By far the greater parts of Pakistan and western India are semi-arid climatic zones. Mean annual precipitation for virtually all of Pakistan is less than 20 inches per year and falls below five inches in the central part of the Indus Valley. Precipitation values increase slightly to the east where the effects of the southwest monsoon become more pronounced; however they still remain low. Significant yearly variation in precipitation also characterizes western South Asia. That is, the long-term rainfall mean is not made up of similar values but is rather the average of significantly different yearly rainfall totals. If it rains, it generally rains hard, but then man times rains completely fail. My own informants on the Saurashtran peninsula of Gujarat State related this factor to their food supply. They feel that within any given ten-year period they will have two or three famine years, four or five hard years and the remain ing will be bountiful.
This precarious rainfall pattern in terms of both absolute amounts and reliability suggests that devices for the control and conservation of water have probably been important features for food producers since agriculturalists first came into Pakistan and western India during the third and fourth millennia B.C. Of course sizable portions of this region are riverine flood plains. The five rivers of the Punjab join to a single stream near Multan where they move en masse to the sea. Sind or the lower Indus Valley would be a desert save for the annual floods of the Indus River. But even here water control problems exist and the populations who successfully mastered the Indus’ riverine environment in prehistoric times must have been skillful hydrological engineers. In the non-riverine areas there is abundant evidence for several types of earthen and/or stone embankments which serve to conserve and channel rainfall runoff and the accompanying alluvium. It is to these features that the following remarks are directed.
Earth and stone structures connected with agricultural practices have been reported from many areas in western South Asia. These range from the stone dams of Baluchistan and Kutch to earthen and brick bunds of Karachi, the Kachi Plain and in Sabarkantha and Bhavnagar Districts in Gujarat. In fact they are so widespread that many scholars have considered them commonplace and hardly worthy of mention. Few people have thought of them in terms of their history. We know that devices such as these are still constructed and used today, while others of them have been abandoned and appear to have a certain antiquity. I will direct my comments to the consideration of chronology after a further discussion of these features in some detail.
Whether considering the stone gabarbands of Baluchistan, the earthen damans in the vicinity of Karachi or bunds elsewhere, we are faced with features which serve essentially the same function: the conservation of water and soil. Thus, while the form of these structures varies considerably and the materials from which they are constructed can be stone, brick or earth, they are comparable.
The gabarbands of Baluchistan have been widely discussed and reported in considerable numbers (see Hughes-Buller 1903-04: 194-201; Stein 1931; Raikes 1965). According to Hughes-Buller the “gabars” are unbelievers or specifically Zoroastrians in modern Baluchi and Brahui parlance. Gabarb ands thus become “dams of the Zoroastrians.” He describes three different types: 1) double parallel dry-stone walls filled with rubble between them, 2) a solid masonry wall and 3) a solid masonry wall as in 2) but buttressed on both faces. As Raikes (1965:5) has noted, “The first technique is fundamentally different from the other two: the third probably represents a refinement of the second.” The inside or up-stream areas of these structures are always filled with rubble. This presumably lessens the force of flood waters before they make contact with the “dams” themselves. On plan, gabarbands come in many shapes and sizes. In fact, they are constructed so as to be in a position to catch runoff water and the suspended silts. Thus shapes tend to be determined to a great extent by local contours. The accompanying illustration shows four gabarbands near Toji-damb in northeastern Kharan District. The “L”-shaped structure to the north is quite typical for many of these facilities, and a schematic representation of it is presented here.
I have noticed in my own travels through Baluchistan that gabarbands rarely occur singly. Many times five or six of them line the nallas or seasonally active gullies of this area. Thus they form a series of “stairs.” Given the extremely barren nature of the region today, it is evident to anyone who visits them in the field that they were constructed as much for the preservation of soil as for water. By ponding the silt-rich runoff behind them, the local agriculturalists thus create rich alluvial fields which are seasonally and naturally irrigated.
Professor R. N. Mehta of the University of Baroda described some time ago a series of eighteen earthen, stone and brick bunds in Sabarkantha District on the north Gujarat plain. Known in Gujarati as pales, these structures were again situated so as to collect soil and be seasonally irrigated from local nallas. Palas are still utilized in Gujarat. I have personally seen them being constructed. But Dr. Mehta’s have by and large been abandoned. In fact one of them has constructed on it two small Hindu temples which can be dated to not later than the 15th century A.D. But I am already digressing to chronology and I have one further instance of pale construction to note.
In the course of archaeological exploration in Bhavnagar District, Gujarat I came across a large earthen enclosure near Khilosera Village. The bund itself is situated on an agricultural plain about four miles from a river. It is oriented so as to catch the precipitation runoff and alluvium from a moderately sized hill complex. The earthen enclosure is over one kilometer (0.62 miles) in length, and is on the average four meters (12 feet) high and approximately eight meters (25 feet) thick at the base. The area enclosed by this earthen structure is approximately thirteen hectares or thirty-two acres, although it catches rainwater from an area two or three times as large. It is quite evident from the structure itself and from the fact that alluvium has built up in the lower portions of this bund that its purpose was identical to the other constructions under consideration.
Chronology
These bunds, which are found in several forms, obviously play a significant role in the agricultural adaptation to a semi-arid region. Thus a chronology for their construction would lend much to understanding the culture history of the region.
The first point to be made in this regard is that bunds of this type have a considerable history. I have already noted that they are in use today. One of the Sabarkantha bunds can be securely dated to before the 15th century A.D. To my knowledge the earliest reference to a bund in this region is to be found in the vicinity of the well-known Ashokan rock edict at Girnar in southern Saurashtra. Ashoka ruled in the 3rd century B.C. (ca. 272-232 B.C.) from a capital on the eastern Ganges plain in north India. Whether his domains can be called an empire in the modern sense of the word is unclear; however, he had prepared a series of inscriptions which cover most of India, and extend as far west as Afghanistan. Although Ashoka’s inscription proper does not refer to the bund and Sudarshana Lake, they are noted in two later inscriptions. The earliest of these dates to the 2nd century A.D. and the other to the 5th century. These were carved by later rulers who attributed the original dam, which was several times damaged and rebuilt, to the reign of Chandragupta Maurya, Ashoka’s grandfather. The inscription of Rudradman (ca. A.D. 150] does however note Ashoka’s improvements to the structure. Thus we can be assured that large-scale irrigation facilities were being engineered and constructed at the opening of the Christian Era in western India. We have an early, although not prehistoric, base line for chronology of these structures in Gujarat,
As I noted above, the bunds in Sabarkantha can be dated at least to the 15th century A.D. based on the presence of the temple. I can further suggest that they are unlikely to be older than about the 2nd or 3rd century B.C. Extensive archaeological exploration in north Gujarat has failed to produce a single village farming community before the appearance of ceramic known as Red Polished Ware. This pottery type appears not to have been made before about 200 B.C. The other parts of Gujarat were inhabited by agriculturalists as early as approximately 2300 B.C., but north Gujarat was not, at least as betrayed by no small amount of archaeological exploration. It has been suggested (Possehl 1974) that this pattern of settlement has its roots in subsistence adaptation to dry cropping the fertile, moisture retentive black cotton soils. These black loams are found everywhere in Gujarat save for the northern plain which is a very different sandy alluvium. But explanation aside, intensive observation has failed to produce an agricultural settlement in Sabarkantha District which predates the early historic period. It is therefore unlikely that agricultural facilities on the scale of these pales predate the same epoch. Thus, at least some of the Sabarkantha bunds can be assigned to a period before the 15th century A.D. but not older than about 200 B.C.
R. N. Mehta in his discussion of the chronology for the bunds notes that several of them were made of brick or brick and earth. It is well known that brick sizes vary widely. Many times a particular size can be shown to have been manufactured for only a specific length of time and can therefore be used for dating purposes. Bricks similar in size and proportion to those used in some of the bunds “… were noted in different excavations at Baroda, Vadnagar, Devnimori, and Shamalaji where they were noted in the layers ascribed to the early centuries of the Christian Era.” (Mehta 1963: 363.) He also mentions two images of Mother Goddesses which on stylistic grounds could be assigned to the 5th-6th century A.D. But these were found lying on the ground in the vicinity of one of the pales and there is no firm association. That is, these female statues could have been there for many centuries, or arrived just the day before Mehta visited the site. However, the bricks, if they were not re-used in the construction of the bund, suggest a date surprisingly close to Rudradman’s Girnar inscription.
There is no stratigraphic evidence to date the Khilosera bund. It is known however that people associated with the Post-urban Phase of the Indus Civilization (ca. 1900-1300 B.C.) lived along the Kalubhar River about four miles from the bund. My reconnaissance in this area attempted to assess the criteria which people utilized when they selected settlement locations. In doing this I intensively explored the riverine areas close to the streams as well as the areas between them where this pale is located. I found that there were many prehistoric village farming communities along the banks of the two rivers which were explored. Yet on the fertile plain between them, which in this area is ten miles wide and approximately fifteen miles long, not a single settlement predating the introduction of Red Polished Ware was found. Since it seems unlikely that people would walk four miles to such a facility, the Khilosera bund cannot be assigned to a period before about 200 B.C. But pales are still in use in the area around Khilosera and it is possible that this particular feature is quite recent; although it is not presently in repair. It is, however, almost certainly not a prehistoric construction. In fact the broad agreement between the dates for the Girnar dam, the Sabarkantha bunds and that just discussed near Khilosera, suggests a time horizon in the vicinity of 300 B.C. or slightly earlier for introduction of this agricultural technique into Gujarat. In this region at least it appears not to be prehistoric. The situation is not as clear for the gabarbands of Baluchistan.
In a fine paper on gabarbands Robert Raikes (1965) has noted that their distribution appears to be coincident with the distribution of Nal, Kulli and Togau pottery types. He then suggests that: “It is probable therefore that the gabarbands go back to as early as Togau times, at the end of the fourth millennium B.C.” (1965: 11.) This would certainly extend the chronology of these facilities far beyond those found in Gujarat. There is however an alternative chronological hypothesis.
I want to stress from the outset that this alternative is just as hypothetical as that put forth by Raikes; however, it too deserves an airing. My hypothesis is formulated around three new thermoluminescent dates which were run in the laboratories of the Museum Applied Science Center for Archaeology (MASCA) at the University Museum. The ceramic type which was dated is called Ghul ware. It was excavated from the site of Kili Ghul Mohammad in the Quetta Valley, Baluchistan. This pottery is a stylistic variant of another pottery type known as Lando ware which has been widely reported in Baluchistan. Londo ware was originally dated to the early first millennium B.C. through a stylistic comparison with the pottery from Cemetery B at the famous site of Sialk in western Iran (DeCardi 1951). Two factors have changed that chronology. First, the thermoluminescent dates I received from MASCA range from approximately 400 to 1000 A.D. Second, a clear variety of Londo/Ghul ware has been found at Tepe Yahya in eastern Iran in what the excavator calls Partho-Sassanian levels assignable to the period between the 1st century B.C. and the 7th century A.D. (Lamberg-Karlovsky 1972). Given the fact that we are dealing with some stylistic variation and that the Quetta habitation could easily have begun later and ended later than that at Yahya, these two bits of chronological information are not so far out of alignment one from the other. It is also essential to note that the distribution of the Londo/Ghul ware is largely coincident with the distribution of gabarbands in both Iranian and Pakistani Baluchistan. In fact my own site records indicate that the association is better for the latter ceramic type than for the earlier Togau/Nal/ Kulli series which do not extend into Iran in their “pure” forms. In any event, the present discussion is a very general one which seeks to place the gabarbands in some reasonable time framework, and a century or two either way is certainly an acceptable margin of error at this point in the research.
There are two further bits of culture history which can be noted to support my proposed association of Londo/Ghul ware with the manufacturers of gabarbands. First, the Parthians and Sassanians were Zoroastrians or “non-believers” in Muslim eyes. Thus the preservation of the meaning “Zoroastrian dams” in gabarband may be a significant piece of linguistic data. Second, it is known that the Parthians and Sassanians made substantial investments in irrigation and other agricultural facilities in all parts of their empire. Many of the first qanat systems in Iran can be dated to these times. In Baluchistan gabarbands may have been just the thing to which the Persian engineer turned as an answer to increased agricultural production. At least, the widespread occurrence of these facilities in the eastern borderlands of the Partho-Sassanian empire is in certain agreement with what we know about their investments in agricultural production.
Based on these points I would tend to think that gabarbands in Baluchistan do not predate the 3rd century B.C. The association between these facilities and the distribution of prehistoric ceramic types and even sites (e.g. Toji-damb, Kulli, Fairservis’ Hab site, and the Harappan sites reported in the Malir oasis near Karachi, etc.) are discounted as spurious and non-specific. Some of these, for instance those in the Malir area, are in use today and it is therefore unlikely that they were originally constructed over 4000 years ago.
It may be coincidental but I am here struck by the fact that the bund chronology suggested for Baluchistan is in such close agreement with that developed for Gujarat. It still requires the manipulation of a hundred years or slightly more but the appearance of bunds in these two regions is quite close, at least if the reasoning used here has some substance.
There is certainly a good deal which could be done on these bunds in the way of new field work. First, accurate maps and drawings are required to chart their distribution, density and features of construction. It seems highly likely that rigorous documentation of this kind would vastly improve the credibility of all arguments put forth. Second, a comparison of the stone and brick construction of the bunds with that used for architecture within settlements might yield significant insights. If there is variation in dam building technique through time and if this variation can be tied to similar changes in other architectural sequences, then a very strong case for closely seriated chronology could be formulated.
But whatever the chronology, these agricultural facilities are clearly means of adaptation for the local farming population. They have been shown to have been a part of the agricultural system in western South Asia for many centuries, and this is sufficient documentation to demonstrate their historical importance.