On Some Tantrik texts studied in Ancient Kambuja

·期刊原文
On Some Tantrik texts studied in Ancient Kambuja

By P.C. BAGCHI

The Indian Historical Quarterly

Vol.VI, No.1, 1930.03 pp. 97-107

--------------------------------------------------------------------------------

p. 97

II

The Agamas and their influence--The inscriptions
of Kambuja abound in references to the Saivite canon.
Various inscriptions refer to Sivasastra, (1)
Saivagama,(2) Sarvagama,(3) and Saiva-vyakarana.(4)
Agama means the oldest Saivite canon which conformed
to the Vedas and had not entirely separated from the
Vedic religion like the later Saiva sects. Sastra was
a term synonymous with Agama. Agamas are generally
believed to be 28 in number but we have already
discussed the text of the Nisvasatattva Samhita,
itself an Agama, which mentions only 18. We have also
pointed out that these 18 Agamic texts must have
existed long before the 8th century. The references
to Agamas in the inscriptions of Kambuja, the oldest
of which go back to the beginning of the 9th century,
confirm the same view. One of the four texts
mentioned in the inscription of Kambuja, viz. the
Nayottara at least belong to the Agama proper while
the three other belong to the canon which grew later
on under its inspiration. In the inscription of
Angkor vat(5) we find another reference to an Agamic
text: it is the Paramesvara (tasmin kuru mahadyagam
yathokatam paramesvare). It is the Paramesvaratantra
also called Paramesvaramatatanntra which is one of
the 18 Agamas mentioned in the Nisvasasamhita list.
It is the 25th of the 28 Agamas mentioned in later
literature.(6) We have already seen that there is a
ms. of the Paramesvaratantra copied in 859 A.D. The
work was certainly much older, as it is mentioned in
the Nisvasasamhita list of which we have a manuscript
of about the middle of the 8th century.

In my last article I have tried to show that the
original Saivite

----------------------
1. Inscription of Phnom Sandak of about the end of
the 9th century A.D. Bergaigne -- Inscription de
Campa et du Cambodge, II, P.157.
2. Inscription of Angkor vat, Ibid. p.392.
3. Ibid. p.389.
4. Ibid. p.392.
5. Bergaigne, Inscription etc. p.390; also p.384 with
the note of Barth.
6. Gopinath Rao--Hindu Iconography, II, part I, pp.
367-368.

p. 98

canon which contained the 18 Agamas was of
North--Indian origin as according to them, the best
Sivacaryas were the Brahmins of Aryavarta. But the
people of the surrounding countries, Kamarupa,
Kasmira, Kalinga, Konkana, Kanci, Kosala, Kaveri and
Rastra were not eligible to that position for their
physical deformities. By physical deformities we have
to understand that their statures did not follow the
prescribed standard and were either too tall or too
short (atidirgha atihrasvaka). Such a conclusion is
also substantiated by other evidences. The Tantrasara
which is a famous compendium of Bengal Tantrism says
an the authority of Kriyasarasamuccaya, Yamala and
Vaisampayana-samhita that persons with physical
deformities of various description, and persons who
are diseased, immoral etc. cannot be gurus (Ibid.
p.3)--atha nindyagurumaha--

Kriyasara- samuccaye-- Svitri caiva galatkusthi
netrarogi ca vamanah/

kunakhi syavadantas ca strijitas cadhikangakah//
hinangah kapati rogi bahvasi bahujalpakah/
etair dosair vihino yah sa guruh sisyasammatah//
Yamale--abhisaptam aputran ca kadaryam kitavam tatha/
kriyahinam sathan capi vamanam gurunindakam//
jalaraktavikaran ca varjayen matiman sada/
sada matsara-samyuktam gurum tantrena varjayet//.
Vaisampayana-samhitayam--
aputro mrtaputras ca kusthi ca vamanas tatha.....//

The same compendium again says on the authority
of Jabila (quoted by Vidyadharacarya) that the
quality of the gurus differ according to the
countries in which they are born. According to it the
best gurus are found in the countries of Madhyadesa,
Kuruksetra, Nata and Konkana (or Nata-Konkana?),
Antarvedi, Pratisthana, and Avanti. The Madhyadesa is
Aryavarta. The gurus of the second quality are found
in Gauda, Salva, Sura(?), Magadha, Kerala, Kosala and
Dasarna. The worst gurus are those who belong to the
countries of Karnata, Narmada, Rastra,(1) Kaccha,
Kalinda Kalamba and Kamboja(2) (Ibid. p.10-11); tatha
Vidyadharacaryadhrtam Jabalavacanam--

----------------------
1. It's evidently the same name as quoted in the list
of the Pingalamata. Through mistake I connected it
with Kaveri and took it to mean Kaveri-rastra. It
seems to be a different country and probably is
meant for Surastra.
2. Konkana which is amongst the forbidden countries
in the Agama list here is placed in the first rank.
Nata Konkana may however

p. 99

Madhyadesa-Kuruksetra-Natakonkanasamb
Antarvedi-Pratisthana Avantyas ca guruittamah//
Madhyadesa Aryavartah/
Gaudah Salvah Suras caiva Magadhah Keralas tatha/
Kosalas ca Dasarnas ca guravah sapta madhyamah//
Karnata-Naramda-Rastra- Kacchatirodbhavas tatha/
Kalindas ca Kalambas ca Kambojas cadhama matah//

This list was certainly drawn up at a time when
the authority of the orthodox Agamas was a little
undermined by the rise of the heterodox schools. But
it still shows the old tendency according to which
the acaryas of North Indian origin were given the
first place.

This throws some unexpected light on the
recruitment of Sivacaryas in different countries
including ancient Kambuja. We have seen that
Hiranyadama came with the new Sastras from a
janapada, which was most probably a janapada in
India. The family of Sivakaivalya, who was initiated
to these Sastras, was long established in Kambuja.
The history of this family, recorded in the
inscription of Sdok kak Thom is of great interest The
members of this family enjoyed the priesthood of the
king through succession since the time of Bhavavarman
(middle of the 6th century A.D.). They were
Sivacaryas and were guardians of linga established in
different places, The succession of the priests was
determined according to the matrvamsa "i.e. maternal
lineage" (tanmatrvamse yatayas striyo va jata
vidya-vikrama-yukta- bhavah/ tad-yajakas syuh........
BEFEO, 1915, p.62) which implied that the succession
was to go to the children of the sisters (bhagineya)
or to those of the daughter of the sisters, or the
elder brother. There are several cases of such
succession recorded in the inscriptions (Ibid, ,
p.54). It is difficult to explain the necessity of
such an arrangement. Barth in 1901 thought that such
an arrangement was necessary because the royal
priests used to take the vow of celebacy and
therefore they had to choose their successor from the
line of their sisters. But M. Finot (Ibid., p.56)
says that it is difficult to admit this explanation
as

----------------------
be a mistake for some other country. The countries of
Kalinda and Kalamba are not known. Kalinda (certainly
not Kalindi) seems to be a mistake for Kulinda.
Kamboja does not seem to be the ancient country of
the Kamboja-Gandhara group. It may be the country of
the people called Kam-po-tsa in the Tibetan sources
and located in Assam. These people seem to have been
the predecessors of the modern Koch.

p. 100

we hear of priests (though of very late times--IIth
century A.D.) who were married. It is however clear
that the intention was to avoid difficulty in finding
a successor because when the branch lines are counted
the family has an unlimited scope. But what was the
necessity of sticking to a particular family for the
selection of priests? The only explanation that
occurs to my mind is that according to the Agamas the
Sivacaryas had to be chosen preferably from the
Brahmanical families of North Indian origin. Such
families were not numerous in Kambuja. The family of
Sivakaivalya was probably a rare one and priests had
to be chosen from that family and its branch lines,
as the members of them alone were fit to be
Sivacaryas. In the inscriptions of Kambuja we have
several other references to the families of North
Indian origin, of which the members attained the
position of royal chaplain. Thus we hear of the royal
chaplain Bhatta Divakara who came from the banks of
the Kalindi (Yamuna) and was thus an expert in the
Vedic sacrifices (Bergaigne --Inscription I, p.81ff.)
In an inscription of Angkor vat we are told that the
royal priest Sarvajnamuni who was a special adept in
the Saivite rites came from the Aryadesa.
(Bergaigne-- Inscriptions etc. 1xv. 9. p.388.
Aryyadese samutpannas Sivaradhanatatparah/ yo
yogenagatah Kamvudese...). In the same inscription we
hear that a descendant of Sarvajnamuni filled the
country called Madhyadesa (here a part of the ancient
Kambuja) with Brahmins versed in the Veda and Vedanga
(Ixv. 22. cakara desam namnemam, madhyadesam
janakulam/ vedavedangavidvipram....). There seems to
be a reference here to the immigration of Brahmins
from India. In the inscription of Prah vat we find
mention of a Brahmin, named Agastya related to the
royal family, who originally came from the Aryadesa.
(Bergaigne--Inscriptions etc. xliv.5; p.184--atha
dvijo' gastya iti pratito, yo vedavedangavid
aryyadese...). Such practices were known in India
too. The great Cola king Rajendra Cola who built the
Rajarajesvara temple at Tanjore is stated to have
''appointed Sarvasiva Pandita-Sivacarya as the priest
of that temple and have ordered that thenceforth the
Sisyas and their Sisyas alone, belonging to the
Aryadesa, the Madhyadesa and the Gaudadesa shall be
eligible for the office of chief priest,'' (South
Indian Inscription II, I. p.105, wrongly referred to
as II, 2. p.153 in Hindu Iconography II, I. pp.5-6).
We also know that the Malla kings of Bhatgaon (Nepal)
had Brahmins from Bengal as their priests. These
Brahmin families used to come to Bengal from time to
time to contract their marriages in order to

p. 101

maintain the purity of their family tradition. This
was however the custom most probably in the pure
Sivasadhana i.e. Agamanta Saivism. For the heterodox
Saiva sects like the Pasupatas and others the
practice was different. Thus in Nepal the priests of
Pasupatinatha were recruited only from amongst the
South Indian Brahmins (S.Levi, Le Nepal I,
p.364-365).

The influence of the Agamas can also be traced in
the Saivite cult practised in Kambuja and Campa.
There are ample evidences in the ancient inscriptions
to prove that the constructions of the Sivalingas
were made according to the prescription of the canon.
According to the Agamas the Iingas can be of two
kinds, the cala i.e. moveable and the acala, i.e,
immoveable. The cala lingas are again of different
types: mrnmaya, earthen; lohaja, metallic; ratnaja,
of precious stones; daruja, wooden; sailaja, of
stone; and Ksanika, those made for temporary worship.
The lohaja i.e. metallic lingas are made of 8 metals:
gold, silver, copper, bell-metal, iron, lead, brass
and tin and the ratnaja ones are made of pearls,
coral, vaidurya, topaz, emerald and bluestone.(1)

The acala or sthavara lingas are of 10 kinds,
Svayambhuva, Purva, Daivata, Ganapatya, Asura, Sura,
Arsa Raksasa, Manusa and Bana. The Makutagama calls
them Sthira lingas and divide them into four classes:
Daivika, Arsaka, Ganapa and Manusa.

In ancient Campa Saivism was the predominant
religion and Siva was worshipped mostly in the form
of a linga. A linga established by king Bhadravarman
towards the close of the 4th or the beginning of the
5th century A.D. became a sort of national deity for
the people of Catnpa. This linga is differently
called in the inscriptions--Bhadresvara,
Sambhu-Bhadresvara, and most probably also as
Srisana-Bhadresvara (see R.C. Majumdar--Campa,
pp.177ff.). The inscriptions do not generally speak
of the materials used for the construction of the
lingas. Many of them, specially the mukhalingas were
certainly curved from stone. But we have some
references to other types of lingas too. An
inscription of Po-nagar, dated 965 A.D. (Majumder II,
n" 47), speaks of the gold and stone images of the
goddess i.e. Bhagavati (haimi and sailamayi pratima;
inscr.

----------------------
1. See Gopinath Rao, Hindu Iconography II, (I)
pp.75ff: The Agamas from which he derives the
information are: Suprabhedagama Karanagama,
Kamikagama, Makutagama and the Kirannagama. See
Ibird. II(2) App. B. p.3 ff.

p. 102

n" 45 kaladhautadeha), erected by king Indravarman.
This shows that both gold and stone was used in the
construction of the images of deities in Campa.
Another inscription is more explicit on the point.
The Yang Tikuh Inscription of Indravarman I (dated
721 saka =799 A.D.) contains two stanzas which have
not been correctly iuterpreted till now. The stanzas
in question are (See Majumdar, II, n" 23, viii and
ix):-

[VIII]

[ IX ]

Dr. Majumdar translates the stanzas thus:
"Indravarman also installed an earthen linga of the
God, which therefore came to be known as
Indrabhadresvara. He also established in the year of
the Sakas Sasi yam adri(721), two treasures for the
god, the one composed of moveable and immoveable
property, and the other moveable and with a mouth
(priests?)."(1) The last part of the translation is
evidently unintelligible. There is no question of
"property" in the text and "a moveable treasure with
a mouth(?)" does not convey any meaning. Kosa here,
as in many other cases in these inscriptions, should
be taken in the sense of linga-kosa. Kosa was
apparently an outer covering of the linga, and was
used probably for decorative purposes. The
inscriptions of Campa very often record the gifts of
kosa made by the kings to the lingas. These Kosas
were often golden and decorated with costly gems. The
kosas had sometimes faces and Kasa with six faces are
twice spoken of. We find mention of Urddhvakosa which
was most probably a detachable one (See Majumdar,
Campa I, p.182). If in the present case we take kosa
in the sense of linga-kosa, the text becomes clear.
It should then be translated: "Indravarman also
installed an earthen-parthiva-linga of him (the god
which therefore came to be known as Indrabhadresvara.
He

----------------------
1. Bergaigne --Inscriptions etc. II, p. 33 et 37--
VIII-IX "Sri-Indra varman a erige aussi un linga
terrestre de ce dieu, qui a ete appele desormais
d' un antre nom Indrabhadresvara. II a aussi
constitue pour lui deux tresors: l'un compose de
biens meubles et doue d'elo-quence" (les pretres
du temple--Barth).

2. Cf. Ibid. I, Inscr. Il. 10, XV. B- 26, XVII. B.26,
XVIII.D. 27, B. 24 and II, Inscr. LXI. C. II.

p. 103

also established, in the saka year sasi-yama-adri
(721), two kosas, one cara i.e. moveable and the
other sthira i.e. immoveable. The moveable (cara)
Kosa had a face (or faces)." The linga was an earthen
one (parthiva) which corresponds to the minmaya-linga
mentioned by the Agamas and it had two kosas, of
which one was moveable and the other, probably a
simple cylindrical one was a fixed one. The cara
kosa, had a face (or faces) and thus when fixed to
the lingas used to convert them into mukhalingas. The
two words cara and sthira naturally remind us of the
two types of linga, cala, moveable acala, immoveable,
also called sthira or sthavara in the Agamas. In
ancient Kambuja the lingas used to be made of metal
as well as precious stones. We have references to
lingam haimasobham, suvarna mayalinga, svarnalinga,
kaladhauta-linga, sphatikalinga and manilinga. The
materials used for the construction of these lingas
therefore were chosen in accordance with the
prescription of the Agamas. They all were of the type
known as calalinga and fell under its subdivisions:
sailaja, lohaja and ratnaja.

The four faces of Tumburu--I have already tried
to establish that the four Tantrik texts sirascheda,
vinasikha, sammohana and nayottara mentioned in the
inscription of Sdok Kak Thom were authentic
Saivasastras being studied in India in the 7th and
8th centuries A.D. if not earlier. These texts
constituted the "vaktracatuskam"(1) of the god

----------------------
1. We have now a definite text before us which
supports the. identification of Tumburu with Siva.
In the Yogavasisttha-Ramayana (Nirvana-prakarana,
I. xviii 23-26), compiled before the 9th century
A.D., we find the following verses:

ity astaisvarya-yuktas ta mataro raudra-cesthitah/
kadacinmilita vyomni sarvah kenapi hetuna//
utsavam paramam cakruh paramarthaprakasakam/
vamasrotogata etas Tumburum Rudram asritah//
pujayitva jagatpujyau devau Tumburu-Bhairavau/
vicitrarthah kathas cakrur-madira-madatositah//

These things are spoken of the eight matrkas who
were one day out for amusement. They are here
characterised as the followers of the left current
(vamasrotogata) and related to Rudra who is Tumburu
i.e. the Tumburu aspect of Rudra (Tumburum Rudram).
The eight-matrkas are here made to worship the two
gods Tumburu and Bhairava. This passage clearly
mentions Tumburu as an aspect of Rudra. It should
also be noted that Siva is often referred to in the
inscrip-

p. 104

Tumburu and were introduced in Kambuja for
establishing the mystic rites known as devaraja
(siddhih........ devarajabhikhya). Tumburu evidently
had some sort of connection with the Devaraja cult.
Devaraja was a phallic representation (lingaraja) of
Siva--and we have already seen that Tumburu was an
emanation of Siva himself. The inscription of Sdok
Kak Them tells us that the first temple of Devaraja
was built by Jayavarman II (802 A.D.) in his new
capital Mahendraparvata (Phnom Kulen), and the royal
chaplain Sivakaivalya was appointed priest. The deity
was subsequently taken to Hariharalaya where the
capital was shifted. Afterwards when the king
Paramasivaloka (i.e., Yacovarman 889-910 A.D.) built
his capital at Yosodharapur (Angkor Thom) he brought
the deity to the new capital and placed him in the
temple of Vnam Kantal (lit. the central mount) which
was built in the centre of the city for receiving the
deity. This central edifice erected by Yasovarman was
for a long time believed to be the Bayon which is
situated just in the centre of Angkor. But M.Finot in
his recent studies (Etudes Asiatiques, vol.I
p.245ff.) has tried to show that the inscription of
Sdok Kak Thom has told a lie. A detailed examination
of the sculpture of Bayon has led M.Finot to believe
that Bayon could not have been originally a Saiva
temple. He thinks that the newly built capital of
Yasovarman was not placed under the protection of the
linga Devaraja, the national deity of Kambuja, but
under that of the Bodhisattva Avalokitesvara.
Necessarily he was led to conclude that Angkor Thom
and Bayon were not built by king Yasovarman, as the
inscription would have us believe, because he was a
Saiva, but by his predecessor Jayavarman II who was a
Mahayanist. Yasovarman according to him played the
part of a vandal and changed Bayon into a sanctuary
of the linga. The principal reason for starting this
theory was that the sculpture of Bayon is almost
entirely Buddhist. But it might be argued that the
temple was begun as a Buddhist one and finished as a
Saiva one. But to this objection M. Finot answers
that even in several niches of the towers the central
figures were originally those of Buddha. They were
later on deliberately destroyed and replaced by
linga. Another serious difficulty remained to be
explained away. Each tower of Bayon is decorated with
four colossal faces turning towards the four cardinal
points. In 1911 M.Finot interpreted them

----------------------
tions of Kambuja as Caturanana, Caturmukha etc. Cf.
Bergaigne Inscriptions etc. II, n" LXIV (P.377);
n XLIV (P.183); n LV (P.213).

p. 105

as the architectural translation of a
caturmukhalinga, He, however, gives up that
explanation in the light of later researches and now
thinks that they represent the faces of the
Avalokitesvara. He is aware of the fact that no such
architectural representation of Avalokitesvara is at
present available but he still supposes that the
architect wanted to represent Avalokitesvara as
looking in the four directions and thus protecting
the city on all sides. M.Finot would therefore
conclude that the inscription of Sdok Kak Thom has
distorted the facts. Bayon was not originally a Saiva
temple and Yasovarman, who was a staunch Saiva, could
not be its founder. It was founded in the time of
Jayavarman II (802-869 A.D.) who was a Buddhist king.

But M. Phillip Stern in his study on the
evolution of the Khmer Art (Le Bayon d'Angkor et
l'Evolution de l'Art Khmer, 1927) has questioned the
hitherto admitted chronology of the monuments of
Angkor on grounds of style. According to him, Bayon
did not exist in the time of Yasovarman. Therefore,
the central mount (Vnam kankal) of Yasovarman has to
be searched for elsewhere. He thinks that it should
be identified with the Phimanakas, which in all
appear ance, occupied the central position in the old
city. The city developed in course of subsequent
centuries and its centre was naturally removed.
According to the chronology proposed by him, Bayon
could not have been built before the time of
Udayadityavarman II (1049-1052 A.D.) or that of his
predecessor Suryavarman I (1002- 1049 A.D.). The
outer walls of the city would belong to this period.

But M. Coedes in a recent study(BEFEO, XXVIII,
pp.81 ff.) has tried to prove that Bayon was built
still later during the reign of Jayavarman VII
(1182-1201 A.D.). According to him the outer walls of
Angkor Thom and some other buildings, which is of the
style of Bayon, were constructed in the same period.
Thus both M. Stern and Coedes agree in placing Bayon
and the outer walls of the city in the same epoch
though they do not assign the same date to their
construction. Both of them disbelieve in the
testimony of the inscription of Sdok Kak Thom which
clearly attributes the foundation of Yasodharapura
(i.e. Angkor Thom) and Vnam Kantal, "the central
mount," to Yasovarman.

This inscription, we have already seen, was
composed in Saka 974(= 1052 A.D.) The date of its
composition therefore falls in the period to which M.
Stern would attribute the construction of Bayon. It
seems strange that a contemporary inscription would
mean by Vnam

p. 106

Kantal any other edifice except the Bayon. What is
possible is that the traditional history of religious
foundations, which it records, is confused. Its
attribution of the foundation of Bayon to Yasovarman
may therefore be easily questioned but Bayon was
certainly considered as a sufficiently old edifice in
the middle of the IIth century for affording scope
for confusion about its real founder to the author of
the inscription.

According to M. Coedes and M. Stern, the
construction of Bayon and the outer walls of the city
would fall in the same period. The towers of Bayon
and those of the five city-gates are all decorated
with four colossal faces. What do these four colossal
faces represent? Are they the representations of the
faces of Avalokitesvara, as M. Finot is thinks? Even
admitting that Jayavarman II, if not directly but
through his tradition, influenced the construction of
Bayon, it is difficult to believe with M. Finot that
he was a Buddhist king. M. Finot takes him to be a
Buddhist--firstly, because he came from Java or from
Srivijaya which was a great centre of Mahayana
Buddhism in this period and-secondly, because he
founded the city of Amarendrapura, formerly
identified with the ruins of Bantay Chmar which is
completely a Buddhist city to judge from the
sculptures. But the identification of Amarendrapura
with Bantay Chmar has been reasonably doubted by M.
Stern (loc. cit.). We should also bear in mind that
the posthumous name of Jayavarman II is Paramesvara
(the Supreme Lord=Siva). The cities which he
built--Mahendra- parvata, Hariharalaya and
Amarendrapura are all connected with the names of
Siva. The last name seems to be only a different form
of Devaraja. The priest whom he chose as his
chaplain, Sivakaivalya was a Saiva and came from a
Saiva family. It was again he who authorised
Hiranyadama to introduce the texts of Saivagama along
with the Saiva cult of Devaraja into Kambuja. He
really made it the religion of the state, erected its
temples and granted lands to the priestly family for
its maintenance. Besides it would be wrong to say
that the sculptures of Bayon have no trace of
Saivism. An important bas-relief of the first gallery
of Bayon (See Comaille, Guides aux Ruines d'Angkor
P.135, n.36) represents three temples in one row, of
which the towers bear tridents (trisula) and the
deity in the centre is a Sivalinga. In the face of
these facts it is difficult to admit that Jayavarman
II was a Buddhist king and that he introduced
Mahayana from Srivijaya into Kambuja. There is no
reason to suppose that the four Tantrik texts brought
by Hirayadama had

p. 107

anything to do with Mahayana. Jayavarman II was a
Saiva. If any of the edifices (for example Bantay
Chmar, Bayon, etc.) containing some Buddhist
sculptures can even be proved to have been
constructed in the time of Jayavarman II, the only
possible explanation is either that he was a tolerant
king and allowed Mahayana to flourish in the country,
or that he had employed artists who had come from the
neighbouring territory of Srivijaya and had
Mahayanist training. It will be wrong to suppose that
Mahayana Buddhism of the 8th-9th century A.D. was
very much antagonistic to Tantrik Saivism. Though the
sculpture of the temples partly seem to be
Mahayanist, the indwelling deity was no doubt Siva.

It seems difficult to admit that Bayon was not
originally a Saiva temple. The state religion of
Kambuja was always the cult of Devaraja. A temple
like Bayon, which is situated just in the centre of
the city, could not therefore have been meant for any
other deity except Devaraja. If in some of the niches
of the towers of Bayon the figures of Buddha have
been deliberately destroyed and substituted by linga
we must attribute that work of vandalism to a period
when the king was a very orthodox one and did not
even tolerate the sculptural representation of Buddha
in the temple of Devaraja as his predecessors used to
do. It is therefore necessary to go back to the older
theory of M. Finot that the four faces of the towers
of Bayon (as well as those of the towers of the city
gates) are the sculptural representation of the four
faces of Siva. Devaraja was in all probability a
mukhalinga and it was quite natural that the towers
of its temple and those of the city-gates constructed
in the same period would bear the mukhalinga symbol.
This explanation seems to have a strong support in
the inscription of Sdok Kak Thom which says that the
four sastras which prescribed the cult of Devaraja
constituted the four faces of the Tumburu. It may not
be therefore improbable that the four colossal faces
on the towers are architectural translation of the
four faces of Tumburu, Tumvuror vaktracatuskam,
mentioned in the inscription, because, it is through
those four faces that the god originally communicated
the four fundamental texts which prescribed the
religious rites of the king and his people. They are
the symbol of the different amnayas of the Saivite
Canon.