Fortifications in Ancient India DURGASHASTRA

 

 Fortifications in Ancient India

 A. Cunningham and his associates, between 1862 and 1885, have identified several locations of buried fortified sites. British archaeologists at the beginning of the 20th century, followed by Indians from the twenties onwards, were mainly concerned with the Copper and Bronze Age. This is the reason why so many Harappan sites in Pakistan and in Western India have been systematically excavated and also why early historic sites did not attract the attention of scholars with the same eagerness.

2It is true that the task is not an easy one. Vestiges of ramparts are buried under enormous amounts of debris corresponding to different periods of construction and, moreover, enclosures have undergone modifications through the centuries, which make the restoration of the original structures very difficult.

• 4 G.R. Sharma, Excavations at Kausambi, 1957-59. M.S. Mate, in ‘Early Historic Fortifications in the (...)

3The excavations have often been partial, limited, and the outcome has been of varying quality. The recent reports, particularly those which concern the early historic period are generally mere accounts, with a few plans, forcing us to use the old drawings of fortified towns published by Cunningham and his associates at the end of the 19th century. Even the systematic and comprehensive publications pay more attention to the stratigraphy, the classification of layers and strata than to the specific characteristics of the defence sites (archaeologists are not necessarily experts in military technology and architecture!). The most typical example is the detailed report by G.R. Sharma on Kausambi, one of the best excavated sites in India, in which there is no general plan of the site, no drawing of the gate described, no mention of the curtain walls and towers, or of the location and spacing of the salients; the section in the rampart shows only the outside revetment and the late additions.4 It is therefore difficult to give a consistent, systematic presentation of the ancient military works.

4However, in this section based on the study of 30 sites (8 from the Harappan period, 22 from Early Historical India), an attempt is made to analyse the main features of some ancient Indian fortifications.

5Two phases are to be considered: the first one, from the third to the first millennium corresponding to the brilliant urban development of the Harappan civilisation; the second, from the middle of the first millennium to the historical period, representing what is called “the second urban revolution” in India.

A. Harappan Fortifications

6The body of material at present available on the fortified cities of the Harappan period, which are all located to the north-east of the subcontinent, is entirely based on the results of archaeological excavations.

7The plans of the cities show that their layout usually consisted of two distinct parts: on the one hand, the actual urban zone; on the other hand, the citadel, generally located at a higher level, whose defensive works varied depending on the sites.

The Great Sites

8a. In the Indus valley, both settlements were separated.

• 5 EIA, vol. II, pp. 194-196; B.K. Thapar, ‘Synthesis of the Multiple Data as obtained from Kalibangan (...)

9- At Kalibangan, the city is divided in two parts, both quadrangular (of 8.64 ha and 2.88 ha respectively), surrounded by mud brick enclosures. The lower city is enclosed by a wall strengthened by rectangular towers at the comers. The citadel, consists of two parts reinforced, at regular intervals, by rectangular towers.5 (see fig. 2, a).

10- At Harappa and Mohenjo Daro, the citadel only is surrounded by a wall. This is an enormous bund built up of mud and debris with a nucleus of mud brick.

• 6 R.E.M. Wheeler, ‘Harappa 1946: The Defences and Cemeteries’, Ancient India, No. 3, January 1947, pp (...)

11In the first site, having the shape of a parallelogram of 400 x 200 m roughly (i.e. slightly less than 8 ha), the enclosure, made of a bund on which stood the main wall, 12 m thick at its base and 10 m high, was externally revetted with a baked brick facing and reinforced by rectangular towers higher than the curtain walls, representing an elaborate system of enfilade6 (see fig. 2, b).

Zoom in Original (jpeg, 496k)

Fig. 1. Map of the fortified sites.

• 7 J.M. Casal, op.cit., pp. 100-103.

12In the second one is also found an immense bund revetted with mud brick, 12 m wide and 4 m high. At its south-east corner, a gate flanked by two towers of baked brick has been exposed.7

13b. In the Harappan settlements of Gujarat, unlike those of the Indus valley, city and citadel are both enclosed by the same wall.

• 8 EIA, vol. II, pp. 257-260; S.R. Rao, Lothal 1955-62, vol. I, pp. 135-136 and pl. XXXVIII.

14- At Lothal, a massive mud wall, reinforced by baked brick walls on the northern side encompasses a 12 ha quadrilateral area in which three high platforms represent the acropolis, but there is no salient8 (see fig. 2, c).

• 9 EIA, vol. II, pp. 424-425; J.P. Joshi, ‘Excavations at Surkotada’, in D.P. Agrawal & A. Gosh, op.ci (...)

15- At Surkotada, the township was a single unit, divided into two equal parts: the residential area and the citadel; the enclosure, rectangular in shape and covering 0.86 ha, is made not only of mud and mud brick but also of rubble. It is a massive wall, with a basal width of 7 m and a height of 4.5 m, flanked by quadrangular towers at the comer and pierced by intricate entrances9 (see fig. 2, d).

• 10 EIA, vol. II, pp. 119-121 IAR, 1963-64, pp. 10-12 and pl. VII.

16- Slightly larger (1.30 ha), the settlement of Desalpur is enclosed by a wall having a basal width of 4 m and rising to an extent height of 2.5 m, with a veneer of sandstone blocks on its outer and inner sides, and defended, at the comers, by square towers.10

• 11 J.W.A. Fairservis, The Roots of Ancient India, pp. 179-182.
• 12 EIA, vol. II, pp. 45-46; R.S. Bisht, ‘Excavations at Banawali 1974-77’, in G.L. Possehl, Harappan C (...)

17- The same type of defensive work is found at some other sites, such as Kot Diji11 or Banavali.12 At the latter place, where an extensive squarish tower has been exposed, the wall is from 5.40 m to 7.50 m wide.

Gateways

• 13 The following paragraph is principally based on the observations made by Arun Kesarwani, ‘Harappan (...)

18The enclosure walls were pierced by elaborately designed gateways.13

19-At Kalibangan (fig. 2, a & 3, a), the citadel area is divided into two parts by a partition wall. In the northern half three entrances flanked by a tower were exposed; in the southern portion, there were two only: the southern one, located between two towers, is a simple 2.60 m broad passage without guardrooms (fig. 3, a, 1), the other one, located on the partition wall flanked by two towers is approached from the north by a broad stairway (fig. 3, a, 3).

20In the lower city, two gateways have been found; on the western side it is a passage, 5 m broad, provided with a guardroom (fig. 3, a, 2), to the north-west, the entrance leads to two streets (fig. 3, a, 4).

21- At Harappa (fig. 2, b & 3, b), the northern gateway looks like a gap in the fortification: it was just a simple entrance without guardroom (fig. 2, b); the gates exposed to the west (fig. 3, b, 1, 2, 3) are all situated in a curved part of the wall: the first one (b, 1) is a narrow passageway which leads to a stair, the second is provided with a guardroom where door sockets are seen (h, 2) and the third, the most complete (b, 3), forms a long entrance turning sharply towards the south.

22- At Surkotada, one of the most important Harappan settlements, two gateways, made of mud and mud brick, both located on the southern face of the enclosure, are particularly elaborate and significant (figs. 2, d & 3, d): the first one, to the west, leading to the citadel (jig. 3, d, 1), forms a complex with a 10 x 23 m projection, provided with a stairway and two guardrooms on either side; the second (fig. 3, d, 2), to the east, used for communication with the residential area, consists of a wide passage with two rectangular structures believed to be guardrooms on both sides.

23- At Mohenjo-Daro and Lothal, no well-built gateways have so far been found, except in the latter one, where there is a simple oblique opening in the southern wall.

Assessment of the Defence System

24a. It is difficult to determine the actual function of these edifices for various reasons. It should be noted, first, that, in several sites, because of limited excavations, many structures have not been subject to adequate probing and that, therefore, very little is known about the exact plan of the constructions, the precise shape and size of the works. Moreover, it should not be forgotten that, as the walls exposed have been rebuilt time and again, we often ignore the exact nature of the enclosures during a particular period.

• 14 Ibid.

25Most of the archaeologists do not consider these constructions to be defensive works, but regard them either as protective embankments against floods or as structures erected for social functions.14

• 15 K.M. Srivastava (‘The Myth of Aryan Invasion of Harappan Towns’, in op.cit., p. 439) thinks that th (...)

26They reckon that, at Surkotada, the gates were neither strong enough nor elaborate enough to serve the needs of armed defence, that those found at Kalibangan may have been used as ceremonial pathways or for commercial traffic or were ordinary entrances for the free movement of people and goods and that the long entrances to Harappa fort were used for official formalities, religious or secular and served as processional ways. Thus, they want us to believe that these enclosures had mainly a social function.15

27Regarding Lothal and Mohenjo-Daro, they all agree that the walls were built for the protection of the town from floods.

28b. Now, it is obvious that, at the last two sites, the massive walls of the enclosures were mainly protective bunds, safeguards from the regular inundation of the area. It is also possible that the location of the different gates bore some relation to the daily lives of the people (control of traffic, human and animal movement, processions). But, these factors do not explain fully why such massive ramparts were erected.

Zoom in Original (jpeg, 408k)

Fig. 2. Harappan Fortifications (A), Enclosures: a, after B.K. Thapar, in D. Agrawal & A. Gosh, Radiocarbon and Indian Archaeology, p. 268; b, after R.E.M. Wheeler, in Ancient India, No. 3, pl. XV; c, after S.R. Rao, Lothal 1955-62, vol. I, pl. XXXVII; d, after J. P. Joshi, in D. Agrawal & A. Gosh, op. cit., p. 178. Note the plans quadrilateral in shape and the quadrangular towers

Zoom in Original (jpeg, 208k)

Fig. 3. Harappan Fortifications (B), Gates: a, b, d (after A. Kesarwani, in B.B. Lai & S.P. Gupta, Frontiers of the Indus civilisation, figs. 7.4, 7.5). Note the flights of steps through the gates.

29If the interpretations given above are correct, then, why were the walls reinforced by a regular series of massive towers or salients with elaborately designed gateways? (at Banavali, the enormous walls are provided with extensive square towers, some of them near the openings). These projecting portions of the rampart, particularly at the comer, are not merely buttresses; they were built to ensure the permanent flanking of the curtain walls.

30From this observation it can be inferred that the enclosures of the cities were erected for the purpose of strengthening the human settlements against a possible intrusion by wild animals, marauders and particularly an offensive military operation. In places liable to the overflow of a great body of water, these defensive structures were reinforced and also served as protective embankments.

31Thus, in Harappan sites, are found the earliest features of a defence system in the Indian subcontinent, i.e. massive enclosures, built of mud brick or rubble and usually revetted with baked brick, quadrilateral in form, flanked by quadrangular towers, a design which was adopted and perfected by the builders of strong places during the early historic period.

B. Early Historic Fortifications

• 16 A. Gosh, The City in Early Historical India, pp. 2, 11, 51, 61, 62-67, 80-81, 90.
• 17 M.S. Mate, op.cit., pp. 58. Regarding the rising of an urban civilisation and the growth of fortifi (...)

32After this period there was a chronological and cultural void followed by an intense building activity initiated in the middle of the first millennium, sometimes regarded as the “second urban revolution”.16 For fortifications it is a significant period, particularly around 600 B.C. and also in the years 200-100 B.C., after the decomposition of the Mauryan empire and the advent of regional principalities. At that time could be seen «the rebirth of military architecture» in the Indian subcontinent17: the fortified cities which have been excavated are not only citadels or walls enclosing limited residential areas, as in the Harappan period, but large human complexes taking advantage of the defensive potentialities of natural sites.

Sources

33a. To study them we depend essentially on the investigations carried out by archaeologists in the different sites (though, in many cases, detailed excavation reports are not available). For our analysis we have selected 22 significant sites. A list and short description of each one of them is given, along with bibliographical references, in the appendix (the numbers within brackets which follow the names of strongholds in our demonstration - example Kausambi (No. 8) - refer to the numbers of this list).

34b. To these sources may be added literary and iconographie data.

• 18 A. Ray, Villages, Towns and Secular Buildings in Ancient India, c. 150 B.C - 250 A.D., pp. 46-63, n (...)

35Cities and their enclosures are abundantly described in ancient literature.18

• 19 See R.S. Betai, A Reconstruction of the original interpretations of the Manusmṛti.
• 20 Devī Purāṇa, ch. 72, 11.42-44; Agni Purāṇa, ch. 106, 11.8-10; Matsya Purāṇa, ch. 217, 11.24-28.
• 21 Arthaśāstra, II, 3; see The Kautiliya Arthaśāstra, ed. R.P. Kangle.
• 22 Mānasollāsa, ch. II, si. 544-584, cd. G.K. Shrigondekar.
• 23 Mānasāra, an Encyclopaedia of Hindu Architecture, ed. P.K. Acharya, vol. VII, pp. 226-229.
• 24 Mayamata, X, 36-61, ed. B. Dagens, Mayamata, 1st Part, pp. 152, 160-170. This information has been (...)

36The technical treaties on architecture and urbanism generally devote at least a chapter to the erection of fortifications. Besides the Manusmṛti19 and the Parāṇa,20 the old treatises which give the most details are the Arthaśātra21 and works of this category (Mānasollāsa22 etc.), as well as the treatises on architecture or Śilpaśātra, such as the Mānasāra23 or the Mayamata24

37The descriptions given in these books are to a great extent conventional; they are found with variants in most of the books dealing with urban centres where, for example, the specifications suggested by the Arthaśāstra are reproduced. Thus, the descriptions given by them sound rather theoretical.

• 25 In the bas-reliefs of Amaravati, Bodhgaya, Mathura, Nagarjunakonda, Sanchi, as well as those found (...)

38Iconographic representations, almost entirely Buddhist (2nd c. B.C. - 4th c. A.D.),25 shed some light on the stereotyped descriptions of the ancient texts, because they show, in a simplified but probably faithful manner, urban constructions and particularly the superstructures of the defensive works, not found in the vestiges exposed by the archaeological excavations.

39The history of the different sites considered here will be given in the appendix. Let us now try to classify them. Of the numerous old classifications proposed by the technical treatises on the urban settlements, we keep in mind those that have been selected according to the sites, then to the layout. They are the most commonly mentioned.

Typology of the Sites

• 26 Arthaśāstra, II, 3.1; Mayamata, X, 36-38.
• 27 Arthaśāstra II, 3.21.

40The first classification distinguishes the mountain forts (giridurga, pārvatadurga), the water forts (jaladurga, audakadurga), the forest forts (vanadurga), the desert forts (dhānvanadurga, irinadurga).26 The first two (i.e. those located on the slopes or on the top of a mountain or near a water surface) are considered to be the best situated to defend large human settlements.27

41The fortified towns exposed by excavations are often situated on flat land, at a low or mean altitude, with hillocks or rocky outcrops, such as Ahichchatra (No. 1) or Sankisa (No. 14) in Uttar Pradesh (fig. 8, a & b), or they are found close to protective elongated eminences, such as Sirsukh (No. 18) in Panjab, or near a hill slope, such as Sirkap (No. 17) (fig. 10), or also in more abrupt places where the walls follow a chain of hills surrounding a flat basin, such as Rajgir (No. 12) in Bihar (fig. 9).

42But most of them are located on the terraces bordering a stream course, large or small, because of the defensive value offered by this natural obstacle. This is the case, in the Ganges valley, with Mathura (No. 9) (fig. 5, a) and Kausambi (No. 8) (fig. 4, a), on the Yamuna, with Sravasti (No. 20), on an old arm of the Rapti (fig. 5, b); in Orissa, it is also the case with Sisupalgarh (No. 19), on a branch of the delta of the Mahanadi (fig. 7, c); in Madhya Pradesh, with Ujjain (No. 21), on the Sipra (fig. 7, a), with Besnagar (No. 4), on the Betva (fig. 4, b) or with Eran (No. 6), on the Bina (fig. 6, a); in Andhra Pradesh, with Nagarjunakonda (No. 10), on the Krishna (fig. 6, b) or Satanikota (No. 16), on the Tungabhadra; in Karnataka, with Banavasi (No. 12), on the Varada, and Sannati (No. 15), on the Bhima.

43Mathura and Sravasti are both located on the concave bank of meander which forms the longest side of the strongholds (fig. 5, a & b). Besnagar stands at the head of confluence of two rivers surrounding the settlement on three sides (fig. 4, b), which is also the case with Eran wedged in the loop of a meander (fig. 6, a).

44The other reason for this choice are the facilities offered by this continuous flow to supply the ditches with water. It is the case at Sisupalgarh (No. 19) where a river has been diverted on to the ditches of the enclosure (fig. 7, c), at Kausambi (No. 8), where the old side ditches correspond to the streams which flow into the river (fig. 4, a), finally, at Ujjain (No. 21), where the two arms of the ditch are connected with the river for water supply (fig. 7, a).

Typology of the Layouts

• 28 Mayamata, X, 13-14.
• 29 Mayamata, X, 42-43. According to Devī-Ayni Kālikā Purāṇa, it seems that square or rectangular forms (...)

45Regarding the plans of enclosures, the Śilpaśāstra distinguish several types of rampart: square (caturaśra), rectangular (āyataśra), round (vṛtta), elliptical (vṛttāyata) or perfectly circular (golavṛtta)28, triangular (trikoṇa), octogonal (astāśṭra) or semicircular (nemikhaṇḍa).29

46The vestiges exposed during excavations are almost always geometrical: an irregular polygon at Ujjain (No. 21) (fig. 7, a) and at Sirkap (No. 17) (fig. 10), various quadrilaterals at Rajgir (No. 12) (fig. 9), a kind of trapezium at Kausambi (No. 8) (fig. 4, a), a parallelogram at Sirsukh (No. 18) (fig. 10), a rectangle at Vaisali (No. 22) (fig. 7, b), a square at Sisupalgarh (No. 19) (fig. 7, c), an isosceles triangle at Ahichchatra (No. 1) (fig. 8, b), or it is a circular or rather oval plan at Sankisa (No. 14) (fig. 8, a), at Banavasi (No. 2), semicircular at Sravasti (No. 20) (fig. 5, b) and Sannati (No. 15), crescent-shaped at Mathura (No. 9) (fig. 5, a).

Zoom in Original (jpeg, 1,4M)

Fig. 4. a. River front site, trapezoid in shape, 322 ha (after CSR, vol. I. pl. XLVIII, f.p.301); b, confluence head site, oval in shape, 176 ha (after CSR, vol. X, pl. XII).

Zoom in Original (jpeg, 340k)

Fig. 5. a, River front site, crescent-shaped, 370 ha (after EIA, vol. II, p. 284); b, river site, semicircular in shape, 150 ha, Nausahra Gate shown (after ASI-AR, 1907-08, f.p. 80).

Zoom in Original (jpeg, 436k)

Fig. 6. a. River front site, mango-shaped, 15 ha, late Medieval fortification represented (after CSR, vol. X, pl. XXIII; b, river site, polygonal in shape, 50 ha, western gate shown (after IAR, 1957-58, p. 7). Note that there are no salients.

Zoom in Original (jpeg, 336k)

Fig. 7. a, River Iront site, polygonal in shape, 100 ha (after IAR, 1956-57, p. 21); b, flat land site, rectangular in shape, 12.5 ha (after ASl- AR, 1903-04, f.p. 88); c. river site, square in shape, 132 ha, western gate shown (after Ancient India, No. 5, fig. 1, f.p. 66, pl. XXVII).

Zoom in Original (jpeg, 296k)

Fig. 8. a, Flat land site with rocky outcrops, oval in shape, 265 ha, citadel in the middle (after CSR, vol. I, pl. XLV, f.p. 271); b, flat land site with hillocks, 169 ha (after CSR, vol. I, pl. XLIII, f.p. 257). Note that the fortifications represented are of a later period.

Zoom in Original (jpeg, 700k)

Fig. 9. Rajgir: dissected hill tract site, enclosures forming irregular quadrilaterals around old and new Rajgir, 200 ha and26 ha respectively (after ASI- AR, 1905-06, f.p. 86).

Zoom in Original (jpeg, 488k)

Fig. 10. Sirkap: hill slope site, forming an irregular polygon, 73 ha; note the pentagonal towers at the corners; Sirsukh, flat land site, forming a parallelogram, 134 ha; note the semicircular towers (after J. Marshall, A Guide to Taxila, f.p. 132). The layout of the fortifications of Sirkap has been corrected according to the map given by A. Gosh in Ancient India, No. 4, f.p. 84.

47What were the characteristics of these enclosures?

Principles of Defence

• 30 Arthaśāstra, II, 3; Mayamata, 36-48.

48From the ancient technical treatises three main features of the defence system have to be considered:30 the rampart-ditch, the flanking towers and the gates.

• 31 Kautilya (Arthaśāstra, II, 3) mentions expressly that the rampart (vapra) must be built with the pi (...)

49- first, the rampart-ditch, prākāra-parikhā, which results from the same operation, i.e. the digging of an excavation in the ground, the earth from which is thrown up behind to serve as an embankment,31 strengthened by a wall;

50- then, the projecting parts of the rampart forming quadrangular towers, attached at regular intervals to the main enclosure, often accessible by stairs or a ramp;

51- finally, the monumental gates, whose number is often mentioned (generally there is one in the middle of each face and sometimes more).

• 32 “That the moat and rampart were the results of the same operation, namely the heaping up of the mat (...)

52Excavations confirm the treatises. What the ancient texts say regarding a high and thick bund corresponding to a deep and broad ditch is seen in the section through the defensive system of Ujjain (fig. 11). The diggings carried out in other sites32 also testify to this dichotomy, except in places where it was not needed.

Zoom in Original (jpeg, 140k)

Fig. 11. Dichotomy ditch-rampart: the Ujjain example (after IAR, 1956-57. p. 22).

53The embankments, then, were simply raised and strengthened by masonry works to become true ramparts, revetted with brick or stone, better adapted to the military technology of the time. The ditches were provided with water coming from periodical rains or neighbouring rivers through canals.

54Most of these works, requiring an immense labour, were gigantic. It is therefore not surprising that they had an extraordinary longevity, having been modified and even rebuilt several times.

55Figure 11 shows how archaeologists have been able to identify the different levels of the succeeding deposits and to recreate the evolution of the sites. At Ujjain (No. 21) precisely, excavations have revealed a continuous occupation from the 7th century B.C. to the Muslim period in Malwa, i.e. until the 14th century A.D. (roughly 2000 years!). At Kausambi (No. 8), the history of the fortifications can be followed from the second millennium B.C. to the 6th century A.D!

Curtain Walls

56Many of these fortifications cover huge areas of more than 100 ha (370 à Mathura (No. 9), 320 at Kausambi (No. 8), 200 for the inner enclosure of Rajgir (No. 12) and in almost all the strongholds the same character of massiveness is found.

57Walls in all the places are bulky. In the alluvial plains, where stone is scarce, especially along the river banks, it seems (though their dimensions are often not given in the excavation reports) that, at their base at least, they were roughly 30 m in width: 33 m at Sisupalgarh (No. 19), 34 m at Nagarjunakonda (No. 10), 45 m at Eran (No. 6). At Ujjain (No. 21) (fig. 7, a), the rampart is 74 m thick on its western side and 105 m on the riverside where it is reinforced by wooden logs and sleepers. These embankments are generally revetted with brick, several metres thick, with a strong batter (at Kausambi (No. 8), it is 15° and 40°).

58In other sites, where stone is abundant, walls made of stone blocks separated by earth and nibble filling are of a lesser thickness: from 4.2 to 5 m at Rajgir (No. 12) (fig. 14), from 4.5 to 6.4 m at Sirkap (No. 17) (fig. 12). 4.4 m at Sirsukh (No. 18) (fig. 13), 3.2 m at Satanikota (No. 16).

59In all these sites the curtain walls are solid, i.e. having the interior completely filled up, except at Sirsukh (fig. 13), where they are pierced by loopholes

Towers

60On the salients or projecting outward works flanking the curtain walls many obscure points remain because of the inadequacy of the excavations and also owing to the rudimentatry descriptions given by archaeologists.

• 33 Arthaśāstra, II, 3, 7, 10; Mayamata, X, 13-16, 44-46.

61Classical treatises maintain that ramparts were broken by rectangular towers, separated by spaces of 30 daṇḍa or about 60 m.33 However, they seem to be missing in some fortresses, such as Nagarjunakonda (No. 10), where serious excavations have been conducted, or at Banavasi (No. 2).

62In fact, in the reports published on the enclosures of earth revetted with brick, quadrangular towers are mentioned only at Kausambi (No. 8) where they are attached to the curtain walls, at regular intervals, and yet their dimensions are not given. At Katragarh (No. 7), two watch-towers, one accessible by stairs were found. It is likely that systematic excavations conducted at other sites would reveal the existence of such works. At Sisupalgarh (No. 19) (fig. 7, c), heaps of debris at the four comers appear to be the vestiges of towers.

Zoom in Original (jpeg, 276k)

Fig. 12. Sirkap: solid rectangular towers on the curtain walls, pentagonal towers at the comers (after A. Gosh, op.cit., f.p. 84); north gate shown (after J. Marshall, Taxila, vol. III, pl. 10).

Zoom in Original (jpeg, 224k)

Fig. 13. Sirsukh: curtain walls with loopholes, hollow semicircular towers (after J. Marshall, op.cit., vol. III, pl. 42.

Zoom in Original (jpeg, 804k)

Fig. 14. Rajgir: curtain walls and rectangular towers on the ridge (photo by M.H. Kuraishi & A. Gosh, op.cit., cover).

Zoom in Original (jpeg, 184k)

Fig. 15. Pataliputra: double wall made of wooden sleepers thought to be a part of the palisade seen by Megasthenes (after ASI- AR, 1926-27, pl. XXX, c and d).

63In the stone enclosures in Bihar or Panjab, on the other hand, projecting towers have been well preserved. At Rajgir (No. 12) (fig. 14), on the outer enclosure, there remain 16 rectangular towers, with a salient of 10 to 12 m and a width of 14 to 18 m. At Sirkap (No. 17) (fig. 12), towers flanking irregularly the rampart project 5 m and are 7.7 m wide; those at the comers are pentagonal in shape with a salient of 20 m and a width of 15 m.

64As for Sirsukh (No. 18) (fig. 13), the curtain walls are flanked, at intervals of 37 m, by towers, semicircular in plan, accessible from the interior by means of a narrow passage carried out through the thickness of the wall. These towers are hollow within instead of solid, and are furnished with loopholes placed at a height of 1.5 m above the floor level.

65We should also be reminded that Pataliputra (No. 11) was protected by a high fence made of heavy wooden sleepers placed vertically in a double row, thought to be the wooden palisade seen by Megasthenes

66Thus, except in the North-West of the Indian subcontinent and at Pataliputra, in all the other excavated sites in India, massive ramparts, flanked by solid quadrangular towers, are found.

Gateways

67In the defence system, openings through the enclosures appear to have been the chief concern of the builders, for gates are very elaborate, with defensive outworks.

68Most of the gaps made in the ancient earthen embankments (when they are not the work of erosion) correspond to holes permitting passage through them, but, for want of stratigraphical trenches perpendicular to the rampart, it is difficult to determine how they were built.

69- At Nagarjunakonda (No. 10), in the valley of the Krishna, the western gate (fig. 6, b) forms a rectangular courtyard whose projecting part outwards is pierced by a corridor, 5 m wide; it was provided, on both sides, with guardrooms.

70- At Satanikota (No. 16), on the Tungabhadra, an elaborate gateway, facing south, was exposed; it is characterized by a flight of 5 steps, with a width of 3 m, flanked by a running parapet partly of baked bricks and of stones; significant are two pairs of sockets cut into blocks of stone, distributed evenly on either side of the gate: the first set inside the gate was carved to carry two leaves of the door; another set was found on either side of the outer edge of the gateway, probably meant for the heavy posts of a draw-bridge.

71- At Kausambi (No. 8), in the Ganges valley, the eastern gateway is protected by a mud bund, 106 m long, which served as a curtain for the entrance, and also two towers at the top of the rampart facing an outer tower; situated on the other side of an extensive moat, 144 m wide! It is unfortunate that we do not have a plan of this structure because it appears to be a carefully laid out defence complex

72The best examples of elaborate gateways, of which we have precise plans, are found at Sravasti, Sisupalgarh and Sirkap.

Zoom in Original (jpeg, 368k)

Fig. 16. Sravasti: Nausahra Gate, double curve inwards so as to form two salients at both sides (after ASI-AR, 1901-08, pl. XXXI, f.p. 112).

Zoom in Original (jpeg, 340k)

Fig. 17. Sisupalgarh: west gateway, long and narrow passageway positioned between two rectangular blocks (after Ancient India, No. 5, p. 76, fig. 4).

• 34 The same system existed in Burma where a double curve inwards in the rampart is also found. At Sri (...)

73- The Nausahra Gate, one of the main entrances of Sravasti (No. 20) (fig. 16) is made of a double curve inwards,34 so as to form two quadrangular brick structures, leaving between a space of 18 m; this opening is reinforced outside by two projecting quadrangular walls, 10 m long, on both sides of the curves; in the interior, are two rectangular rooms enclosed by solid walls serving as guardrooms, in which iron clamps and nails probably used in the vanished wooden doors of the gateway were found.

74- At Sisupalgarh (No. 19) (Jig. 17), it is not an inward but an outward structure that is found. The excavated western gateway, built of large slabs of dressed latente, is an imposing complex comprising an outwork made of two massive L-shaped structures, 8.4 m thick (the length of the longer arm of the L is 48 m, that of the shorter 19 m); between these two walls, is a long passageway, 8 m wide, leading to the interior of the fort through two separate entrances: the outer one and the inner one, provided, at the origin, with wooden doors, as can be inferred from the sockets found at both places; behind the outer entrance there was a guardroom in the southern flank, while the northern flank, near the inner gate, was pierced by a narrow ancillary passage for pedestrian traffic; the top of the flanks could be ascended, inside the curtain wall, by flights of stone steps at the end and on either side of this long corridor. It is indeed a remarkably elaborate structure.

75- In the northern gateway of Sirkap (fig. 12), the first opening, parallel to the main wall, leads to a courtyard (18.6 m x 10.5 m), forming a kind of barbican; to the west of this open area, and communicating with it, are two guardrooms, located on the inner and outer faces of the rampart; worthy of note is the fact that the outer opening of the courtyard is not set directly opposite the main street, but a little to the east: it means that the inner entrance was masked from view and that it could check any sudden rush of the assailant, a feature which will be adopted later and perfected.

76We ignore the shape of the gates in other sites, but probably they were similar to those just described, perhaps with some regional variations in features.

77What emerges from the few examples given supra is that gateways are no more simple rectangular openings in the plan of the curtain walls, they are now large and complex structures, skilfully designed, situated at the far end of a narrow bend in the curtain walls (Sravasti), where the attackers are vulnerable to the projectiles of the defenders; forming protruding double walls forcing the assailant to enter a corridor (Sisupalgarh), where they are under fire from both sides with open courts or corridors defended by guardrooms; or provided with entrances purposely not aligned (Sirkap), so as to block the attack of an enemy rushing through such a passage.

78With the adoption of the principle of double flanking, the gates have become major pieces in the defence system.

79Unfortunately archaeological evidence is incomplete. Vestiges exposed represent a kind of ground plan only, with no information on the parts of the fortifications above their basement, which is a serious deficiency for a military architecture principally based on the high command of the defence system.

80What was the original height of the enclosures? Were the ramparts and the towers crowned with a parapet? And if so, were they provided with battlements and projecting galleries?

81We shall answer these questions by referring to ancient treatises and analysing the iconographic data.

Superstructures

• 35 Arthaśāstra, II, 3.7, 10; Mayamata, X, 13-16, 44-46.
• 36 Arthaśāstra, II, 3, 7; Śukranītī, I, 1.479.

82Technical treatises shed little light on the superstructures. The information they give on how fortifications should be erected is incomplete and varies from book to book. Regarding the dimensions of defence works, for instance, the figures given for the height of the ramparts vary from 12 to 24 hasta (from 6 to 12 m).35 Crenellations are not specifically mentioned, except in the Arthaśāstra, where it is said that the top of the ramparts should be “crowned with (merlons in the form of) drums and monkey heads”.36

• 37 See, supra, note 24.
• 38 We have tried to reproduce faithfully, though in a simplified manner, the defence works represented (...)

83On the other hand, the narrative of the bas-reliefs made around the Christian era give us representations of fortified towns which appear to correspond to the vestiges analysed by us.37 Obviously, the sculptors of the Bodhgaya, Sanchi, Nagarjunakonda or Amaravati schools show the massive works of the peninsula and the Gandhara artists, the strongholds of the North-West.38

Military Works Represented on the Great Stupas of the Peninsula

• 39 Examples of plastered masonry or dressed stones at the basement of buildings are found at Bodhgaya (...)

84The fortifications shown in the great Buddhist monuments of peninsular India (figs. 18, c, d, e, 19, a, c, d, 20, b)39 seem to be brick structures, a building material exposed in excavations made in alluvial zones. The height of the ramparts appears to be that prescribed by the texts, i.e. from 6 to 12 m. At the top of the enclosure, the parapet is sometimes a simple wall, without any openings at Amaravati (fig. 19, c), but, in most cases, it is crenellated with serrated merlons at Sanchi (fig. 18, b, c), rectangular merlons put close together at Sanchi and Mathura (figs. 18, d, 20, a), or widely spaced at Amaravati and Nagarjunakonda (fig. 19, b, d, e).

85The ramparts seem solid, because no loopholes are seen on the curtain walls and parapets.

86At Bodhgaya are represented two parallel barriers consisting of a series of blocks with the shape of lotus petals which could represent the extremity of wooden sleepers placed vertically in a double row, as in the famous palisade of Pataliputra (fig. 15).

Zoom in Original (jpeg, 300k)

Fig. 18. Types of fortification represented in bas-reliefs, (A): a, double row of pointed merlons which could be the extremities of the sleepers of a palisade (see fig. 15); b and c, serrated merlons; d, rectangular merlons; e, gate surmounted by a gallery (after J. Auboyer & J.F. Enault, op. cit., pls. 1, 1, 5; 36, 1; A.K. Coomaraswamy, Early Indian Architecture, pp. 217, fig. A, 219, fig. C).

Zoom in Original (jpeg, 308k)

Fig. 19. Types of fortification represented in bas-reliefs (B): a, rectangular towers; b, towers topped with a hemispherical dome; c, wall without battlement; d, towers with loopholes above the curtain walls; e, rectangular merlons (after J. Auboyer & J. Enault, op.cit., pls. 39, 6, 1, 4, 5; 1, 4; 41, 2; A.K. Coomaraswamy, op.cit., p. 214, figs. 13, 15, 17; A. Ray, op. cit., fig. 66).

Zoom in Original (jpeg, 228k)

Fig. 20. Types of fortification represented in bas-reliefs (C): a, quadrangular towers and rectangular merlons; b, hollow semicircular tower; c, towers pierced by sagital loopholes and merlons topped with a roof (after K. Coomaraswamy, op.cit., p. 184, fig. 13, p. 212, fig. 7; J. Ph. Vogel, La sculpture de Mathura, pl. XX, b).

Zoom in Original (jpeg, 196k)

Fig. 21. Types of fortification represented in bas-reliefs (D), Gandhara Art: a, b and c, gates with overhanging turrets; b and d, merlons topped with a roof; d, towers pierced by loopholes (after A. Foucher, L’art grécobouddhique du Gandhara, vol. I, figs. 229, 234, 288; for c, K. Fisher, ‘Gandharan Sculptures from Kunduz and environs’, Artibus Asiae, vol. XXI, pp. 234-235).

Zoom in Original (jpeg, 203k)

Fig. 22. Types of fortification represented in bas-reliefs (E), Gandhara Art: a and d, towers pierced by loopholes; b, gate with three overhanging turrets; a-d, merlons topped with a roof (after A. Foucher, op. cit., vol. I, figs. 292, 269, 235, 31).

87As for the flanking towers, well represented, especially at Amaravati (figs. 19, a, b, d, 20, a), they are all quadrangular in shape, without any opening up to the parapet; they are capped by a wooden structure forming a gallery with a roof in the form of an upside down hull (figs. 18, c, d, e, 19, a, d, 20, a). Gates are usually built between two projecting rectangular towers provided with rectangular openings indicating the existence of guardrooms inside. On the passageway is one (sometimes two at a higher lever) projecting railed platform projecting from the wall, joining up the towers.

88At Sanchi is represented a curious long balcony above the gate (fig. 18, e), looking like a hoard or covered wooden gallery built out from the parapet of the gate supported on corbels providing for vertical defence of the area below.

89Obviously these magnificent sophisticated edifices are befitting for a palatial zone and not for the austere constructions of an ordinary stronghold: they were the defences of the palace area of a city, as can still be seen today at Hampi where the royal zone is also surrounded with a wall. However, the representations of the different types of battlements should be retained, because the main ramparts of the cities were probably provided with the same crenellations.

90Now, in the representations found in the bas-reliefs of Mathura and in the Gandhara art are shown new elements in the towers and parapets, which are the expression of a different defence system.

Military Works Represented at Mathura and in Gandhara Art

91As shown by the excavations, the fortifications in the Northwest consisted of random mud infill and facing of coursed rubble and not brick as in the sites of the Ganges valley. In Greco-Buddhist representations,

92- the curtain walls are pierced by triangular and sagital loopholes (figs. 21, d, 22, a, d), the parapets are made of merlons surmounted by a roof with four sides and provided with triangular loopholes (figs. 20, b, c, 21, b, d, 22, a, c, d),

93- the towers, little projecting, are either rectangular (figs. 20, c, 21, d) or semicircular (fig. 20, b, 21, c, 22, a, d); they are obviously hollow, with openings in their upper part used as archery rooms (fig. 21, d) or sagital loopholes (figs. 20, c, 21, c, d, 22, a, d); their revetment is often decorated with triangular motifs (fig. 21, c).

94- the double doors tend to be broader towards the bottom and are trapezoidal in shape (figs. 21, a, b, d, 22, a, b, c, d); they are defended by lateral towers, and above all, they are provided with small overhanging turrets open towards the sky (figs. 21, a, d, 22, a) or boxes with a roof (fig. 21, c); in the case of projecting gates, three of these boxes are found, one in front and the two other on the sides (fig. 22, b): these structures are true brattices intended to batter the foot of the wall by means of holes made in the floor through which missiles could be thrown down on the enemy below.

95Now, it is surprising to find in the bas-reliefs of Mathura some elements shown in Gandhara art: a curtain wall forming a curve, square merlons covered with a roof with four sides (fig. 20, c), and a hollow semicircular tower (fig. 20, b).

96Obviously, in Gandhara art also, the works represented are the defences of the palace areas of a city. Nevertheless, what is to be retained from this observation is the special shape of the merlons covered with a roof and the presence of boxes built on the crests of gates, which are the first representations of brattices in the Indian subcontinent

The Central Asia System of Fortification

• 40 They correspond to the Rushan defence works built elsewhere and described by H.P. Francfort, Les fo (...)
• 41 On the problems of datation and the question of the shape of the towers (quadrangular or circular), (...)
• 42 E.I.A., vol. II, p. 286.

97The main features of the fortifications represented in Gandhara art are in conformity with the vestiges excavated in Panjab, at Sirsukh (No. 18), built by the Kushans (fig. 13)40 Now, it is assumed that they are the people who brought to the Northwest of the Indian subcontinent the traditions of fortification of Central Asia, with curtain walls pierced with long loopholes and hollow semicircular towers,41 adapted to the military techniques of horsemen and bowmen people coming from the Steppes. If we find this type of defence in Mathura (No. 9), it is probably because it was one of their capitals at the beginning of the Christian era; in fact, in the excavations conducted at that place, archaeologists found that the towers of the inner enclosure were probably semicircular42: is it not a striking coincidence?

98This defence system, however, does not seem to have been adopted anywhere else in India. Solid (not hollow) towers are erected in all the other sites; as for semicircular or circular works and shooting devices (protruding boxes or galleries), they appear in the peninsula only in the Muslim Deccan kingdoms in the 15th century and, evidently, there is no relation between those two types of fortification, since, as will be shown, infra, the Muslim defence system probably originates in contacts with Arabs and Turks from the Middle East.

The Indian System of Fortification

• 43 A. Gosh, Taxila (Sirkap) 1944-45, Ancient India, No. 4, p. 44; J. Marshall, Taxila, vol. I, p. 1 14 (...)
• 44 P. Leriche, Fouilles d'Aï Khanoun, V, Les remparts et les monuments associés, p. 98.

99The Indo-Greek and Indo-Parthian site of Sirkap (No. 17) (fig. 12), dating from the 1st century A.D., with its massive ramparts and solid towers, is considered to be purely hellenistic,43 though not built according to the techniques then in usage in Central Asia, where curtain walls tended to be lighter and towers more projecting.44 In fact its defence system does not differ essentially from that of the other sites in the peninsula considered here (except for the unusual pentagonal shape of the towers at the angles).

100Whether built in brick as in alluvial zones, or in stone in rocky places, the fortifications are designed on the same principles.

101Excavations broadly corroborate the information furnished by the literary sources and show that the most persistent features of the early fortifications are massive and compact earthen walls, provided with a baked brick or stone facing, surrounded by wide and deep ditches. These ramparts are, in some cases, reinforced, at regular intervals, by solid quadrangular towers. Regarding the crenellation of the curtain walls, the narrative reliefs of the great Buddhist stupa show that the merlons were triangular or rectangular. Gateways are very elaborate and probably the strongest parts of the defence: in certain places, the opening is strengthened by two powerful towers with an open courtyard, in others, the wall turns inwards so as to form two quadrangular structures or it turns outwards and joins the heel of a massive double wall. The design of the northern gate of Sirkap, with a courtyard, described supra, is particularly interesting, because it is perhaps the model followed by the builders of the historic period, who constructed gateways composed of open courtyards, pierced by openings, which are not on the same alignment so as to force the assailants to follow a zigzag course, and thus make them vulnerable to the projectiles of the defenders.

102In this system the defence is not through the loopholes of the ramparts, as in the strongholds built-up on the Central Asian pattern, but it is from the top, from the battlements, a high position which enables the defenders to dominate the besiegers.

Defence against Floods

• 45 See M.S. Mate, op.cit., pp. 58-69.

103It has been suggested,45 with good reasons, that ramparts were not only meant to be defence walls but were in some places protective embankments against floods. It is evident that at Kausambi (No. 8) or at Ujjain (No. 21), for example, there was no need for defence purposes of such a large mass of walls, with a basal width exceeding 50 m. If they were built it was also as a protection against floods. In these sites threatened by the overflowing of rivers, embankments were thus used as ramparts.

Conclusion

104If, in almost all the fortified ancient sites, massive structures are found, it is because the ancient military engineers believed in a system of protection, based on the accumulation and the enormity of the obstacles opposed to assailants and considered that these entranchements would guarantee the best security.

105As proved by excavations, this concept of fortification was already defined from the middle of the first millennium and it is manifest that defence architecture in the Indian subcontinent has retained its distinctive features for more than one thousand years, except in the Northwest, more sensitive to foreign influences during the Greek and Kushan periods.

ANNEXES

Appendix. Catalogue of 22 Selected Early Historic Fortified Sites46

(For each site we quote EIA, in which an almost exhaustive list of references is found; in addition are mentioned studies which bring precise details discussed in the present analysis or give illustrations).

No. 1. Ahichchatra (Bareilly district) (see fig. 8, b)

Ahicchatra, the ancient capital of the Panchalas, in the district of Bareillly, in Uttar Pradesh, stands in a flat region where there are outcrops of rocks. Its enclosure surrounding three rolling mounds has roughly the shape of an isosceles triangle covering 169 ha.

Initially built of mud around 100 B.C., the rampart was raised with a brick wall, 5 m wide and 2.6 m high, before being protected by a mud cover and later buttressed. The towers exposed during excavations are rectangular; those shown in Cunningham’s plan, semicircular in shape, date from the 18th century.

Ref.: EIA, vol. II, pp. 7-9; A. Gosh & C. Panigrahi, ‘The pottery of Ahichchatra, district Bareilly, U.P.’, Ancient India, No. 1, Jan. 1946, pp. 37-40; IAR, 1963-64, pp. 43-44, pls. XXVIII-XXIX; 1964-65, pp. 39-42; plans: CSR, vol. I, pl. XLIII, f.p. 257; IAR, 1964-65, p. 40.

No. 2. Banavasi (Dakshina Kannada district)

At Banavasi, situated on the left bank of the Varada river, a tributary of the Tungabhadra, in Karnataka, the rampart, oval in shape and covering an area of about 42 ha, is pierced by two openings. It is a brick structure on a rubble foundation, surrounded by a deep moat, belonging to the Satavahana period (1st c. B.C. - 2nd c. A.D. It was reinforced by large laterite blocks after the 4th c. A.D. Apparently the enclosure was not flanked by towers.

Ref.: EIA, vol. II, p. 45; S.K. Joshi, Defence Architecture in Early Karnataka, pp. 40-47; plans: ibid., figs. 7 and 8.

No. 3. Bangarh (W. Dinajpur district)

At Bangarh, on the eastern bank of the Punarbhaba, in West Bengal, the site was in continuous occupation from the 3rd c. B.C. till the invasion of the Turks in the 13th c. A.D. Surrounded on three sides by a ditch, the fortress occupies an area of 24.3 ha. The rampart, consisting of mud revetted with bricks was repaired several times. On the eastern side of the enclosure, a gate and a causeway, about 60 m long, leading from the ditch into the city were found. The excavators mention also two circular towers, which are probably later additions.

Ref.: EIA, vol. II, p. 47; K.G. Goswami, Excavations at Bangarh (1938-41), pp. 1-5, 34-35, pl. XVIII.

No. 4. Besnagar (Vidisa district) (see fig. 4, b)

At Besnagar, the ancient Vidisa, to the north-east of Bhopal, in Madhya Pradesh, the fortress, covering an area of 176 ha, is situated in the fork between two rivers, the Bes and the Betva. On the western side, there is a huge rampart strengthened by a ditch which runs right across the neck of land between the two rivers; commanding the reach of the Bes, there is a tower which rises to a height of 15 m.

Recent excavations brought to light different layers dating from 1300-500 B.C, but no transversal sections of the rampart are available.

Ref.: EIA, vol. II, p. 62; D.R. Bhandarkar, ‘Excavations at Besnagar’, ASI-AR, 1914-15, pp. 66-88; plan: CSR, vol. X, pl. XII.

No. 5. Campa (Bhagalpur district)

The fortress of Campa, situated on the right bank of the Ganges, in Bihar, is surrounded by a deep moat on three sides, the fourth one being the bed of the river. The rampart is characterized by a thick wall of blackish soil with a gentle slope on the outer side, dating from 500 B.C. It was reinforced during a second phase (150-50 B.C.) by the dumping over of yellow and red soil and finally raised to a height of 5.80 m with a brick wall.

Ref. :EIA. vol. II, p. 90.

No. 6. Eran (Sagar district) (see fig. 6, a)

The fortified settlement of Eran, situated inside a deep curve of the Bina river, an affluent of the Betva, in Madhya Pradesh, oval in shape, covers an area of 15 ha. The rampart, built of black and yellow clay, 6.40 m high and 47 m broad, was girdled on three sides by the river and on the fourth by a moat 36 m broad and 5 m deep. The settlement seems to have been in use down to historical times (the round towers shown in Cunningham’s plans are of a later period).

Ref: EIA., vol. II, pp. 134-136; IAR, 1962-63, pp. 11-12 & pl. XXX; 1963-64, pp. 15-16; plan: CSR. vol. X, pl. XXIII.

No. 7. Katragarh (Muzaffarpur district)

At Katragarh, in Bihar, the fortress may be dated to the Sunga period. The rampart built of mud was later revetted with brick, with sloping sides. Two (rectangular?) watchtowers, one with a flight of steps, were exposed.

Ref: EIA, vol. II, p. 210; IAR, 1977-78, pp. 15-16 & pl. XII A.

No. 8. Kausambi (Allahabad district) (see fig. 4, a)

The enormous ruins of Kausambi, one of the most famous cities in Ancient India, are situated on the left bank of the Yamuna, in Uttar Pradesh. Thanks to the excavations conducted at this site, the different phases of occupation of the city from the first millennium B.C. to the 6th century A.D. are known to us.

Its impressive rampart, quadrilateral in shape, encloses an area of 322 ha. It is surrounded by a ditch, with a width that varies from 144 to 480 m !

The walls are flanked, at regular intervals, by a series of salients and towers and are pierced by 5 principal gateways, one of which is described supra and 6 subsidiary ones. The average height of the walls is 10.6 m, that of individual towers, more than 22 m.

Constructed of clay, it was first erected in the last decades of the 2nd millennium and provided with a baked brick revetment with a prominent inward batter. In the second phase (10th c. B.C.), weep-holes were dug up and seepage water drained and, during the third one (6th-5th c. A.D., the rampart was raised to a further height of 2.4 m with mud blocks and the brick revetment heightened by 90 cm. Hundred years later, the wall was again raised and a secondary rampart created. Before the Christian era several guardrooms and towers were erected. Finally, the whole complex was destroyed by the Hunas in the 6th c. A.D. and the defences abandoned.

Ref.: EIA., vol. II, pp. 212-215; G.R. Sharma, The excavations at Kausambi, 1957-59), pp. 24-43, pls. 21, 23; IAR, 1954-55, pl. XXXI; 1956-57, pl. XXXVI; 1960-61, pl. LIV; 1961-62, pl. LXXX; plans: C.S.R., vol. I, pl. XLVIII, f.p. 31; G.R. Sharma, op.cit., pl. 5, aerial photograph.

No. 9. Mathura (district of the same name) (see fig. 5, a)

Mathura, in Uttar Pradesh, another river site, is situated on the Yamuna. During the 4th c.

B.C., the place was fortified by a massive mud and kaṅkar wall, originally about 6.5 m high, forming a longish crescent, covering 370 ha, over an undulating ground. During 200 years, the enclosure seems to have been neglected, but from the 1st to the 3rd c. A.D., under the Kushans, the old structure was enlarged, and, on the northern side, a new inner enclosure, made of mud strengthened by a wall of brick, circular in shape, with a basal width of 17 m, was built with towers probably semicircular.

Ref. : EIA, vol. II, pp. 283-286; plan: ibid., p. 284.

No. 10. Nagarjunakonda (Guntur district) (see fig. 6, b)

The site of Nagarjunakonda, on the right bank of the Krishna, in Andhra Pradesh, has been the subject of large excavations because of the construction of the reservoir of Nagarjunasagar. The citadel, covering an area of 50 ha, reached its peak in the 3rd and 4th c. A.D. Its rampart, made of mud, 24 m wide at the base, was later strengthened by a revetment of baked brick, 2.7 to 4.2 thick, built on the bare rock surface. It was surrounded by a ditch varying from 22 to 39 m in width.

Ref.: EIA, vol. II, pp. 299-303; IAR, 1957-58, pp. 5, 6, pl. II; plan: ibid., p. 7, fig. 4.

No. 11. Pataliputra (Patna district) (see fig. 15)

Pataliputra, situated on a strip of land between the Ganges and the Son, in Bihar, exposed to floods, was the capital of Magadha since the 5th c. B.C. According to Megasthenes, the enclosure, shaped like a parallelogram, 14.5 km in length and 2.4 km in breadth, was girded by a wooden wall with 570 towers, 64 gates and a ditch about 182 m in breadth and 14 m in depth.

Excavations at Bulandibagh resulted in the discovery of a wooden structure, 137 m long, made of heavy wooden sleepers, which could be a part of the famous enclosure, which protected the city.

Ref.: EIA, vol. II, pp. 334-336; J.A. Page, ‘Bulandibagh near Patna’, ASI-AR, 1926-27, pp. 136-138, pl. XXX, c & d.

No. 12. Rajgir (Nalanda district) (see figs. 9 & 14)

The most amazing site perhaps and, in any case, the only one which, until recently, was believed to be prior to the Mauryas, is Rajgir, Rajagriha, the capital of Magadha until the 5th c. B.C., before the creation of Pataliputra.

It is a dissected hill tract site, with elongated hills oriented south-west-north-east, overlooking the narrow valley of the Banganga which gets wider, forming a basin where the settlement developed.

An outer enclosure with a circuit of about 40 km runs over the top of each hill, on both sides of the river. The rampart is built of massive undressed stones carefully fitted and bonded together with a core composed of smaller blocks without mortar. The wall is about 3.5 m high, from 4 to 5 m thick; inside, it is provided with stairs or ramps, so as to give access to the top; it is flanked by rectangular towers, from 14 to 18 m broad, projecting 10 or 12 m (16 of them have been found, along with some watchtowers erected at various vantage points on the hills).

At the bottom of the valley, the old town is girdled by a long ridge of earth and stone, without trace of towers or stairs, having the shape of an irregular quadrilateral of 200 ha.

Towards the north, the new town, occupied from the 5th c. B.C. to the 1st c. A.D., was surrounded by a thick mud rampart, having the shape of an irregular pentagon, strengthened by a ditch about 30 m broad. Recent excavations have shown that this earthen rampart, from 40 to 53 m thick, was revetted with brick, then later raised by a wall of the same material.

The south-west portion represents a kind of citadel, having the shape of a quadrilateral covering 26 ha. The walls, built of earth and stone rubble faced with huge unhewn blocks of stone, are from 4 to 5 m thick; they are flanked by semicircular towers of the same masonry built at irregular intervals, probably a later addition

Ref.: EIA, vol. II, pp. 362-365; J.H. Marshall, ‘Rajagrha and its Remains’, ASI-AR, 1905-06, pp. 86-90; M.H. Kuraishi, The Antiquarian Remains in Bihar, pp. 432-472; M.H. Kuraishi & A. Gosh, Rajgir, 41 pp.; plans: ASI-AR, 1905-06, f.p. 86; M.H. Kuraishi & A. Gosh, op.cit., fig. 1 & pl. IX.

No. 13. Sanghol (Ludhiyana district)

At Sanghol, in Panjab, the fortress (2nd half of the 1st c. to 5th c. A.D.) is a massive mud rampart with a basal width of 29 m, surrounded by three ditches, one outside the walls and the other two inside (details of new excavations are not known).

Ref.: E.I.A., vol. II, pp. 390-391; IAR, 1970-71, pp. 30-31 & pl. XLVIII B.

No. 14. Sankisa (Farrukhabad district) (see fig. 8, a)

Sankisa is situated in the Ganges-Yamuna Doāb, slightly to the north of the Kalinadi river, in Uttar Pradesh, in a flat land with rocky outcrops. The city comprising these mounds is enclosed by an enormous earthen rampart, oval in shape, covering an area of 265 ha, cut by three big openings said to be the positions of the three gates of the city. Inside the town, there is a mount 12 m high, rectangular in shape, which served as a citadel before the Christian era. However, nothing is known about the fortification since no excavations have been carried out in the rampart

Ref.: EIA, vol. II, pp. 391-392; plan: CSR, vol. I, pl. XLV, f.p. 271.

No. 15. Sannati (Gulbarga district)

A Buddhist site dating between 1st c. B.C. to 3rd c. A.D., Sannati, situated in a deep meander of the Bhima river, in Karnataka, is surrounded by two enclosures, semicircular in shape, linking the the two curves of the loop.

The outer enclosure is 600 m long, from 12 to 15 m thick, with two openings; the inner one is only 450 m long, with a single entrance. As the city is strongly protected on three sides by the river there was no need of a ditch.

Ref.: EIA, vol. II, p. 392; S.K. Joshi, Defence Architecture in Early Karnataka, pp. 47-49; plan: ibid., fig. 16.

No. 16. Satanikota (Karnul district)

At Satanikota, situated on the right bank of the Tungabhadra, in Andhra Pradesh, the enclosure (1st c. B.C. to the middle of the 3rd c. A.D.) follows the contour of the natural elevated rocky ground. Its wall, measuring 3.2 m in width, built of stone slabs laid in mud mortar, is provided with a 1.45 m baked brick facing. It is skirted by a rock-cut ditch, 3.20 m deep and 4.25 m wide. The elaborated gateway complex found there has been described supra.

Ref.: EIA, vol. II, pp. 399-401; IAR, 1977-78, pp. 3-11 & pls. 2 & 3, section across fortification, fig. 2.

No. 17. Sirkap (Rawalpindi district) (see figs. 9 & 12)

The site of Sirkap, in Panjab (Pakistan), founded by the Indo-Greek in the 2nd c. B.C., lies between the western spurs of a hill range and the eastern bank of a small stream.

Its enclosure, 5.6 km in length, covering an area of 73 ha, straight on the north and east part, has an irregular alignment broken by various salients on the western edge of the plateau. The wall, composed of random rubble in mud revetted with hard limestone, with a thickness varying from 4.5 to 6.4 m, is strengthened, at irregular intervals, by solid rectangular towers, 7.7 m wide, and projecting 8.3 m, but, at the comers, with towers pentagonal in plan, 15 m wide and projecting 20 m. The northern gateway has been described supra.

Sirkap was abandoned by the Kushan invaders towards the middle of the 2nd or 3rd c. A.D. who selected a new site, about 1.5 km to the north-east: Sirsukh.

Ref.: J. Marshall, Taxila, an Illustrated Account of Archaeological Excavations carried at Taxila, vol. I, pp. 113, 118, vol. III, pls. 1 & 10; J. Marshall, A Guide to Taxila, pp. 4-7; A. Gosh, ‘Taxila (Sirkap), 1944-45’, Ancient India, No. 4, pp. 41-84 & figs. 1, 2 & pls.V, VI, VII; plans: CSR, vol. II, pl. LVII, f.p. 111; J. Marshall, Taxila, vol. III, pl. 1; A Guide to Taxila, f.p. 132; A. Gosh, op.cit., f.p. 84.

No. 18. Sirsukh (Rawalpindi district) (see fig. 9 & 13)

The plan of the fortress is roughly a parallelogram covering an area of 134 ha (half of it is obliterated today). It is a massive construction, made of rough rubble faced with limestone masonry, 5.4 m in thickness, with walls pierced by loopholes and strenghthened, every 27 m, by hollow semicircular towers, as said supra.

Ref.: J. Marshall, Taxila, vol. I, pp. 217-218, vol. III, pl. 1; A Guide to Taxila, pp. 6-7; plans: CSR, vol. II, pl. LVII, f.p. III; J. Marshall, Taxila, vol. III, pl. 1; A Guide to Taxila, f.p. 132.

No. 19. Sisupalgarh (Bhubaneshvar district) (see fig. 7, c)

The most striking ancient fortified town is undoubtedly the imposing site of Sisupalgarh, situated at about 700 m of the Bhargavi river, to the south-west of Bhubaneshvar, Orissa, a site “unfolding to a remarkable degree the story of an integral culture that had its own stages of growth, decline and decay.”

The fort, square in plan, covers an area of 132 ha, with comer towers and 8 large gateways, 2 on each side, suggesting a regular planning. It is circumscribed by the waters of a rivulet, producing a ditch with a perennial supply of water.

In the first quarter of the 2nd c. B.C., the defences consisted of a massive clay rampart, 32 m wide at the base and 7.5 m high, on which were a series of circular holes, arranged at regular intervals of 55 cm, the exact purpose of which is difficult to determine.

This structure was reinforced by a thick layer of latérite and then surmounted by two brick walls, 8 m apart, towards the middle of the 1st c. A.D.

Finally, the site was abandoned somewhere about the middle of the 4th c. A.D. Its remarkable western gateway has been described supra.

Ref. :EIA, vol. II, pp. 412-413; B.B. Lal, ‘Sisupalgarh 1948: an Early Historical Fort in Eastern India’, Ancient India, No. 5, January 1949, pp. 64, 73-77, fig. 4, Western gateway, pls. XXXI & XXXIII; plans: ibid., p. 65, fig. 1, f.p. 66 & pl. XXVII, aerial photograph.

No. 20. Sravasti (Bahraich district) (see fig. 5, b)

Sravasti, the capital of ancient Kosala, in Uttar Pradesh, is situated on the southern bank of an ancient bed of the Rapti river. Its earthen enclosure, crescentic in shape, covering an area of 150 ha, was defended on the longer curve of the crescent by a ditch, the remains of which yet exist as a swamp.

The walls considerably vary in height, those to the west being 10.5 to 12 m high, while those on the south and east not more than 7.5 to 9 m; they are pierced by a series of openings giving access to the interior; most of them are only gaps in the rampart, but there are some which are well designed monumental gates; one of them (the Nausahra Gate) has been described supra.

These fortifications were built in successive phases, from the end of the 3rd c. to 50 B.C., early with the first construction of a mud rampart, afterwards topped by walls of baked brick at regular intervals, and subsequently raised by brick walls. They fell into disuse in the early centuries of the Christian era.

Ref.: EIA, vol. II, pp. 419-420; J. Ph. Vogel, ‘Excavations at Sahêth-Mahêth’, ASI- AR, 1907-08, pp. 81-131 and pl. XXXI, Nausahra Gate; J.H. Marshall, ‘Excavations at Sahêth-Mahêth,’ ASI- AR, 1910-11, pp. 1-24; plans: CSR, vol. I, pl. L, f.p. 330; J. Ph. Vogel, op.cit., f.p. 80.

No. 21. Ujjain (district of the same name) (see fig. 7, a)

Ujjain, well-known as the capital of Avanti, is situated on the bank of the Sipra river, in Madhya Pradesh. Excavations revealed a continuity of occupation divided into four successive periods.

During the first phase (750-500 B.C.), the rampart was built by the dumping of dug-up yellow and black clay to form a thick wall, having the shape of an irregular pentagon, covering an area of 100 ha, with its western side running alongside the Sipra river.

The enclosure, 74 m thick towards the east and 105 m (!) on the riverside, was protected by a ditch, 46 m wide and 8 m deep, connected at its extremities with the river for water supply. To protect the mud fort from erosion, it was strengthened by a network of logs of wood following its curved outline and serving as a buffer against floods. Several openings in the rampart appear to have been gateways; a fact confirmed by the exposure, to the north-east, of a road running through one of these holes.

During the second phase, the fortification was breached by floods at least three times; it was repaired by the construction of a brick revetment and raised a second time. The rampart survived as a city wall during the third phase (200 B.C. - 1300 A.D.), but ceased to be of utility during the last one (1300 - 1500).

Ref.: EIA, vol. II, pp. 447-449; IAR, 1956-57, pp. 20-28, 1957-58, pp. 34-36, pls. XXXIX, XLI; plan: IAR, 1956-57, p. 21, schematic section across rampart and moat, ibid., p. 22.

No. 22. Vaisali (district of the same name) (see fig. 7, b)

At Vaisali, near Patna, in Bihar, the ancient fort is of smaller dimensions. It forms a rectangle covering an area of 12.5 ha only, surrounded by a ditch with a width varying from 24 to 50 m.

Between 350 and 150 B.C., during the Maurya period, a mud rampart was erected, strengthened in the Sunga period with courses of mud brick; afterwards, a massive rampart about 21 m in width at the base and 4 m in extant height was made of rammed earth and, subsequently, in the pre-Gupta period, a brick rampart 2.4 m wide was constructed with military barracks made of brick.

Ref.: EIA, vol. II, pp. 457-458; T. Bloch, ‘Excavations at Basarh’, ASI-AR, 1903-04, pp. 81-131; K. Deva & V. Mishra, Vaisali Excavations: 1950, pp. 1-14; plans: ASI-AR, 1903-04, f.p. 88; K. Deva & V. Mishra, op.cit., fig. 1.

List of Maps and Figures

Fig. 1. Map of the fortified sites.

Fig. 2. Harappan Fortifications (A), Enclosures: a, after B.K.. Thapar, in D. Agrawal & A. Gosh, Radiocarbon and Indian Archaeology, p. 268; b, after R.E.M. Wheeler, in Ancient India, No. 3, pi. XV; c, after S.R. Rao, Lothal 1955-62, vol. I, pl. XXXVII; d, after J. P. Joshi, in D. Agrawal & A. Gosh, op. cit., p. 178. Note the plans quadrilateral in shape and the quadrangular towers.

Fig. 3. Harappan Fortifications (B), Gates: a, b, d, after A. Kesarwani, in B.B. Lal & S.P. Gupta, Frontiers of the Indus civilisation, figs. 7.4, 7.5). Note the flights of steps through the gates.

Fig. 4. a, River front site, trapezoid in shape, 322 ha (after CSR, vol. I, pl. XLVIII, f.p. 301); b, confluence head site, oval in shape, 176 ha (after CSR, vol. X, pl. XII).

Fig. 5. a, River front site, crescent-shaped, 370 ha (after EIA, vol. II, p. 284); b, river site, semicircular in shape, 150 ha, Nausahra Gate shown (after ASI- AR, 1907-08, f.p. 80).

Fig. 6. a, River front site, mango-shaped, 15 ha, late Medieval fortification represented (after C.S.R, vol. X, pl. XXIII; b, river site, polygonal in shape, 50 ha, western gate shown (after IAR, 1957-58, p. 7). Note that there are no salients.

Fig. 7. a, River front site, polygonal in shape, 100 ha (after IAR, 1956-57, p. 21); b, flat land site, rectangular in shape, 12.5 ha (after ASI- AR, 1903-04, f. p. 88); c. river site, square in shape, 132 ha, western gate shown (after Ancient India, No. 5, fig. 1, f.p. 66, pl. XXVII).

Fig. 8. a, Flat land site with rocky outcrops, oval in shape, 265 ha, citadel in the middle (after CSR, vol. I, pl. XLV, f.p. 271); b, flat land site with hillocks, 169 ha (after CSR, vol. 1 pl. XLIII, f.p. 257). Note that the fortifications represented are of a later period.

Fig. 9. Rajgir: dissected hill tract site, enclosures forming irregular quadrilaterals around old and new Rajgir, 200 ha and 26 ha respectively (after ASI- AR, 1905-06, f.p. 86).

Fig. 10. Sirkap: hill slope site, forming an irregular polygon, 73 ha; note the pentagonal towers at the corners; Sirsukh, flat land site, forming a parallelogram, 134 ha; note the semicircular towers (after J. Marshall, A Guide to Taxila, f.p. 132). The layout of the fortifications of Sirkap has been corrected according to the map given by A. Gosh in Ancient India, No. 4, f.p. 84.

Fig. 11. Dichotomy ditch-rampart: the Ujjain example (after IAR, 1956-57, p. 22).

Fig. 12. Sirkap: solid rectangular towers on the curtain walls, pentagonal towers at the comers (after A. Gosh, op.cit., f.p. 84); north gate shown (after J. Marshall, Taxila, vol. Ill, pi. 10).

Fig. 13. Sirsukh: curtain walls with loopholes, hollow semicircular towers (after J. Marshall, op.cit., vol. Ill, pi. 42.

Fig. 14. Rajgir: curtain walls and rectangular towers on the ridge (photo by M.H. Kuraishi & A. Gosh, op.cit., cover).

Fig. 15. Pataliputra: double wall made of wooden sleepers thought to be a part of the palisade seen by Megasthenes (after ASI- AR, 1926-27, pl. XXX, c and d).

Fig. 16. Sravasti: Nausahra Gate, double curve inwards so as to form two salients at both sides (after ASI-AR, 1907-08, pl. XXXI, f.p. 112).

Fig. 17. Sisupalgarh: west gateway, long and narrow passageway positioned between two rectangular blocks (after Ancient India, No. 5, p. 76, fig. 4).

Fig. 18. Types of fortification represented in bas-reliefs, (A): a, double row of pointed merlons which could be the extremities of the sleepers of a palisade (see fig. 15); b and c, serrated merlons; d, rectangular merlons; e, gate surmounted by a gallery (after J. Auboyer & J.F. Enault, op. cit., pls. 1, 1, 5; 36, 1; A.K. Coomara-swamy, Early Indian Architecture, pp. 217, fig. A, 219, fig. C).

Fig. 19. Types of fortification represented in bas-reliefs (B): a, rectangular towers; b, towers topped with a hemispherical dome; c, wall without battlement; d, towers with loopholes above the curtain walls; e, rectangular merlons (after J. Auboyer & J. Enault, op.cit., pls. 39, 6, 1, 4, 5; 1, 4; 41, 2; A.K. Coomara-swamy, op.cit., p. 214, figs. 13, 15, 17; A. Ray, op. cit., fig. 66).

Fig. 20. Types of fortification represented in bas-reliefs (C): a, quadrangular towers and rectangular merlons; b, hollow semicircular tower; c, towers pierced by sagital loopholes and merlons topped with a roof (after K. Coomaraswamy, op.cit., p. 184, fig. 13, p. 212, fig. 7; J. Ph. Vogel, La sculpture de Mathura, pl. XX, b).

Fig. 21. Types of fortification represented in bas-reliefs (D), Gandhara Art: a, b and c, gates with overhanging turrets; b and d, merlons topped with a roof; d, towers pierced by loopholes (after A. Foucher, L’art gréco-bouddhique du Gandhara, vol. I, figs. 229, 234, 288; for c, K. Fisher, ‘Gandharan Sculptures from Kunduz and environs’, Artibus Asiae, vol. XXI, pp. 234-235).

Fig. 22. Types of fortification represented in bas-reliefs (E), Gandhara Art: a and d, towers pierced by loopholes; b, gate with three overhanging turrets; a-d, merlons topped with a roof (after A. Foucher, op. cit., vol. I, figs. 292, 269, 235, 31).

NOTES

1 Revised version of my article entitled: ‘Etudes sur les fortifications de l’inde, 1. Les fortifications de l’Inde ancienne’, Bulletin de l’Ecole française d’Extrême-Orient 79.1, 1992, pp. 89-131.1

2 To simplify the bibliographical references, the following abbreviations are used:
ASI-AR: Archaeological Survey of India, Annual Reports (since 1902),
CSR: A. Cunningham, Archaeological Survey of India, Reports, 1862-1884,
EIA: A. Gosh (ed.), An Encyclopaedia of Indian Archaeology, 2 vols., 1989; (Volume II, Gazetteer of explored and excavated sites in India, presents alphabetically all the sites where excavations have been conducted and gives, for each one, an almost exhaustive bibliography),
IAR: Indian Archaeology A Review (since 1953-54).
To designate the projecting parts of the enclosures, archaeologists generally use the word bastions which corresponds to a system of fortification with low defence works; we prefer to adopt the word used by experts in fortifications of the 18th and beginning of the 19th century who called towers and not bastions the quadrangular or semicircular constructions flanking the curtains of the Indian defence works (see E. Lake, Journals of the Sieges of the Madras Army, passim).

3 Until the twenties of the last century, the only known example of defensive works prior to the Mauryas were the cyclopean walls of Rajgir in Bihar.

4 G.R. Sharma, Excavations at Kausambi, 1957-59. M.S. Mate, in ‘Early Historic Fortifications in the Ganga Valley’, Puratattava, No. 3, 1969-70, p. 60), rightly observes: “Information published on the remains at this site is available in plenty. However, it is piecemeal and no cogent picture emerges from it. This is mainly due to the failure on the part of the excavator to publish a comprehensive plan of the site indicating the mud walls and ramparts”.

5 EIA, vol. II, pp. 194-196; B.K. Thapar, ‘Synthesis of the Multiple Data as obtained from Kalibangan’, in D.P. Agrawal & A. Gosh, Radiocarbon and Indian Archaeology, pp. 264-271 and fig. 1.

6 R.E.M. Wheeler, ‘Harappa 1946: The Defences and Cemeteries’, Ancient India, No. 3, January 1947, pp. 58-130, particularly pp. 63-65 and pl. XV; J.M. Casal, La civilisation de l'Indus et ses énigmes, pp. 94-100.

7 J.M. Casal, op.cit., pp. 100-103.

8 EIA, vol. II, pp. 257-260; S.R. Rao, Lothal 1955-62, vol. I, pp. 135-136 and pl. XXXVIII.

9 EIA, vol. II, pp. 424-425; J.P. Joshi, ‘Excavations at Surkotada’, in D.P. Agrawal & A. Gosh, op.cit., pp. 173-181 and fig. 3.

10 EIA, vol. II, pp. 119-121 IAR, 1963-64, pp. 10-12 and pl. VII.

11 J.W.A. Fairservis, The Roots of Ancient India, pp. 179-182.

12 EIA, vol. II, pp. 45-46; R.S. Bisht, ‘Excavations at Banawali 1974-77’, in G.L. Possehl, Harappan Civilisation, a Contemporary Perspective, pp. 113-124; ‘Structural Remains and Town-Planning of Banawali’, in B.B. Lal & S.P. Gupta, Frontiers of the Indus Civilisation, pp. 89-97 and fig. 9.1 & 2.

13 The following paragraph is principally based on the observations made by Arun Kesarwani, ‘Harappan Gateways: a Functional Reassessment’, in B.B. Lal & S.P. Gupta, op.cit., pp. 63-73.

14 Ibid.

15 K.M. Srivastava (‘The Myth of Aryan Invasion of Harappan Towns’, in op.cit., p. 439) thinks that the walls were probably erected for social reasons (separation of the classes on the basis of occupation and profession): “Thus it was the class division, functional division and the distinction of the elite group from the rest which brought into existence the settlement of the so-called citadel".

16 A. Gosh, The City in Early Historical India, pp. 2, 11, 51, 61, 62-67, 80-81, 90.

17 M.S. Mate, op.cit., pp. 58. Regarding the rising of an urban civilisation and the growth of fortified cities in the Ganges valley, see, beside M.S. Mate’s study mentioned supra, the remarkable article by Federica Barba, ‘The Fortified cities of the Ganges Plain in the First Millennium B.C.’, East and West, vol. 54, No. 164 (December 2004), pp. 223-250. In this study, the author sets out to revise settlements dynamics and cultural development in the Ganges valley and consider different reconstructions concerning chronology and typology of the fortifications. She thinks in particular that, in terms of urban facilities, cities did not arise before the 4th century B.C. But, she does not consider the various elements of fortification from the point of view of military architecture and technology.

18 A. Ray, Villages, Towns and Secular Buildings in Ancient India, c. 150 B.C - 250 A.D., pp. 46-63, nagara-saniveśa, pp. 64-78, durga-sanniveśa; A. Gosh (in op.cit., pp. 43-58), in a chapter entitled: “The city in literature”, analyses the names of cities, their types, their characteristics and insists on the conventional character of the descriptions found in the texts.

19 See R.S. Betai, A Reconstruction of the original interpretations of the Manusmṛti.

20 Devī Purāṇa, ch. 72, 11.42-44; Agni Purāṇa, ch. 106, 11.8-10; Matsya Purāṇa, ch. 217, 11.24-28.

21 Arthaśāstra, II, 3; see The Kautiliya Arthaśāstra, ed. R.P. Kangle.

22 Mānasollāsa, ch. II, si. 544-584, cd. G.K. Shrigondekar.

23 Mānasāra, an Encyclopaedia of Hindu Architecture, ed. P.K. Acharya, vol. VII, pp. 226-229.

24 Mayamata, X, 36-61, ed. B. Dagens, Mayamata, 1st Part, pp. 152, 160-170. This information has been used in modem books such as B.B. Dutt, Town Planning in Ancient India, pp. 70-108; S.K. Joshi, Defence Architecture in Early Karnataka, pp. 121-150; P.V. Begde, Forts and Palaces of India, pp. 3-32.

25 In the bas-reliefs of Amaravati, Bodhgaya, Mathura, Nagarjunakonda, Sanchi, as well as those found in Gandhara. See A.K. Coomaraswamy, ‘Early Indian Architecture: I. Cities and City-Gates’, Eastern Art, pp. 209-235; A. Foucher, L'art gréco-bouddhique du Gandhara, vol. I, figs. 31, 229, 234, 235, 268, 269, 288, 289, 292; H. Ingholt, Gandharan Art of Pakistan, figs. 2, 103, 162, 464. The best restorations will be found in J. Auboyer & J.F. Enault, La vie publique et privée dans l'Inde ancienne, fascicule 1, L'architecture civile et religieuse, pp. 44-46, pls. 1, 36, 39, and in A. Ray, Villages, Towns and secular Buildings in Ancient India, c. 150 B.C.-c. 250 A.D., pp. 64-78, sketches and drawings, Nos. 60-72.

26 Arthaśāstra, II, 3.1; Mayamata, X, 36-38.

27 Arthaśāstra II, 3.21.

28 Mayamata, X, 13-14.

29 Mayamata, X, 42-43. According to Devī-Ayni Kālikā Purāṇa, it seems that square or rectangular forms were preferred (see B.B. Dutt, op.cit.. pp.102-105).

30 Arthaśāstra, II, 3; Mayamata, 36-48.

31 Kautilya (Arthaśāstra, II, 3) mentions expressly that the rampart (vapra) must be built with the piled-up earth (khāta) coming from the ditch.

32 “That the moat and rampart were the results of the same operation, namely the heaping up of the material scooped out to form the moat, has been recognised at all excavated sites where the rampart can be identified, except those where the defences are of brick or masonry." (A. Gosh, op.cit., p. 51).

33 Arthaśāstra, II, 3, 7, 10; Mayamata, X, 13-16, 44-46.

34 The same system existed in Burma where a double curve inwards in the rampart is also found. At Sri Ksetra, one of the ancient capitals of Burma, situated on the left bank of the Iravadi, in the 5th-9th centuries, the city, circular in shape, with a perimeter of 13 km, was also provided with gates, in the form of a double curving inwards ending with a double row of rectangular structures comprising towers and guardrooms (see Aung Thaw, Historical Sites in Burma, pp. 16-20 and in particular p. 19, plan of one of the gates.

35 Arthaśāstra, II, 3.7, 10; Mayamata, X, 13-16, 44-46.

36 Arthaśāstra, II, 3, 7; Śukranītī, I, 1.479.

37 See, supra, note 24.

38 We have tried to reproduce faithfully, though in a simplified manner, the defence works represented in Gandhara Art (figs. 21-22), laying stress on their main features (curtain walls, towers and gates). As regard the strongholds represented elsewhere in the great stupas, we found that the restorations made by modern authors are not faithful on certain points; for example, in the bas-reliefs of Mathura (see fig. 20, b, c), the sagital loopholes on the towers have been omitted and the merlons on the walls, wrongly described, since they are shown covered with a rectangular block, whereas they are in fact covered with a roof with four sides, similar to those shown in Gandhara sculptures.

39 Examples of plastered masonry or dressed stones at the basement of buildings are found at Bodhgaya (1st c. B.C.) (see J. Auboyer & J.F. Enault, op.cit., pp. 29-30, pls. 1,2, 3).

40 They correspond to the Rushan defence works built elsewhere and described by H.P. Francfort, Les fortifications en Asie centrale de l’âge du bronze à l’époque kouchane, pp. 31-37.

41 On the problems of datation and the question of the shape of the towers (quadrangular or circular), see G. Fussman, ‘Ruines de la vallée de Wardak’, Arts Asiatiques, vol. 30, pp. 89-92; F.R. Allchin & N. Hammond, The Archaeology of Afghanistan, pp. 263-266.

42 E.I.A., vol. II, p. 286.

43 A. Gosh, Taxila (Sirkap) 1944-45, Ancient India, No. 4, p. 44; J. Marshall, Taxila, vol. I, p. 1 14; H.P. Francfort, op.cit., p. 29.

44 P. Leriche, Fouilles d'Aï Khanoun, V, Les remparts et les monuments associés, p. 98.

45 See M.S. Mate, op.cit., pp. 58-69.

46 The brief descriptions of the sites given here arc based on the documents cited at the end of each note.