Chapter #3: Cave 26 as an Inaugural Monument: Early Features

If one agrees with Bakker (1999,41) and many other scholars that "Cave 26...may be of a later date" than many or most of the other caves at the site, then my own view of the cave's development must be incorrect. And if this is incorrect, the whole "Short Chronology" must be too wrong. For this reason a proper understanding of Cave 26's development-a justification of its position as an inaugural monument, as well one whose history ultimately speaks of the fall of empire-is essential.

K. Khandalavala (1991, 102) expresses the case for the opposition most decisively. "The inscription in Cave 26 is a bed-rock of any sensible chronology of the Ajantā Caves.... All the far-fetched attempts, with no evidence in support, which have been made by Dr. Spink to surmount what the inscription in Cave 26 states can only be regarded as exercises in sheer futility for no amount of ingenuity can ever displace the fact that Cave 26 could only have been excavated after the fall of the Vākāṭaka dominion."

To better explain my own position, I will make a listing of some of the Cave 26 complex's never-noticed early forms and features, and link them with the site's overall development. As a result of this we will be better able to reconstruct the remarkable history of the whole site which the complicated development of Buddhabhadra's cave so tantalizingly mirrors.

1. Astylar Plans

The astylar plan of Cave 26's four wings is revealing as a notably early feature, showing the early dependence on Hinayāna prototypes at the site. As in most Hinayāna caves, each wing's central hall, as planned, was to be surrounded by monks' cells on left, rear, and right; but this notably simple arrangement, representing the original plan of all four of Cave 26's wings, undergoes major later adjustments in all cases, as conventions at the site develop.

2. Porch End Development

All of Cave 26's wings were conceived without porch end cells, a retardatory (Hinayāna) feature found throughout the site but wisely abandoned after 465 in order to add cells at these convenient locations in all caves. Starting in 466, simple single cells were cut wherever possible in these previously "wasted" areas, rapidly becoming conventional (and invariable) features for the next couple of years. In 467 cells with pillared vestibules come (also invariably) into fashion, replacing the single cells whenever possible, just as the single cells, where possible, had been cut into the still-earlier plain endwalls of the porches.

The surprising adjustments made to Cave 26's left wing, under the dictates of rapidly changing taste, show this process of transformation. Originally this wing must have been planned to precisely mirror its counterpart on the right, preserving the careful and conventional symmetry of the whole complex's facade arrangement. But as we can see in the earlier-defined right wing, this involved a porch too shallow to allow the addition of porch-end cells, when (in 466) such cells came into fashion; and in any case, the excavators had already cut a cistern at the right end of the right wing's little porch-something they would never have done in 466 or after, for the cistern could easily have been cut in some area where it would not take up this space now needed for the porch end cell.. By the same token, there was no possibility of putting a cell at the other (left) end, an area where the needs of the main hall would have had priority; presumably, in 466, the porch end cells there would already have been conceived (See "Defining Features" chart).

Since the lower left wing of the caitya complex had not been started as early as the right, in 465 or 466 the caitya complex's planners, to accord with new conventions, made the decision to shift the still not fully revealed court of the caitya hall farther to the left (west). This was done in order to make room in the set-back porch of the left wing (Cave 26LW) for a single cell at either end--to accord with new conventions. But in 467/468 the still merely roughed-out right porch cell of Cave 26LW was converted--again to accord with changing conventions--into the a new "cell complex", in which a pillared vestibule fronts the residence cell behind. That the vestibule was previously cut out as a cell is revealed by its uncharacteristically deep dimensions, and by the equally uncharacteristic thinness of the fronting pillars, which were, in fact, cut out of the previously defined front wall of the now-converted cell! Their thinness partly explains why they later broke away; in fact, the present restorations are somewhat thicker than they should be.

Such a fundamental change, from plain porch end, to the addition of single cells, and then to pillared cell complexes, all taking place during the course of excavation, provides insight into the flexibility of the planners as well as the force of the rule of taste at the site. The complex cell L1 in Cave Upper 6, and the cell at porch right (PR) in Cave 27 were started as simple cells before the Recession, and then "upgraded" to complex types after 475. Cell R1 in Cave Upper 6, however, due to lack of time, remained in its pre-Recession state until an intrusive donor took it over in 479, and turned it into his own personal shrine. Although Caves 21 and 23 now have typically late pillared cell complexes in their porches, we can tell from the too-narrow area of scarp between them that single cells, or no cells at all, must have been originally planned here when the two caves were first laid out-the earlier (Cave 21) in c. 465 or c. 466. Hardly a year later, because the deeper cell complexes had just come into fashion, the space between Caves 23 and 24 was very significantly widened.

3. Pillars Sitting Directly of Floor

Although the standard convention which was developed at Ajantā in the Vākāṭaka period was to support the pillars in the cave with a monolithic floor beam, over which visitors often trip today, it would appear that in the earliest excavations the pillars were set directly on the floor. This is notably evident in the multipillared Cave Lower 6, even though at the upper level the pillars "support" the beams which in turn "support" the ceiling. Although the use of supporting floor beams came into use very early, and of course persisted, they are significantly absent in the very early Right Wing of the Cave 26 complex, which can be seen as yet another confirmation of the very early date at which it was undertaken. By contrast, the Left Wing, under excavation a few years later, shows the beams under the porch pillars. The very distinction between them-for they started out (as planned) as parallel excavations-reveals the rapid developmental changes which the whole complex was undergoing.

Cave 25, the upper right wing of the Cave 26 complex was probably started before 26RW (the lower right wing) as work progressed downward when the whole complex was being exposed. It almost certainly was planned with the "primitive" feature-the pillars connecting directly with the floor. Indeed, that is what we see today, even though in 478 much of the long abandoned and unfinished porch was re-cut, possibly because it was being expediently converted into an anomalous shrine area. We can see that the pillars, originally planned (and only roughed out) with heavy octagonal shafts, as expected in the early 460s, were later being recut in a more modern style when all Aṣmaka patronage was interrupted late in 478. The floor between them, though surely still rough when these late revisions started, probably had already been enough defined so that it was finally finished as originally intended-that is, without a floor beam.

It seems to be true, also, from old photographs, that the front pillars of the two porticoes of Cave 7-one of the earliest excavations started-had no "supporting" monolithic beams; the fact that they were reconstructed recently in this way would seem to confirm this, since reconstructions generally respect the traces of old forms. However, more rearward pillars do have beams, probably because they were exposed slightly later, and thus conform to the new convention. It is not surprising for conventions to change very rapidly. For instance the porch pillars toward the right of Cave 7-that area having been done slightly in advance of the left side-have the earliest variety of ribbed capitals, showing five and a half ribs. The "rib convention" has already changed on the left side, where we find four and a half ribs, as on the porch colonnade of Cave 16. Slightly later.... ------------------------------

Another related convention, which makes its appearance somewhat later, but undergoes a similar change, is the absence of beams under certain??early pillared cells ?????????

4. From Shrineless to shrines

Just as in all of the caves at the site which were underway during the first four years or so of activity, the very early four wings were planned without any image shrines, again following Hinayāna precedents. These shrines, which in the earliest Vākāṭaka caves such as these were always added features, were consequently not without problems of placement; we will later discuss the adjustments which had to be made to accommodate such previously unplanned-for images. The reason that the shrine Buddhas did not choose to take up "residence" at the site until 466, by which time at least 90% of the viharas were already underway, is still unexplained, and should be a challenge to Buddhologists. If the viharas at a site as important as Ajantā were originally planned without shrines-that is, as mere dormitories-does this reflect the contemporary situation in structural viharas elsewhere as well? Was this crucial feature in fact one of Ajantā's innovations?

Although shrines were planned for by 466, at least in the major caves, it seems likely that the first separate Buddha images not appear were not finished and dedicated until perhaps the middle of 469. Except for the Buddha image in Cave 11, which had been started (as the fronting element of a stūpa) and then abandoned in 468 before being made into an independent image in 469, the Buddha images within them were not even undertaken-and then in considerable haste-early in that troubled year. However, the reason that we know that shrine images had to be planned (at least in the major caves) much earlier than they were actually carved is from the widening of the intercolumniation between the central pillars of the front aisle, to provide an axial focus for the shrine. However, such widening never appears in the earliest excavations which have pillared halls, namely Caves 2, Lower 6 and Upper 6, showing that, like the various astylar halls, they were originally planned (always before 466) with no thought at all of shrines.

Since many of Ajantā's earliest excavations (8,15,20,25,26RW,26LW,27) are astylar, they provide no evidence in this regardand the small Cave 2 has too few pillars to provide evidence of spacing. Although much of Cave U6's interior was excavated after 475, the front aisle colonnade was apparently defined (and set the pattern) very early. The nearby cells, L1 and R1, which would have been started somewhat later, were almost certainly in the A-mode and thus were not defined after 467. The present windows, for instance, are late (unfinished) conversions from very early vertical types like those of Caves 15 and 20.

The idea of adding a shrine to the viharas was probably one of the many concepts transmitted to Ajantā from Bagh, where the friable nature of the relatively low sandstone scarp apparently disallowed the making of the expected caitya hall. Therefore, although the earliest of the Bagh viharas, Cave 2, (which has no axial focus) had apparently been started as a mere "dormitory", it could be converted to a "caitya hall" by the addition of a shrine at the rear. And being the very first such converted vihara, at least in central India, it is hardly surprising that the "resident" caitya was a stūpa, not a Buddha image, following the long established convention for halls of worship.

Shortly thereafter, at Ajantā, when the idea of adding shrines-probably at first only to the largest caves-developed, it was probably also stūpas, not Buddhas, that were originally intended for them. Of course, in these privileged major caves, such as 4, 16, 17, and the small (but royal!) Cave 20, the original plans (without shrines) had to be adjusted to provide for this important (indeed revolutionary) new feature. In these larger and more important caves the new conception must have developed by no later than 466, before the pillars of the front aisle had been started, or while they were still so rough that their central intercolumniations could be widened, as new conventions required. Only Cave 1, begun much later (c 466) could be excavated as planned, at least in this regard. But it too may well have been planned with a stūpa as its focus. Indeed, it image was not even started until 475, although its shrine had been penetrated earlier.

In the larger and more important viharas, such as Cave 1, 16, 17, and 4, the plans would have been adjusted well before the excavators had reached the shrine area; and when, in the course of excavation, they were starting on the shrine itself, Buddhas instead of stūpas had become the "caitya" of choice. So their past has been obscured: indeed, we would not know, seeing them today, that their Buddha shrines "replace" once-intended stūpa shrines, and that the stūpa shrines themselves would have been cut where residence cells were once intended. cellsa cell or two was once going to be put where the shrine/shrine antechamber is today, and that even when the old-fashioned idea of having nothing but cells was renounced, the plan was to put a stūpa in this area.

King Upendragupta's beautiful Cave 20 seems to be the single exception to the "rule" that the smaller excavations were somewhat slower in developing the shrine concept. This is undoubtedly because of its high priority royal character. We know that its shrine area (if not the shrine proper) was planned very early-probably starting in 467 (or possibly even 466)-because its shrine antechamber projects out into the astylar hall, and the latter contains many very early features, the most striking being its "primitive" doorways, doorsteps, and door fittings. However, its position of privilege did not alleviate its obvious financial difficulties as the Recession progressed, for both its image and its shrine doorway had to be rushed to a most expedient completion when Upendragupta's troubles caused him to have to abandon his connection with the site late in 471. For a full discussion of the development of Cave 20's shrine,

In lesser caves (other than the royal Cave 20) it would appear that it took two or three years for the idea of adding a shrine to have its impact, for even though in the smaller caves it took less time to reach the rear of the cave in the course of excavation, shrines were never begun in them either until at least 468.

The earliest example may be Cave 11, where the shrine was expediently made from a converted cell; significantly, it contains a monolithic stūpa fronted by a Buddha image; but the Buddha image soon become the sole focus, and the stūpa is paid no heed whatsoever. Not only does the Buddha block the view of the stūpa-the halo, in particular, seems to have been made particularly large just for this purpose--but the Buddha alone was worked on and carefully finished in 469. It seems likely that the combined stūpa/Buddha, before the "victory" of the latter, had been conceived following the lead of the influential plan for a similar combination in Buddhabhadra's great hall of worship where, for the first time in any such caitya hall, an image was to front the stūpa.

Of course, for reasons of ritual and tradition, one could hardly "abandon" the stūpas in the caitya halls, as was so abruptly done in Cave 11. However, the developments both in the shrine of Cave 11 and in the purely Buddha oriented shrines of Caves Lower 6, 7, and 15 reveal the irrelevance of the stūpa in the vihara setting. Indeed, the same could be said for the ritual of circumambulation, for in spite of what scholars generally assume, there is not the slightest reason to think that the ritual of circumambulation was ever practiced or even intended in any of the Vākāṭakas viharas.

With the interest in stūpas as a focus gone, the patrons and planners of the viharas were increasingly burdened by the convention of the "centralized" shrine format. This made sense, indeed was necessary, when the shrines were conceived for stūpas, but when the stūpa concept (never truly realized) was rejected in favor of Buddha images the old format imposed unacceptable restraints, particularly as the Buddha groupings expanded. This is painfully evident in the constrictions imposed by the old format on the developed image groups of Caves 1 and 4. The ultimate goal was to spread out the image group against the rear wall, as in the anomalous Cave 20; but this ideal was not "legitimately" achieved until after 475, in the shrines of Caves 2, 21, Upper 6, and 26L.

At the same time, it is logical to believe, that just as in Bagh Cave 2 and probably Bagh Cave 4, the patrons and planners first conceived of their shrines as having stūpas only. Indeed, it may well be the case that when the Vākāṭaka caitya halls Caves 26 and 19 were first planned-very much under the influence of the Hinayāna Caves 10 and 9 respectively-that they too were going to have stūpas alone as their focus, and that the idea of fronting Buddha images only developed, both at Ajantā and Bagh, as excitement mounted at the sites, and as new influences, including the desire for image worship flooded in. As discussed elsewhere (\\\\) the first vihara shrine undertaken, in Cave 11, apparently appropriated the same already-conceived combination. But it was still unfinished at the time of the Recession, at which point the Cave 11 stūpa was totally abandoned, while the Buddha image, along with those in Caves Lower 6, 7, and 15, becomes the lone focus. The insistent switch to Buddha images surely reflects a trend toward image worship throughout the subcontinent in this general period. But in the case of these first shrines, forced by circumstance to be expediently rushed to the point where their Buddha images could be dedicated, time and money may have played some part in the decision to focus on the images alone.

However, stūpas were recommended of course by tradition and convention; and the fact that the Prime Minister put a relief stūpa as a kind of substitute altar in his Ghatotkacha vihara, when he discontinued its excavation in 468, suggests that the preferences were still somewhat in balance, as they so clearly were in the shrine of Cave 11 when the stūpa and the attached Buddha were left unfinished at the end of 468. But sometime in 469, the Buddha image had clearly won out, not only being the sole part of the combination continued in Cave 11, but the sole image made in the shrines of Lower 6, 7, and 15-even though they surely had all been originally conceived for stūpas.

Similarly, as we shall see, the development of the plans in the lower wings of Cave 26 proves that by 468 (after which these Aṣmaka caves had to be abandoned for the next six years) these viharas were being adjusted (26RW) or extended (26LW) to include shrines. Needless to say these caves-originally mere dormitories-had not been planned with shrines in mind, a fact which tested the architect's creativity now that he had to add them. These new introduced shrines may well have been conceived for stūpas in 468, but the places were still in the process of being prepared for them, when the Aṣmakas were ousted. Then, by the time they were finally completed, in 478, they were provided with typically developed Buddha images, just as we would expect.

The discussion above only tells us how and when shrines, and ultimately shrine images, first made their appearance; it does not tell us why this remarkable development took place or why the eighteen mere dormitories already underway before 466, were suddenly turned into places of worship. Judging from the spectrum of different painting styles in particular, Ajantā drew its forms and features, and its ideas, from many different parts of the Indian subcontinent. In this regard the established Buddhist sites in both the northwest and the southeast deserve special attention, both being linked to Ajantā by the routes of trade as well as by their strong Buddhist traditions. For instance, the bhadrāsana pose of the Buddha, perhaps unknown in central India and wider regions beyond until its appearance at Ajantā, appears much earlier in Andhradessa, while the frontal projection framing the Buddha in the Ajantā caityas 19 and 26 may well reflect the "false gables" common in Gandhara. If the idea of putting images (or even stūpas) in viharas was current in the wider world at this time, would we not find shrines in the viharas at Ajantā and Bagh from the start? Is it possible that this revolutionary concept was first developed at the Vākāṭaka caves themselves? If not, then how do we explain it?

5. Excavations Procedures

When Cave 26 was first under excavation, the vault, as expected, was revealed first, since excavation logically proceeded downward. However, at this early date, the excavators failed to reserve matrix for the upper levels of the stūpa, so that it was necessary to attach the umbrellas (now missing) separately. This was hardly very practical, and is in clear contrast to the technically more sophisticated-and later--treatment of the stūpa in Cave 19, where the umbrellas were cut from matrix which had been sensibly reserved. This was also to be the case in the significantly later (469) Caitya Cave 29, where the excavation of the unfinished vault, characteristically, has stopped at the point where the stūpa was to be carved; had work continued, the upper portions of the stūpa could have been monolithic, following the later and more sophisticated mode.

Another early feature seen in Buddhabhadra's Cave 26 complex is the still diffident mode of excavation of the entire complex, which was wholly blocked out before any decoration was started. This is notably evident in various early but very incomplete caves. The very unfinished Cave 26's upper right wing (Cave 25) shows the process very clearly, despite some obscuring late additions and adjustments made in 478. Its main hall was being totally shaped out before any cells, even those toward the front, were started; and while the excavation was being opened up not a bit of imagery or decorative carving was added to it. In fact this was true of the whole great Cave 26 complex which, when its development was forcibly interrupted at the end of 468, had not a single such motif on any of its surfaces. The treatment of the Prime Minister Varāhadeva's Cave 16 appears to have been similar. Although it was among the first caves started (c. 462), and had been deeply penetrated (though hardly finished) a half-decade later, the porch doorway decoration had to be hurriedly (but beautifully) completed when the Recession began in 469, while work on the elaborately carved front aisle ceiling was also going on at that time.

By way of contrast, Buddhabhadra's Cave 24, which had only been barely penetrated prior to 475, the cells were being cut sequentially from front to back, as the interior was gradually opened up (see Cave 24 Plan), a very different process from that seen in the much earlier Cave 25. Although Cave 24's great hall remains largely unfinished, the pillars of its front aisle are in various stages of completion and decoration, and its nearly completed porch is already richly detailed. Even in Upendragupta's aborted Caitya Cave 29, started in 469, we find that certain facade and vault details were being defined even while excavation was in its initial stages-a more developed system than appears to have been used earlier, when Cave 26 was initially underway.

6. Door Fittings

Monolithic pivot projections-which make their first appearance in the Vākāṭaka caves in 468--are found in the very last cell doorways cut in the Cave 26 complex before work broke off with the Recession. The projection in the doorway of the left cell (L1) of Cave 26 RW gives striking proof that this little cave, like the great hall to which it is attached, was well underway when the Aṣmaka patrons were ousted from the site late in 468. As we might expect, cell R1, although mostly broken away is also B-mode; and like L1 was converted to the D-mode when work was renewed on the long-abandoned complex in 475. Early projecting fittings, paired for double-doors, were also used for the Cave 26's main doorway, just as they were for early shrine doorways and cistern-chambers throughout the site.

Most cells in the Cave 26 complex, having been roughed out prior to 468, had primitive "A-mode" doorways; but this is not immediately evident, since all of them (like the B-mode doorways in the Right Wing) were converted to the more efficient D-mode after the renewal of work on the complex. However, the cells in the Cave 26 complex typically have thin walls, and this reveals the early (468 or before) date of their cutting, despite their conversion. Cells with original (non-converted) D-mode doorways conventionally have very thick walls; none are found in the Cave 26 complex except in a few cases in Cave 27, where the cells had not been penetrated before the Recession. Their thickness probably resulted from conventions transmitted from Bagh with the return of many workmen after 475. There the thickness was required by the friable nature of the sandstone, but it also was particular suited for the insertion of the D-mode pivot holes in a deep recess at the rear of the doorway. The problem of converting early doorways is evident in the conversion of the particularly thin-walled B-mode doorway of Cave 26RW's to the D-mode after 475. The added recess is minimal, and necessarily uses the obsolete projection as a strengthening element; the same difficulties are seen in the relatively early and thin-walled doorways of Cave 1 where, apparently for esthetic reasons, the planner tried to preserve as much depth as possible in the doorways' surrounds. We might also note how, considering the relative size of the caves, the B-mode doorway of Cave 26 is so much thinner than the significantly later doorway of Cave 19.

7. Converted Features

The Caitya hall's complex pillars have an underlying octagonal "Hinayāna" format which can still be seen in the eight undecorated pillars in the cave's apsidal end. They were probably left thus to save time and money, often a factor even in such prestigious caves. However, just as in Cave 2's porch, where (as opposed to those in that same cave's interior) octagonal pillars must already have been roughed out when work was first interrupted, the Cave 26 pillars were able to be "modernized" after 475 and made into more complex sixteen-sided forms. However, since they had already been roughed out in an octagonal format it was now impossible to supply them with the large square bases which came into vogue during later years of activity at the site. This convention may have started as early as 469, when the interior pillars of Caitya Cave 19 were probably defined, followed by the porch pillars of Cave 1.

Significantly, although Cave 2's porch pillars show their octagonal origins, the cave's interior pillars are of the conventionally post-Hiatus square-based types. This notably violates the convention of having the porch pillars and those of the interior of essentially the same design. But in Cave 2, this can be explained because the interior pillars must still have been in a very roughed-out state when work was interrupted on the cave in 468. (See conjectural plan as of 468) Thus, when work on Cave 2 was continued after 475, it was still possible to make the interior pillars into the up-to-date square-based types. As if to confirm the fact that Cave 2 is not a wholly early cave, as is generally assumed, we should note that by 468, when work was interrupted, the interior had been penetrated deeply enough, for the doorway filling of cell R2 to have the cutting of a revealingly early B-mode (projecting) door fitting underway . This in turn explains why the porch pillars would have been roughed out in the early octagonal format by that time, even though the work on the interior pillars had not progressed so far. On the other hand, as in other early excavations, the central intercolumniation of the front (and rear) aisle colonnades was not widened, as they were by "squeezing" in Caves 4, 16, and 17; probably there was not sufficient matrix, in 475, to both squeeze these pillars outward and to then convert them to square-based types.

It seems clear that Cave 26's four uncrowded porch pillars were relatively larger in format than those of the interior when they were first roughed out by 468. This helps to explain why, when finally finished after 475, they could be completed in the late square-based mode. However, these square bases are smaller than expected (compare the roughly contemporaneous pillars in the same patron's Cave 24!), almost certainly because this was necessitated when the originally octagonal forms were cut back to a square format. By the same token it was necessary to reduce the width of the supporting "floor" beam beneath them, when that was finally finished; its originally intended width is revealed by the now "meaningless" cutback on the court side of the related pilasters. Thus, although the evidence is typically subtle, we can see the difficulties which the planners encountered, after 475, in bringing the old cave "up-to-date". It is only through such observations that we can understand its true "two-phase" history.

8. Development of the Cave 26 Stūpa

The present bhadrāsana image in Cave 26 was added to the stūpa no earlier than 477. In fact, the Cave 26 stūpa was originally planned to reveal a very different image on its front; it was probably to be a standing Buddha similar to that in Cave 19. This need not seem surprising, since the two sponsoring feudatories had not yet come into conflict at that time; in fact the Aṣmakas could only have been included among the inaugurators of the site with the approval of the local Rishika king. Furthermore, the two images in question were, as far as we know, the very first ever used in Caitya halls in India-a revolutionary development conceived by Ajantā's planners in the mid-sixties, even though the images were actually carved only when the excavations were more fully developed. Therefore, it is easier to believe that a single image form-in this case a standing Buddha-would have been conceived at this inaugural moment than that two significantly different types would have been developed for these essentially contemporaneous and related undertakings.

Weiner, speaking of the possible sources of this fusion of stūpa and Buddha, states: "Thus the stūpa and the Buddha are combined in a single structure, a solution which in part may have been influenced by the use of Buddha images in the northwest to adorn the stūpas, and in part, by the example of the great Amaravati and Nagarjunakonda stūpas to which had been added standing images at the four cardinal points." This is indeed reasonable, and reminds us of the significance of both these sources which, linked by trade route connections, in so many ways lie in the background of Ajantā and contemporary monuments.

Of course, at this early date, shrine Buddha images were never attended by Bodhisattvas at Ajantā, even though attendant figures had long been common in India's sculptural tradition. This important feature makes its appearance in the site's shrines only in about 470, quite possibly first in the now-broken attendants of the Buddha image in Cave 19. Such bodhisattva attendants become conventional thereafter, in the shrines of the same royal patron's Caves 17 and 20, at least until the Bodhisattvas finally yield to attendant standing Buddhas in 479/480. The shift is particularly striking in the ambulatory of Cave 26, where the earlier (479) Buddha panels (those nearer the front, in better light, on better wall surfaces) have attendant Bodhisattvas, while the later (480, and often unfinished) panels have attendant Buddhas. Panel L2, which in terms of its positioning might well be seen as midway between the types, intriguingly shows Avalokitesvara on the proper left, yielding pride of placd to a standing Buddha on the proper right.

We must remember that the new pralambabādāsana image in Cave 26 was a very late concept developed nearly a decade after the stūpa had been roughed out. Therefore it had to be placed in the too-constricted fronting space which was originally (we assume) intended to hold a standing image, and so its now-"required" Bodhisattvas had to be squeezed onto the stūpa drum, where they are not even visible when one stands in front of the stūpa. Similarly, kneeling devotees, also now a "required" post-Hiatus feature, had to be most expediently treated. In 477, when the image was being carved, there was still enough uncut matrix on the floor at the left to carve a group of kneeling devotees, even though the constraints of the available stone resulted in their being smaller than expected and set too far rearward, for in normal course the floor would have been more fully trimmed right up to the stūpa. The fact that so much matrix remained adjacent to the stūpa would confirm the assumption that the stūpa itself, and its fronting projection, was still only roughed out when work was taken up again on the cave after 475.

On the right the floor had surely been more fully exposed when work was interrupted by the ouster of the Aṣmakas in 468. This is probably why the remaining matrix on this side was less than on the left, and therefore insufficient for the carving of the "necessary" monolithic devotees. Instead it is evident that a separately carved group was "plugged into" a slot in the floor; needless to say only the slot remains! But by its position we can tell that the group it held was more appropriately located vis-a-vis the image, being closer and more to the front. This of course destroys the perfect symmetry which is generally so dominant a feature in such compositions, but Ajantā's planners were typically able to adjust convention in cases like this, to allow for a significant correction.

9. Penetrationof Cave 24 at Expense of Cave 26's Right Wing

Because the huge Cave 24, planned by the powerful monk Buddhabhadra in about 466, was going to boldly (and knowingly) penetrate deep into the precincts of Cave 25, work on the latter half-finished cave (and on the similarly affected lower right wing, Cave 26RW) had to be abandoned at this point. All of the porch and hall cells on Cave 24's left side would knowlingly cut deeply into the right portion of the Cave 26 complex, although since work on Cave 24 was interrupted by the Recession, such cutting of the cells did not actually take place until 475 and after. However, the courtyard must have been cut down already by 468, making the creation of a shrine in the rear wall of Cave 26's right wing impossible.

The fact that these "violated" wings of Cave 26 were already significantly underway by 466 alone proves that these two caves (and consequently the Cave 26 complex as a whole) must have been started by 462 or very shortly thereafter.

The adjacent viharas, Caves 21, 23, and 24, were originally planned, in about 466, with single cells at their porch ends, as was conventional at that time. Although their excavation was undertaken sequentially, with Cave 21 started first, the group must have been planned together, since before the expansion of their porch cells they would have fitted perfectly into the expanse of scarp available between Cave 21 and the already established Cave 26 complex. (see reconstructed plan) But when extended porch cell complexes with pillar-fronted vestibules came into fashion in 467, all of these caves now required more room. This could be found only by shifting them all to the left, since Cave 21 (the earliest of the triad) was already significantly underway, thus making any shift to the right impossible. Consequently, as explained above, the upper and lower left wings of the Cave 26 complex had to be "sacrificed", if the splendid later Cave 24 was to be properly completed in the way which evolving taste now demanded.

This "destruction" of the right wings of the great Caitya complex, to accommodate the extended Cave 24 was a decision as drastic as it is surprising. However, given the fact that every good space here was already taken, it was in fact the trading off of embarrassingly obsolete older excavations for a new conception in which the donor could take an understandable pride. And this donor of the impressive and up-to-date Cave 24 could have been no one other than the "owner" of the caitya hall itself, namely the powerful monk Buddhabhadra, whose force of character and access to funds-already evident in his remarkable caitya complex--is revealed by his dramatic decision.

Indeed, since the similarly expanded Caves 21 and 23 were similarly implicated in the shift leftward, they too must have been his own excavations, or under his clear control. That is to say, Buddhabhadra appears to have appropriated (or received) the whole western extremity of the site as an Aṣmaka complex, perhaps from the time the caitya hall was begun in 462, even though Caves 21, 23, and 24 were not undertaken, in that order, until four or five years later, soon being interrupted by the Recession.

If one crosses the river and views the layout of Caves 21, 23, and 24 from afar, one can understand the repositioning of these caves better. From a distance, it is immediately evident that the mass of rock between Caves 21 and 23 is much less wide than that between Caves 23 and 24. This is because single cells at the porch ends were the mode when these major caves were conceived; so the amount of space reserved was very adequate at that time. However, by 467, when the porch of the earliest of the caves, Cave 21, was underway, the new porch end treatment, with a pillared vestibule fronting the residence cell, was suddenly de rigeur throughout the site. Therefore, the excavators of Cave 21 extended (in effect doubled) the porch-end cells, necessarily consuming more of the limited mass of rock separating Cave 21 and 23, leaving an insufficiency for Cave 23.

Unfortunately, the excavators of Cave 23 had already excavated the court area back to the degree where they could expose the originally anticipated facade plane, which typically was established as a flat surface, quite understandably, before the cutting of the porch colonnade was begun. This plane can be seen, still uncut, at the extreme right of Cave 23's facade, where it now exists as a meaningless stretch of smoothed rock some ten feet in width. The explanation for its existence lies in the fact that Cave 23's porch also now had to be expanded to allow for the new cell complexes, in particular at the right end. The whole cave had to be shifted to the left. Fortunately the porch colonnade had still not been begun at this point, so the excavators were able to "displace" the facade plane an equivalent amount to the left, before starting the reveal the pillars of the porch in their new (more leftward) positions. It is evident, of course, that this had to have its effects on Cave 24, which when earlier planned (with the then-standard single cells at its porch-ends) fitted closely but comfortably into the space between Cave 23 and the Cave 26 complex. (see reconstruction of area as of 468).

Cave 21 was well under excavation-at least the porch was essentially revealed-when the Recession caused the temporary abandonment of all of these "Aṣmaka" caves at the site's western extremity at the end of 468. By then the excavators of Cave 21 realized, even though Cave 23 had been displaced to the left, how constricted the space was between the two caves, it now had to contain not only the two residence cells, but the pillared vestibules which fronted them. Therefore, although they cut the right residence cell and its vestibule in the normal way-for there was no problem on this side-they notably reduced the depth of both of these elements at the left, knowing that if the excavators of Cave 23 did the same thing with the abutting complex (at Cave 23's right) it would be possible to fit everything in, since Cave 23 had already been shifted to allow more space. It seemed that the problem was solved.

However, at the end of 468 the Recession occurred, and all work in the western extremity had to be abandoned until 475, by which time the Aṣmakas had won control of the site for themselves. Then work started up again. But since many of the workers, and probably the planners too, had left the site during the Recession and the following Hiatus, it seems apparent that a new crew took up the work at this time; and although they of course knew that Cave 23 was to have the now standard pillared complexes at the porch ends, they obviously were not told that they should do as the Cave 21 crew had earlier done, and to carefully squeeze these elements into a space which was in fact very constricted. They merely went ahead in the normal way, starting the problematic complex at the right in a way which was destined for disaster.

The disaster indeed happened when, in cutting the inner cell, the excavator suddenly broke through into the inner porch-end cell at the left of Cave 21. At this point it was necessarily abandoned, surely to the great embarrassment of those in charge. We can be sure from such evidence that, although the porch of Cave 21 was well underway by 468, the porch of Cave 23-at least the revealing of its porch end complexes-was not even being roughed out until 475. And by that time the new workers quite understandably assumed that they could proceed in a normal fashion, not realizing that time had laid this trap for them.

When the facade plane of Cave 23 was shifted leftward, just before the Recession, it is evident that the new convention for porch-ends had already been established. (This facade shift was hardly earlier than 468, or the porch cells would have been started before work was interrupted.) Of course Cave 24, where the courtyard alone was being excavated at this time, was to have such complexes too. This explains why, in order to contain the new cell complexes the stretch of stone separating Caves 23 and 24 is nearly twice as wide as that between Cave 21 and 23. But at the same time, the result of all this accumulating leftward shifting meant that (and explains why) Cave 24 had to deeply and destructively penetrate into the outmoded right wings of the Cave 26 complex.

It is of course fascinating that such a drastic course of linked action, first involving Cave 21, and then 23, and then 24, resulting inevitably in the destructive penetration of the Cave 26 complex, could have been sanctioned. And since the ultimate impact is on the vast and sacred caitya hall-which alone sustained (and allowed) the "damage", it is evident that such drastic actions could not have been done without the approval of Buddhabhadra, and at his own initiative. One should not discount, of course, his auspicious connection with the powerful Aṣmaka minister, "who was attached to him (the monk) in friendship through many successive births" (Cave 26 inscription). Although by 478, when the cave was inscribed, his powerful friend had died, to be succeeded by his son ("an equally foremost personality") it seems safe to assume that Buddhabhadra's good friend was alive and committed when the great caitya complex was started in 462, over fifteen years before. In any case, whether or not Buddhabhadra himself raised the funds for all of the caves in the site's western extremity, it is evident that nothing in the area would have been done without his approval or his initiative, the expansion and shifting leftward of Caves 21, 23, and 24 being a clear case in point.

The above analyses prove that Cave 21, the porch (or at least the facade) of Cave 23, and the court area of Cave 24 were all underway (to varying degrees) by 468, when the Recession forced the Aṣmaka's sudden departure. The same decision that authorized the expansion and shifting of these three viharas required the abandonment of work on the consequently corrupted right wings of Cave 26, allowing us at the same time to see, at least in general terms, how far work had progressed on the caitya 26 complex as a whole when the decision was made to halt any further excavation on its right side. It is reasonable to assume that the decision to expand the porches of the three major viharas was made in 467, for the space between Caves 21 and 23 was sufficient for only single porch-end cells, which were still in fashion in 466.

Cave 21 was probably started in that same year (466), with the porch adjustments underway in 467 and finished by 468, to "update" the cave's conception. If the porch cells-the left one showing such careful adjustments, of which the later excavators of Cave 23's porch were innocent-had been more or less fully roughed out by the Recession, we can assume that the exposure of Cave 21's main hall was also well underway. As for Cave 23, it had also been started in 466, as the original positioning of the facade makes evident. However, the porch was not penetrated until at least 467, since its new (leftward) positioning shows that more complex porch-end units were being planned for when the colonnade was being roughed out. Admittedly, However, it is quite possible that there was a gap between the time that the original facade plane was defined and the time that it was moved to the left, for this required a major rethinking. It is even possible that it was put off for the time being, while work progressed vigorously on Cave 21. There is no way of knowing if the facade plane was moved to the left, and the penetration of the porch begun, before the Recession or not. All we can be sure about is that the porch-end complex at the right (and probably its untroubled counterpart at the left) was not begun until after the Aṣmaka restoration in 475, by which time the need to carefully "squeeze it in" had been forgotten. At that time, of course, work began again with great vigor throughout all of the Aṣmaka caves in the site's western extremity. Thus all of these caves (except the very late new undertakings such as Cave 22 and 28) reveal a dramatic two-phase pattern, first continuing up until the end of 468 and then again from the beginning of 475 until after Harisena's death, precisely reflecting the developmental pattern of the great Cave 26 complex itself, with which they were all associated.

Finally, it is of interest to note that when the facade plane of Cave 23 was first reveals, most of the matrix in the court had not been fully cut away-the focus of work being to get the plane defined and then the colonnade started. Then when it became necessary to shift Cave 23 to the left, a large portion of the uncut matrix in this now "useless" area was merely left, as a low priority concern. This is the reason that it could be utilized for the steps up to Cave 22, which was probably not even begun until 477.

Chapter #4: Cave 26's Development: Reflecting Ajantā's History

462-468: Ajantā's First Flourishing Under Rishika: Aṣmaka also Vigorously Involved in Ajantā's Patronage

The vagaries of the Aṣmakas' involvement in Ajantā, and in the course of the empire, form the most gripping, and painful, chapter in Ajantā's "illustrated history". In 462, when the site was inaugurated by a powerful and courtly consortium of patrons, the Aṣmakas were clearly on good (or at least workable) terms with the local feudatory of the Ajantā region, Upendragupta of Rishika. (See Map) In fact, the whole area must have been at peace, to provide a political environment in which such an ambitious undertaking as Ajantā could have been begun.

The peaceful connection between Aṣmaka, Rishika, and Anupa, early in Harisena's reign, can probably be ascribed to the fact that they were part of the domains which he inherited from his father, Devasena. Thus it is not surprising that these central areas, all bordering on western Vidarbha, are notably absent from the list of presumably conquered and/or dominated territories listed in his Prime Minister's Cave 16 inscription.

However, the relations between the local king and Aṣmaka had eroded by the end of 468. It was at this point that the development of the Cave 26 complex (indeed of the whole "Aṣmaka" series of caves at the western extremity of the site) came to an abrupt end when the Aṣmakas were precipitously expelled from the region. This assumption is supported by the evidence that Ajantā supplies, which shows that after a few years of strife (already smoldering in 469 but breaking into flames by 472) the tables were turned and the Aṣmakas, now victorious, had in fact taken over control of the site. The emperor Harisena, although powerful, had apparently not tried to prevent this. Even though the Aṣmaka's strength and aggressiveness must have been a matter of concern at the central court,for the present Harisena (or his powerful prime minister, Varahadeva), perhaps intentionally opted for a policy of divisiveness, trusting that such a strife might weaken a potential aggressor. Rulers-the case of Akbar springs to mind-have often improved their own position by tolerating and even encouraging the internecine rivalries of some of their vassals. Actually, if such a policy of "divide and rule" on the part of Harisena was indeed the case, it may not have had the desired effect. Upendragupta, the local king, had, it is fair to say, wasted his birthright on good and pious deeds, and was ill-prepared for the Aṣmaka troops when they came marching over the hill. The brief (c 472-c.474?) conflict, instead of slowly eroding the Aṣmaka power, must have significantly enhanced it, as they so easily took over Upendragupta's rich domains. If we can judge Upendragupta's Cave 17 record, and indeed the evidence of his luxurious donations, Upendragupta was far from destitute; so his great losses must have contributed directly to the war effort which the Aṣmakas may already have been planning against their Vākāṭaka overlord.

This strife between Aṣmaka and Rishika caused Ajantā's Recession, roughly from 469 through 471, after which their conflict erupted into war; and the manner in which the Recession is manifest in different caves at the site is significant. Only the grand Cave 1, Harisena's own donation, continued to develop without apparent problems, at least until the war put a temporary stop to its development. Not surprisingly, the caves of the local king (17-20; 29) continued underway during the Recession also. However, if Harisena's great Cave 1 shows a relatively untroubled development of Cave 1, Upendragupta's involvement was much more subject to haste and expediency; for after all it was Upendragupta, not Harisena, who was being troubled.

As for most of the other caves in the main area, only in that of Varahadeva, Harisena's prime minister (Cave 16) was major work continued vigorously for a somewhat extended period-probably throughout most of 469. However this was as much involved in subtraction as in addition, in the interest of saving time. This remarkable phenomenon, a process of simplification unique in India's excavated art, will be discussed in due course.

But Varahadeva clearly misjudged the situation which, at the beginning, he was planning to transcend. As a consequence, unlike the lesser patrons of Caves Lower 6, 7, 11, and 15, who had rushed their Buddha images to completion by the middle of 469, he could not even get his image started by the end of that difficult year. Prime Minister or not, he finally had to yield to circumstance. And his reward, when he was finally able to continue work on his cave again nearly a decade later, was that he was able to sponsor the most staggering Buddha image at the site-an image that was, in the end, nothing like the simple one that he had earlier planned.

The Prime Minister had a similar disappointment with his related Ghatotkacha vihara; because it was so hopelessly unfinished at the time of the Recession. Therefore, early in 469 he decided to hurriedly carve a projecting stūpa--a kind of expedient shrinelet--out of the mass of still uncut matrix at the right end of the front aisle. The "sanctity" of this unusual shrinelet is evident from the fact that in the Period of Disruption it attracted even more intrusions (including the seated Buddha carved on its face) than did the cave's main image.

Other lesser patrons, also afflicted by Aṣmaka's threat, apparently had two options. In those caves where the shrine image could be rushed to (or at least toward) completion in the few weeks allowed, this was done. It is evident that starting, completing and then dedicating the main Buddha image was the patrons' greatest concern in these troubled times; and such images were completed, even if with a sobering expediency, in Caves Lower 6, 7, 11, and 15, where we can assume they were successfully dedicated, particularly since this was still a time of peace.

The shrine in Cave Lower 6 was the only one essentially completed as planned, as far as its basic design is concerned. However, it must have been originally planned for a stūpa, rather than for the Buddha image carved there, to satisfy changing taste, in 469. Being of obvious high priority, the Buddha image was carefully finished, despite the fact that so many of the wall surfaces at the rear of the cave had never been properly finished when, in 469, they were hurriedly surfaced with a thick layer of plaster and then painted. The shrine walls and ceilings were probably finished last-the scaffolding for the image being in the way in 469-and never got fully completed. Cave Lower 6's fine shrine doorway was probably well underway in 468, but must have been finished in 469, because of its high priority. Therefore Cave Lower 6's Buddha figure, like those of Caves 7, 11, and 15, belongs to the opening weeks or months of the Recession; but it is the only one whose plans were not too significantly aborted by the pressures of time. It might be noted that Cave Lower 6's cells had still not been fitted out at this time (469); their doorways-originally A-mode-were re-cut in the late D-mode after 475.

As for the less fully excavated caves, where the image had not even been reached, they were abandoned completely at this point, although work on them would start up again after the Aṣmakas had finally taken over control of the region by 475.

Significantly, the situation is very different in the Aṣmaka's Cave 26 complex which, despite its importance, now suffered a harsh reversal. By 468 three of its shrine areas (the stūpa in the main hall as well as the shrine areas recently added to its lower left and lower right wings) had already been blocked out, and were ready to receive their images, as soon as the "decoration phase" of the complex would be started. The shrines in the two viharas, neither of which had been planned when the wings were originally conceived, reveal in their positioning the insistent demands of this new requirement--the need for now providing a "residence" for the Buddha in every vihara. The shrine in the Left Wing is "attached" by 468to what (still in 466) was to have been a "Hinayāna" type (both astylar and image-less) residence hall.

The shrine in the Right Wing is even more expedient. It was not started until 467 or more probably 468, since the clearly associated and set-back adjacent cell has a B-mode fitting, first introduced to the site in the latter year. And by this time, the presence of the complex cell at the main cave's right porch end made it impossible to cut the shrine back into this area. Nor could it be cut on the opposite wall where the stone was so corrupt that even one of the expected residence cells could not be excavated. Nor could it be cut, as would be conventional, in the cave's rear wall, because by 468 the floor of the adjacent Cave 24's extensive court had cut away so much matrix that the roof of any such shrine beneath it would have immediately collapsed. This is particularly evident since the presence of a bad flaw high up along the Right Wing's rear wall would have required the shrine ceiling to be raised up so high that a breakthrough would have been inevitable.

Since work on both of Cave 26's upper and lower wings was abandoned by 467 in order to allow Cave 24, with its up-to-date porch complexes, to expand as it would, it is likely that the planners of Cave 24 were not concerned about the problem their court floor would cause in 468. Indeed, it seems likely that work on Cave 26's Right Wing was taken up again at that time mostly in order to provide another location for a Buddha shrine, since at the very most the little cave could now contain only two of the standard six cells originally intended. In any case, because a shrine was now required, the decision was made to cut back the thick matrix along the left wall to form a central projection into which the intended early image--never as such realized--would be carved. As mentioned above, this explains the curious "set-back" of the adjacent cell, while the fact that on the left the wall could not be penetrated (because the area had already been utilized in 466 or 467 for Cave 26's right porch cell) may have suggested utilizing the space for a bed. It is reasonable to believe that the main hall had priority over the problematic right wing, and the fact that the cells of the main hall originally had A-mode doorways, while the right had B-mode doorways, supports this view. However, this bed, like the bhadrāsana "shrine" image finally carved in the left wall's projection in 477 or (more probably) 478, had not yet been carved--and surely not even conceived--in this early phase, because of the uniquely abrupt way that work broke off in the Aṣmaka caves at the end of 468.

Surely, considering the importance of their patron and of the complex itself, and the fact that their locations were already essentially prepared, these three images would have been completed even more quickly than those in Caves Lower 6, 7, 11, and 15, had the Aṣmakas been able to continue work. But they were not; and so these three shrine areas are unique in not being hastily supplied with their intended Buddha images at this stressful time. As if to explicate the political situation in which the Aṣmakas now found themselves, work on these Aṣmaka shrines just suddenly stops--as suddenly as the Aṣmakas hurried (or were hurried) out of the region. By the time that the aggressive Aṣmakas returned and took over the site by 475 and , a few years later, had finally completed these "missing" images, almost a decade had passed; so understandably the new images were very different from those originally intended. Cave 26 and 26RW have notably late (477-478) bhadrāsana images; the padmāsana image in Cave 26LW, like the very similar image in Cave 2, probably was begun somewhat earlier in 477; although the addition of leonine throne legs to its base (heightened because of a flaw), the late character of its shrine doorway, and the fact that its shrine antechamber pillars were expediently finished, suggests that it was still underway in 478.

Drawing historical conclusions from this revealing art historical evidence, it seems clear that the Aṣmakas were abruptly expelled from Ajantā (and of course the surrounding region as well) at this very point. The reason for this must have been their developing aggression toward Upendragupta, the local king, who now, under an all too real threat of conquest, drastically limited the ongoing work at the site, presumably in order to prepare (too little, too late) for the likely neeeds of war. It is fair to assume that, with a (Buddhistic?) policy of "peace in our time", he had used his vaunted "power of the expenditure of wealth" for precisely the wrong purpose. His numerous excavations at Ajantā were obviously only part of his lavish, even Olympian, program to "adorn the earth with stūpas and viharas and to cause the joy of supplicants by conferring gifts (on them)". (Ajantā Cave 17 inscr., verse 220 But this did not save them from the Aṣmakas.

As Bakker tellingly suggests, analyzing the attitudes revealed by the inscriptions, the courtly patrons such as Upendragupta were hardly disposed to deal with such harsh realities as an attack from Sparta. For they had been living in a luxurious "world of courtiers in which the ministers took over the day-to-day worries of government in order to allow the king to become 'free from care' so that he could engage 'himself in the enjoyment of pleasures, acting as he liked'", using the privileges of the gods rather than the responsibilities of the hero as his model. Fascinatingly, Bakker and Dandin, though still not on speaking terms, are in remarkable accord in their analysis of the ills of empire during the troubled reign of Harisena's successor, Sarvasena III. The royal court apparently took on a fin de siecle character as compelling as it was ominous; and in fact the indulgences appear to have multiplied to such a degree that the fin de siecle--the end of this great century--arrived some twenty years before its time was due.

469-471: The Recession: Aṣmaka Caves Abandoned-Many Workmen Leave as Patronge Reduced and Fear of Aṣmaka Invasion Increase

War, as the Recession developed from 469 to 471, was increasingly imminent. This is suggested by the more and more expedient fashion in which the local king's sumptuous caves, upon which he "expended abundant wealth" (Ajantā Cave 17 inscr., 25) were rushed toward completion and were dedicated, but were finally not properly finished at all. When, late in 471, the workers of the local king, with uncharacteristic haste, cut Cave 17's left porch pilaster back in order to make more room for his dedicatory inscription (probably just composed, formatted, and sent down from the capital), the boast that in some earlier time he and his brother "had attacked repeatedly (?) rich countries such as Aṣmaka" (Ajantā Cave 17 inscr., 10 trans. Bakker 1997,25)J was by now but a useless example of braggadoccio. The reference is suspiciously apotropaic--seeking to prevent what was surely going to happen in the weakening present by invoking the lost glory of the past. But it would not protect the little kingdom from the rising Aṣmaka tide.

That the Aṣmakas were about to "come over the hill" at this point is evident in the manner in which all of the local king's major caves were now nervously rushed toward completion, as if the overriding new criterion was just to get them done. The plastering in the later (generally rearward) portions of Upendragupta's ambitious Cave 17 covers shockingly rough wall surfaces and cell doorway openings. This must have pained the highly discriminating Upendragupta, just as he must have been pained, late in 471, to have to rush his beautiful image into dedication, when the surrounding shrine walls were still devoid of decoration. Similarly, the image group in Cave 20, along with its shrine doorway--both forms closely related to those in Cave 17--had to be roughly completed with plaster rather than fully carved. Indeed, the whole of the lavishly conceived little cave, with many of its uncarved details expediently plastered and painted (or half painted), is a sobering study in how to try to finish very quickly what in better times would have been completed with care.

The caitya Cave 19, on the other hand, was more fully finished before Upendragupta's world began to fall apart. It was obviously his "priority" monument. But even here portions of the court decoration, including the expected large yakṣa (or alternatively the yakṣi Harītī) which (according to convention) should have been paired with the beautiful seated nāgarāja across the way, were never finished when time ran out in 471. In the case of the yakṣa (or yakshi), only the honorific roof-motif was blocked out. This confirms the hypothesis that the location had earlier been assigned to the expected subdivinity; the "vestigial" roof-motif has nothing to do with the bhadrāsana image cut beneath it in 479, at which point it was merely used as a location for garland hooks (now missing).

This breakage from pulling out the hooks was quite different from that caused by the more gradual damage caused by changing the garlands. The latter is seen in one instance in the cave, however. This is at the center of the front aisle ceiling where the knocked-away (but recently restored!) plaster around the hole for the now-missing garland hook proves a degree of ritual use. However, this usage is more likely to have taken place during the Period of Disruption (479-480) than during the brief period of frenzy when Upendragupta was hurrying to come the cave-failing even to get the inscription cut into the panel prepared for it on the hall's front wall.

It is also relevant to note that, although doors were hung in the right rear court cell in 470 and in the left court cells in 471, as their fittings and related shelves reveal, they could not have been used more than briefly, if at all, in Upendragupta's day. The pivot holes in the doorways of Cave 19's court cells give evidence of considerable wear. It is possible that the cells (like those of Cave 17) these also continued in use, for practical reasons, even after Upendragupta lost his control over the region. However, it is more likely, considering how the Aṣmakas forbade worship in the cave and also"desecrated" the two front cells (see below), that the evident wear dates from the period of the site's breakdown in 479-480 (the Period of Disruption) and for a few year thereafter.

C-mode cell doorways appear at the site only in 470 and 471, as is evident from their very late (471) positioning in Cave 17 and Cave 1. Excavated niches, as in the left cells, never were cut prior to 471, as is again revealed by their positioning in Cave 17. The right rear cell, apparently dating from 470-for its vestibule as well as the connected cell were finished while the similar area at the left remained unfinished--has pegs for a (unique) wooden counterpart of the left cells' niches, which had not yet come into usage in 479.

Although we must believe-from the presence of intrusions alone-that the main Buddha image in Cave 19 was dedicated, the inscription planned for placement over the doorway inside the cave apparently did not get sent down to the site in time. Perhaps there was by now no time (or literate carver) to inscribe it. Nor could the cave have been in worship more than very briefly, before (with the Aṣmaka takeover of the region) it received the same summary treatment that had earlier been accorded to the "rival" caitya hall Cave 26. This is why this beautiful caitya hall, which Upendragupta had planned as the ritual center of the site, shows not a trace of soot from incense or oil lamps; while even a few of its garland hooks were never inserted along the pillar tops, as would have been done in this insistently organized cave if its royal patron had had a chance to properly finish and use it. In fact, even those which were already in place were probably never used at all, since the plaster around them shows none of the damage typically caused (in many other caves) by the careless changing of such garlands with a "pole-hook". The damage that we see today was undoubtedly caused in more recent times by people pulling the useful iron hooks out, causing some of the plaster (which was applied after the hook was in) to break away.

This breakage from pulling out the hooks was quite different from that caused by the more gradual damage caused by changing the garlands. The latter is seen in one instance in the cave, however. This is at the center of the front aisle ceiling where the knocked-away (but recently restored!) plaster around the hole for the now-missing garland hook proves a degree of ritual use. However, this usage is not likely to have taken place during the brief period of frenzy when Upendragupta was hurrying to complete the cave-failing even to get the inscription cut into the panel prepared for it on the hall's front wall. It is more likely to have taken place during the Period of Disruption (479-80) when many devotees were making intrusive donations there.

We can locate the near-completion of the complex court cells of Cave 19 very precisely in time. They may not even have been planned when the splendid Caitya Hall was originally conceived; and they certainly were not started until very late. This is evident because, like the very late nāgarāja and (merely planned) yakṣa across the court, they are not conventionally and properly aligned (See Plan). They respond instead to the surprising curvature of the facade, whereby the left front face of the originally precisely engineered cave is drastically (and surprisingly) twisted to achieve a solstitial alignment. _ Thus the fact that the vestibuled cell complexes are aligned not with the facade as would have been originally planned, but (like the late nāgarāja relief also) with the torque of the facade, proves that they were conceived and started rather late, probably in about 470.

It is clear that cave 19's facade was not "adjusted" to connect with the (winter) solstitial axis--which it does, at its point of most extreme curvature-- until the cave had already been deeply reamed out with no thought whatsoever of such an orientation. It was then that an order must have come from the king and his advisors in the capital to do what was in effect the impossible, for the original orientation of the rock-cut hall is angled some (\\\\) degrees from the solstitial axis. Had the perpetrators of this impossible demand realized the excavation problems involved, they would probably not have made it; but the planners at the site, following orders from above, did it anyway-or at least tried to do it.

Cave 26's excavation had progressed to nearly (but not quite) the same degree when the new idea imposed itself on that cave too; but the task was somewhat easier, since by chance it was already nearly aligned with the summer solstice--being a mere three degrees off. ((\\\\kim's axis is 3 degrees off; but wrong anyway!!!)) By contrast when, a few years later, the local king, having aborted the development of the Aṣmaka's caitya hall, took perverse pleasure in ordering a summer solstice caitya hall of his own, it was carefully oriented from the start; for by then--in 469--the solstitial orientation could be built into the plans. _

In Cave 19, the harried planners did what they could. The stūpa, still with some matrix around it, could be wrenched to the left with great difficulty, but could by no means achieve a solstitial alignment. If one stands in the court and looks directly through the cave's portico, the stūpa, instead of being on axis, appears very much to the right; and though there are distinct even if necessarily (leftward) minor adjustments to the shape of the dome and the positioning of the standing Buddha, they hardly make a difference in bringing the stūpa into alignment, for its position and shape had surely been fixed too early to allow the deserved changes. _ The planners, forced to wrench the whole hall around as much as they could, even though it had already been roughed out, had better success with the pillared portico, almost surely because it was reached later in the course of cutting than the stūpa, the upper parts of which and the umbrellas would have been revealed when the cave's vault was being reamed out. Particularly since the portico would have been blocked out with much extra matrix, it could be somewhat shifted into alignment. But even so, its adjustment is far from sufficient.

One might have thought that the plane of the still roughed-out facade could merely be trimmed back at the needed angle, with great depth at the left diminishing gradually to a relatively thin wall at the right. However, there were problems. At the right, the stone could not be cut back sufficiently because of the presence of the rather useless interior niche-like inset near the right end of the front aisle; had that not been already been in place, the hard-pressed planners could have defined the new plane of the facade as a simple angle rather than with its present twisted curve. (See Plan)

Since Cave 19 is so influenced by the Hinayāna Cave 9, one might expect that these two niche-like insets were originally intended to be small windows. However, since their cutting obviously started from within the front aisle rather than in the normal way, from the outside, it would appear that the planner decided that windows were unnecessary in this small cave with its great caitya window, and would interrupt the elaborate program of carving on the facade. And so (not quite renouncing the connection with Cave 9) they were "replaced" with these niche-like insets, probably for the same reason that extra vertical beams were attached to pillars L1 and R1: namely, that the architect was "blindly" following the precedent of the Hinayāna Cave 9.

It is clear from the difficulty that the right niche-like inset created relative to the efficient adjustment of the facade plane in that area, that these insets had already been started when the drastic facade revision was authorized. Like Cave 26, the whole of Cave 19 must have been more or less fully roughed out prior to both this adjustment and to the many other adjustments inside the cave. _ This is evident from the fact that the decorative overlay of the cave, inside and out, was carved on the "adjusted" elements and not before, and therefore could not have been started until they had first been very roughly blocked out in the "normal" fashion, and then forced (no matter how unsuccessfully) into the new orientation.

Such observations require us to believe that all of the cave's sculptural overlay (iconic as well as decorative) was concentrated into the last few years of the king's patronage of the cave. If we allow three years for this work, which of course was still not finished by the time that the king's connection with the cave suddenly ended late in 471, then the (attempted) reorientation of the cave would have taken place in about 467 and 468. At this time it would have been already adversely affected by the presence of the niche-like insets, which we can thus assume had been penetrated in 466 or earlier, in the pre-adjustment phase of the excavation, which had been started perhaps as early as 462 but surely by 463. _

Features of Cave 19's cells clearly reveal the date when work on the nearly (but not quite) completed cave came to an end because of the impending takeover of the region by the rival Aṣmakas. C-mode cell doorways appear at the site only in 470 and 471, while excavated shelves, as in Cave 19's left cells, never were cut prior to 471. The right rear cell in Cave 19's court apparently dates from 470--its context is somewhat more finished than that of its counterparts at the left, which would suggest it was completed first. At this point (470) excavated niches had not been developed and, presumably to satisfy the need for such a functional feature, what we can assume was a double-leaved wooden shelf (see above p. \\\\ RED?)was affixed to the better-lit right side wall of the cell by means of four wooden pegs (only the holes remain).

By contrast, in 471, both here and in the later area of Cave 17, niches now came into use, being cut (in Cave 19) into the better-lit walls of the cells at court left. (The left front cell was partly destroyed by the Aṣmakas, but traces of the niche are still evident.) These niches and the related shelf in the right rear cell were similarly and unconventionally placed, on the better lit side wall, to take advantage of the light reflected into the pillared vestibules from the court. _ If we ask why Cave 19's planner did not order a niche for the right rear cell, when such features came into fashion in 471, we might hazard the guess that the two areas were the responsibility of different contractors and that once the job on the right had been paid for, a revision would have involved a further unexpected (and unauthorized) payment. There are many features at the site where changes would have been desirable, but such changes may have involved administrative (and pecuniary) complications. As to why it was not adjusted later, after more funding was authorized, we can assume that time had already run out for the local king and his works, due to the Aṣmaka invasion at just this time.

The pivot holes in the doorways of the intact left and right rear cells show considerable wear. This might at first suggest that Cave 19's cells (like those of Cave 17) continued in use, for practical reasons, even after Upendragupta lost his control over the region. It is more likely, however, considering how the Aṣmakas "desecrated" the two front cells (see below), and otherwise dishonored Upendragupta's beautiful hall, that the evident wear dates from the period of the site's breakdown in 479-480. This is when the fine intrusive panels and related paintings were added to the vestibules and when (conventionally) the cells were plastered, which further reveals that they were in use at that time, and probably for a few years after.

Cave 19's right front cell may already have been converted into a special chamber by 471, by which time a doorway would have been expediently cut in its east wall, making a direct and convenient connection between Upendragupta's pillared cistern chamber, Cave 18-"delightful to the eyes"-and Cave 19 itself. Thus the "sweet, light, clear, cool, and copious water" would have been brought from the cistern (Cave 18) through this converted cell and then through the converted cell doorway into cave's the right pillared vestibule.

When the Aṣmakas brutally cut into Cave 19's cells to make a passage to their own complex, they would have taken advantage of this convenient opening in the cell's east wall, with which they aligned their new "destructive" cuts, one through the west wall of this same converted cell, and one through the corresponding east wall of the front left cell, and finally one through the west wall of the latter; this of course further suggests that the opening in the east wall of the cistern chamber was already there when the Aṣmakas took over and decided to make their passage.

The same east wall has been cut back to reveal a low bench or platform (sometimes found in cistern chambers), while the now-unnecessary cell door was never hung; no latch was ever cut and there is no wear in the pivot hole cut into the upper projection. Since at this time (470-471) C-mode doorways were always cut, it seems likely that the expected lower projection either was eroded away (like that in the left front cell), with any traces obscured by the thick layer of cement covering the floor of the cell. Since it is generally true that door fittings (as well as pole holes and niches) were cut only at the time that the doors were about to be hung, the presence of the (unused) pivot hole would suggest that the conversion of the right front cell to a cistern chamber was made just at the time that the doors were being fitted in the other three cells.

Although it would appear that water was brought from the adjacent cistern in Cave 18 it is conceivable that an old cistern, now filled in, lies under the presently cemented floor of Cave 19's converted cell. One might well expect this particularly important cave to have had such a cistern. The connection with Cave 18 also provided a convenient pathway between Upendragupta's Cave 17 and 19, even though we must believe that another "external" path led from the great vihara outside Cave 19's courtyard wall (which preserves traces of a railing motif) to join with Cave 19's (now totally missing) staircase up from the river. This probably steep ascent led one into the court past the beautiful life-size guardian naga, which can be more easily seen from the ravine below; the other naga, at the left of the now blocked entrance, has fallen away. This "external" path from Cave 17, must have continued somewhat below the "barrier" at the left front of Cave 19's court and then over to Upendragupta's associated Cave 20, for it is unlikely that the king or his cohort (to say nothing of the resident monks) would have been required to go down to the river to get from Cave 19 to Cave 20. And of course the brutally convenient passage that the Aṣmakas made when they had taken over the site by 475 would have been unthinkable when Upendragupta was excavating the caitya hall.

It would be of interest to remove the modern cement at the left of Cave 20's courtyard too, for this area (now much broken away) probably obscures another nearby, and associated, water supply. The small platform at the left end of the porch plinth of Cave 20 would be a characteristic cistern feature, while the evidence of Cave 20's porch cell, started well after the cistern, being surprisingly elevated strongly suggests this, for the porch-end cell (placed higher than that at the right) was not conceived and begun until 465 or 466.

Along with cutting through Cave 19's monks' cells, and apparently forbidding the use of not only the focal Cave 19 but also Upendragupta's Cave 20, the Aṣmakas may well have forbidden the representation of the local king's "signature" image--the standing Buddha with abhaya mudra--which was the focus of Upendragupta's caitya hall, and was reflected in a few images on the Cave 17 and Cave 20 shrine doorways. It is only during the Period of Disruption (479-480), when the Aṣmakas had given up their patronage of the site, because of the developing demands of war, that such images flourish once again. _

472-474: The Hiatus Begins, Aṣmaka Invades Rishika. All Work Stops at Ajantā. Remaining Workmen Leave for Bagh and Other Peaceful Sites

It is clear that there was no patronage activity whatsoever at the site for a short period (roughly 472-474) after both the local king and the emperor (the patron of Cave 1), had had to abandon their ambitious undertakings. This Hiatus--to which not a single chisel-cut or brush-stroke can be ascribed--surely represents the period of conflict when the Aṣmakas were struggling to take over the region; and their ultimate victory is evidenced by the reinauguration of work, in 475, on their long-abandoned (since 468) Cave 26 complex. At the same time, work was taken up again, as workmen rapidly returned to the site, on most other caves at the site as well.

By contrast, all further development of Upendragupta's caves was now clearly disallowed. An understandable and informative exception to this prohibition was the "upgrading" of the doorways of cells L5 and L6 in Cave 17 to the post-475 D-mode; but this was a purely practical consideration, to better secure these cells for their evident storage function, since the cave continued to house monks now, just as before. This may explain the other exception too-our previously discussed cell at the left end of the porch. This was first planned with an A-mode doorway, but unlike many other cells in Cave 17 was not converted to the B-mode sometime between 468 and 471, probably because, with its anomalous "fill" at the upper level, it was a "special case" and so escaped immediate attention. However, later on, after 475, when its disadvantages became obvious, it was expediently fitted out in the D-mode. _

It is important to recognize that there was indeed this distinct break in Ajantā's development immediately after the Recession, for this short Hiatus represents the shift of control from the local king to the Aṣmakas. But we cannot just say it, we have to see it. We can of course observe that work on Upendragupta's caves came to an abrupt end at some point midway in the site's development, and we can also observe that the Aṣmaka caves, which had been long abandoned, had a relatively late "renaissance" at some point after work broke off on the former caves. But how do we know that the later Aṣmaka work did not merely succeed the final stages of work on the excavations of the Rishika king, as if the excavators of Cave 26, coming to the site, merely passed the excavators of Cave 19 on the road as the latter were leaving? That is, how do we know that what I call the "Hiatus"--and blame on war--did indeed involve a distinct and decisive break in the site's development?.

To answer this question, it is useful to turn to the evolution of technology at the site, in particular considering the changing manner in which cell doors were hung.

A Discurus on Doorfittings: A Proof of the "Hiatus"

I have said that there was a total break in the site's development between 472, when workmen had to leave the site, and 475, when many of them returned, particularly from Bagh, bringing certain new forms and features. It is evident that all relatively early caves have windows which were shuttered from the outside and were invariably undecorated, for the shutters would hide the decoration and in fact would slam against the surrounding paintings, which did indeed continue beneath the shuttered areas. In contrast, all later windows were planned to shutter conveniently from the inside, a happy change made possible in part by the much thicker walls which they in turn demanded; they also could now have decorated frames.

The interior closure, the thicker walls, the new frame ornamentation, and even the invariably square format, are all features which must have been brought back from Bagh by returning workmen after the Hiatus caused by the attack of the Aṣmakas on the weak forces of the local king. But how do we really know that the "old" type of window in Cave 1 is clearly separated from the "new" window in Cave 2 by the interruption of the Hiatus? Is it not possible that the Cave 1 window was cut in 475, and that of Cave 2 the following year? How can we decide that there was a distinct temporal break between two different forms?

We can ask the same question when we consider the two up-to-date large standing Buddhas on the Cave 26 facade. These almost certainly "replace" intended standing yakṣas, which were earlier planned to parallel those on Cave 19 but were later shifted to lesser adjacent positions to give the standing Buddhas precedence; but it would be possible to argue (even if incorrectly!) that the change in conception might have occurred overnight rather than over the course of a decade. Or, in the case of a cell such as the one at the right end of the porch in Cave 26's Left Wing, we can, by studying the overall evolution of the site, conclude that the porch end was first conceived with no cell at all, then changed to incorporate a simple cell, and finally transformed into a pillared and vestibuled cell complex; however, even though this can surely be proved, it is not immediately obvious.

But when we realize that a certain type of doorway (C-mode, with monolithic projections for the pivot holes both at the bottom and the top) was invariably cut, or else was underway, in all of the royal caves when work suddenly stopped midway through the site's development (at the "Hiatus"); and when we equally realize that all later caves invariably utilize a far more practical fitting (the D-mode, where the pivot holes are set within a simple recess at the rear); and then when we turn to Cave 1, and realize that the former C-mode door fittings (both those already finished and those left unfinished when the "Hiatus" occured) have been almost invariably converted to the D-mode by an expedient recutting when work started up again after the Hiatus, we have what we can see as clear evidence of a decisive temporal disjunction. And it is from such a break, just as from the appearance of a distinctive stratum in an archaeological dig, that we can make out the chapter divisions in the site's history. _

This temporal break can be clearly seen in the local king's Cave 17, where the projecting C-mode is the latest mode used, and where work broke off before all of them had been completed. Yet even though Cave 17 continued in use for housing monks even after the Aṣmakas had defeated the local king, it was essentially abandoned as far as being properly improved and completed; and this was equally the case with the defeated local king's other caves also. But if Upendragupta's Caves 17, 19, 20, and 29 would seem to say that the beautiful city, after having been destroyed, was never built again, the great Emperor Harisena's Cave 1, when we dig it out of the ashes, tells us that, for it, the disaster was not permanent; it rose again even more splendid than before. This is because (to come back to Ajantā) the Emperor's cave, after the local war which had forced work to stop on all the caves at the site, started up with vigor once again, and within a few years had become the more sumptuous vihara not only at the site, but in the whole of India. In fact, all of the site's excavations-with the striking exception of those of the local king-could now progress again, after a long period (the Recession plus the Hiatus) and did so according to the prestige and the piety and the personal force (including the financial capabilities!) of the various lesser donors. But the break is unique in the emperor's Cave 1; for it has the distinction of being the only cave at the site which continued to flourish uninterruptedly during the Recession, when the site was controlled by Upendragupta, the local king, and then again, after the Hiatus, under the victorious Aṣmakas. (See Time Chart) In other words, since both of the contending rivals honored him (and him alone) by letting his patronage continue, it seems evident that Cave 1 must be the cave of their mutual overlord--the emperor Harisena. Only when the region was in flames, during the Hiatus from 472 through 474, did work on this imperial cave necessarily stop. And for us this break is the more telling because, during this brief period Ajantā's workmen necessarily moved to Bagh and other peaceful areas for employment. Then when they returned, starting in 475, they brought back significant new forms and concepts which in a very evident way impinge upon developments at Ajantā, very noticeably upon the emperor's Cave 1; but they are of course not seen in the now-abandoned caves of the local king.

Such a conclusion is supported by a host of other factors. _ Up until the time that work stopped at Ajantā in 471, due to the Hiatus, two distinctive types of door-fittings had been recently carved at the site, to improve upon the impractical A-mode, where no provision at all was made for affixing the doors. _ The first of these improved forms was the B-mode, where a single monolithic projection holds the upper end of the door pivot, while the next was the C-mode, where there are monolithic pivot holders both above and also below. A study of such fittings throughout the site proves that the B-mode came into use just before the Recession (i.e. in 468 according to my necessarily arbitrary sequence); it is commonly found (sometimes not even fully defined) in the latest portions of the caves upon which work broke off at that point (end of 468). It continued to be used in the "privileged" royal caves-the only ones still underway--early in the Recession (i.e. in 469), but by 470 it had invariably yielded to the C-mode, as noted above. The latter type (often unfinished too, but invariably found in later contexts than the B-mode) was used through 471, when all activity stopped with the Hiatus. _

The fact that these distinctive technological features appear in the cells of Cave 1--the earlier B-mode toward the front of the cave, the C-mode toward the rear--is alone sufficient evidence to date the major portions of Cave 1's excavation to the pre-Hiatus period (i.e. prior to 472) even though the carving of the Buddha image (as opposed to the earlier roughing out of the image block) and the work on most of the cave's decoration dates from 475 onward. _ The evolution of these practical features is so precise that as each better-functioning form is invented, the previous one is invariably abandoned, never to be used again-like the proverbial mousetrap.

After the Hiatus, when work on Cave 1 continued, the totally new and very sensible D-mode fittings came suddenly into invariable use. Knowledge of the new type had surely been brought back to Ajantā by workmen returning from their interim employment in the related caves at Bagh, where the nature of the sandstone disallowed monolithic (B or C mode) projections for use as fittings, and also required the thicker cell walls (and hall pillars) that become conventional at Ajantā too after 475. In this--the new D-mode--the back of the doorway is merely recessed, with the pivot holes neatly cut into the deep set-back. The very proof of the reality of the Hiatus can be seen in the fact that in 475, before finally fitting the swinging single doors in Cave 1's cells, the workmen totally disregarded the old B-mode and C-mode projections, which had been previously (from 468 through 471) cut on nearly all of the doorways, and used the new recessed mode instead. To do this, they often had to cut the old projections partially away--though happily for our investigations, the removal of the vestigial fittings was apparently not in the contract, so they generally remain either intact or somewhat sliced back in order to properly adjust the new fitting. Thus the evidence of the recutting is manifest.

The most fascinating example of the conversion is in Cell R3, where the old B-mode, because of the relatively high placement of the old projection, did not need to be cut back when the recessed D-mode was added. But in this particular case the earlier B-mode shows distinct signs of wear in the pivot hole--something either definitely not the case or no longer visible in other doorways in the cave, many of which never received a pivot hole in the earlier period in any case. Since there is significant wear in the D-mode's pivot hole too, it is clear that a cell door was originally hung sometime after 468 or 469 (when the B-mode fitting was cut) and then refitted in or very shortly after 475, when the still rather diffidently rendered new D-mode was cut to replace it. Thus the B-mode doorway was used for a maxiumum of about eight years, during which time the pivot hole shows significant wear.

Of course we do not know whether the occupant(s) moved in right away, how often the door was opened and closed, etc.; but this still gives us some evidence about usage; and has implications regarding how long monks may have continued residing at the site after Ajantā's collapse. It also counters the possible argument that such wear-because made by wooden pivots-suggests that such doors were used for decades rather than years. Few of the pivot holes in the latest D-mode doorways fitted out at the site--those dating to 477--show much more wear than this; and this would suggest that the monks who remained at the site as their world fell apart (where else were they to go?) probably did not live in them for very long. _

Incidentally, one can note that the wear almost always is far greater to the front right side of the upper pivot hole, due to the pull of what was obviously a heavy door-probably made of solid (or cleated) wooden planks with the pivot-pins cut from the same board as an integral part of the door. It is interesting too to see how, as the lower pivot gradually ground away, the lower one would change position, making two (or more) smooth concentric whorls in the upper pivot hole. It is harder to see, or feel, these in the bottom hole, since the first pattern left by the door's continual turning was generally smoothed down more evenly, as the level of the hole (and therefore of the door) was ground downward.

This gradually "sinking" of the door, as the lower pivot was worn away, was a particularly problem in the B-mode, where the lower door pivot could easily wear away enough so that the door would scrape on the well floor. This problem was faced, and to some degree solved, by the C-mode fitting, where the doorway is lifted up off the cell floor by being set in a lower projection. Typically (and expectedly) these lower holes are also very smooth; although in perhaps fifty percent of the cases this evidence is obscured by the apparent compulsion, in recent times, to fill all such holes with cement. It is also interesting to note how the doors were put in place. In many cases (necessarily in the B-mode,where the lower hole was in the floor next the the door opening) the upper pivot hole is higher than what must have been the length of the upper door pivot, so that the whole door could be lifted up and then dropped into place-into the lower pivot hole. But in many cases (especially in the later doorways), instead of this lift and drop method, a channel was made leading to the lower pivot hole, and the lower pivot could be slid into this until it was in the proper position; then a stone or wooden block was either pounded into the channel, to keep the lower pivot in place, or in many cases, it was firmly locked in place by means of a retaining block set into a slot horizonally oriented to the channel.

A further problem of interest concerns who it was that was living in Cell R3 (and possibly a few other cells in Cave 1) during the 470s, when the cave was far from being fully excavated and was probably still devoid of nearly all carved decoration to say nothing of the paintings, which would not have been started until late 476 or more probably 477. It seems unlikely that a mere workman or workmen would be residing in a cell already fitted out; it may be more likely that some monks were allowed the opportunity, for in the 470s residence facilities were scarce to say the least. _

As I have suggested above, it is clear that much of the work in Cave 1 (and other caves) was done by contract. In Cave 1, it seems clear that the same crew or crews that recessed the doorways, to convert them to the D-mode, also cut the surprisingly uniform niches at the rear of the cells, and almost certainly chiselled the holes on opposing walls for the insertion of clothes-poles or shelf-supports; indeed, such procedures were typical throughout the site. _ This also explains why (fortunately for our observations) the old projections were only cut away to the degree necessary to allow the shaping of the new D-mode recesses, rather than treated according to the predilections of an unconnected group of workers. In the latter case the approaches surely would have been more varied, with some workers finding it easier easy or more desirable to remove such vestigial features. Actually, in many cases (but by no means all) the projection, even though it had to be cut partially back, clearly strengthened the area near the new D-mode pivot hole, so it was wise to retain what was left of it. Cave 1's D-mode recesses are not only the earliest ones to have been made after the Hiatus, but also, because these were probably the first conversions to the D-mode at the site, they are particularly shallow, as if the workers approached this totally new form with diffidence; and for this reason, some care had to be taken in locating the pivot holes. Actually, the thinness of Cave 1's relatively early cell walls surely contributed to this diffidence, which is not to be found in caves where the cells were excavated after 475, for (as at Bagh) these were invariably provided with the thicker walls more appropriate for the recessed D-mode fittings. _

The relevance of this discursus on door-fittings for our understanding of the site's troubled history is that the changes in these functional features--as in various others--clearly show that a distinct time gap must have occurred between the creation of the earlier and the creation of the later types. But I am using it too, intentionally, as a telling example of how revealingly even such a minor feature can be probed, as part of a total analysis of the site's development. Indeed, every door fitting at the site can be located along the site's continuum on what is effectively a year by year basis. But if this merely refers to the time that the fittings were actually cut, we can, with very few exceptions, be equally precise about when each doorway was penetrated, when the pivot holes were added to the fittings and when each door was hung, and when the inner recesses (of certain doorways) were plastered. In many (but not all) caves we can also note that the niches, pole-holes and hook-holes, were cut when the doors were finally fitted out, again suggesting contractual labor.

We must also consider the significant changes in the character of the doorways' latches, peg-holes, drop-slots, and door-stops, which equally change over time. Furthermore, the situation is complicated (and therefore ultimately elucidated) by the great number of revisions (and also some repairs) which the earlier fittings underwent, to make them more functional and up-to-date. Finally, the plastering of the inner recesses of the doorways can give us further information, even though, like the plastering of the cells, this was never done prior to 477 and generally not until 479 or 480; for surprisingly it continued, along with the donation of the spate of intrusive images, during the Period of Disruption. The late character of this plastering is evident, for instance, from the fact that in Cave 1 the hasty plastering of the door recesses sometimes is carelessly spread over the painting of the doorway reveals, which had been painted during the heyday of the site, in 477. _

Cave 1 was started too late to include the earliest doorway type (the A mode) but mention should be made of them, if only to show the primitive characteristics of these inaugural forms. These A mode entrances were of the simplest possible form-the doorway merely opens in the front wall of the cell, with no provision for fittings, as if this was something the planners could not conceive of at the start, but would deal with in due course.

Actually, there was only one cave-the primitive Cave 8-where such simple A-mode doorways, found in part or fully in no less than twelve early Vākāṭaka caves, were actually fitted out. This was probably done before 468, for by then one would expect conversions to the more functional B-mode, as in Cave 11, etc. Perhaps because the planners could not make up their minds about how to fit out Cave 8's plain doorways, there is little consistency in their expedient solutions, involving mere pivot holes in the reveals, or notches cut into the lintel and the door base. Because of much breakage of the rock in the cave, most such fittings are hard to see; but it is clear that even as these more primitive doorways were being so summarily fitted out, Cave 8's excavators were responding to the introduction of the more convenient B-mode, which appears to have been invariably used at the site starting in 468 and continuing through 469. Thus we find that when the excavators were cutting out Cell L2, at the cave's left rear, they introduced the new B-mode fitting type, with its monolithic pivot-holder conventionally placed at the upper left corner of the doorway as viewed from the inside. However, it seems clear that the cell was never put into use, or was used for hardly any time at all, for there is no discernible wear in the pivot hole. This is hardly surprising, for although the patron may have rushed the cave's image (now lost) to dedication, when the Recession started in early 469, the cave was not plastered and painted until at least 477. So it may be that the cave never got into use as a residence in 468, even though the cell may have been fitted out.

As it happened, the impractically simple A-mode doorways, found in the earliest excavations were often converted to a quasi-B mode by means of an expediently applied projection, presumably of wood, held in place (normally at the upper left, as viewed from inside the cell) by two teak pegs, the stumps of which often remain. Except in the relatively untroubled royal caves such conversion from A mode to B mode always occurred in 468, just before the Recession-the year in which the monolithic B mode had just been developed; it is unlikely that the conversion would have been done in 469, at least in caves 7, 11, and 15, because the total effort in 469 in those caves (as in Cave Lower 6, where all conversions were done much later, involving a shift to the D-mode) was in rushing the images to completion. It seems reasonable to conclude that these converted cells were first occupied-in caves 7, 11, 15, and 17 at this time (468), and that only the very early Cave 8 may have had residents, at least in fitted out cells (i.e. cells with the expected doors and accompanying features) before this time, for no others give evidence of having had doors previously. The very expediency of Cave 8's door fittings shows by its very clumsiness why so many of the site's A-mode doorways were converted to the B-mode as soon as that new type had been developed in 468. After 475, of course, when the D-mode had been developed, a better and simpler mode of conversion was used; the old A-mode, and even B and C mode doorways, were turned into quasi-D mode types merely by receiving recesses at the back. Such late conversions are found variously in Cave 1, Lower 6, Upper 6, 7, 15, 16, 17 (PL, L5, L6), 20, and the Cave 26 complex.

----End of Discursus on Cell Dorways----

472-474: The Hiatus: War Between Riṣika and Aṣmaka

The conclusion that there was a drastic Hiatus at the site (472-474), during which work stopped completely after a brief period of exclusively royal activity in 469-471 (the Recession), perfectly accords with the assumption that the Aṣmakas did indeed first menace and then finally take over the Ajantā region from their once friendly (or at least originally tolerant) rival, the local king. Even the compelling evidence that many of Ajantā's workmen had migrated to Bagh (and presumably many other peaceful areas such as eastern Vidarbha, the Konkan (Kanheri caves) , and Kuntala (Dharasiva caves) during these troubled times helps us to understand the political situation. For Bagh was in Anupa, part of Harisena's inherited central domains, and (like these other areas) was not involved in the Aṣmaka/Rishika strife, as its untroubled development. And, as is evident when work starts up again at Ajantā in 475, some of these "interim" excavations influenced Ajantā, just as the migration of workmen from Ajantā had earlier influenced them.

We know from the manner in which the various excavations closely abut one another during the Vākāṭaka (as opposed to the Hinayāna) phase, obviously to avoid any unnecessary waste of space, that there must have been strict administrative controls at the site. This may also explain the surprising fact that during the Hiatus, when the consistent patronage of the caves stopped completely, there were no intrusions added whatsoever, even though we might well expect "uninvited" devotees to have taken advantage of the break to add their own votive images, as they did so flagrantly and so disruptively in the later Period of Disruption, when the site was on the verge of collapse and the former controls no longer functioned. Whether, because of the needs of war, there was a threatening royal prohibition against such merit-making private endeavors, or whether the resident monastic community had taken it on itself to protect the "abandoned" donations, we have a situation which is of considerable interest as a reflection of the concerns of those involved in the site's care and development.

Of course if, as I have argued, such intrusions would be put only in caves which were already dedicated, this would significantly reduce the areas available for any such uninvited offerings which might have been put there during the Hiatus, or even the Recession. This group would comprise those dedicated in early 469 (Lower 6, 7, 11, 15) and Caves 17, 19 and 20. Lower 6 was essentially filled up by 469, while Cave 7's many intrusions are so connected with the work of 477 and 478, that they can hardly be earlier than the years just after that. Both Cave 11 and Cave 15 have intrusions, but in terms of style and iconography they fit comfortably into the 479-480 context, rather than to some earlier date. Various carved Buddhas were added to the court area of Cave 17; most of generic seated images, but the major panel under the inscription is clearly a developed type, dateable to 479 or 480. Similarly, the presence of various carved bhadrāsana Buddhas around the court area of Cave 19 makes it clear that they and all of the images associated with them, belong to the Period of Disruption. The same must be true of the intrusive bhadrāsana Buddha in Cave 20, while the two Buddha panels in the porch also are late in type. Thus it is fair to say that every intrusive image at Ajantā was put there in the Period of Disruption, which started in 479 and ended abruptly shortly thereafter-the end of 480 seems reasonable. There were never any intrusions cut or painted at the site at any earlier date, even in times of trouble or abandonment.

475-477: Ajantā's Renaissance Under Aṣmaka Feudatories

In 475, with Aṣmaka control of site now established, work was vigorously begun again, after nearly a decade, on the Aṣmaka's Cave 26 complex and on other caves at the site's western extremity. These other caves, notably Caves 21, 23, 24, and subsequently 22, 23A, 24A, and 28 were all apparently under Buddhabhadra's patronage. This seems evident from the fact that the powerful monk obviously had to authorize the surprising penetration of the splendid Cave 24 into the upper and lower right wings of his Cave own 26 complex--a drastic decision which totally disrupted their originally intended development. He obviously did this to give priority to the richly conceived and very up-to-date Cave 24. ; but such a drastic action would have been impossible if he had not been the "owner" of both of these caves. If anyone cuts a garage deep into your kitchen, it must be you yourself, not some neighbor, who has decided on such a drastic action.

The fact that, in contrast to the situation in other caves at the site, work proceeded in a relatively normal way in Caves 21 and 23 (as in Cave 26) throughout the whole difficult course of 478, further shows that these caves were clearly under Aṣmaka--probably specifically Buddhabhadra's--control. Work on the grand Caves 24 and 28 did not, however, continue in 478; they were just abandoned, since they were both hopelessly incomplete at the time of Harisena's death. Even the attempt to complete Caves 21, 22, and 23 in 478 was highly expedient, and ultimately unsuccessful, proving that even these donations by the now ruling Aṣmakas were also afflicted by the terminal troubles which were starting to destroy the site. In Cave 21, the walls were plastered, in preparation for painting, before they (and some of the cell doorways) had been properly defined. Furthermore the antechamber brackets that had been started so confidently in 477 now, in 478, were being peremptorily chiselled off, in recognition of the fact that the cave would never be able to be finished in the elegant fashion originally planned. But the fact that here (and similarly in Cave 23) much lower priority work was being done while work on the high priority Buddha image was progressing only slowly, shows that the sudden ending of Aṣmaka patronage was not sufficiently anticipated. Buddhabhadra and his associates obviously recognized that the future was dark, but they did not realize how quickly the darkness would descend. By contrast, in the Vākāṭaka (i.e.non-Aṣmaka) caves, the entire focus at this time was on finishing the Buddha images.

In all of these Aṣmaka caves, which occupied Ajantā's western extremity, work proceeded with particular vigor from 475 on, now that the Aṣmakas, though still feudatories of Harisena, were now in effective control of the site. The overlay of carvings on Cave 26's main facade--all slightly too early to include any bhadrāsana images--effectively masks the typically early warping of the facade plane. Inside, the disposition of the colonnades (not evident on plan) is subtly adjusted in an attempt to align the cave to the rays of the rising sun at the summer solstice. _ Even more obviously, the stūpa--remarkably--has been shifted slightly to the right and at the same time somewhat warped for the same reason.

Equally significant, this must be the only caitya hall in India where the space behind the stūpa is not the same as that to right and left. The reason for such an unprecedented forward adjustment must again be to pull it better into alignment with the solstice--a requirement obviously planned when the excavation was still in its early stages, and such changes were still possible. only a few years after the cave's inauguration. The inner frame of the great caitya window has also been adjusted for the same reason; its width is significantly greater on the left than on the right, thus shifting the central point through which the solstitial axis would pass._

The whole interior is splendidly decorated, and with such subtlety that one is hard-pressed to see the very early original structure beneath the now highly elaborated pillars. Like the comparable porch pillars of Cave 2, they had been blocked out as "old-fashioned" octagonal forms some ten years before. Similarly, the newly introduced bhadrāsana image is carved upon a stūpa front which was not originally planned to contain it. This new type of image, probably begun in 477, like its impressive counterpart in the prime minister's Cave 16, has already sponsored three small "offspring" of the same new bhadrāsana type--two in the frieze above the pillars, and one directly over the main stūpa image. It is reasonable to assume that all of these bhadrāsana Buddhas, along with the four major panels in the ambulatory, were started shortly after the central image, which was cut from the stupa's too narrow fronting projection, which had been roughed out a decade earlier, probably to contain a standing Buddha.

Considering the overriding popularity of this new form, as it rapidly develops over the course of the next few years, it is interesting to note that these central Buddha images in Cave 16 and in the Cave 26 caitya hall are the sole examples of such shrine images started during the main phase of the site's patronage. These two images were completed no earlier than 478, for Varahadeva had to rush to get his image dedicated in the first few months after Harisena's death, and Buddhabhadra clearly made work on his cave a priority during the whole course of that same difficult year. But the impact of these great images is immediate. Already in 479, during the Period of Disruption, the bhadrāsana type has become the image of choice, to be used anywhere and everywhere, and the popularity of the type continues and greatly increases in the caves of the next centuries.

The Aṣmaka takeover of the site by 475 and the brief stability which they brought to the region was responsible for a new flourishing equivalent to that of the seven years or so of Ajantā's inaugural activity; but now the planners and the artists and the patrons had gained confidence from both praise and experience, and approached their tasks with a growing exuberance. Although some of the lesser patrons apparently had to delay or reduce their efforts until sufficient workmen had returned to the region, work on all of the caves at the site now could begin again, except on the disallowed (and still unfinished) caves of the defeated local king.

Even with this proscription, it is interesting to note that Upendragupta's Cave 17, which had been occupied by monks since 471, continued to be used as a much-needed monastic residence, probably for purely practical reasons. Remarkably, when activity at the site was renewed in 475--a mere three years before Harisena's death--the small and very unfinished Cave 11 was the only vihara other than Cave 17 where essentially all of the cells had been fitted out for proper residence, so housing was obviously a serious problem at this time. _ Of course monks (or workmen) may have utilized some of the cells in caves still under excavation, but if so they were in no way properly fitted out._ Considering this great need for residence cells in the expanding site, we might conclude that one of the reasons that Buddhabhadra gave such priority to the huge Cave 24, instead of merely reducing it in size to fit in between Caves 23 and 26, was that it was planned to house forty four monks comfortably (two to a cell), if the space at the rear was fully utilized. By contrast, the old Cave 25 (which Cave 24's expansion "destroyed") was originally planned to hold only nine. Cave 26's troubled Right Wing, just beneath, finally was able to hold only two, largely because Cave 24's aggressive proximity; it had originally been planned for six cells. _

478: Year of Annxious consolidation. Vākāṭaka Patrons Rush to get Images Dedicatea dna then Flee when (of just before) the Aṣmakas Declare Independence from their Vākāṭaka Overlorder. Aṣmaka Caves Continue with More Confidence Until end of Year when Suddenly Abandoned.

On "December 31, 477" the sudden death of the emperor Harisena immediately threw the site into a turmoil, for all too quickly the Aṣmakas, taking advantage of the ineptitude of Harisena's successor, had declared their independence. This sea-change in the political situation is evident from Buddhabhadra's Cave 26 inscription which, written in the year after Harisena's death, tellingly omits mention of a Vākāṭaka overlord, praising instead "the mighty king of Aṣmaka" and his "excellent minister". Although Harisena's sumptuous Cave 1--the finest vihara in India--could easily have been finished within a few weeks, it was now totally abandoned, while in all of the other vigorously developing caves (with the significant exception of those of the Aṣmakas) the normal course of work was abruptly and dramatically affected _

Despite the trauma now visited upon the site by Harisena's death, Ajantā's established patrons were obviously driven to get their shrine Buddhas dedicated, an act from which particular merit must have accrued. The evidence of their hurried efforts makes it clear that although the situation was turbulent, there were still a few weeks or months of "grace"; for in every case where it was reasonably possible to bring their shrine images to some kind of expedient completion, they managed to do so, even if they could not properly decorate (or sometimes even properly prepare) the walls around them. Remarkably, not one of these late images(unlike that in Cave 1, finished the year before) was completed in normal course. _ And if this is true of the main images, it is easy to understand why the caves themselves remained invariably unfinished. _

This period of "grace" can be identified as the few months between the time of Harisena's unexpected death and the time when the unruly Aṣmakas, scorning the Vākāṭaka successor, Sarvasena III, rejected the Vākāṭaka overlordship, a decision reflected in Buddhabhadra's revolutionary inscription, which in the press of events can be dated to about the middle of 478. Like the other major inscriptions at the site, this elaborate composition must have been written in the patron's capital, as its "political" content would suggest. It probably came down to the site with an order that it should be placed on the main porch wall over the right aisle doorway, for its format precisely fitted it for that position. However, the officials who, from afar, chose the location, apparently did not know that there was both a serious horizontal flaw and a serious vertical one, which might eventually make parts of the inscription illegible. So it would appear that a kind of compromise was made; the inscription was shifted to the right enough to minimize any problem from the vertical flaw, but otherwise was placed where (we assume) it was intended, with deleterious results: verse 17, which appears to have provided significant information about the donor himself and his gift, is unreadable. But the evidence it presumably provides about connections between the site and the capital is relevant to our understanding. _

It seems evident that the Aṣmakas had not actually broken their feudatory connection with the Vākāṭakas until at least a few months, in the ominous year of 478, had already passed. If this were not true, we could not explain how Varahadeva, as the Vākāṭaka Prime Minister, with considerable responsibility--he claims in his Cave 16 inscription to have himself "governed the country righteously"--would have been able to rush his own great image to completion in 478, just as so many of the other patrons working in the main (or "Vākāṭaka") part of the site also tried to do. _However, from the worried activity in the "Vākāṭaka" early in 478, we can conclude that a deep fear for the future afflicted the site as soon as Harisena died. Consequently, since the Vākāṭaka patrons thought that time was short, they immediately shifted the focus of their work to the completion of the main Buddhas in their caves. However, no matter how intense their involvement, we can reasonably assume that, except in the Aṣmaka caves themselves, all of the anxious patrons' last minute efforts to complete their main Buddhas ended abruptly with the Aṣmakas' assertion of independence in about the middle of 478.

That Harisena died too precipitously to get his own image dedicated is "proven" not merely by the absence of a dedicatory inscription and by the lack of any evidence of ritual use in the shrine, but also by the fact that this most significant cave is bereft of any later intrusions. There are hundreds of such intrusive images at the site. Most of them date to 479 and 480, although in the abandoned caves of the disenfranchised Vākāṭaka patrons, this takeover process probably began in the latter part of 478. But what is relevant to our present considerations is that these intrusions are notably absent in caves where the main image had not been put into worship, no matter how ideal the location might otherwise seem to be. This alone strongly argues that the Cave 1 image was never dedicated.

By contrast, in the Aṣmaka's Cave 26 complex, and in other relatively complete Aṣmaka caves nearby (Caves 21, 22 and 23), the overall programs of excavation and decoration continued, even if in some cases very expediently. It is evident that the frantic urgency that typifies the rush to complete the main images in the rest of the site was far less a factor in this "Aṣmaka" area, for the Aṣmakas were now controlling both the site and the situation. However, the expediency seen in the continuation of work in 21, 22, and 23 suggests that priorities were rapidly changing from pious to military concerns, and this assumption is supported by the fact that direct Aṣmaka patronage of the site ended--apparently abruptly--after this final year (478) of their own continued involvement. The magnificent Dying Buddha in Cave 26, poignantly recalling the death of Buddhabhadra's patronage in 478, together with the splendid Temptation scene, the three beautiful "Sravasti Miracle" panels in the right aisle along with panel L8 at the back, and the Eight Buddhas (seven plus Maitreya) are the last great programmed images at the site, reflecting the greater care and creativity that Buddhabhadra expended on this particularly sacred hall. _ They also sadly announce the tragedy of Ajantā's ending, proclaiming that a new sculptural force and inventiveness was just coming into its own. If, like Mozart, Ajantā had not died so young (in 478, it was still in its early maturity) one is hard pressed to imagine what delights, as the rewards of devotion, it could have provided. _ As it is, to see the potential of the developing tradition come to a further expression, one must seek out the work of its heirs some two generations later at sites like Elephanta, or in such "transitional" sites as Kanheri and Jogesvari.

479-480: Established Patronage Now Tatally Abandoned due toPreparation for War."Intrusive" Donors Cover put their Helterskelkelter Votive Imagery in or on all Caves in which the Main Image had already been Dedicated.

The dramatic starts and stops which Ajantā underwent in 462, 468, 469, 471, 475, 477 and 478, during the period of its consistent patronage, directly reflect the developing political realities which were to lead to its own downfall, and to the fall of the great Vākāṭaka house. And the ensuing Period of Disruption (from mid-478 to 480) may give us still further insights into the deteriorating situation. The situation in the Aṣmaka caves at the western extremity of the site--and particularly in Buddhabhadra's great Cave 26 Caitya hall complex--is particularly revealing; for it shows how, even in the caves of these newly "independent" rulers of the region, patronage was finally seriously disrupted, even if not as dramatically as elsewhere at the site.

When we consider the community involvement in so many Buddhist sites (including Ajantā in its Hinayāna phase), it is remarkable that the work at Ajantā between 462 and 478 was both elitist and exclusive. For more than a decade and a half, while the site flourished, there was not a single painted or scultured image donated by anyone other than the major donors themselves, whose careful programs are never disturbed by private votive offerings made by anyone not a part of this exclusive and elite group. Remarkably, this is true even during periods such as the Recession and the Hiatus, when the site's highly controlled development first drastically suffered and then, at least for a few years, stopped altogether. But from the time that Aṣmaka rejected its own feudatory status late in 478, and caused the major Vākāṭaka patrons' involvement in the site to be aborted, the gates were finally opened, and a spate of new and lesser donors started to sponsor hundreds of their own intrusive images. These were overwhelmingly carvings or paintings of the Buddha, but included some relief stūpas, and a number of figures of Avalokitesvara as Lord of Travellers--the latter undoubtedly offered for protection upon the once flourishing but now-dangerous routes which those leaving the dying site would have to take. _

These helterskelter donations typically usurped the most desirable areas in the caves, as long as these locations were still available; but, as already mentioned, it is important to realize that they were added only to those caves which (unlike Cave 1) either were already dedicated. Not surprisingly, most of these scattered intrusions were sponsored by the monks still resident at the site who, up to 479, along with the general public, had never been allowed to make a single donation of their own. _ So it is hardly surprising that, threatened by the rapidly darkening times, they now availed themselves of the chance to make merit--surely very cheaply--at this eleventh hour, when the remaining hoards of workmen had no other work to do at the site, and little choice of anyplace else to go.

We need look only at the facade returns of Cave 26 to witness graphically Buddhabhadra's loss of authority over his own cave. This was caused of course by the Aṣmaka's shift of priorities from preparations for "the attainment of...supreme knowledge" (Cave 26 inscription) to preparations for war. Because this was an Aṣmaka cave, and indeed the cave of the great Buddhabhadra, work had gone on here in a relatively normal manner in 478. However, by 479, when intrusive donors took over, neither the returns of the facade nor much of the ambulatory had been decorated according to Buddhabhadra's plan. Therefore now there was what must have been a rush among these new "patrons" to claim the cave's unused spaces and to make some last-minute merit while they could. Understandably those first to take advantage of the situation chose the best spots available for their unwelcome additions, which had essentially no connection with the decorative and iconographic program originally planned.

It is reasonable to assume that the first of these intrusive images are the two large standing Buddhas carved high up on either side, just next to the main facade, which itself had been almost fully decorated shortly after 475, following Buddhabhadra's commendably organized plan. The excellent visibility of these two intrusive images, their impressive size, their location adjacent to the cave front proper, and the fact no previous intrusions hindered the appropriation of such important corresponding spaces, all add up to a weighty argument for their primacy. It seems likely that at the very start of this Period of Disruption (479-489), the financial situation must surely have been better than at the end. This might explain the impressive size of these two images, to say nothing of the unusual fact that they both have incised (as opposed to painted) inscriptions. The inscription under the large left image refers to "the religious donation of the Sakyabhiksu reverend Gunakara" and it would appear from the identical character of the remaining portion of the right inscription, and the similarities of the images and of their placement, that he was responsible for both, and surely proud of his achievement.

One can by and large determine the sequence of the other intrusive images on these facade returns, most of which are carved. Intrusive painted icons, most of which have long since been washed away in the monsoons, due in part to the breakage of the facade overhang, probably filled the few related spaces around or between them. Significantly, there is only one bhadrāsana panel on the cave front, and this "late" iconography, plus the fact that it was in a "low-priority" area, not visible (prior to the collapse of the porch roof) from the courtyard, reveals it as one of the very last images in this area. _ Indeed, it is even later than the two small standing Buddhas to its left, for their presence has displaced it to the right, necessitating the unusual positioning of the farther bodhisattva at a right angle on the main facade surface. And the two standing images themselves must have been among the last panels cut on the facade return, for they too are in a visually undesirable position; that is why they were squeezed as much as possible into an area where they would be marginally visible, leaving the otherwise useful surface still uncut and thus available for the bhadrāsana group.

The fact that the bhadrāsana Buddha found in this very low-priority location is the single example of this newly popular form, which from the very start takes over the ambulatory in 479 and 480, suggests that the intrusions on the cave front can all be assigned to 479, and were being finished at the time when the bhadrāsana type started to be utilized for such intrusions. The many Buddha images along the fronting base of the caitya arch and within the vault also are of the standard earlier types; and since they, like those on the returns, are all fully finished, it is clear that all were completed before this Period of Disruption ended. This too would support the dating of all this abundant intrusive material on the facade to 479.

By contrast, the alternating seated Buddhas and floral panels painted under the vault belong to Buddhabhadra's own program, and probably date to 477 or 478, when the whole cave was being painted. However, looking just below, it may at first be surprising to see the obviously intrusive rows of Eight Buddhas, etc., carved just below at the springing of the vault, since no matter where intrusions were put, they never cut or cover any previous Buddha images. But the explanation is that these lower vault areas, which could not be seen from the courtyard, never were painted with images, they were merely covered with plain colors, if with anything at all. Therefore, having been left "blank" to save time and money, they were perfectly satisfactory areas for intrusions, whose donors would be willing to sacrifice visibility for any relatively auspicious location, particular one on the great caitya hall. A few images slightly higher up were carefully located only where the floral motifs--not the painted Buddhas--had to be cut away. This is not immediately obvious, in part because the mudpacking of a major flaw has fallen away; but the situation can be reconstructed without difficulty. The low "wall" directly beneath the (now-screened) caitya window--again not visible from the courtyard--was also used for a crowded host of intrusions. By contrast, the now-exposed rear wall of the cave's porch which (like the porch ceiling above) was clearly part of Buddhabhadra's original painting program, shows no intrusions. Decorated after the porch doorways, dating to c. 477, were ornamented, it probably showed the expected paired Bodhisattvas flanking the main doorway.

The new "intrusive" donors, who started in 479 to take over the facade areas, were mostly monks, judging from the preponderance of their votive inscriptions throughout the site. Some surely had been previously associated with Buddhabhadra himself, even though his control over the cave had ended after 478. And it was only after all or most of the spaces on the cave front were taken that they turned their attention to filling up the ambulatory. _ This is suggested by the fact that the ambulatory's intrusive iconography, compared with the intrusions on the cave front, is notably later in character from the start. All of the first intrusive panels cut in the right ambulatory--those in the more frontal, better lit, less flawed, and still available areas--show the Buddha in the very recently introduced bhadrāsana pose. _

The fact that small scale bhadrāsana images never appear in intrusive contexts until the very last year of activity at the site--480--is one of Ajantā's many curious but dependable idiosyncracies. Admittedly, such small bhadrāsana images were carved as part of the original program in 477/478 (see page__ above) while a few appear in similar positions in the closely related Aurangabad Cave 3. However, when the Period of Disruption started in 479, although there were many small padmāsana panels donated, all of the bhadrāsana images were relatively large in scale. Perhaps the sculptors found it difficult to carve such images, with their projecting knees, in a small compass. But, knowing the rule of taste at Ajantā, it seems more likely that the smaller format type merely did not become conventional until this time. In any case, it forms a useful criterion for the dating of such intrusions.

The fact that there are no unfinished intrusive images in or around the front of Cave 26 makes it evident that those areas had all been filled up before time began to run out. It is reasonable to date the facade additions all to 479, the first year intrusions appear, at least in the Aṣmaka caves; in the caves of the patrons allied to the now-rejected Vākāṭaka house, the first intrusions may well date from mid-478, since by that time most of those caves in the main area of the site had been rushed to dedication and then abandoned by their threatened courtly patrons. Thus these "Vākāṭaka" caves were surely the first to be taken over for the random votive donations of the "intruders".

The finished character of all of these external images would itself suggest that they should be dated to 479 rather than 480, a conclusion suggested by the fact that there is only a single intrusive bhadrāsana image on the cave front, as noted above. Further confirmation that this lone bhadrāsana image should be dated to 479 rather than to 480, is the fact that it has bodhisattva attendants, rather than the attendant standing Buddhas seen in the late (i.e. 480) intrusions found toward the rear of the right ambulatory, where the light was bad and the rock very faulted.

It is just when these latest intrusions are being carved, that we begin to find a significant number of unfinished images, for time was obviously running out and in fact came to a sudden halt. Almost the last of these must be the barely started standing Buddha at the very front of the right aisle wall. One would at first assume that this frontal and well-lit location would have been appropriated very early; but the fact is that it was not a good position at all, because the image would be obscured (and perhaps damaged) when the right aisle doorway was flung open. The great Dying Buddha in the other aisle is placed much closer to the front end of the wall for the simple reason that the aisle doorway there-conventionally-opens on the same (left) side, and so would not touch it.

By the end of 480, this "intrusive" phase of work itself comes to a sudden halt. In the ambulatory of Cave 26 alone, even excluding images, which may never have gotten painted, there are no less than twelve unfinished (carved) Buddha figures. These relatively simple images which, like the standing Buddha just discussed, could surely have been finished in a matter of days, are to be found in the least desirable areas in terms of location (often being at a higher level, as well as at the rear), lighting, and rock quality, confirming their very late date; and all show notably late iconographic features--features which change with a remarkable rapidity in the late phase of work at the site, and help to justify our dating not merely on a year to year basis, but in fact require us to think in terms of "early" and "late" in years such as 479 and 480, which are so crowded with change. But we cannot think past 480, for sometime during the late evening or night of December 31^st, it is clear that the site's remaining carvers and painters decided not to return. The fact that all of these abandoned images, all to be found in the latest--i.e. least desirable-positions, must have been underway at the very same time, clearly shows how rapidly and urgently and anxiously the new patrons, eager to make merit while time allowed, had been filling up the cave's ambulatory in the last few weeks of that fateful year.

Thus the evidence of the insistent, and surely anxious, intrusive work on and in Cave 26--which could be confirmed by reference to such work on many other caves as well--proves that work at the site stopped suddenly at the end of 480. The presence of so many contemporaneous unfinished images in Cave 26 compels us to recognize this break as decisive; and this conclusion is supported by a review of the intrusive Buddhas in other caves, notably Cave Upper 6, Cave 22, Cave 21, and Cave 4 (shrine), as well as in the Ghatotkacha vihara, where the latest images were similarly left unfinished. By contrast, there are no unfinished images among the abundance of intrusive forms in Caves 2, 7, 11 (porch), 16, and 19, where the available space was filled up before the crisis occurred.

The reason that these intrusive donations abruptly stopped must be that the huge insurrection against the Vākāṭaka house, which the Aṣmakas were fomenting, was now moving from the stage of political intrigue to that of serious preparation for military action. Perhaps there was even a sudden flaring up of a local conflict in the Ajantā region, initiated by the banished Vākāṭaka patrons who, through their benefactions, had developed such a connection with the region. It is, after all, most unlikely that the site could have continued to flourish, even in this rank and undisciplined way, in the shadow of the great troubles which were already beginning. And the site clearly tells us that it indeed did not flourish; it once again explains its own history.

Thus, just as it is reasonable to see 478 as the year in which the Aṣmakas declared their independence from their overlords, it is reasonable to see the "Period of Disruption" (479-480)--when these "eleventh hour" donors were making merit while they could--as the time when the Aṣmakas were developing their plans to take over the empire itself. This poisonous process must have already begun as soon as the vulnerable Sarvasena III acceded to the throne in 478, and may have started in local skirmishes even before the total impact of the eager insurrectionists came down with full force upon the anxious empire in the early 480s.

In fact, the remarkably precipitous ending of work at the site-the fact that on the morning of January 1, 481 not a single artist showed up to continue work on the many images already underway-may be the first hint of the more global conflict that was to come. Since it is clear that the Aṣmakas were out to make trouble anyway, the simplest and most reasonable way for them to start was to first consolidate their hold over Risika (the Ajantā region) itself. They had already declared themselves its overlords in Buddhabhadra's Cave 26 record of mid 478, but this is not to say that the so rudely disenfranchised Vākāṭakas had agreed to their own debasement. So it is reasonable to assume that it was in this battle, with its flaring start on early on New Year's Day 481, that the confrontation was resolved and Ajantā's fate at last fully and sadly sealed.

After 480: All Artistic Patronge has Ended, Some Monks stay on for a Few Years, Until Local support Becomes Infficient

Ajantā's patronage includes not only the carefully controlled productions, which ended so urgently shortly after Harisena's death, but also the helter-skelter intrusions, which so vigorously afflicted the site in the subsequent Period of Disruption. However, after 480, even these late scattered offerings had no progeny. One could assume that, bereft of support, the artists now moved away, probably returning to their villages, or possibly being conscripted to satisfy the mounting demands of the military. The manner in which the vigorously ongoing intrusive work ended in mid-course late in 480, would seem to reflect the trauma which brought this helter-skelter patronage to a decisive end-sounding a signal for the artists to depart.

In the stormy climate of the times, it is reasonable to assume that most, or at least some, of the monks stayed on at the site after 480, for there was really no other place for them to go. It is clear that for at least a short while, a number of them continued to reside in the caves, presumably depending upon the villagers from the surrounding areas for their reduced support. We can tell by the evidence of significant wear in the pivot holes of doorways with very late (c. 477) fittings, that some of the cells at the site must have been used as residences for a decade or so, well into the 480s. But after that, the chanting surely stopped.

Although the caves were essentially abandoned by the end of the 480s, we should note that occasionally, over the course of later centuries, they were used as residence by a few sadhus and the like. In Cave 11 there are a few symbols such as tridents scrawled on the walls, and in both Cave 11 and Cave 7, the ceilings of the porches are badly smoked up. This was surely the result of cooking rather than of worship. for while the site was flourishing, this would never have happened. Also in both Cave 11 and in the Left Wing of Cave 26, a few cells were plastered by (or for) such occupants, using a thin surfacing which (despite opinions to the contrary) appears to be partly composed of cow-dung, which was probably never used in the more generous plastering of the cells during its Vākāṭaka phase. The proof that this plastering has nothing to do with the original work in these caves is to be seen in the fact that this thin plastering in Cell R2A of Cave 26LW is slathered right up into one of the long-abandoned pivot holes of the doorway, while in the cell at the left end of the porch of Cave 11 it once covered (and still partially covers) the broken-off teak wood "pins" which affixed the wooden projection which had been added in 468 to "convert" the primitive A-mode doorway to the more up-to-date projecting B-mode._

It seems likely that the Aṣmakas, who destroyed the Vākāṭakas and Ajantā too, must have been hoist by their own petard. They may have been so weakened by their own insurrection that they were unable to continue their patronage even after the war was done. The fact that even their late excavations at Aurangabad, developed (but never finished) during the heyday of their patronage between 475 and 478, were located closer to their own center of power and yet were never taken up again, would seem to confirm such an assumption.

Just possibly the Aṣmaka's power was ceded to another ruler, after the great insurrection, and the new authority was Hindu in its orientation rather than Buddhist. One suggestion that indeed the Aṣmakas may have lost control of the region is to be found in an elusive and now nearly illegible inscription carved on the wide wall between Cave 26 and its left wing. _ Scholars generally agree that this long and highly visible record belongs to one Nannaraja, a Rastrakuta king ruling (possibly) in the late sixth century. But it is in no sense a donative inscription, nor does it have any discernible connection with the site other than the fact that it was cut on a convenient wall between Cave 26 and Cave 26LW. It must be an advertisement of power by a king who was probably ruling over the Ajantā area at this time.

When we ask what Nannaraja was doing here in the old Aṣmaka domains, perhaps a slight light emerges from a reference to an earlier Rastrakuta king, Mananka, who "frightened the countries of Vidarbha and Aṣmaka by his policies". _ Admittedly Mananka is generally considered a fourth century king but (even if this is true) it could suggest a history of conflict, as well as contiguity. What is far more certain than this is that there is no way that this non-donative inscription can be used to "testify against Spink's assumption...(that) all work at the site should have come to a halt almost immediately after Harisena's death." (Bakker 1997, 41) Rather, the record is more likely a self-aggrandizing record of a visit, a bit like that of the tiger hunter who came upon the site by chance nearly two centuries ago, and scratched across the chest of a painted Buddha on pillar R13 in Cave 10: "John Smith, 28th Cavalry, 28th April, 1819". (Dhavalikar 1968, 147-153).