dānasāgara is essentially a compendium of quotes on rituals of dāna from various Purāṇas. Ballālasena includes in his introduction some comments on why he used some texts and excluded others when he made his selections. I believe that he rejects at least some texts because they advocate the practice of certain rituals with which Brahmins should never be involved in his opinion. These rituals have to do with worship in temples, with consecration of images, for example, and with other rituals that we associate with priests who also worship in temples, for example the Tantric initiation rituals or dīkṣā.20 Ballālasena specifically mentions the consecration of images and initiation and rails against certain texts because they are tainted by association with heretics, pāṣaṇḍas.

  • 21 For a discussion of the basic rituals of what I am calling the Hinduism of the Āgamas and Tantras s (...)
  • 22 See Bakker et al., eds., 1998 in the Prolegomena to his edition of the Skandapurāṇa. This text is d (...)
  • 23 See for example Yāmuna’s Agamaprāmāṇya.

21It is not possible to identify Ballālasena’s ‘pāṣaṇḍas’ with one hundred-percent certainty. Indeed scholarship on the Senas has seen the dynasty as pro-Brahmin and anti-Buddhist, responsible in no small degree for the decline of Buddhism in Bengal. But I think that at least in these verses Ballālasena is not attacking the Buddhists; he is attacking Hindus and Hindu texts that he feels cross the line, a line that is a ritual boundary. Ballālasena hates certain texts because they include material that he feels should be irrelevant to the religious role of the Brahmins, the subject of his Dānasāgara. He rejects certain rituals, like initiation and consecration or at least the way that these rituals, which are perhaps the central rituals of Āgamic Hinduism, are practiced by some people.21 Given the way in which Purāṇas continued to develop over the centuries, we can never know exactly what texts Ballālasena had that displeased him.22 Ballālasena in the Dānasāgara does acknowledge the practice of image worship in the public setting of the temple; the concluding sections of the text deal with offerings of lamps and cows in Śaiva and Vaiśṇava temples. But for Ballālasena in this text participation in temple worship is restricted to gift giving. In these chapters there are no instructions for rituals to consecrate images and there is no praise of cleaning or sweeping the temples, rituals that orthodox priests would object to in other texts.23

  • 24 Jalāśayānāṃ dānāni tathā ca suraveśmanām//noktāni dānāni samyag uktāni pratiṣṭhāsāgare yataḥ [verse (...)

22Ballālasena notes that he had written elsewhere, in the Pratiṣṭhāsāgara, now lost, about the gifting of bodies of water, tanks and wells, for example, and temples.24 It is impossible to guess what rituals he had included in his text and thus to assess more precisely what it was about pratiṣṭhā and dīkṣā that he found unacceptable. Was it the involvement of Brahmins in these rituals at all that disturbed him or was it the way that certain texts described the rituals that bothered him? We will never know for certain. The Kṛtyakalpataru of Lakṣmīdhara of roughly the same date includes a section on pratiṣṭhā. There is a long quote from a Devīpurāṇa (p. 66), one of the texts that Ballālasena particularly dislikes. According to this quote, the Devīpurāṇa assigns the role of officiating priest to those who are knowledgeable in the Śaivasiddhānta for Śiva and the Pāñcarātra for Viṣṇu. It also mentions the Bhūtatantra and the Vetālatantra and states that a consecration performed by someone outside the Āgamic tradition is harmful. Perhaps then Ballālasena objects to the growing influence of Tantric and Agamic forms of worship that we see in this passage from the Devīpurāṇa. Perhaps he objects to considering the priests of the Pāñcarātra or Śaivasiddhānta as orthodox. In such an interpretation, his voice then would be one of the voices against whom Yamuna was forced to argue.

  • 25 The complex factors that motivated Brahmin priestly interest in image cults require a separate stud (...)

23Texts like Yamuna’s and perhaps Ballālasena’s show us that orthodox priestly circles did not embrace image worship and the rituals of the Śaivasiddhānta or Pāñcarātra with unquestioned enthusiasm; Yamuna had to argue vigorously for the inclusion of image worship into the repertoire of sacred duties of Brahmin priests and Ballālasena seems to have had difficulty with Āgamic and Tantric rituals, as far as the evidence permits us to judge.25

24There are, of course, many questions raised by this hypothesis of Brahmin priestly incursion into an alien territory of image worship. Perhaps the most important question concerns the identity of the disparaged image-worshipping priests. There is a passage in the 6th-century Bṛhatsaṃhitā of Varāhamihira that suggests that priests were of different sorts. Varāhamihira describes the consecration of images in which Brahmins play a role; they are to recite mantras and are honored by the traditional dakṣiṇā or gift for their services. But then he tells us,

  • 26 Viṣṇor bhāgavatān magāṃśca savituḥ śambhoḥ sabhasmadvijān /mātrṇām api maṇ-dalakramavido viprān vid (...)

Know that the priests are as follows: for Visnu, the Bhāgavatas; for the sun, the Magas; for Śiva, the twice-born who smear themselves with ashes; for the Mothers, those who know the rites of the maṇḍala; for Brahmā, the Brahmins; for the one devoted to the welfare of all beings, the calm-minded Śākyas; for the Jinas, the naked ones. Each group of worshippers is to carry out the rituals in their own way for the deity they worship. [19]26

  • 27 On the multiplicity of ritual specialists see also Colas 1996: 352ff.; Tarabout 1997: 253-284.
  • 28 Cited by Grünendahl, ed., 1989, 3: 51.
  • 29 Das Gupta (1967, vol. II: 547) suggests that image worship belonged to the non-Vedic Bhāgavatas, bu (...)

25This suggests that worship was not entirely uniform and ritual specialists were diverse27. It may also suggest that there was a division between the basic consecration ritual, in which Brahmins participated, and other rituals, which were left to other officiants. Such was the case, we may recall, in the story of the origin of the temple at Prabhāsa, as told in the Skandapurāṇa, which I cited earlier. But even more important, this verse in singling out the priests who tend the images for Viṣṇu as Bhāgavatas may also be offering us a clue to help unravel the complex history of shifting attitudes towards image worship in one sectarian group. There is other evidence that the lines Yamuna draws, between the prāsādapālakas, the ‘custodians of the temples’, with their menial tasks of sweeping and cleaning, on the one hand, and the combined Bhāgavatas/ Pāñcarātrins with their more lofty ritual role, on the other hand, may be a later interpretation, which represents a stage when Brahmanical priestly incursion into image worship was already well established. Earlier, in the 7 th century, not too long after Varāhmihira’s Bṛhatsaṃhitā, Bāṇa in the Harṣacarita, chapter 8, had listed Bhāgavatas and Pāñcarātrins as two entirely separate groups.28 I would like to pursue this suggestion that the Bhāgavatas were the image specialists and that they were distinct from the Pāñcarātrins at one point in time. It was only gradually that the Pāñcarātrins appropriated the Bhāgavata concerns with image worship, so that eventually Yamuna could consider the two as one, opposed to the low-caste temple custodians.29

  • 30 For a detailed discussion of possible dates and the complicated history of the text see Grünendahl, (...)
  • 31 Viṣṇudharmāh, 3: 51.

26Such an hypothesis is not entirely wanting in evidence, although we shall see that it leaves many questions unanswered. In his edition and study of the Vaiśṇava ritual text known as the Viṣṇudharmāh, Reinhold Grünendahl has suggested that the Viṣṇudharmāḥ was originally a Bhāgavata text that was later appropriated by the Pāñcarātrins through the addition of a frame.30 He argues further that it was not difficult for the Pāñcarātrins to do so, since on the evidence of a later text like the Viṣṇurahasya, the Pāñcarātrins and Bhāgavatas would seem to have shared the same rituals. According to Grünendahl, the Viṣṇurahasya is unmistakeably Pāñcarātra in its theology, while its rituals focus on such things as vows (vrata) and gifts (dāna), the very rituals emphasized in the Viṣṇudharmāḥ. Indeed they have much of their material in common. Grünendahl further notes that both the Viṣṇudharmāḥ and the Viṣṇurahasya differ from the Pāñcarātra Saṃhitās, which show, he says a “strong emphasis on a (pseudo-) Vedic fire cult.”31 His conclusion is that Bhāgavatas and Pāñcarātrins were originally separate groups with considerable ritual overlap. He does not speculate about the significance of his observation that the Saṃhitās propose a very different ritual paradigm.

  • 32 Brahma tasyāṃ samāropya manasā tanmavo bhavet/tām arcayet tām praṇamet tāṃ smaret tāṃ vicintayet. I (...)

27I would like to do just that. The Viṣṇudharmāḥ offers us important information, and I think that its significance is far reaching for a study of the reception of the image cult in orthodox circles in early medieval India. My hypothesis in this paper is that there was a time when image worshipping Vaiśṇava specialists (we may call them with Varahamihira Bhagavatas) and Pāñcarātrins were separate groups. Furthermore, the Pāñcarātra Saṃhitās represent at least in part the orthodox Brahmin effort to appropriate these image worshipping cults by offering a distinct set of rituals that only its own priests were qualified to perform. A reading of the Viṣṇudharmāḥ suggests that the Bhāgavatas of this text had expanded on the devotional religion of the Bhagavadgītā, developing more explicitly image worship as a method of expressing devotion (2.66-83). It is fascinating to note that the ways the text suggests one should worship images correspond closely with the despised tasks that Yamuna ascribes to the non-Brahmin priests, smearing the temple with dung, sweeping the temple, painting its walls (chapter 31). There is even a long story about a śūdra who became the glorious king Mandhātṛ because he served in the temple, cleaning it and washing the floor (82). The rituals of worship elsewhere in the text include making offerings, particularly of lamps and flowers, but there is no hint of the complex rituals of the Saṃhitās, neither of the fire altars nor of mantras laid on the body (nyāsa) to turn the worshipper into god. Instead the worshipper is told in 99.28, “Imagining the image to be Brahman, one should in one’s mind become totally identified with Brahman; one should worship the image, bow down to the image, meditate on it and think constantly of it.”32 Like the Bhagavadgītā, this text is also intensely interested in death-bed thoughts (68) and rituals for the death-bed (79); it has numerous references to the God of Death, Yama, and how to conquer him and his minions.

  • 33 It is clear that the paradigm of image worshipping Bhāgavatas on the one hand and Pāñcarātrins need (...)

28I would like to suggest that a text like the Viṣṇurahasya, or the better known and more easily accessible, Viṣṇudharmottarapurāṇa, which is regarded as a sequel to the Viṣṇudharmāḥ under discussion, represents a stage of early Pāñcarātra appropriation of the temple cult. At this stage there was considerable speculation about the nature of God and only gradually an effort to distinguish themselves from the Bhāgavatas by the creation of new rituals and new ritual markers. The Viṣṇudharmottara still glorifies the rituals of the despised image-tending priests, as does the Viṣṇurahasya, from the description of it given by Grünendahl. It is difficult to imagine what Yamuna, with his clear distinction between Brahmin temple priests and their rituals and the low-caste temple priests who sweep and clean the temple, would have done with this anomaly.33

29The observation that what is distinctively Pāñcarātrin about a text like the Viṣṇudharmottara is less the ritual sections and more the theology, leads us to a consideration of theology and the diversity of problems the tradition saw in image worship. By no means all of the objections to image worship focussed on the caste or social status of its priests. Some of the philosophical problems that arose with images are also explicitly discussed in the texts. For example, some of the Purāṇas contain extended discussions about the legitimacy of image worship. Such passages alert us to the different areas which the tradition found problematic. While it is beyond the scope of this paper to deal comprehensively with objections to image worship, below I do discuss a section of the Viṣṇudharmottarapurāṇa on the worship of God in the form of an image. I have chosen this section of the text because its arguments are consistent with the gaps I will point to in the ritual texts themselves, which in turn suggest to me the ‘collision’ of ritual cultures and ritual groups, and the presence of at least two different groups of ritual specialists, whom I am calling here the Brahmin priests and the image priests.

  • 34 While I am aware that the term ‘ritual inconsistency’ raises the spectre of asking of the texts tha (...)

30It is in fact these ritual texts that to me offer the most interesting evidence for a slow process of acceptance of image worship in orthodox circles. It is precisely these Āgamas to which modem scholars often turn for iconographic information and for clues to the interpretation of the ‘meaning’ or ‘symbolism’ of temples and their images in medieval India. After I consider the evidence of the Viṣṇudharmottarapurāṇa I shall turn to a consideration of some early Āgamas, like the Pāñcarātra Jayākhyāsaṃhitā, in an effort to show that in fact they are not yet entirely comfortable with the cult of the permanent image in a temple; their rituals are more appropriately directed towards impermanent ritual objects. Moreover their rituals do not really make sense when directed toward the permanent image of the temple; indeed, there is an inconsistency to some of these texts that suggests that the theology of the text was not entirely compatible with the cult of images. I will select a few examples of what I call ‘ritual inconsistency’, where a given ritual is in conflict with another ritual or a widely held religious belief in the text.34 My hypothesis is that we can make sense of these moments of ‘ritual inconsistency’ if we see in these early texts the beginnings of a priestly appropriation of image worship and the conflict of two different belief systems. The ramifications for modern scholars, I think, could be profound, if it is the case that image worship was really somehow alien to the dominant religious world of the Brahmin/Vedic or Pāñcarātra priests; at best we would need to exercise great caution in using the theology of these texts to interpret images well until late medieval times, when the priestly appropriation of image cults seems to have been well established (we need to recall that Ballālasena in the 12 th century still seems to have trouble with it and that Yamuna in the 11 th century still had to argue hard in favor of priests’ participation in image worship). My own study of the ritual texts has just begun; what I will present here are a few tantalizing cases that I hope can establish that a more detailed investigation is of interest. I turn now to consider the Viṣṇudharmottarapurāṇa and philosophical objections to image worship, followed by a discussion of ritual inconsistency in the ritual texts themselves.


  • 35 I have discussed this material elsewhere in greater detail (Granoff 2003). The date of the Viṣṇudha (...)

31The Viṣṇudharmottarapurāṇa is generally considered to be a Pāñcarātra text. It has been mined for its wealth of detail about the building of temples and construction of images. Despite this profusion of information about image worship, the text also contains an important debate about the legitimacy of worshipping images.35

  • 36 On this question see Malamoud 1990: 183-191.

32The objections in the Viṣṇudharmottarapurāṇa (3.108) stress the incongruity between a god that is described as formless and all-pervasive somehow being worshipped in a body that has a visible form and in rituals where it is necessary to summon the god into the image. The chapter opens with this second point; what is the purpose of the ritual of āvāhana, or summoning the god into the image at the beginning of each worship ceremony or pūjā. An omnipresent God is already there by virtue of his omnipresence. One might add that a properly consecrated image also should not require further invoking of the god into the image, while Purāṇic stories about images describe a contract between a deity and a worshipper in which the deity agrees to abide permanently in a particular place. We shall see, when we move to a discussion of what I have called ‘ritual inconsistency’ in the ritual texts themselves, that the rituals of summoning the god and dismissing the god (visarjana) remained problematic. I shall argue that the persistence of these rituals in image worship may be seen as evidence of a complex historical process by which image worship eventually came to be appropriated by the Vedic priestly circles in which the ritual summoning and dismissing the deity was appropriate. It has been convincingly argued that the very notion of the god residing in a place permanently, a notion that would seem to be essential to the idea of image worship, was contradictory to Vedic religion, in which the gods were only occasionally summoned to be present at the sacrifice.36

33Having begun with the ritual of invocation, the Viṣṇudharmottarapurāṇa goes on to question the entire worship ceremony that follows the invocation to the god; since the god has all his wishes fulfilled, what could the god possibly gain by the worship we offer? The reply concedes that it is fundamentally a contradiction to summon a deity who is already there, since he is all-pervasive, and that worship cannot serve to satisfy the deity, who is complete in and of himself. Nonetheless, neither ritual is without purpose; they bring satisfaction to the worshipper.

  • 37 Āvāhanaṃ svaṭrptyarthaṃ yathāsthitam / tathāpyarcā mahābhāga svatuṣtyarthaṃ prayojanam // [15] nity (...)

Just as it has been shown that the ritual of invocation serves to satisfy the worshipper, so the ceremony of worship has as its purpose satisfying the worshipper. [15] If you ask what God, who is always satisfied, can gain from worship, the reply is that he accepts worship from his devotees simply to satisfy their needs. [19]37

34This is repeated in the next verse, where it is said, “tasyārcākāraṇaṃ viddhi kartṛprītiprayojanam”, “Know that the reason for worship is that it brings joy to the worshipper.”

35There are other questions. If the highest God has no form, which is the way he is described in many authoritative texts, then what are we worshipping when we worship an image? In chapter 46 we read:

You have said that God is devoid of qualities like color, taste and odor; devoid of touch or sound. How can he then have a physical form?

36Mārkaṇḍeya replied,

  • 38 Rūpagandharasair hīnaḥ śabdasparśavivarjitaḥ / puruṣas tu tvayā proktas tasya rūpam idam katham // (...)

The Highest Soul has two forms, one that is his original form and one that is his transformed (or derived) form. The form that cannot be seen is called his original form. The form that has a sensible physical dimension is called the transformed form. The whole world is in fact his transformed form. His transformed form is the only form that can be worshipped or can be the object of meditation. One should worship the highest soul in the proper way; it is difficult for mortals, who themselves have bodies, to worship something that is unmanifest. For that reason, of his own free will, the highest soul in his various appearances displays a sensible form which can be seen; the other gods do the same. [1 -5]38

  • 39 Nāmopāssva brahmeti brahmabuddhyā. yathā pratimā viṣṇubuddhyopāste tadvat.

37The reply borrows the terms prakṛti and vikṛti; like the nirmāṇa kāya of the Buddha, which he creates at will, there is a transformation of the god’s body in the Viṣṇudharmottapurāṇa. Indeed it is only this transformed body that can be the object of meditation or worship of any kind. This last point should be borne in mind. For this text at least, it would not make sense to consider any physical form of the god as representing in any way his highest form. The highest form is inaccessible to worship. At best this seems to be a grudging acceptance of image worship, in which the essence of the god remains inaccessible to the worshipper. One might recall the great philosopher Sankara’s interpretation of image worship. In his commentary to the Chāndogyopaniṣad 7.1.4, in which it is enjoined to worship name, nāman, as various things, from the Vedas to Brahman, Sankara explains that this means that one should think of nāman as Brahman, etc., in the same way in which one worships an image, conceiving in one’s mind that it is God:39 ‘ “worship nāman as Brahman’, means with the notion that it is Brahman, just as one worships an image with the notion that it is Viṣṇu.”

38What is obvious from this is that just as nāman is not really Brahman but is only conceived to be Brahman far the purpose of worship, so the image, which is not really Visnu, is imagined to be Viṣṇu for the purpose of worship. Nothing could be further from the language of the inscriptions or Purāṇic stories in which the image is clearly god.

39It seems undeniable that apologies for image worship, such as the Viṣṇudharmottarapurāṇa offers, regard the image as a derived form of the god, something other than the highest God. In addition the text is aware of inconsistencies between theology and ritual. A god who is omnipresent does not need to be summoned and a god whose needs are all met can have nothing to gain from worship. The answer that the text gives, that these things are for the satisfaction of the worshipper, is again belied by the countless Purāṇic stories in which it is the god who is satisfied by the worship and comes to offer a boon to the worshipper. This kind of cognitive gap suggests to me that somehow we are in very different religious worlds. Theology and practice, in this case the worship of images, somehow are at odds. I turn now to a further discussion of such ‘gaps’, this time in the ritual texts, and what we can learn about them. My argument will be that at least in some cases, gaps such as these suggest to us that the texts have a complex history and the practices they describe did not originally belong together.


  • 40 See for example Brunner 1986: 89-132; 1993: 1-61; 1990: 4-24.
  • 41 For the Śaiva evidence see Brunner 1990 and the texts she cites. As will be clear below, my own int (...)
  • 42 On the date of this text as between 600-850 A. D., see Rastelli 1999: 25-27. It is first cited in t (...)
  • 43 Ārādhito ‘si bhagavan sādhakānāṃ hitāya ca // tvayā ‘py anugrahārthaṃ ca vastavyam iha sarvadā [20. (...)
  • 44 For a parallel from a Śaiva text see the Ajitāgama 18.264 (p. 167). For my references to the Śaiva (...)
  • 45 See the Cintra praśasti of Saraṅgadeva cited above, Epigraphia Indica, I: 271 ff. “sākṣād bhavanti (...)

40Hélène Brunner in her extensive studies of the Śaiva Āgamas has noted more than once that there are often inconsistencies in the ritual texts.40 Here I return to the rituals of āvāhana and visarjana, which we saw the Viṣṇudharmottarapurāṇa felt called upon to explain. A number of ritual texts, Śaiva and Vaiṣṇava, seem to be acutely aware that there is a problem with the rituals of āvāhana and visarjana.41 I will center my discussion around the Jayākhyasaṃhitā, which is considered by scholars to be an early Pāñcarātra ritual text. The Jayākhyasaṃhitā is well aware that there is a problem with āvāhana, the first of the series of daily rituals.42 Given its preoccupation with ritual, the Jayākhyasaṃhitā deals with the problem of the āvāhana as a problem of conflicting rituals, rather than an incompatibility between doctrine and ritual, which we saw in the Viṣṇudharmottarapurāṇa. To begin with, images, in order to be worshipped, must be duly consecrated and enlivened. The implication is that through the ritual of consecration the god is made to reside in the image for a duration of time beyond the ritual of consecration itself. The consecration rituals in the text make this explicit. We are told that the purpose of the ritual is to ensure that the god is present for all times (yena sannihitaḥ sadā, 20.208). Indeed the ritual turns the image into God, “By this means the image becomes the god of all” (yena sarveśata vipra bimbasyāsya prajāyate, 20.221). At the end of the ritual the sādhaka addresses the image as God, saying, “O Lord! I have worshipped you for the welfare of the sādhakas; now in order to shower us with your favor, you must reside here forever.”43 The text seems unambiguous on this point: once an image is consecrated, it is God; no further ritual should be required to bring God into the image.44 That this belief was widespread is further confirmed by the language of medieval inscriptions. In a 13th-century inscription at the Śiva temple at Somanātha, for example, we are told that even stones consecrated by the proper hands turn right into gods.45

  • 46 See for example Schopen 1990: 181-219; Granoff 1993: 66-94. The question must arise whether the ear (...)

41Nonetheless, as if in direct contradiction to these fundamental beliefs underlying the consecration ceremonies, the description of daily worship in the Jayākhyasaṃhitā, as in other ritual texts, both Śaiva and Vaiṣṇava, begins with the ritual of āvāhana, inviting the god to be present in the image. The simplest explanation for this collision of rituals seems to me to be that these two groups of rituals, the rituals for consecration and those for daily worship, could not have been intended for the same ritual objects. The rituals for consecration as we have them in texts like the Jayākhyasaṃhitā with their prayers to the god to abide permanently in the image have a close parallel in medieval discourse about images as God that we find in the inscriptions and Purāṇic stories. In Purāṇic stories, for example, the god is requested to reside permanently in a place, just as we have seen that the god is requested at the close of the consecration ceremony.46 By contrast, the rituals for daily worship must have been devised with ritual objects other than the permanent image of a temple in mind, that is, for impermanent ritual objects. Further, I propose that the incompatibility of rituals like these in the ritual texts is evidence of a clash of religious values that can be best explained by assuming that the rituals surrounding the permanent images of the temple originally belonged to one group of ritual specialists, while the rituals of daily worship belonged to another group. That the rituals come into open conflict suggests that the one group, with its rituals and their implied assumptions, was moving into the sphere of activity of the other group.

  • 47 Rastelli 1999: 284 fn. 1265. Brunner (1993: 28) suggests that in fact the earlier Śaiva Āgamas deal (...)

42In fact there is evidence within the text itself to support such an hypothesis. Although the Jayākhyasaṃhitā mentions that its rituals can be applied to the worship of an image in a temple, it drops this mention fairly quickly and applies its rituals for the most part to temporary ritual objects made for the ritual in question. The reader notes fairly early on that the rituals are not really applicable to or intended for temple images.47 In addition, the text makes it explicit that at least some of its rituals cannot be applied or should not be applied to permanent images in the temple. It should come as no surprise that the ritual which it says should be omitted for temple images is the āvāhana. The Jayākhyasaṃhitā states that this ritual is not necessary for an image that bas been consecrated by sages, gods or devoted human beings (13.60). This effectively excludes all temple images from the first ritual of daily worship.

  • 48 G. Colas raised the interesting point here that even today a temple may quickly be considered divin (...)
  • 49 Colas 1996: 298. I hope in the future to make an in-depth comparative study of these rituals across (...)
  • 50 Brunner (1993: 12) in a discussion of the relationship between ritual and doctrine in the Śaiva Āga (...)

43It is interesting to see how this exception is treated in later texts. The later Viśvāmitrasaṃhitā 10.108 says that you do not need to perform the ritual of invitation only for those images that have been consecrated by the gods, siddhas or munis. What this new version of the exception to performing the āvāhana does is to ensure that the āvāhana is performed for man-made images and images consecrated by normal human beings; in other words, for most temple images.48 To me this implies a growing encroachment of priestly rituals onto the temple cult: now all but a few images require the full panoply of rituals, including the incoherent one that invites a god who is already present in the image! The Vaikhānasa texts also note that āvāhana is a problem for the main image; they direct many of their rituals to a temporary image and the divine presence is invited from the main image into the temporary image.49 It seems to me odd indeed to imagine that these ritual texts, which limit the role of the main image of the temple in their rituals, originally developed in the context of temple worship, no matter how they are used today. Once again I think we have an indication that the rituals of our priestly texts and temple worship are not perfectly compatible.50


  • 51 Brunner (1993: 116) in her discussion of the Śaiva texts concludes that the rituals of āvāhana coul (...)
  • 52 In her discussion of the treatment of śakti in the rituals, Brunner (1993: 22) notes further incons (...)

44I have suggested that the incompatibility centers on the worship of permanent objects which was done by one group of ritual specialists and the worship of impermanent objects, which was the sphere of a different group of specialists.51 I have further suggested that the ritual specialists of the āvāhana ritual paradigm were gradually moving into the area of image worship; they were by their own self-description Brahmin priests and I would like to argue that their ritual paradigm was most at home in the religious world of the Vedic sacrifice in which the gods were summoned to the sacrifice and then dismissed and in which ritual objects were impermanent. By contrast, the worship of permanent images in the temple belonged to another group entirely, perhaps that of Yamuna’s prāsādapālakas, whom we met at the start of this paper.52

  • 53 That this is only one of many competing discourses on image worship is clear from Tarabout 1997: 25 (...)

45These groups also had very different understandings of god; in the priestly circle god is much like the absolute of the Upaniṣads, formless, infinite in extension and eternal, while the god of the image worshippers, if we may judge from their stories and temple inscriptions, was concretely present in his image. It is the new priests then who had to struggle with a number of dissonant concepts: with rituals more appropriate to impermanent objects into which the divine presence was summoned and dismissed, and a theology of an all-pervasive and formless deity who now somehow had to be made to fit into the confined and physically defined space of the image in a way that would make image worship a meaningful religious activity. I would add that it was not just the Brahmin priests who treated deities as beings who could be invited and then dismissed when they have performed their function. This was an understanding of the gods and how they functioned that transcended sectarian affiliations and persisted well into early medieval times. A Jain story from the Nāyadhammakahāo 1.16 tells about clever prince Abhaya who wants to fulfill his mother’s pregnancy longing. His mother wants to enjoy the sight of a forest in the rainy season, but unfortunately the rainy season has not yet arrived. Abhaya realizes that only divine aid can help him and so he goes into the posaha hall and practices austerities. A god to whom he had a prior connection comes in response to Abhaya’s summons. He makes it rain and the pregnancy longing is fulfilled. Abhaya then dismisses the god. The term used is padivisajjeti, from the same verb as the ritual visarjana. I would add that it is the same word used for the dismissal of servants, whom one summons and then dismisses when they have received their orders. This suggests to me that there were many contradictions involved for the priests who sought to impose their own ritual structure on image worship; inviting and dismissing a god may well have had other connotations that were not compatible with an ideology of image worship such as that expressed in the Purāṇas in which it is the devotee who is the eternal servant and the god the eternal master.53

46I think that there are other indications in Āgamic texts that reveal a major gap between the Āgamas and the image worshipping cults, some information about which can be gleaned from the Purāṇas. The goal of worship is an interesting question. It is well known that Āgamic and Tantric ritual texts demand of the worshipper that through elaborate procedures of nyāsa the worshipper becomes god. Only then can he worship god. But if we read Purāṇic accounts of the merits of worshipping images or deities, we learn that the worshipper, quite fallible and mortal, becomes god as the result of his worship. Viṣṇu comes in a chariot to whisk him off to heaven, or he is absorbed into the Śivaliṅga and becomes one of Śiva’s eternal companions. The worshipper takes on the form of the god he has worshipped. The stories are striking and unambiguous: the divine transformation only occurs at death, and often after a battle between the god and Yama’s messengers who come to claim the soul for Yama, the king of the Dead. The Jayākhyasaṃhitā seems to be aware of the gap between its ritual of making a divine body at the start of worship and the Purāṇic goal of becoming divine at death. It offers this comment on why the practitioner must turn himself into god before worshipping god:

  • 54 Yanmayas sādhako vipra dehasthas sāmprato bhavet / tanmayo dehapātāt syād ity etat kathitam mayā // (...)

O Brahmin! The form that the sādhaka achieves while he is alive will be the form that he has after death. [13.9]54

  • 55 On this question see Malamoud 1988: 15-32.

47On the surface of things, this seems to be a strange statement. But if you know Purāṇic stories about the glories of worshipping Śiva or Visnu, with the reward of becoming like Śiva or Viṣṇu on your death, the comment at least seems to have a reason behind it. The goal of Tantric worship, making the sādhaka immortal here and now, giving him a glorious adamantine body here and now, is in this passage assimilated to the Purāṇic goal of a god-like after life. That the author feels the need to explain why the practitioner must get a divine body in the ritual suggests to me that he was aware that it somehow did not fit in with some other fundamental assumption about the goal of worship. I would add that the notion of getting a temporary divine body in a ritual context is perfectly familiar from Vedic rituals, in which the sacrificer gets a divine body through the consecration or dīkṣā, leaving his profane body in the care of the officiating priests.55 Perhaps we have here another indication of a priestly incursion into alien territory.

  • 56 Desai 1996 and the articles referred to therein. pal (1975: 66) had argued that the Vaikuṇṭha image (...)

48A slightly different example of inconsistency, I think, can be found again in the Jayākhyasaṃhitā 6.73-76 when it discusses the dhyāna or meditation formula for the main or mūla mantra. The description is of the form of Viṣṇu known as Vaikuṇṭha and has been used as the textual source for images of Vaikuṇṭha by art historians. The problem is that the anthropomorphic form of the mantra is described as having four faces, but when the mantras are given for the faces there are only three mantras: varāha, narasiṃha and kapila. What makes this particularly interesting is that there is a sculptural tradition of three-headed Vaikuṇṭha images, two of which are at Khajuraho. In addition, the long inscription in the Vaikuṇṭha temple there describes a three-headed deity (despite the efforts of scholars to misread the first verse to make the deity four-headed!).56 There is an inconsistency here in the description of the deity as four-headed and the mantras for the heads being only three. I believe that it can also tell us about the complex history of these texts and their relation to other kinds of religious practices. I suspect that the original cult in the temple was of the three-headed deity, who was then brought into line with the abstruse theology of the four-vyūhas and four-headed Viṣṇu of the Pāñcarātra texts. Here in the mantras we have a case where the appropriation of an iconic form by the theologians was still incomplete.

49I think that all of these inconsistencies, the problems with the āvāhana ritual, its clashing with Purāṇic ideology and with another important set of Agamic rituals, those used to consecrate images; the comments like that of the Jayākhyasaṃhitā on why the worshipper turns himself into god; the three mantras for the four heads, can tell us something very important about the history of the ritual texts and the groups who promoted them. I would like to see in these inconsistencies a growing appropriation by priestly groups of practices that were not originally theirs.


  • 57 On the introduction of Vedic rites into the Vaikhānasa ritual, see Colas 1999: 41-50.
  • 58 Rāghava Bhaṭṭa in his commentary to the Śāradātilakatantra, p. 1.

50There is a variety of evidence of different types and from different genres of texts that suggests that orthodox priestly circles were uncomfortable with professional worship of images and the priests who devoted themselves to image worship even at a relatively late period. At some point in time, Brahmin priests, or at least priests who wished to be considered as orthodox Brahmins, began to be occupied with image worship. Initially their connection may have been limited to their serving as officiants at rituals of consecration of temples and images and building of temples. Their contribution to the worship of images was manifold; they introduced a complex religious symbolism and cult that had developed in the context of Vedic rites and that initially had not been part of either the worship or the conceptualization of images.57 One of the earliest of these groups of Vedicizing image worshippers is the Pāñcarātra, whose texts tell us of their coming to terms with image worship. A Pāñcarātra apologist, Yamuna, also tells us that the rituals these orthodox priests practiced were different from those of the other, despised image-worshipping priests. The new rituals were modelled on Vedic rituals, to such an extent that a late commentator on a Tantric text could describe the Vedas as three-fold: the karma kāṇḍa of the Vedic ritual, the jñāna kāṇḍa of the Vedanta, and an upāsanā kāṇḍa, consisting of Tantric and Agamic texts.58 The Gītabhāṣya attributed to Sankara had described the Vedas as two-fold, making no mention of any upāsanā kāṇḍa.

51The growing priestly interest in images and temples ensured that we have today an impressive body of texts containing iconographical information and theological speculation on the meaning of images and temples, but I hope that I have shown here that some key speculations about images and important rituals in these texts are diametrically opposed to each other and to what we know from other types of texts, inscriptions and Purāṇic stories, about how images were regarded.

52The practical implications for this hypothesis are several. Most importantly, it means that when we seek textual explanations for images and temples we must treat these texts of our Vedicizing Brahmin temple priests with some measure of caution, at least until well into the late medieval period. An early text like the Jayākhyasaṃhitā, if it is as early as the 9 th century, indicates that priestly appropriation of these matters was just beginning. A text like the Viṣṇudharmottarapurāṇa may in fact indicate that early priestly appropriations did not draw the ritual boundaries that Yamuna would draw; the new priests initially kept the old rituals.

53As is the case with all hypotheses, this one also must stand up to the hard evidence of historical fact as it may be determined from sources such as inscriptions. We might expect that the picture that will emerge will show regional variation and will be far more complex than this one hypothesis allows; the hypothesis may well need to be refined as research progresses. Nonetheless, inscriptions from Orissa, for example, do suggest that there is little evidence of the association of Brahmins with temples, while there is evidence of the functioning of non-Brahmins at temples in a priestly capacity. In a study of the inscriptions of Orissa from the 5 th to the 12 th century A. D., Upinder Singh suggests either that the orthodox Brahmins were not associated with temple religion to a significant extent for the period, or that the inscriptions did not mark this aspect of their activities, as he notes, “an association with temples not being considered with favour in the orthodox tradition.” The inscriptions prefer to depict the Brahmin as Vedic scholar, performing the rituals sanctioned by the Law Books, for example, studying, teaching, sacrificing and receiving gifts (Singh 1994: 295). Leslie Orr in her recent studies has highlighted the lack of interest in Brahmins in temple inscriptions of the Chola period. She has also cited evidence for considerable non-Brahmin participation in ritual roles and suggested that our understanding of the ‘orthodoxy’ of the great Vaiśṇava and Śaiva centers is more appropriate to late medieval times. Indeed, she notes that on the few occasions when Brahmins are specifically identified as such in the inscriptions, they are not involved in temple management or ritual services. They are more commonly temple patrons and landowners (Orr 2000: 23-25, 30). In a study of early medieval Gujarat, Marlene Njammasch (2001: xx) has suggested much the same, namely that Brahmins had little or no role in temple cults until a relatively late date. Early medieval inscriptions, for example, never identify temple priests as Brahmins. She dates the efforts of Brahmins to identify themselves with the temple cults to as late as the 10 th century A. D.

54Now we are still much in the dark about the gradual process of Brahmin interest in temple worship; we are also in the dark as to who exactly the original image worshippers were and what they thought.

  • 59 Śastrābhilāṣāt sutarām mahātmā saṃtyajya deveśvaradāsyakṛtyam / kāmcīpure lakṣmaṇadeśikendras tasyā (...)

55A Sanskrit hagiography of Rāmānuja, the Prapannāmṛta written in the 13 th century by Anantācārya, also suggests that there was a distinct division of labor between priests who tended images and the Brahmins, who were custodians of sacred texts. The Prapannāmṛta further suggests that the image-tending priests were śūdras. Ramanuja first meets the śūdra who serves Viṣṇu on Hastigiri after he has fallen out with his Brahmin teacher Yādava, who had indeed plotted to kill the young Ramanuja. Under the tutelage of this Kāñcīpūrṇa, who was part tribal, or Śabara, Rāmānuja begins to serve the god by bringing him water from a well in a golden vessel (5.35). But Ramanuja is intent on learning the śāstras, the sacred texts, and so he abandons the service of the god and returns to Yādava, in the words of the text, “And because he was desirous of learning the sacred texts, the great one, Lakṣmaṇendra, abandoned his tasks as servant of the lord of lords in the city of Kāñcī and went to Yādava as before” (6.2 8).59 Rāmānuja will have another disagreement with Yādava, this time over the interpretation of scripture, and he will return again to serve the god at Hastiśaila under the tutelage of Kāñcīpūrṇa, the śūdra (8.21-22 and 9.3), but eventually he will leave the service of the god and return to the fold of the Brahmins, when Kāñcīpūrṇa as a śūdra is unable to give the Brahmin Ramanuja the five Vaiṣṇava saṃskāras (10.40ff.).

  • 60 On the late history of the priests of the temple at Śriranagam see Jagannāthan 1994. It is worth no (...)

56The 13th-century Prapannāmṛta here depicts a distinct division between types of sacred duties. The śūdra, part tribal, is dedicated to the service of the lord, that is, the image, performing such tasks as bringing water for the god from the well. It is duties such as these that we read about in the Purāṇic stories, and that Yamuna in the Agamaprāmāṇya had in fact associated with the prāsāda pālakas, the low-caste priests. In the Prapannāmṛta, the Brahmins, on the other hand, read and study sacred texts, and formally initiate believers into the fold through rituals like the conferring of the five saṃskāras. Perhaps we see in this text the efforts of the Brahmin priestly community both to dissociate itself from the image-tending priests and to build a new role for Brahmins in the context of the temple service. The subject requires further study.60

  • 61 The Jain texts are reproduced in the edition of the Gaṇakārikā.

57There is also some tantalizing evidence about Śaiva practices. Medieval Jain descriptions of the Pāśupata tradition, for example the Ṣaddarśanasamuccayas of Haribhadra and Rājaśekhara, note specifically that their priests or pūjakas were not Brahmins; they are called bharāṭakas, a group better known in medieval literature as the brunt of off-color jokes. This is in sharp contrast to the practitioner of the Pāśupata ritual in the 10th-century Gaṇakārikā, who is definitely called a Brahmin. If the Jain evidence can be relied upon, we have here another indication that image worship belonged to a group different from the Brahmin theologians.61

58Evidence that may help corroborate the general hypothesis expressed here must thus be carefully gleaned from a variety of sources, from a variety of stories, religious biographies, inscriptions, ritual texts, and debates about image worship in the Purāṇas and philosophical texts.

59If the underlying assumption of this paper can stand the test of further investigation, we would have before us a case study of how ritual boundaries were gradually constructed by religious groups in medieval India who sought to define themselves as distinctive from their competitors through the creation of new rituals.

60P. G.



Sanskrit Texts

Agnipurāṇa. Kashi Sanskrit Series, 174. Varanasi ( “Chowkamba Sanskrit Series”), 1966.

Agamaprāmāṇya of Yāmuna, edited by R. M. Sastri. Pandit, 29, 1900.

Ahirbudhnyasaṃhitã, edited by Pandit M. D. Ramanujacarya. Madras, Adyar Library and Research Centre, 1986.

Ajitāgama, edited by N. R. Bhatt. Pondicherry, Institut francais d’Indologie, 1964.

Ballālacarita of Ānanda Bhaṭṭa. Calcutta ( “Bibliotheca Indica”), 1904.

Bṛhatsaṃhitā of Varāhamihira, edited by M. Ramakrishna Bhat. Delhi, Motilal Banarsidass, 1982

Chāndogya Upaniṣad in Ten Principal Upaniṣads with Śāṅkarabhāṣya. Delhi, Motilal Banarsidass, 1964.

Dānasāgara of Ballālasena, edited by Bhabatosh Bhattacharya. Calcutta ( “Bibliotheca Indica”), 1953.

Gaṇakārikā, edited by Chimanlal D. Dala. Baroda, Oriental Institute ( “Gaekwads Oriental Series” 15), 1966.

Harṣacarita, edited by Suranan Kunjan Pillai. Trivandrum, University Manuscripts Library, 1958.

Kṛtyakalpataru of Lakṣmīdhara, edited by K. V Rangaswami Aiyangar. Baroda, Oriental Institute ( “Gaekwad’s Oriental Series” 167), 1979.

Matsyapurāṇa. Pune, Anandāśrama Sanskrit Series, vol. 54, 1981.

Prapannāṃrta of Anantācārya, edited by Svāmi Rāmanārāyaṇācārya. Varanasi, Somānī Trust, 1966.

Rāmāyaṇa of Mādhava Kandalī. Delhi, Ajanta Prakashan, 1985.

Śāradātilaka Tantra with commentary of RĀghava Bhatta, edited by Sri Mukundajhā Bakhśī. Varanasi ( “Chowkhamba Sanskrit Series Office”), 1963.

Śivapurāṇa. Kāśī, Paṇḍita Pustakālaya, 1963.

Skandapurāṇa. New Delhi, Nag Publishers, 1986.

Skandapurāṇa, edited by Hans Bakker Gröningen, Egbert Forsten, 1998.

Vaikhānasagṛhyasūtra, edited by W. Caland. New Delhi, 1989. (Repr. of the 1941 ed.)

Viṣṇudharmāh: Precepts for the Worship of Visnu, edited by Reinhold Grünendahl. Wiesbaden, Otto Harrassowitz, 1989, 3 vols.

Viṣṇudharmottarapurāṇa. New Delhi, Nag Publishers, 1985.

Viśvāmitrasaṃhitā, edited by Undemane Shankara Bhatta. Tirupati, Kendriya Sanskrit Vidyapeetha, 1970.

Secondary Sources

Altekar, A. S. (1941-42), “Six Saindhava copper-plate grants from Ghumli”, Epigraphia Indica, XXVI, pp. 184-227.

appadurai, A. (1983), “The status of brahmin priests in Hindu India”, South Asian Anthropologist, pp. 43-56.

Brunner, H. (1986), “Les membres de Siva”, Asiatische Studien, 40, pp. 89-132.

Brunner, H. (1990), “Ātmārthapūjā versus parārthāpūjā”, in T. Goudriaan, ed., The Sanskrit Tradition and Tantrism. Leiden, E. J. Brill, pp. 4-24.

Brunner, H. (1993), “Jñana and Kriyā: Relation between theory and practice in the Śaivāgamas”, in T. Goudriaan, Ritual and Speculation in Early Tantrism: Studies in Honor of André Padoux. Delhi, Sri Satguru Publications ( “Sri Garib Dass Oriental Series” 163), pp. 1-61.

Brunner, H. (1994), “The place of yoga in the Śaivāgamas”, in P.-S. Filliozat et al., eds., Pandit N. R Bhatt Felicitation Volume. Delhi, Motilal Banarsidass, pp. 425-461.

Bühler, G. (1892), “The Cintra praśasti of the Reign of Saraṅgadeva”, Epigraphia Indica, I, pp. 271-287.

Colas, G. (1986), Le Temple selon Marīci. Pondicherry, Publications de l’Institut français d’indologie.

Colas, G. (1990), “Sectarian divisions according to the Vaikhānasāgama”, in T. Goudriaan, ed., The Sanskrit Tradition and Tantrism. Leiden, E. J. Brill, 1990, pp. 24-32.

Colas, G. (1994), “On the Bauddhāyanagṛhyapariśiṣṭasūtra and the Vaiṣṇavāgamas”, in P.-S. Filliozat, et al., eds., N. R. Bhatt Felicitation Volume. Delhi, Motilal Banarsidass, pp. 511-527.

Colas, G. (1995), “Cultes et courants du vishnuisme en Inde du Sud. Quelques observations à partir des textes”, Puruṣārtha, 17: Les Ruses du salut. Religion et politiques dans le monde indien. Paris, Éd. de l’EHESS, pp. 111-139.

Colas, G. (1996), Viṣṇu, ses images et ses feux. Les métamorphoses du dieu chez les vaikānasa. Paris, Presses de l’École française d’Extrême-Orient.

Colas, G. (1999), “The reworking of ‘Vedic’ paradigm in Medieval liturgies”, in J. Assayag, ed., Les Ressources de l’Histoire. Tradition, narration et nation en Asie du Sud. Paris/Pondicherry, École française d’Extrême-Orient, pp. 41-50.

Das Gupta, S. (1967), History of Indian Philosophy. Cambridge, Cambridge University Press, 2 vols.

Davis, R. (2001), “Indian image-worship and its discontents”, in J. Assmann & A. Baumgarten, eds., Representation in Religion: Studies in Honor of Moshe Barasch. Leiden, Brill.

Desai, D. (1996), The Religious Imagery of Khajuraho. Mumbai, Project for Indian Cultural Studies.

Dvivedī, V. (1984), Āgama Aur Tantra Śāstra. Dillī, Parimala Pablikeśan.

Fuller, C. J. (1984), Servants of the Goddess: The Priests of a South Indian Temple. London, Cambridge University Press.

Goodall, D. (1998), Bhaṭṭa Rāmakaṇṭha’s Commentary on the Kiraṇatantra. Pondicherry, Institut français de Pondichéry, EFEO.

Goudriaan, T. (1965), Kāśyapa’s Book of Wisdom (Kāśyapa-jñānakāṇḍah): A Ritual Handbook of the Vaikhānasas. The Hague, Mouton & Co.

Granoff, P. (1993), “Halāyudha’s prism: the experience of religion in Medieval hymns and stories”, in V Desai & D. Mason, eds., Gods, Guardians and Lovers: Temple Sculptures from North India A. D. 700-1200. New York, Asia Society, pp. 66-94.

Granoff, P. (1998), “Defining sacred space: priestly conflict, compromise and control of the sacred site in Hindu Māhātmya texts”, Annals of the Bhandarkar Oriental Research Institute, 79, pp. 1-28.

Granoff, P. (2000), “Other people’s rituals: ritual eclecticism in early Medieval Indian religions”, Journal of Indian Philosophy, 28 (4), pp. 399-424.

Granoff, P. (2001), “My ritual, my gods: ritual exclusivism in Medieval India”, Journal of Indian Philosophy, 29 (1-2), pp. 109-131.

Granoff, P. (2003), “The absent artist as an apology for image worship: an investigation of some Medieval Indian accounts of the origins of sacred images”, in O. Qvarnstrom, ed., Jainism and Early Buddhism. Essays in honour of Padmanah Jaini. Part 2, Berkeley, Asian Humanities Press, pp. 187-221.

Gupta, S., Hoens, D. J. & Goudriaan, T. (1979), Hindu Tantrism. Leiden, E. J. Brill ( “Handbuch der Orientalistik” sect. 2, vol. 4).

Jagannathan, Smt. Sarojini (1994), Impact of Sri Rāmānujācārya on Temple Worship. Delhi, Nag Publishers.

kane, P. V (1962), History of Dharmaśāstra, vol. V, part II. Pune, Bhandarkar Oriental Institute.

Kreisel, G. (1987) “Caturmukhaliṅga and Maheśamūrti: the cosmic aspects of Siva”, Tribus, vol. 32, pp. 105-116.

Malamoud, C. (1988), “The contractual body of the gods: remarks on the Vedic rite of Tānūnaptra”, in R. Gombrich, ed., Indian Ritual and its Exegesis. Delhi, Oxford University Press ( “Oxford University Papers on India” 2), pp. 15-32.

Malamoud, C. (1990), “Sans lieu ni date. Note sur l’absence de fondation dans l’lnde Védique”, in M. Detienne, Tracés de Fondation. Louvain/Paris, Peeters/EPHE, pp. 183-191.

Nandi, R. N. (1979-80), “Client, ritual and conflict in early brahmanical order”, Indian Historical Review, VI, pp. 64-118.

Njammasch, M. (2001), Bauern, Buddhisten und Brahmanen: Das frühe Mittelalter in Gujarat. Wiesbaden, HarrassowitzVerlag (Asien- und Afrikastudien der Humboldt-Universität, Berlin 2).

Orr, L. (2000), Donors, Devotees and Daughters of God: Temple Women in Medieval Tamilnadu. London, OUP.

Pal, P. (1975), The Bronzes of Kashmir. Graz, Akademische Druck.

Rastelli, M. (1999), Philsophisch-theologische Grundanschaungen der Jayākhyasaṃhitā. Wien, Verlag der Osterreichischen Akademie der Wissenschaften.

Reiniche, M.-L. (1989), Tiruvannamalai, vol. 4. Paris, École française d’Extreme-Orient.

Rocher, L. (1986), The Purāṇas. Wiesbaden, Otto Harrassowitz ( “A History of Indian Literature” 2-3).

Sanderson, A. (1992), “The doctrine of the Mālinīvijayottaratantra”, inT. Goudriaan, ed., Ritual and Speculation in Early Tantrism: Studies in Honor of André Padoux. Albany, State University of New York Press, pp. 281-232.

Schopen, G. (1990), “The Buddha as owner of property and permanent resident in medieval Indian monasteries”, Journal of Indian Philosophy, 18, pp. 181-219.

Shinohara, K. (2004), “Stories of miraculous images and paying respect to the three jewels: A discourse on image worship in Seventh-Century China”, in P. Granoff & K. Shinora, eds., Images on Asian Religions: Texts and Contexts. Vancouver, UBC Press, pp. 180-222.

Singh, U. (1994), Kings, Brāhmaṇas and Temples in Orissa: An Epigraphic Study. New Delhi, Munshiram Manoharlal.

Smith, H. D. (1978), The Smith Āgama Collection: Sanskrit Books and Manuscripts Relating to Pāñcarātra Studies. Syracuse, Maxwell School of Citizenship and Public Affairs.

Stietencron, H. von (1971), “Orthodox attitudes towards temple service and image worship in Ancient India”, Central Asiatic Journal, 2, pp. 126-138.

Strong, J. (1989), The Legend of King Aśoka: A Study and Translation of the Aśtokāvadāna. Delhi, Motilal Banarsidass.

Tarabout, G. (1997), “Maîtres et serviteurs, commander à des dieux au Kérala (Inde du Sud)”, in A. de Surgy, A. Padoux & P. Lory, eds., Religion et pratiques de puissance. Paris, L’Harmattan, pp. 253-284.


1 I thank Gérard Colas and Gilles Tarabout for inviting me to participate and for their extensive comments on the paper.

2 Vaikhānasagṛhyasūtra, p. 62. The passage is discussed in Colas 1996: 22-26. See also Colas 1994: 511-527 and 1995: 119-139; Goudriaan 1965: 198-199. While Colas 1996: 24 notes that the strict distinction between private and public worship in terms of svārtha (worship for one’s one sake) and pārārthapūjā (worship for the sake of another) is a late medieval contribution to the discussion, there is no question that the image in the Vaikhānāsagṛhyasūtra is not the large image of the temple but is part of domestic worship. My argument in this paper will be that priestly interest in image worship in the temple setting is a later development and one that is fraught with problems.

3 The Purāṇas are notoriously difficult to date. For some information on the dating of various Purāṇas, see Rocher 1986.

4 Ed. Anandāśrama Sanskrit Series, vol. 54, 1981, chapter 266.

5 Pāṣaṇdinam yas tu karoti bhaktyā vihāya viprāñ chrutidharmayuktān/guruṃ pratiṣṭhādiṣu tatra nūnaṃ kulakṣayaḥ syād acirād apūjyaḥ / [267.34]; sthānaṃ piśācaiḥ parigṛhyate vāpūjyatām yāty acireṇa loke/vipraiḥ kṛtam yacchubhaṃ kule syāt prapūjyatāṃ yāti ciram ca kālam // [267.35].

6 The hypothesis that public image worship was not originally a Brahmin concern is not entirely new. Goudriaan (1965: 198-199) in his study of a Vaikhānasa ritual text has suggested that image worship made its way into the ‘literary Brahmanical tradition’ only gradually. In his interpretation it makes a brief appearance at the stage of the gṛhyasūtras and only slowly replaces the earlier Śrauta ritual. Goudriaan also seems to consider image worship something distinctively different from the Brahmin ritual world, when he speaks of it as the ‘indigenous image traditions’. See also Njammasch 2001: xx.

7 I have written on the topic in several papers. See Granoff 1998 and Granoff 2003. See also Davis 2001: 107-133. I would suggest that if we look carefully we will see that anti-image rhetoric was far more widespread than our accounts of Indian religions might have led us to suspect. I do consider here one text, the Viṣṇudharmottarapurāṇa, with its more general objections to the very notion of image worship, since some of its arguments are reflected in the ritual texts and are relevant for the development of the hypothesis that the ritual texts indicate a gradual encroachment of Brahmin priestly interest in image worship.

8 The evidence for medieval Buddhism in India is less abundant, although some early stories do clearly indicate a discomfort with image worship and Tantric Buddhist texts often contain statements about the uselessness of image worship and the importance of inner practices of meditation and visualization. One might also view the elaborate discussion on the nature of the Buddha body as at least in part motivated by a desire to come to terms with image worship. For an early story see the account of Mara appearing to Upagupta in the guise of the Buddha in the Aśokāvadāna (Strong 1989: 193). See also Shinohara 2004, which discusses some early Chinese Buddhist efforts to make a place for image worship in monastic circles.

9 Much of this material has been well studied. See Appadurai 1983: 43-56; StietenCron 1971: 126-138; Reiniche 1989: 170-173; Fuller 1984. These are cited in Colas 1996: 133. See also Granoff 1998.

10 Śivapurāṣa, koṭirudrasaṃhitā, chapter 13.

11 Seethe Cintrā praśasti of Saraṅgadeva, Epigraphia Indica, 1: 271 ff., verse 48. Bühler, who edited and translated the inscription, does not seem to have realized that baṭuka here is a technical term, referring to an attendant who serves the temple and as the verse tells us, lives off the offerings, the naivedyānna.

12 For a parallel story in the Vaikhānasa tradition see Colas 1996: 140.

13 I have also discussed this in Granoff 1998.

14 Niveśya nagaraṃ somaḥ śrīsomeśvarasamnidhau / svakarmaṇāṃ pracārārtham athābhyarthayata dvijān // [68],

15 Uvāca Kauśikas teṣu gotrāṇaṃ prathamo dvijaḥ / sādhūpadiṣṭam asmākaṃ dvijarājena sarvathā // [84] sarvaw etat kariṣyāmaḥ kiṃtu kiṃcin niśāmaya/ niyogataḥ pūjayatām śivanirmālyasevinām // [85] pātityaṃ jāyate ‘smākaṃ śrutismṛtivigarhitam / śrutismṛtī hi rudrasya yasmād ājñādvayaṃ mahat // [86] kas tad ullaṃghayen mūḍhaḥ prāṇaiḥ kaṇṭhagatair api // [87] aṣṭamūrteḥ punar murtāv agnāu devamukhe makhāṇ kurvāṇāḥ śrutimārgeṇa prīṇayāmo ‘khilaṃ jagat // [88],

16 It is not only temple priests who were the object of distrust in the Purāṇic accounts. There are passages which can be interpreted as evidence of a more general distrust of Brahmin priests who congregated at sacred places. Thus we hear in the Agnipurāṇa, Gayāmāhātmya, chapter 114, that when the sacred site was first established Brahma made for the Brahmins a city that was very much like heaven; the villages had mountains of gold, while the rivers flowed with curds and honey; the lakes were made of ghee and curds and there were mountains of food (34). He gave them the wishing cow and wishing trees and homes of gold and silver, so that the Brahmins would never have to ask anything of those who were their inferiors (35). But the Brahmins were greedy and accepted gifts of money and other things at the sacrifice and so all those Brahmins at Gayā were cursed by Brahmā to be devoid of wisdom and always filled with desires. The rivers lost their milk and honey and the mountains became ordinary stone (36-37). When the Brahmins begged Brahma to give them some means to live upon, he agreed that they would live off the holy place. Those who came to Gayā and gave things to these Brahmins would ensure that their ancestors went from hell to heaven (38-41). While it is true that Gayā, which is associated with funerary offerings in the Purāṇic tradition, may indeed be a special case, I think that it is also possible to understand the passage as critical of the very notion of Brahmin participation at pilgrimage sites. Like Brahmin interest in image worship, Brahmin interest in pilgrimage sites may well have been a later and secondary development. For a discussion of the development of Brahmin priestly involvement in pilgrimage sites see Nandi 1979-80: 62-118. The clear split in roles between Brahmins, whose job it is to recite the Vedas, and temple priests, whose job it is to worship the gods, is also clear in many vernacular texts of a late medieval date. An example can be found in the 14th-century Assamese Rāmāyaṇa, Ayodhyākāṇda, verse 2005, describing Ayodhyā after Rāma has left the city. The temple priests no longer worship the gods, while the Brahmins no longer recite the Vedas. There then follows a long list of castes and their abandoned duties.

17 Epigraphia Indica, XXVI, pp. 212-217.

18 Āgamaprāmaṇya [75-76]: na ca tāvatā teṣāṃ takṣnāṃ caikajātitvasaṃśayaḥ, evam ihāpi bhagavacchāstrasiddhapāñcakālikānuṣṭhātṛṇām anyeṣām āyatanaśodhanādi kurvatāṃ prāsādapālakāvaranāmnāṃ antyānāṃ ceti, “But that does not mean that the priests who officiate at the Vedic sacrifice and the craftmen who make the implements for the sacrifice are of the same caste; similarly there is no question that the priests who worship with the rituals to be carried out five times a day in a manner ordained in the texts propounded by the Lord and the out-castes who perform such functions as cleaning the temple and are known by the undistinguished term ‘temple custodians’ are of the same caste.”

19 For some information on the text see Kane 1962: 868-872. Kane emphasizes Ballālasena’s dislike of Tantric ritual. There is an odd biography of Ballālasena by Ananda Bhaṭṭa in the 16 th century which tells of a fight that developed between the king’s Brahmin priest, the purohita, and the priest who takes care of a temple. The source of their conflict is simple: the purohita wants a share in the jewels that have been donated to the temple. The text uses the old term of opprobrium for the image priest, calling him a devala. The purohita ultimately triumphs and the king banishes the devala from his kingdom. One might assume that the purohita then took over the duties of the devala in the temple and with those duties, the temple wealth. This story suggests that there was hostility on the part of the king to the temple priests and a struggle between the temple priests and the Brahmins over control of the wealth of the temples.

20 On the Tantric dīkṣā see Gupta, Hoens & Goudriaan 1979: 71-89.

21 For a discussion of the basic rituals of what I am calling the Hinduism of the Āgamas and Tantras see Gupta, Hoens & Goudriaan 1979. I use the terms Āgama and Tantra together to describe the religious practices of a group of diverse texts that go under these titles, which are often used interchangeably. See Dvivedī 1984.

22 See Bakker et al., eds., 1998 in the Prolegomena to his edition of the Skandapurāṇa. This text is different from and considerably earlier than the Skandapurāṇa that has been published many times before. Similarly the Devīpurāṇa discussed below has nothing to do with the printed text of that name.

23 See for example Yāmuna’s Agamaprāmāṇya.

24 Jalāśayānāṃ dānāni tathā ca suraveśmanām//noktāni dānāni samyag uktāni pratiṣṭhāsāgare yataḥ [verse 55, page 6], “I have not said anything here about the donation of reservoirs or temples since I have discussed all of that in the Pratiṣṭhāsāgara.”

25 The complex factors that motivated Brahmin priestly interest in image cults require a separate study. There are hints in the texts like the Pāñcarātra Ahirbudhnyasaṃhitā that the elevation of the status of image worship and its priests may well have had something to do with the pursuit of royal patronage and a growing use of rituals in the service of protecting the realm. Thus the text tells us that one should not employ its rituals for trivial purposes; they should be directed to protecting the world, the kingdom or the king (20.49ff.). On the date of this text see Smith 1978: 8. Although some scholars have argued for an early date, the text is not cited until the 13th-14th century. One might also consider the comment of an objector in the Āgamaprāmāṇya (p. 8) that Sātvatas are of low caste and that they have been assigned to worship in temples by royal order. The statement is a quote; its source is unknown to me.

26 Viṣṇor bhāgavatān magāṃśca savituḥ śambhoḥ sabhasmadvijān /mātrṇām api maṇ-dalakramavido viprān vidur brāhmaṇāḥ / śakyān sarvahitasya śāntamanaso nagnān jinānām vidur / ye yaṃ devaṃ upāśritāḥ svavidhinā tais tasya kāryā kriyā // [LX. 19, vol. 2, p. 571].

27 On the multiplicity of ritual specialists see also Colas 1996: 352ff.; Tarabout 1997: 253-284.

28 Cited by Grünendahl, ed., 1989, 3: 51.

29 Das Gupta (1967, vol. II: 547) suggests that image worship belonged to the non-Vedic Bhāgavatas, but he identifies them with the Pāñcarātrins. As Gérard Colas reminds me, the exact referent of the term Bhāgavata is extremely problematic. I take it here to mean image-worshipping Vaiśṇava priests, but who exactly they were and with what texts we may identify them, remain important and troubling questions. Nonetheless, there is some evidence that they were not identical to the Pāñcarātra initially. See also Colas (1990: 24-32), who cites the Yajñādhikāra as describing the Bhāgavatas as having no vedamāryādā. He cites other texts in which the Bhāgavatas are distinguished from the Pāñcarātra. The issue is clearly complicated in these medieval texts.

30 For a detailed discussion of possible dates and the complicated history of the text see Grünendahl, ed., 1989, III: 34-63.

31 Viṣṇudharmāh, 3: 51.

32 Brahma tasyāṃ samāropya manasā tanmavo bhavet/tām arcayet tām praṇamet tāṃ smaret tāṃ vicintayet. I understand the term samāropya here to mean something like, ‘imagine’ or ‘mentally construe’ in the same sense as Śamkara’s viṣṇubuddhyā cited above. On the technical sense of the term see Colas 1996: 280ff. and Colas 1986: 79.

33 It is clear that the paradigm of image worshipping Bhāgavatas on the one hand and Pāñcarātrins needs to be further refined as more texts are critically examined.

34 While I am aware that the term ‘ritual inconsistency’ raises the spectre of asking of the texts that they adhere to a standard, a ‘consistency’, that exists in the mind of the modern reader, I would argue in support of the term that the texts themselves seem to be aware that there is an incompatibility or at least that there is something amiss between certain rituals.

35 I have discussed this material elsewhere in greater detail (Granoff 2003). The date of the Viṣṇudharmottara like the date of the Purāṇas as a whole is uncertain; it is later than the Jayākhyasaṃhitā, which is not cited until the 10th century. See Kreisel 1987.

36 On this question see Malamoud 1990: 183-191.

37 Āvāhanaṃ svaṭrptyarthaṃ yathāsthitam / tathāpyarcā mahābhāga svatuṣtyarthaṃ prayojanam // [15] nityatṛptasya kiṃ tasya kāryaṃ bhavati cārcayā / kevalaṃ sa bhaktatuṣtyarthaṃ sa bhaktebhyah pratīcchati // [19].

38 Rūpagandharasair hīnaḥ śabdasparśavivarjitaḥ / puruṣas tu tvayā proktas tasya rūpam idam katham // [1] Mārkaṇḍeya uvāca: prakṛtir vikṛtis tu tasya rūpeṇa paramātmanaḥ // alakṣyam tasya tad rūpaṃ prakṛtis tu prakīrtitā // [2] sākārā vikṛtir jñeyā tasya sarvaṃ jagat smṛtam / pūjādhyānādikaṃ kartuṃ sākārasyaiva śakyate // [3] svatas tu devasyākāraḥ pūjanīyo yathāvidhi / avyaktā hi gatir duḥkham dehabhṛdbhir avāpyate // [4] ato bhagavatānena svecchayā tat pradarśitam /prādurbhāveṣv athākāraṃ tam diśanti divaukasaḥ // [5].

39 Nāmopāssva brahmeti brahmabuddhyā. yathā pratimā viṣṇubuddhyopāste tadvat.

40 See for example Brunner 1986: 89-132; 1993: 1-61; 1990: 4-24.

41 For the Śaiva evidence see Brunner 1990 and the texts she cites. As will be clear below, my own interpretation of the significance of the incompatibility between the āvāhana and other rituals differs somewhat from Brunner’s.

42 On the date of this text as between 600-850 A. D., see Rastelli 1999: 25-27. It is first cited in the 10 th century. Scholars agree that the Viṣṇudharmottapurāṇa is later. See Kreisel 1987: 108.

43 Ārādhito ‘si bhagavan sādhakānāṃ hitāya ca // tvayā ‘py anugrahārthaṃ ca vastavyam iha sarvadā [20.331],

44 For a parallel from a Śaiva text see the Ajitāgama 18.264 (p. 167). For my references to the Śaiva texts I have relied on the articles by Brunner cited here. The dating of all of these texts is extremely problematic. Sanderson 1992: 291 (cited in Goodall 1998: 404) has argued for a late date for the extant Ajitāgama.

45 See the Cintra praśasti of Saraṅgadeva cited above, Epigraphia Indica, I: 271 ff. “sākṣād bhavanti vibudhāḥ kila tadvidhānāṃ hastāmbujena dṛṣado ‘pi kṛtapratiṣṭhāḥ” [21], “Even stones become gods when consecrated by the lotus hands of such people.”

46 See for example Schopen 1990: 181-219; Granoff 1993: 66-94. The question must arise whether the earliest rituals included the plea for permanent presence; my guess at this stage is that they may not have done so; see for example the Vaikhānāsagṛhyasūtra cited above.

47 Rastelli 1999: 284 fn. 1265. Brunner (1993: 28) suggests that in fact the earlier Śaiva Āgamas dealt with private worship and that those dealing with public temple worship are later.

48 G. Colas raised the interesting point here that even today a temple may quickly be considered divine. Purāṇic stories about images support this; the origin stories virtually all give a divine origin for the main cult image. One interpretation, then, of the Viśvāmitra passage is that the category of man-made images is not mentioned in the exclusion since it really did not exist. I doubt this interpretation, however. The texts devote considerable energy to consecration rituals, another category of rituals not required of divine images; I find it difficult to believe that the rituals were never meant to be conducted, which would have been the case if all images were regarded as of divine origin! I would also regard the origin stories as later than the images and told only of those images or temples that had achieved a certain status; in other words, the consecration and worship had already been carried out by the time the stories began to circulate. Colas (personal communication) also suggested to me that in the Vaikhānasa texts the division is between movable and immovable rather than permanent/temporary ritual objects. I would like to see the Vaikhānasa distinction as a later reworking of an earlier division.

49 Colas 1996: 298. I hope in the future to make an in-depth comparative study of these rituals across a variety of ritual schools.

50 Brunner (1993: 12) in a discussion of the relationship between ritual and doctrine in the Śaiva Āgamas has gathered evidence from Śaiva texts that shows a clear awareness that the āvāhana is not compatible with the doctrine of an all-pervasive deity who should not require being summoned into a particular limited space, such as an image. Brunner notes that the texts are explicit in their attempts to harmonize doctrine and ritual. One solution proposed was to posit that the god first is given a mantra-body that limits him, just as our bodies act as limits on our souls. After the god gets this mantra-body he can be summoned to a particular place. Other texts focus on the fact that the ritual of summoning the god every day contradicts the underlying assumption of the pratiṣṭhā or consecration ceremony at the installation of an image, in which the god is summoned into an image once and for all. These texts interpret the āvāhana not as a physical summoning of the already present god, but as a means for the worshipper to get the god’s attention (sammukhīkaraṇa). See also her references there to the various Agama texts.
A similar interpretation of the ritual as sammukhīkaraṇa is also found in the later Pāñcarātra texts like the Viśvāmitrasaṃhitā cited above, which enjoins the daily ritual, including the āvāhana, for all images. It is clear that both Vaiṣṇava and Śaiva ritual texts are aware of the incompatibility between ritual and theology.

51 Brunner (1993: 116) in her discussion of the Śaiva texts concludes that the rituals of āvāhana could not have been devised by someone who had “in mind the greatness, the allpowerfulness, the uniqueness of Śiva.’’ Instead she offers that they are more appropriate to someone she calls the aspirant, the sādhaka, who worshipped a host of minor gods for the acquisition of powers and who most likely conducted his rituals with an impermanent image. See also Brunner 1994: 456 in which she makes a distinction between ‘yoga-permeated’ ritual and a popular tradition. She traces the yoga ritual to the private cult and the sādhaka. At a later time, she argues, the yoga-oriented cult of the sādhaka reached a wider audience and eventually became part of temple worship. In Brunner 1990 she suggests that the sādhakas, with their yogic practices were eventually structured into a maṭha. These maṭhas had contact with ‘village cults’ of various deities and eventually began to extend their authority over village worship. She raises the intriguing possibility that this was the beginning of ‘proper temples’. She leaves open the question of the introduction of Vedic rites into the temple practices. I came to this article after I had written this paper, but it is obvious that I have come to similar conclusions on one fundamental point: that the worship in the Āgamas stems from different sources. My reading of the Pāñcarātra texts suggests to me, however, a somewhat different interpretation. As I outline in what follows, instead of the opposition between sādhaka and yogin or sādhaka and popular village practice, I see an opposition between two different groups of priests or religious specialists, the orthodox Brahmin/Vedicized priests on the one hand and the non-Vedic image worshipping temple priests on the other. To the orthodox priests belong both the (now anachronistic) ritual paradigm of invocation and the developed theology of the all-pervasive formless god. To the image worshipping priests belongs the belief in the permanent presence of the deity as the image. Their rituals probably centered around sweeping and cleaning the temple and making offerings to the deity, the rituals of texts like the Viṣṇudharmāḥ. In this understanding, orthodox incursion into the temple cults begins only slowly and is accomplished only by leaving gaps and inconsistencies such as I am attempting to analyze here. Orthodox incursion in this model is also into a well-developed temple cult in its own right, and not a primitive village cult.

52 In her discussion of the treatment of śakti in the rituals, Brunner (1993: 22) notes further inconsistencies. She concludes of the arrangement of three śaktis in the temple, “In spite of the inevitable attempts to parallel this triad with icchā, jñāna, and kriyā, it seems here that we are faced with quite another vision, which probably is more or less inspired by the village cults and agrees with the temple ritual much better than the theoretical classifications we have considered up to now.” I am arguing in this paper for a position similar to this but more thoroughgoing, namely that image worship as a whole makes more sense outside the world of these fascinating but often self-contradictory ritual texts.

53 That this is only one of many competing discourses on image worship is clear from Tarabout 1997: 253-284. My argument in this paper is that image worship and reactions to it had a long and complicated history; indeed, the very variety of ways in which it is rationalized and explained suggests that it did not always meet with a favorable reception.

54 Yanmayas sādhako vipra dehasthas sāmprato bhavet / tanmayo dehapātāt syād ity etat kathitam mayā // [13.9].

55 On this question see Malamoud 1988: 15-32.

56 Desai 1996 and the articles referred to therein. pal (1975: 66) had argued that the Vaikuṇṭha images were originally three-headed with the fourth head added later in the light of Pāñcarātra theology. I thank Robert Brown, University of California at Los Angeles, for this reference.

57 On the introduction of Vedic rites into the Vaikhānasa ritual, see Colas 1999: 41-50.

58 Rāghava Bhaṭṭa in his commentary to the Śāradātilakatantra, p. 1.

59 Śastrābhilāṣāt sutarām mahātmā saṃtyajya deveśvaradāsyakṛtyam / kāmcīpure lakṣmaṇadeśikendras tasyāntike pūrvavad āsthito ‘bhūt [6.28],

60 On the late history of the priests of the temple at Śriranagam see Jagannāthan 1994. It is worth noting that among the duties of the Brahmin priests at Śrīraṅgam, where Rāmānuja institutes his reforms, are those that the śūdra priest had performed at Hastiśaila. This raises the basic question as to whether we can interpret Rāmānuja’s activities at Śrīrahgam at least in part as an effort to solidify Brahmin priestly activities in the temple setting.

61 The Jain texts are reproduced in the edition of the Gaṇakārikā.


Professor of Religious Studies, Yale University, Department of Religious Studies. Her field is about early and medieval Hinduism and Jainism.
– 2000 « Portraits, likenesses and looking glasses : Some literary and philosophical reflections on representation and art in medieval India » in J. Assmann & A. Baumgarten, eds., Representation in Religion : Studies in Honor of Moshe Barasch, Leiden, E.J. Brill, pp. 63-107.
– 2003 P.G. & K. Shinohara, eds., Pilgrims, Patrons and Place : Localizing Sanctity in Asian Religions, Vancouver, University of British Columbia Press.
– 2004 P.G. & K. Shinohara, eds., Images in Asian Religions Texts and Contexts, Vancouver, UBC Press (« Asian Religions and Society Series »),

© Éditions de l’École des hautes études en sciences sociales, 2006

Conditions d’utilisation :

Cette publication numérique est issue d’un traitement automatique par reconnaissance optique de caractères.
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search