vastu - About Dravid Temples

7/30/2019 vastu - About Dravid Temples

1/22

41

Drvia Temples in the

Samargaastradhra

AdAmHArdy

The relationship to actual practice of the vastustras (orvstustras) and ilpastras, the canonical Indian textson architecture and sculpture, is a complex one. Scholarlyattitudes to these texts range between an uncriticalassumption that, traditionally, these texts set the rulesfor making buildings and sculptures, thereby holding the

key to understanding them, and complete denial of theirutility, on the basis that they were probably composed byBrahmans who were cut o from practical experience.The truth must lie somewhere in between. To establish theextent to which any part icular text may have been useful forcreating architecture, it must be shown whether it can beused for this purpose if not by actually building, at leastby drawing. This, surely, should be a prerequisite for anysensible discussion of the nature of these texts.

Surprisingly, the one sustained attempt to illustratea vastustra is that of Ram Raz, whose 1834 essay is therst work of modern scholarship on Hindu temples.1 On thebasis of a fragment of the south IndianMnasra,Ram Razwas assisted by a contemporary practitioner in interpretingits prescriptions through lucid drawings, done in a oridlatter-day Drvia style (Figure 1). Successors to thisenterprise are extremely rare.2

This article is an attempt to interpret one vastu textthrough drawing, and in so doing to reach some conclusionsabout its usability. It is a rst fruit of a collaborativestudy of the Samargaastradhra by Mattia Salviniand me. Salvini has transliterated the chapters on templearchitecture and translated them from the Sanskrit,3 andwe have begun to rene the translation through discussion.Our eventual aim is to produce a critical, annotated, and

illustrated translation of these chapters.A large proportion of the text consists of technicalterms, which must always have rendered it meaningless toanyone unable to visualise what is being conveyed. Accessto this vocabulary would be impossible if scholarship in thepast two centuries had not unearthed much of its meaning,especially in the last fty years, and particularly throughthe work of M.A. Dhaky encapsulated in the Encyclopaediaof Indian Temple Architecture (EITA) produced by theAmerican Institute of Indian Studies.4 While it is widely

understood that regional traditions employed dierentterminologies, a relatively standardised vocabulary hasbecome accessible to students of Indian temple architecture.This provides indispensable points of reference from which

1. Vimna consisting of ve Stories, from Ram Raz,Essay

on the Architecture of the Hindus(1834), Plate XXXII, with

alignment and prescribed proportions added.


2/22

42 South Asian Studies 25

AdAm HArdy

to solve the puzzles posed by the Samargaa. Butbetween these footholds, the text wanders widely from themodern academic norms. Moreover, the term for a givenelement may vary not only from chapter to chapter, butalmost from verse to verse, often giving the impression thatelegant linguistic variation is more important than precision,that from time to the reader is treated to riddles, and aboveall that meanings are conveyed as much by the contexts ofwords as by the actual words used. It is only by graspingthe relationships between the architectural elementsdenoted that a coherent picture of the intended templecan be imagined, and this involves knowing and keepingin mind the possible temple styles and compositions, andrecognising when the words t a particular pattern.

Once the parts and their organisation are identied,the main challenge is to understand the measurements andproportions prescribed. Measurements, when given, aregenerally in hastas(cubits) and agulas(digits or inches).On the whole the text is concerned with relative measure,not absolute measure, and the units concerned are bhgas,

padas,and aas. Everything becomes simpler when it is

realised that most of the time these terms are interchangeable,and are varied just to avoid verbal monotony. A given widthor height is divided into so many bhgas or padas,and anumber or fraction of these is then ascribed to its varioussections. Bhga, pada, and aa, therefore generallysignify a part or a module. Stara implies a layer, and thistoo, where vertical divisions are concerned, is often used

synonymously with the other terms.5

The Drvia chapters and their source

The Samargaa traces its authority to the divinearchitect Vivakarman, while proclaiming at the end ofevery chapter that its author is the King of Great Kings,Supreme Lord, Glorious Bhojadeva. This is taken to be thefamous Paramra king Bhoja of Dhar, who ruled c. AD1010-55, and this period indeed accords with the kinds oftemple architecture that are covered. Several chapters (55-57) deal with Ngara temples, comprising both the basicLatina mode and the now established composite, multi-

spired ekhar or Anekaka; although these usefulterms, gaining acceptance in modern scholarship, arenot used. These Ngara chapters, which clearly refer toarchitecture from the broad stylistic zone of central andwestern India, cover similar ground, each with its ownnomenclature for temple types, so that a name such asKailsa, for example, is assigned to dierent temple formsin dierent chapters. One chapter (65) is concerned withBhmija temples, another composite mode, which appearsin the eleventh century in the Paramra realm of Malwaand in surrounding regions. On formal grounds Bhmijatemples can be categorised as a variety of Ngara, thoughthe Samargaa treats them as separate. Chapters 61and 62, devoted to the Drvia temples of south India,are the focus of this article. The two chapters clearlybelong together as a coherent section that has been ratherarticially split. Although this is not the place to arguethe point in detail, the Samargaastradhra gives theoverwhelming impression that, even if it was for Bhojathat it was compiled, it is a patchwork of architectural textsderiving from dierent traditions.

The question therefore arises as to the provenance ofthe Drvia chapters, and what conception of Drvia theyhave in mind. An obvious surmise would be that they seethe Drvia through the eyes of architects from eleventh-century Malwa. The Bhmija chapter of the Samargaashows an explicit awareness of the Drvia which isentirely borne out by actual Bhmija temples. The textmentions the drviakarma ka (Figure 2), a version of theDrvia domed pavilion (ka). The numerous spirelets in aBhmija superstructure (Figure 3), at rst sight resemblingminiature Latina towers, in fact are often composed ofthese drviakarma aedicules, with their curvaceous leafynecklaces that are mutations of the makara monsters of thesouthern oor-with-joist-ends moulding (prati, vylamla).

2. Fragment at Ujjain (Madhya Pradesh), showing a

Drvia compostion of the Paramra period, crowned by a

drviakarma ka(photo by Michael Willis).


3/22

43

drviA TemplesinTHeSamargaaStradhra

3. Smallikharasin the tower of a Bhmija temple, composed of miniaturedrviakarma kas, the Drvia domed pavilionas conceived by the masons of Malwa under the Paramras. Udayevara Temple, Udayapur (Madhya Pradesh), c. 1058-80

(photo by SPA Bhopal).

Temple(no. of storeys)

Width prescribedin hasta (cubits)

Width x 2 Height prescribedin hasta (cubits)

Sum ofascending stages

1 5, 2 (5.08) 7.18 7 7

2 7 9.89 (10) 9

3 11 15.556 15 14,4+x

4 15 21.21 21 21

5 21 29.69 29 29

6 30 42.42 41 41

7 35 49.49 (49) 418 40 56.56 57-3 aas 57

9 51 72.12 72 -

10 56 79.196 79 79

11 65 91.92 92 91

12 67 94.75 95 95

Table 1: showing how the height of each type of temple is prescribed as the width x 2


4/22


AdAm HArdy

t1 t2 t3 t4 t5 t6 t7 t8 t9 t10 t11 t 12 =1 1

rest:3a2

7

2p1

j2a2

1g(1)

j1

9

3p1

j3a2

p1j2

2,4g1+xp1

j2

14,4+x

4p1

j3a2

p1j2

p1j2

3g1p1

j2

21

5

p1j3a2

p1j3

p1j2

p1j2

4g2p1

j2

29

6p1

j4a2

p1j4

p1j4

p1j3

p1j3

5g3p1

j2

41

7

p1j4a3

p1j3v1

p1j3v1

p1j2v1

p1j1v1

5g3p1

j1v1

- 41

8

9 8 8 7 6 5 44g2t3

57

9- - 8 7 (7) 5 5 4

5g2t3

-

1011 10 10 8 7 7 6 5 4

4g2t3

79

1114 12 11 9 8 7 6 5 4 4

4g2t3

91

1214 11 10 10 8 7 7 6 5 4 3

4g2t2

95

Table 2: Left hand column indicates number of storyes (talas, bhms) in the temple, top band indicates the storey concerned

(t1 = rsttala, and so on), right hand column indicates the sum of the storey heights; = ikhara (dome, here termed gha),

p =prastra(entablature, here termedkaprastra), v =ved, j =jagh (thigh, wall, shaft), a =adhihna(base, here

termedpha).


5/22

45


At Bhojpur, site of the unnished Drvia mega-templeattributed to the same Bhoja as supposedly wrote theSamargaa, two of the engraved line drawings on thesurrounding rocks (which we have been documenting inparallel with the present textual study) depict a form ofmaapa found nowhere else (Figure 4, right). The roofis not the tiered Phsan kind familiar at Khajuraho, forexample, and also present among the Bhojpur line drawings,

nor the Savraa type with multiple bell-topped pavilions,which arrived in this region from western India duringthe eleventh century. Instead, this type of maapa roofis composed of a peculiar form of miniature Drvia ordrviakarma pavilion. That at least one hall of this varietyhad been started at Bhojpur, if not completed, is attested bythe survival of small carved stone kasof exactly the kindshown in the drawings (Figure 4, left).

4. Left: fragments ofkasat Bhojpur (Madhya Pradesh). Right: nearby line drawing engraved on rocks, depicting amaaparoof composed of the same kind ofkas(authors sketch on site, as the basis for a measured drawing).

Level/storey (tala) Prescribed height Remaining height (RH) frombottom of level to top of temple

Width of level= RH/2

5thtala 3.50 10.00 7.07

4thtala 4.00 14.00 9.89

3thtala 4.25 18.25 12.90

2thtala 4.50 22.75 16.08

1sttala 4.50 27.25 19.26

Base 2.50 29.75 21.03

Table 3: Calculation of widths of upper storeys (courtesy of Bruno Dagens)


6/22


AdAm HArdy

Various aspects of the Samargaas version of theDrvia corroborate the suggestion that the text is writtenfrom a more northern perspective. Plan forms and, seemingly,dome forms are all based on a square, with no mention of therectangular, apsidal, circular, and elliptical variants usualin the far south. Compared with Tamil proportions, kasare generally squat and the launa section of pilasters is tooshort to accommodate the elegant, vase-like shape that itfollows, for example, in Cola temples. Even without detailedlinguistic analysis, it is clear that several of its basic termsare northern: a shrine is a prsda rather than a vimna, astorey a bhmi rather than a tala, the shaft of a pilaster a

jagh not apada, a moulded base apha not an adhihna,the dome of a ka a gha (bell) not a ikhara.

Yet nothing in the domains where the Bhmija heldsway can compare with the range and complexity of theDrvia architecture described in the text, at least untilone reaches Karnataka and Andhra in the lower Deccan,where both the Drvia and the Bhmija were wellknown. So it would seem very plausible that the Drviaof the Samargaa should be the later Kara Drvia,geographically and stylistically much closer than the Tamil

country to the Paramra orbit. However, the eleventhcentury Kara Drvia (commonly identied as Vesara)is unmistakable on account of its staggered plan forms andinterpenetrating compositional elements (Figure 5), whichnd no reection in the text.

Despite all the northward-pointing clues, theSamargaa prescribes a diversity of forms of mouldedbase found only in the Tamil tradition and its derivatives.Within the range of its plan forms, pride of place is givento unstaggered, ve-projection plans, with the optionof an internal ambulatory; in the eleventh century thispoints to the grand monuments of the later Colas (Figure10). So, too, does the range of elevations from one totwelve conceptual storeys. Kara Drvia temples arevirtually never above four storeys; more than four is rare inTamilnadu, though there are notable eleventh- to twelfth-century exceptions at Darasuram (ve), Tribhuvanam (six),Gangaikondacolapuram (eight), and Tanjavur (fourteen).6

The theoretical range from one to twelve is also that of theMayamata, a south Indian vastustra datable to before theend of the tenth century.7 These aspects x the origin ofthe Samargaastradhras Drvia temples denitively

5. The Siddhevara temple, Haveri (Karnataka), c. 1060s;a late Kara Drviavimna (photoby author).

6. A l aedicule (shrine image crowned by a barrel-roofed pavilion, or l) from the Sagamevara temple,Pattadakal (Karnataka), c. 730. While this element takesup one tier or conceptual storey (tala, bhmi) of the temple,it contains two conceptual storeys within itself, the upperprati marking the oor of the l.


7/22

47


in the far south, even if the text has undergone changesthrough its northward transmission.

Moulded bases

Chapter 61, entitled The Dening Traits of the Five Phas(phapacakalakaa), deals with phas or mouldedbases before continuing on to temple plans. If we compare

the names for phas in the Samargaa with those oftheMayamata and with those selected by the EITA (whichdoes not name its source), we nd names in common, butmostly denoting dierent types of base. All three follow aprocedure typical of southern vastu texts, enumerating everylittle sub-moulding, which makes it more dicult to graspthe principal divisions or mouldings. The latter generallycorrespond to the courses of masonry. Figure 6 is included

7. The ve phas of the Samargaa


8/22


AdAm HArdy

to clarify these divisions, using terms acceptable to modernscholarship but only partly true to the Samargaa. Onthe same basis it also shows the sequence of smaller scalemouldings in the pilasters.

The ve phas of the Samargaa are thePdabandha (or Pdabandhana), rbandha, Vedbandha(or Vedbandhana), Pratikrama, and Kurabandha (orKurakabandhana). The Pdabandha (Chapter 61, verses5-13) is transcribed in Figure 7a. This is an extremelywidespread type wherever Drvia architecture is found,both in the far south and in the Deccan. It consists offoot moulding (jagat), cushion moulding (kumuda, mostcommonly round or faceted), recess or miniature gallery(gala), and eave moulding with dormer windows (kapota)

equivalent to the basic Ngara sequence of kumbha-khura, kalaa, antarapaa, kapotl. For this rst type ofbase the text species, in agulas, thepravea,the relativeprojection and setting back of the mouldings. Here it seemsto be laying out the general principles for all the types.

What remains of the discussion of the Pratikrama(Chapter 61, verses 22-25)is a fragment, probably belongingto a description of the kind of base shown in Figure 7b(missing parts dotted). Incidentally, this corresponds tothe type called Pdabandha in theMayamata, followed inthis one case by the EITA. This is like the SamargaasPdabandha, but with a wide pa ik (llet) instead ofthe kapota. The passage in the Samargaa concerningthe Vedbandha (Chapter 61, verses 19-22) is cut short bythe insertion of the Pratikrama fragment, but is completeenough for Figure 7c to be inferred with condence.

Descriptions of the remaining types, the rbandhaand Kurabandha, are intact, but they contain a surprising

anomaly: the mouldings are in the wrong order. Therbandha (Figure 7d) has the full complement of standardmouldings, the ones shown in Figure 6, including theoor moulding (prati) the cluster dominated here by themakara, and the rail moulding (ved). In the Kurabandha(Figure 7e), only the ved is absent. The full sequence, asin Figure 6, and in that order, is not uncommon in the farsouth, and in the Deccan, from the eighth century onwards,becomes the norm for all but humble shrines. But neverdoes one nd, as one does here, the kapota above thepratiand the ved: the conceptual oor needs to be above theminiature roof of the base, while the railing runs around theedge of the oor platform. Only an inveterate text fetishistwould argue that the Samargaa must be right and all thetemples wrong. Clearly some verses have got out of order.Since the verse runs smoothly, the creases must have beenironed out by a sensitive scribe.8

Plans

Following the ve phas, the text deals with the vekinds of plan (talacchanda, plan rhythm or literally,metre): Padma, Mahpadma, Vardhamna, Svastika,Sarvatobhadra. The rst two, which I have not yet workedout fully, seem to be an interpolation from a more northerlytradition: they are dierent in character and treatmentfrom the others, involving the swinging of chords in theirconstruction. We are given no simple square plan, whichis needed for the one-storey temple described later: thePadma is square with three projections, the Mahpadmaapparently a star with eight points and sharp reentrantprojections like a pigs face (Chapter 61, verse 43).

8. Plans


9/22

49


Then three slightly varying ve-projection plansare given, as shown in Figure 8. A given number ofbhgas is specied for each, and then subdivided. Twoobservations are worth making here. Firstly, there is nota single, all-embracing grid, as is sometimes the case andoften assumed to be universal. In these examples the sidesare divided into fteen or twenty-eight parts, subdividedfor the projections and recesses (salilntara, jalntara,jalamrga), while the square is re-divided into four parts:one for the wall (bhitti), two for the sanctum (garbha).The second point to note concerns the names used for thedierent projections. Rather than kara,pratibhadra, andbhadra for corner, intermediate and central projections,here these are called respectively ka, pajara, and l,showing that they are conceived not just in terms of theplan, but as shrine-images or aedicules rising the fullheight of the rst tier, crowned respectively by square,horseshoe-arched and barrel-roofed pavilions.9. The one-storey temple (ekabhmika-prssda).

10. Mrgasahyevara Temple, Visalur (Tamil Nadu),

c. mid-ninth century; a one-storey (ekatala) minor shrine(alpa-vimna) (photo by Gerard Foekema).

a)

b)


10/22


AdAm HArdy

Finally, the principle of the sndhra plan isexplained. This has an internal ambulatory, as opposed tonirandhra which has none. Here the square is divided intotwelve, with four parts for the sanctum, one for the innerwall, one for the passageway, and two for the outer wall. Thisprocedure, presumably, is to be applied to the previouslydescribed envelopes, in order to make them sndhra.

Elevations

Chapter 62 is about the elevations (rdhvamna,literally upper measurement) of temples of one totwelve storeys (bhmis). Its title Drvia temples(drviaprsdalakaa) would be appropriate forChapters 61 and 62 together, and there was doubtless nosuch break in the original text from which they derive.Verse 1 proclaims:

rdhvamnam atha brmoghantapurapdita (?) |

prama karamnena sarvem evadhrayet || 1 ||

1. I will now explain the verticalmeasurement, starting from what is at thefoot, up to the very top of the gha (bell).

One should ascertain the size of everythingaccording to the measure of the corner(kara).

The second line is crucial. Karamna, the measurementstated to be the key one, is the diagonal of the square ofthe plan,9 and it turns out that in each case this dimension,

11. The ve-storey temple (pacabhmika-prssda) from the general description, three alternative interpretations.


11/22

51


i.e. the width x 2, determines the height of the shrine tothe top of the crowning gha or dome. The arithmeticalgame of making everything add up to this gure seems tohave been more important than making well proportionedtemple towers.

The one-storey temple (Figure 9) corresponds towhat has come to be known in academic parlance as an

ekatala-alpa-vimna (Figure 10). Its width is prescribed as5 hastas and 2 agulas, its height as 7 hastas. Assuming 24digits to the cubit, the width is 5.08 hastas, giving a diagonalof 7.18, close enough to 7. In Figure 9, I have chosen thePdabandha base (Figure 7a) and divided the 2 cubitsprescribed for the base proportionately. All the way up theedge of the shrine, every single element and its size is speltout. If, simply from prior knowledge of the tradition, oneadds horseshoe arches (nss) and intermediate pilasters inthe wall, the result is a passable alpa-vimna. It is, however,crowned by a very squat ka, has an unrealisticallymonotonous number of parts measuring one bhga. Thereis also one strange extra component, the ardhabharaa or

bharardha, among the pilaster mouldings.For each type beyond the one-storey temple, the

procedure is rst to give the overall width and height, andthen to list the height of each storey, culminating in theneck (kaha), dome (gha), and the sequence of partsconstituting the nial. For the shrines up to the one withseven storeys, separate heights are given for the wall andentablature (kaprastra) of each tier; the seven-storeytemple, oddly, has a ved at the base of each storey,assigned a height separately.

From the eight-storey temple onwards, each tieris assigned a single dimension, without breaking it down.These various dimensions are ostensibly in hasta (cubits),but the term bhga (part) is often used synonymously here.Table 1 shows the dimensions for the successive stages, andtheir sum, for each of the twelve temple types. Table 2 showshow the total height deduced in this way is always equal, atleast approximately, to the height of the temple prescribedat the outset, and that this, in turn, is equal or nearly equalto the length of the diagonal of the plan (karamna). In thepursuit of this arithmetical goal, the incremental reductionsin height seem rather arbitrary, and certainly do not followregular arithmetical or geometrical progressions.10 Theseven- and nine-storeyshrines are anomalous, because theseven-storey shrine as described here has in fact only sixstoreys, and the description of the rst two bhmisof thenine-storey shrine is missing from the text. Up as far as theve-storey temple, once the general dimensions have beengiven, the text starts again at the bottom and goes all theway up through every small division, as we have alreadyseen for the one-storey shrine.

The ve-storey temple (pacabhmika prsda)may be taken here as an illustration, and as a furtherdemonstration of the extent to which the text can beuseful as a guide to design. The relevant passage is

appended to this article. This is the only temple inthe series for which the plan type (the Sarvatobhadra)and the type of base (the rbandhapha) are actuallyspecied. For all the others these appear to be open tochoice, although the range does not provide a suitableplan for every elevation. In order to give the ve-storeytemple a Sarvatobhadra plan (Figure 7b), its width oftwenty-one hastas must be re-divided into twenty-eightparts. The prescribed height is 29 hastas, and the stagesup to the top of the dome add up to 29.11 The remaininghalf may be intended for the nia l, but it would only allowfor a stunted one. Sarvatobhadra, incidentally, as well ashaving the technical sense of a plan with pronouncedcardinal projections (and often with four entrances),means beautiful on all sides, and the text contains onesmall and welcome ight into poetry that takes o fromthis idea (Chapter 62, verses 123-126).12

12. Airvatevara temple, Darasuram (Tamil Nadu), mid-

twelfth century, a ve-storey (pacatala) vimna. Unlike

the equivalent temple in the Samargaa, here the ka,

pajara, and l pavilions crowning the rst and second

tiers are two-storey ones (avarga, of six divisions), i.e.

equivalent to the entire l aedicule of Figure 6 (photo by

Gerard Foekema).


12/22


AdAm HArdy

The architectural composition across the elevationis not explained, except inasmuch as it is implied for therst storey by the plan. For the other two ve-projectionplan types, the pattern of aedicules in the rst tier is clearlythe classic ka/pajara/l/pajara/ka , whereasfor the Sarvatobhadra plan the corner and intermediateprojections are equal, and the terms kaand rathikareused synonymously. The intended pattern seems thereforeto be ka/ka/l/ka/ka. One can assume that theve projections would be carried up the tower, potentially

with minor variations in the types of aedicule, as at theAirvatevara temple, Darasuram, illustrated in Figure 12.

Interpreting the elevation depends partly on themeaning ofkaprastra, the term used here instead of themore usual prastra. Prastra is generally taken to meanconceptual beam (uttara) and the eave moulding (kapota),together with the oor moulding (prati) above (see Figure6).13Kaprastra could conceivably mean an entablaturethat includes the entire ka or upper pavilion, as well asthe kapota. This interpretation, followed in Figure 11a,

13. The ve-storey temple (pacabhmika-prssda) from the detailed description, which turns out to have six storeys.


13/22

53


demands that the uppermost tier, below the dome, shouldmirror the others in having a crowning chain of pavilions(a hra, though the term is not found here) a feature,incidentally, that became the norm for the top storey ineleventh-century Karnataka, but remained unusual forthat position in the far south. Several problems are madeapparent by Figure 11a. The rst storey is dumpy; thereis insucient upward diminution, as the second storey isidentical in size to the rst and succeeding entablaturesare all of equal height.14 Moreover, this version gives usuncomfortably compressed pavilions, while, from thesecond storey upwards, an abnormal expanse of pilastershaft is visible.

Kaprastra appears, therefore, to mean anentablature that supports the ka, rather than including it in other words the same as the generally understood sense ofprastra. As shown in Figure 11b, this interpretation allowsthe top storey to take the form, normal in Tamil Nadu, of acrowning alpa-vimna. The interpretation assumes that thekasand ls of each storey overlap with the storey above,with the advantage that one is free to give them comfortableproportions and make them diminish. Here the fact thatthe rst two jaghs are equal is not nearly as disturbing.However, there remains an absence of diminution in theprastras, and as drawn here there is one fatal aw: the

pavilions, even if kept rather low, clash with the pilastermouldings behind them.

In Figures 11a and 11b the shrinking of the widthof the tower from level to level has been gauged by eye,as the text gives no explicit instructions about how to dothis. I am extremely grateful to Bruno Dagens for pointingout the likely implicit method,15 which overcomes variousdiculties and results in a far more convincing elevation.The width of each successive tier can be calculated inproportion to the remaining height, as shown in Table 3provided by Dagens and as illustrated in Figure 11c. Herethere is no clash between pavilions and pilasters. Theproblem with the equalprastra heights remains, however,still attesting to the dominance of the numbers game. Thecorollary of the method for calculating storey widths, thatthe foot of the corner of the rst storey lines up with theequivalent points all the way up the tapering tower, is whatmakes the rst storey rather short.

Interestingly, this very form of all-the-way-downalignment is present to a large degree in the interpretationsof theMnasra drawnby Ram Razs collaborator. In hisVimna consisting of ve Stories (Figure 1), a steeperprole (height = twice the width) allows the rst tala tobe rather taller than its equivalent in the Samargaa.And one need only look at Darasuram (Figure 12) to

14. Interpretation of the three-storey Svastika-vimna described in the Mayamata.


14/22


AdAm HArdy

realise that certain actual temples may well follow thisprinciple and that the eect of a short rst storey may bemitigated by a sub-base or pedestal. Could it be that theintellectual formulations of texts have begun to feed backinto practice? It is worth reecting on how, in practice,this kind of alignment of corners can be achieved. Ina drawing it can be arrived at simply by tracing a line.

For building an actual temple, the widths of the storeyscould be calculated as in Table 3. However, the startingpoint of a storey, if conceived as standing on a prastra,is theoretical rather than tangible, especially where theprole of the superstructure necessitates false or applied(arpita) rather than freestanding (anarpita) parapets.Since it is of dubious use to calculate the positions ofhidden points, the more practical method would be toscale o a drawing.

Figure 13 is drawn, from toe to tip, from theconcluding detailed description of the ve-storey Drvia temple of the Samargaa. I have dutifully drawn therbandhapha with its erroneous sequence, scaled to

the prescribed 2 bhgas. Above this the sequences ofmouldings are largely true to life, though once again theunlikely and monotonous proportion of 1:1:1:1:1 appearseverywhere. Between the dierent storeys there are a fewslight and improbable variations in the sequences and intheir relative proportions, and many variations in the termsused. As one nears the summit an inconsistency looms intoview: the description is actually of a six-storey temple vefully-aedicular bhmisplus a top tier in the form of an alpa-vimna. Is this, therefore, a misplaced description of theabhmika-prsda?

The detailed descriptions of elevations are introducedby such phrases as The number ofhastas(cubits) has beenindicated: now the subdivision will be explained (verse 74).This suggests that the proportions of part to part lain down inthe detailed descriptions are relative to the dimensions givenfor the respective bhmis in the general descriptions. Thatwould account for the fact that, in the detailed descriptionbeing examined here, each tier above the rst (52 parts)is virtually the same height (36-37 parts), i.e. withoutdiminution. However, even if the anomaly of the sixth storeyis ignored, the t between the detailed description andthe general one is weak. If we work out, from the detaileddescription, the ratio in each storey of the kaprastra tothe wall, it does not correspond closely to the ratio impliedby the general description. And if the rst storey in thedesign deduced from the detailed description is ascribed awidth based on the ratio of width to height implied by thegeneral description, then its aedicular components becomeridiculously broad and squat.

Conclusion

A sthapati wishing to follow the Samargaastradhrafaithfully in the design of Drvia temples would nd his

aim complicated by ungainly proportions, contradictions,and occasional solecisms. Many of the perversities in thetext are casualties of its transmission through space andtime. Yet an underlying logic comes across, and this isbased on an ingenious arithmetical game that is partly atodds with Drvia temples as known from the architecturaltradition itself. The person or people who original ly wrote

Chapters 61 and 62 of the Samargaastradhra knewDrvia temple architecture intimately, but expounded ita way that was not always helpful for practice. Perhaps,then, it was written by and for south Indian connoisseurs.And if, as seems certain, the text came from the distantsouth, it is unlikely that, once these chapters had beensewn into Bhojas compendium, the south Indian templearchitecture depicted in them could be imagined byaesthetes at court in Dhar. Compendiousness must thenhave been an end in itself, as this would have been a textthat the literate could not understand and the practicalcould not use.

But before it is concluded that the function of

vastustras was therefore not to be understood or usedfor design, it should be pointed out that some texts aremore usable. A case in point is the Svastika form ofthree-storeyed temple explained by the Mayamata,16which I have interpreted in Figure 14. Here the widthof each storey is specied together with its horizontalsubdivision; hence the whole prole can be worked out,as can the aedicular composition within it. Some optionsare given, so decisions need to be made. There areoptions for the temple height in proportion to its width,and I have chosen the ratio of 1:1, so that a vertical partor bhga is one and a half times a horizontal one. The

prastra height is given, but judgment must be exercisedin deciding the heights of the kasand lsand of theirconstituent mouldings. As the text says that the lsofthe rst storey are taller than the kas, I have chosento use a two-tier or avarga-l (cf. Figure 12). I havebalked only at burdening the upper storey with sixteensmall niches as well as ninety-six small false dormerwindows.17 This top tier is set back to a surprising degree,yet altogether the result, to my eyes, is a pleasant Colaperiod vimna. There has been room for interpretationand invention, and at the same time the satisfaction ofbeing true to the stra.

Vastustra texts, then, vary in their degree ofutility as guides to design. This is true when dierentchapters within the Samargaastradhra arecompared, and I hope to demonstrate in future studiesthat some of the Samargaas instructions for Ngaraand Bhmija temples yield coherent designs. Each textmust be taken on its own terms, and perhaps there canbe no general conclusions as to authorship and audience.Before such questions can be broached for a given text, anecessary rst step is to try to draw the architecture thatit describes.


15/22

55


APPENDIX

The Five-storey Temple(pacabhmika-prsda),from the Samargaastradhra,Chapter 62[This translation by Mattia Salvini is included here to show how the interpretations put forward in the article have been arrivedat. Technical terms have not been given English equivalents, as it is felt that a better understanding can be gained by readingthe text in close conjunction with Figures 11 and 13. This is work in progress; comments and suggestion will be welcomed by the

author of this article and by the translator. The passage begins with the overall description illustrated in Figure 11.]

pacabhaumam atha brma prsda rjapjitam || 106 ||

We will now explain the ve-storeyed temple, worshipped by kings.

vistrea vidhtavya sa hastra(st)n ekaviatim |vibhjayet tathotsedha pdonatriata karn || 107 ||

107. It should be built with a width of twenty-one cubits.18

One should then divide its elevation into thirty cubits minus a quarter.

pha bhgadvaya srdha jagh(srmr)19 tribhgik |kurvta kaprastra srdhahasta ca buddhimn || 108 ||

108. The pha (base) is two bhgas and a half, the jagh (wall zone) is three bhgas.The intelligent should construct a kaprastra of one and half cubits.

jagh dvity kartavy hastatritayam ucchrit |bhyopi kaprastra srdhahasta prakalpayet || 109 ||

109. A second jagh should be constructed, with a height of three cubits.Once again one should construct a kaprastra of one and half cubits.

jagh tty kartavy pdahna karatrayam |srdhahastasamutsedha kaprastra iyate || 110 ||

110. The third jagh should be constructed, three cubits minus a quarter.The kaprastra is accepted as one and a half cubits in elevation.

caturthabhmijagh ca srdhahastadvayocchrit |kaprastraka kuryt prvamnena buddhimn || 111 ||

111. The jagh of the fourth storey should be two cubits and half high.The intelligent should construct a kaprastra of the same size as the previous ones.

pacamy bhuvi kurvta (jagh s?) hi karadvayam |kurvta kaprastra tath prg(gu)dito yath || 112 ||

112. On the fth level one should construct a jagh of two cubits.One should build the kaprastra as explained earlier. 20

kurydd hastadvayotsedha kapotam api buddhimn |caturbhgasamutsedh mahgha vidhyate || 113 ||

113. The intelligent should also construct a kapota with an elevation of two cubits.A great gha is to be built, with an elevation of four bhgas.

uparid bhavet tatra prsde pacabhmike |kumbha tadrdhva kurvta starn ekonaviatim || 114 ||

114. On the upper portion of that ve-storeyed temple,one should construct a kumbha above (the gha), divided into nineteen staras.

sasthnam etat kartavya sarvatobhadrasajakau(ke) |vibhjayed vieea tata staravibhjant || 115 ||


16/22


AdAm HArdy

115. This type of arrangement is to be constructed for the one called Sarvatobhadra.21

[The overall description ends here, and the text continues with the detailed description illustrated in Figure 13]

One should then specically subdivide according to a division into staras.

rbandhapha kartavya srdhahastadvayocchritam |caturdaastara jagh kartavy stambhasayut || 116 ||

116. A rbandhapha should be constructed, with a height of one cubit and a half.The jagh should be built in fourteen staras, endowed with a stambha (pilaster).

kartavy dvistar ml launa starasammitam |vidadhta stara padmakumbhagaasamanvitam (?) || 117 ||

117. The ml should be built in two staras, while the launa measures one stara.One should construct one stara of padmakumbhagaa(?).

ucchla dvistara kury(dio?)bhga vidhyate |dvistara hraka krya pa caiva tathvidh || 118 ||

118. One should construct the ucchla in two staras, while the ia is built as one bhga.The hra should be built as two staras, and the paa should be the same as that.

paik staram eka ca vasanta dvistara tata |vasantapaik bhga kapota tristara tata | | 119 ||

119. The paik is one stara, then the vasanta is two staras;the vasantapaik is one bhga and then the kapota is three staras.

chedam ekastara kuryt staramtra ca mehakam |makara bhgam eka ca bhga (carla?)22 paik(m) || 120 ||

120. One should construct the cheda in one stara and the meha measuring one stara.The makara is one bhga, and the paik is one bhga as well.

kurvta bhgika cheda tata kaha ca bhgikam |kaha cheda tata kaha ca ++++ bhgikam || 121 ||

121. One should build the cheda in one bhga and then the kaha in one bhga.There is a kaha, a cheda, then again a kaha (...), one bhga in size. 23

(vkhypaik?) bhga ved vicakaa |kurvta bhgika cheda tata kaha staradvayam || 122 ||

122. The expert should make the vkhypaik (?) and the ved one bhga in size,the cheda in one bhga, and then a kaha of two staras.

stara stara prakurvta paik padmapaik |kaprastrake kuryn makarnanapacakam || 123 ||

123. One should construct the paik and padmapaik, each being one stara in size.In the kaprastra one should make ve makara faces,

vicitrarpa sarvsu diku sarvagunvitam |rdhvata paikys tu gha pacastar bhavet || 124 | |

124. with a striking appearance, endowed with all good qualities, in all directions.Above, the gha of the paik should be of ve staras,

nsikbhir vicitrbhir atyudrbhir anvit |bhadri yasya dyante ke ke samantata || 125 ||

125. endowed with beautiful, and extremely large nsiks (false dormer windows).


17/22

57


Where, in each ka, on every side, bhadras (or, good things) are to be seen,

sa sarvatobhadra iti prsda ilpin mata |avalambena tadanu stambhaccheda prakalpayet || 126 | |

126. that temple is considered to be the Sarvatobhadra by the craftsmen.As a support for that, one should construct a stambhaccheda.

[Verses 127-133a have been omitted here. These discuss the relative degrees of projection of various mouldings. The ascentthen continues with the second tier.]

mldyair launa caika bharaa kalaas tath |yath ml tathocchla vragaa stara bhavet || 134 ||

134, There should be one launa with ml and so forth, a bharaa, and a kalaa.Just like the ml, so also the ucchla and the vragaa should be of one stara.

ucchlahrake paasame kuryd vicakaa |paik bhgikotsedh vsanta + + k tath | | 135 ||

135. The expert should construct the ucchla and hra of the same size as the paa.The paik should be one bhga high, and the vsanta in the same way (one bhga? 24).

kapota tristarotsedha (chedo?) satryaavarjitam |chedasyrdhe bhaven meho makara paik tath || 136 ||

136. The kapota should be three staras high, and the cheda should be that minus three aas.The meha should be half of the cheda, the makara and the paik should be the same as that.

tata cheda (ca) kaha ca +++ paik tath |mlrdhena prakurvta cchedam eva tato budha || 137 ||

137. Then there should be a cheda, a kaha and a paik. 25

After that, the intelligent should construct a cheda half the size of the garland.

puna kaha prakurvta hrakego(a) samanvitam |pai(k) padmaprv ca tribhgo(ge) na kapotake || 138 | |

138. Once again one should construct a kaha, endowed with a hra.There should be a paik, preceded by a padma, in the kapota. 26

kuryc catu star gha dvbhy kumbha tathopari |puna chedo bhaved bhga jagh kurvta saptabhi || 139 ||

139. One should construct a gha of four staras, and on top of that a kumbha in two staras.

[The third tier begins here, on top of the cheda]

Again, there should be a cheda of one bhga, and one should construct a jagh of seven staras.

(stamh?) vidhtavy mloccu(cco) dvistaro bhavet |launa bharaa kumbho gaa ceti stara staram || 140 ||

140. A stamh (staml?) should be constructed: the height 27 of the ml should be two staras.

The launa, bharaa, kumbha, and gaa should be one stara each.

gaadviguam ucchla hrapaas tathaiva ca |paik staram eka syd vasantapaiksya ca || 141 ||

141. The ucchla should be twice the size of the gaa, and likewise the hrapaa.The paik should be one stara, as also its vasantapaik.

pha da(a)gua kuryc chedamehau stara staram |stara kurvta (rkara?) (tath) makarapaikm || 142 ||


18/22


AdAm HArdy

142. One should construct a pha in ten guas, while the cheda and meha should be one stara each.One should construct the rkara (makara?) in one stara, and so also the makarapaik.28

stara cheda ca kaha ca paik vedik tath |cheda kuryt punar bhga kaa taddvigua tata || 143 ||

143. The cheda, kaha, paik and vedik should be one stara each.Again, one should construct a cheda of one bhga and then a kaha twice that size.

paik staram eka syd vasantapaik staram |catustar bhaved gha ( prggrcakabhit ?) || 144 ||

144. The paik should be one stara, and the vasantapaik should be one stara.The gha should be four staras, (ornate with prggrcaka = prgrvaka?)

tasyopari puna kumbha ghardhenaiva krayet |cheda bhga vijnyj jagh saptik smt || 145 ||

145. Again one should construct on top of that a kumbha half the size of the gha.

[The fourth tier begins here, on top of the cheda]

One should know the cheda to be one bhga, while the jagh is known to be seven aas.

ml dvibhgik kry bhgika launa bhavet |bharaa kumbhaka gaa kuryl launavad budha || 146 ||

146. The ml should be constructed two bhgas in size, while launa should be one bhga.The intelligent should construct the bharaa, kumbhaka and gaa the same size as the launa.

ucchla gaaka caiva hraknta ca bhgikam |srdha bhga bhavet ++ paikrdha stara bhavet || 147 ||

147. The ucchla, gaaka and hraknta should be one bhga each.The (...29) should be one and a half bhgas, while the paik should be half a stara.

++ ta bhgam eka syd vasantkhy ca paik |kapota tristara kuryn nsyukta vicakaa || 148 ||

148. The (...30) should be one bhga, and so also the paik called vasant.The expert should construct a kapota of three staras, endowed with nss (false dormer windows).

chedam aena kurvta (maam aena ?) krayet |makare paik cheda vidadhta stara staram || 149 ||

149. One should construct a cheda of one aa, and should build a maa of one aa.In the makara, one should build a paik and a cheda, one stara each.

kurvta bhgika (kaha) paik vedikm api |bhga kuryt puna cheda tata kaa dvibhgikam || 150 ||

150. One should make a kaha, paik and vedik of one bhga each.Again, one should construct a cheda of one bhga and then a kaha of two bhgas.

paik padmaprv ca vidhtavy stara staram |kurvta gham upari caturbhg vicakaa | | 151 ||

151. The paik and the padmaprv31 should be built one stara each.The expert should construct on top of that a gha of four bhgas.

tadardham rdhvata kumbha chedam ardhena tasya ca |jagh abgik kry (mtt gena suna krayet ?) || 152 ||

152. Above that, there is a kumbha half that size, and a cheda half the size of the kumbha.


19/22

59


[The fth tier begins here, on top of the cheda]

A jagh of six bhgas should be made (...)

launa bharaa kumbha gaam ucchla(va?) ke |hraka ceti kurvta bhgikni pthak pthak || 153 ||

153.One should construct the launa, bharaa, kumbha, gaa, the ucchla and vaka (vedik?), and the hra, each being

one bhga in size.srdhabhga bhavet paa paikrdhastarocchrit |vasanta bhgam eka syd vasantkhy ca paik || 154 ||

154. The paa should be one and a half bhga, while the paik should have a height of half a stara.The vasanta should be one bhga, as also the paik called vasant.

kapota tristara kuryc cheda (tryaonaasakm?) |maako makara caiva paik chedakahakau || 155 ||

155. One should make a kapota in three staras, and the cheda (...).The maaka, makara, paik, cheda, kahaka,

kaha pa ca ved ca ccheda ca syt stara staram |

dvityo dvistara kaho bhgik paik bhavet || 156 ||156. kaha, pa, ved and cheda should be one stara each.The second kaha should be two staras, and the paik should be one bhga.

tathaiva padmasaj ca syd ucchryea paik |gha kuryc caturbhg kumbham ardhena tasya ca || 157 ||

157. The paik called padma should have the same height.One should construct a ghat of four bhgas, and a kumbha half that size,

chedam ekena bhgena jaghma+rdhabhgikm |ml(m e)kena bhgena launa srdhabhgikam || 158 | |

[The anomalous sixth tier begins here, on top of the cheda]

158. a cheda of one bhga, a jagh of (half or one and a half32) bhgas,a ml of one bhga,33 and a launa of one and a half bhgas.

tathaiva bharaa kuryt kumbhocchle stara staram |hraka bhgika kuryt paa srdhastara tata || 159 ||

159. In the same way one should construct a bharaa, and the kumbha and ucchla in one stara each.One should create the hra in one bhga and then the paa in one and a half stara.

paikrdhastara kry vasanta ca stara tata |kapota dvistara kuryd vedm ardhastara tath || 160 ||

160. The paik should be made half a stara in size, and then the vasanta in one stara.One should construct a kapota of two staras, and a ved of half stara.

yath chedas tath mao makara ca vidhyate |paikrdhastara kry chedopy ardhastara bhavet || 161 ||

161. The maa and makara are to be built just the same size as the cheda.The paik should be built in half a stara, and the cheda also should be half a stara.

bhga kaha paik ca ved kry dvibhgik |chedo bhgena kartavya kaa cnyas tribhgika | | 162 ||

162. The kaha and paik should be one bhga, while the ved should be constructed as two bhgas.


20/22


AdAm HArdy

The cheda is to be made as one bhga, and one more kaha should be built as three bhgas.

paik padmapatr ca vidadhta stara staram |tugasya calana krya dvibhgikam anantaram || 163 ||

163. One should build the paik and the padmapatr in one stara each.The calana of the tuga should be constructed contiguously to that, and of two bhgas.

gha kry samutsedh(t) trayastriadvibhgik |sarvatobhadrasayukt candralvibhit || 164 ||

164.The gha should be constructed with a height of thirty bhgas,endowed with a Sarvatobhadra and ornate with candrals.

kurvta tristara padma citrapatrasamanvitam |tasyopari (bhavet) kumbha caturdaavibhgika || 165 ||

165. One should construct a padma of three staras, endowed with beautiful petals.Above that, there should be a kumbha of fourteen bhgas.

grv dvibhgik kry kara caiva tathvidha |bjapra tata krya s(o)bhsayuktam ardhata | | 166 ||

166. The grv should be constructed in two bhgas, and the kara should be of that same size.Then a bjapra should be constructed, beautiful34 and half that size.

padmacakra trila v vidhtavya yathocitam |prottugagrsasayukta ++ makaramehakai || 167 ||

167. A padmacakra or a trila should be made, as appropriate,endowed with prottuga and grsa, () and having makaras and mehakas.

sottugakake kuryd eva diku vidiku ca |bhmau bhmau vidhtavy l sdhyalatoraam (?)35 || 168 ||

168. One should construct a sottuga and kaka in this way, both in the main and intermediate directions.On each storey, a l should be built, with sdhyalatoraas.

koe koe ca ( + kar ) bhadre karikya(ka)rn api |++ kais tribhir yukta caturbhi ca jalntarai || 169 ||

169. On each of the good corners (one should create36) karikara37 (elephants trunks) as well.Endowed with three () kas and four jalntaras (recesses),

kurvta sarvatobhadram evalakaalakitam | 170ab |

170ab. one should construct the Sarvatobhara according to these specic features []

Note: As the author is serving as Editor of South AsianStudies it needs to be noted here that this article has beenpeer reviewed, by Professor M. A. Dhaky and ProfessorBruno Dagens.

ACKNOWLEGEMENTS

This study has been made possible by funding from theArts and Humanities Research Council for the projectentitled The Indian Temple: Production, Place, Patronage.Thanks are due to Mattia Salvini, whose collaboration hasbeen the whole basis of this article. I am very grateful M.A. Dhaky for his support, and to Bruno Dagens for hisinvaluable contribution.

NOTES

1 Ram Raz, Essay on the Architecture of the Hindus (London: Royal Asiatic Society 1834).

2 Exceptions are the reconstructions of threeplans from the Samargaastradhra byStella Kramrisch in The Hindu Temple (Calcutta:University of Calcutta, 1946), pp. 247, 250, 251;and the two elevational diagrams by BrunoDagens in Mayamata: an Indian Treatise on

Housing, Architecture and Iconography, ed. byBruno Dagens (Delhi: Sitaram Bharatia Institute,1985), pp. 144, 139. Mention should also be madeof the drawings in P. K. Acharyas works on the


21/22

61


Mnasra, in a strange, hybrid style that, as far as Iknow, has never been seen in a real building, evenfrom the 1920s.

3 The Sanskrit text was published as SamaranganaSutradhara of Maharajadhiraja Bhoja, ed. by T.Ganapathi Sastry (Baroda: Gaekwad OrientalSeries Vols. 25 and 32, 1924 and 1925). It was

drawn on extensively by Kramrisch, and byD.N. Shukla in Vstu-stra: Hindu Scienceof Architecture (Delhi: Munshiram, 1993). Thetext has recently been translated into Englishby Sudarshan Kumar Sharma, as SamargaaStradhra of Bhojadeva: An Ancient Treatise on

Architecture (Delhi: Parimal Publications, 2007).To me the architectural parts of this translationare totally opaque. An excellent translation andcritical commentary by Felix Otter on the partsof the text concerning domestic architecture has

just been published: Residential Architecture inBhojas Samaranganasutradhara (Delhi: Motilal

Banarasidass, 2009).4 Especially, in this context, Michael W Meister

(ed.), Encyclopaedia of Indian Temple ArchitectureVol.I, Part 1, South India: Lower Drviada, 200BC-AD 1324 (Delhi: American Institute of IndianStudies and Manohar, 1983)

5 Otter takes bhga as a term to form fractions; it maywork in other contexts, but in this chapter, stara andbhga are clearly used as synonyms. See also14.24,where bhga and pada are used interchangeably:pacimottarabhgasth vpm api ca krayet|v(yau?yu)sugrvapadayor gandharvasya cabhyata || (MS)

6 In fact the Bhadvara at Tanjavur (c. AD 1000)set the trend for many-storeyed temples. Formy argument that this temple, strictly speaking,has fourteen storeys, rather than fteen, seeAdam Hardy, The Temple Architecture of India(Chichester: Wiley, 2007), pp. 220-21.

7 Dagens argues that a brief mention in theMayamata of temples with sixteen storeys is alater interpolation taking account of the large Colatemples: seeMayamata, p. v.

8 As suggested to me by Michael Willis.9 See Kramrisch, p. 268.10 Pierre Pichard has shown that the storeys of the

temple at Gangaicondacholapuram do diminishaccording to a geometrical progression: Thanjavur

Bhadvara, An Architectural Study (Delhi:IGNCA and cole Franaise de lExtrme Orient,1995), pp. 84-93.

11 See note 19, which allows for the possibility thatthe stages may indeed add up to 29.

12 Mattia Salvini supports this interpretation, andhas provided the following note: Puns are such a

common feature of Sanskrit literature that indeedit is unlikely that the name Sarvatobhadra wouldnot have been felt to indicate its most literalsense as well as its technical one. Besides, thename is explicitly interpreted in this mannerin another place too (Chapter 55, verse 31):karoti sarvato bhadra sarvatobhadraka kta

Once constructed, the Sarvatobhadracauses good on all sides (sarvato bhadram).To be precise, it is likely that bhadra is used herein a sense akin to meritorious or auspicious.The more technical sense of Sarvatobhadra as [atemple] with bhadras on all sides is given in theetymology oered in chapter 62, verses 125-126,translated in the Appendix.

13 Prastras are shown in this way in Ram Raz, PlateXIX. For the same understanding in a treatise by acontemporary practitioner of Drviaarchitecture,see V. Ganapati Sthapati, Sthaptya Veda (Chennai:Dakshinaa Publishing House, 2005), drawing on

p. 392.14 Note 19 has arrived too late for revisions to be

made, but the rst storey may in fact be half abhga taller than the second. Table 3 would have tobe recalculated, and Figure 11a-c redrawn slightlydierently. Such are the hazards of working withtexts, but the reader must decide whether or not thisbhgas up my entire argument.

15 Jai t surpris quand vous criviez que noinstructions are given for how to diminish thewidth of the tower from level to level, car ce sontdes indications qui sont trs systmatiquementfournies dune manire ou dune autre. Enn ilme semble que le problme est rgl en utilisantvotre hypothse sur le sens de karnamanena, carsi elle permet dobtenir la hauteur partir de lalargeur elle permet aussi le contraire: cela nousdonne donc le tableau que vous trouverez en pice

jointe et apporte une preuve de plus que votrehypothse tient la route (Bruno Dagens, personalcommunication).

16 Mayamata 21.2b-10, Dagens, pp. 128-9.17 Ibid. 21.9.18 Following the suggestion of hastn instead of

hastrn (MS).19 Srmr could be possibly split as sra+mra,

giving a sense akin to pithy mango, which tsneither context nor metre. I would propose that itmay be a corruption ofsrdha-, which would thenmean that the jagh should be made as threebhgas and a half(MS).

20 Accepting the suggestion ofprg udito (MS).21 Accepting -sajake (MS).22 I would propose that carla may actually have been

ca + a two-syllable term qualifyingpaik (MS).


22/22

AdAm HArdy

23 One bhga is tentative. Since four syllables aremissing, it is impossible to exclude other possibilit ies,like two and a halfbhgas (srdhadvibhgikam)or three and a half bhgas (srdhatribhgikam),or one and a halfbhga (kurytsrdhabhgikam)(MS).

24 I would propose that ++k may have been bhgik

(MS).25 Here vasanta-paikm would t the metre (and

perhaps the context) well (MS).26 Here I translate accepting the suggestions of

hrakea,paik and tribhgena. (MS).27 Accepting mlocco (MS).28 Accepting daa- and tath (MS).29 Considering metre and context, I wonder whether

the missing part could be kaam (MS).30 Could the missing bit be vasanta ? Compare

verse 154 (MS).31 This could also mean The paik, preceded by a

padma (MS).32 The measurement of the jagh is here tentative.

Another available reading is jagh marddho

tribhgik. Perhaps it could have been jaghsrdhatribhgikm, in which case thejagh wouldbe three and a halfbhgas (MS).

33 Accepting mlam ekena (MS).34 Reading obha- (MS).35 Sdhyalatoraa probably a corruption (MS).36 Perhaps +kar could have been kurvta ? (MS).37 Accepting the suggestion ofkarikarn (MS).

Documents

vastu - About Dravid Temples