Although temperature is a less important variable than rain- fall for an appreciation of the human geography of South Asia, its bearing on both geography and history is considerable. To cite a single, perhaps oversimple example, one may assume that Europeans' aversion to tropical climates was among the principal reasons they made no large-scale attempt to settle any portion of the Indian subcontinent. Of those intrepid souls who did venture to India, following commercial, military, and other pursuits, high temperatures certainly took their toll— usually indirectly, as a key element in the etiology of tropical disease. Low temperatures, on the other hand, have been im- portant at various times in deciding the fate of military expe- ditions along and beyond South Asia's northern mountain fron- tiers (as with the ill-fated invasion of Tibet by Zorawar Singh in 1841). Conventional maps of temperature show means or mean maxima and/or minima for specific months (e.g., July and January). Maps (b) and (c) of plate III.C.1 vary from this pattern in that they show the mean maxima and minima for the hottest and coolest months, whatever they may be (i.e., from April to July for the maxima and December or January for the minima). While one cannot, therefore, derive from these maps any synchronous picture of the temperature at a particular time of year, one can ascertain the range of tem- peratures to which a resident in any part of South Asia is apt to be subjected during a typical year. Supplementing these maps are the data on monthly means, mean maxima, and mean minima for those fifteen selected stations for which climo- graphs have been provided.

The extent of climatic fluctuations within South Asia over the period of human habitation has been a subject for schol- arly speculation at least since the late 19th century. It is only in recent years, however, that scientific research has provided much empirical support for systematic statements on the gen- eral nature, if not always the magnitude, of changes at partic- ular periods of history. Direct evidence for such statements derives in part from geomorphic, stratigraphic, pedologic, and paleontologic studies in the field, as well as from the laboratory analysis of pollen indicating what types of vegetation grew near specific sites at times that can be dated by radiocarbon methods. Indirect evidence is based on the knowledge that from a climatological point of view the world is a single inti- mately interconnected system and that changes in one compo- nent of the system normally presuppose changes in others. A knowledge of the workings of the system therefore should en- able one to predict changes in one area, given data on specific changes elsewhere.

With regard to major long-term changes affecting prehistoric man during the Pleistocene epoch of geologic history, the reader is referred to the graph on plate II.1, which indicates the successive cold and warm periods, marked over much of the northern hemisphere by either widespread glaciation or a relative absence of glaciation. Although extensive continental glaciers presumably were nowhere in evidence in South Asia, there is abundant evidence of the repeated waxing and waning of mountain glaciers in Kashmir, and one may safely assume that much of the northern mountain wall of the subcontinent was similarly affected. Even in the present postglacial (or in- terglacial?) period, mountain glaciers are found in many parts of the Himalayan system, from Kashmir to Arunachal Pradesh, as well as in the Hindu Kush. In fact, some 40,000 square kilometers of Kashmir alone are still glaciated, the ice fields of that region being the world's largest outside the subpolar re- gions. Above the graph on plate II.1, a table indicates the vary- ing climatic phases of the late Pleistocene from roughly 95,000 to 10,000 years before the present. Although the table specifi- cally relates to the Mediterranean region, it should also be gen- erally applicable to northwestern South Asia, which is closely associated with it climatologically. The final column of that table, in fact, should be more or less applicable to South Asia as a whole. One should, however, be advised that the absolute chronology of both the graph and the table is highly approxi- mate and subject to revision as research progresses.

For the past ten millennia, evidence on climatic change is rather diverse and includes radiocarbon datings that permit much greater specificity about absolute chronology. Analyses of pollen deposits in three salt lakes in Rajasthan¹ point to a rather sudden ending of the prevailing arid conditions roughly 10,000 years ago, followed by a relatively humid phase lasting until about 2000 B.C. and a renewed period of aridity there- after (Singh [1970] and Singh et al. 1972). (The freshwater lakes dried up and became saline between c. 2000 and 1500 B.C.) Figure 1.7 indicates the inferred total annual and sum- mer monsoon rainfall based on the available pollen sequence and the radiocarbon datings therein. While the data shown re- late to only one of three lakes, the trends depicted are verified by the findings from each of the others.

The most striking new finding from the Rajasthan pollen analysis—one that may have a major impact on the study of neolithic domestication and the diffusion of agriculture—is that the advent of grain cultivation in South Asia appears to have occurred during a period of unusual wetness about 9500 B.P ("before the present," or roughly 7500 B.C.), which is virtually coeval with the earliest known grain cultivation in the Middle East. Conceivably, then, seed agriculture might have diffused from South Asia westward as easily as in the opposite direction, as was hitherto almost universally believed.² Simi- larly, it appears that the Indus Civilization had its beginnings not long after the onset of yet another period of wetness within the much longer period of generally humid conditions depicted on the graph.³ Further, one is struck by the temporal corre- spondence of the decline of the Indus Civilization and the sub- sequent absence of fixed human habitation in the Indus Plain (judging from the archaeological record), with the phase of desiccation setting in about 4000 B.P. (c. 2000 B.C.). Within the historical period the reestablishment of a fixed but precari- ous agricultural society seems to have been linked with a mod- est increase in precipitation.

For the past millennium a rather ingeniously derived set of estimates relating to rainfall in northwestern India has been made available by means of a regression equation derived by correlating known fluctuations of the temperature in Iceland with measured precipitation in the Rajasthan Desert (Bryson 1975 and oral presentations noted under Sources . . .). The unusually high correlation between the two is explained by the fact that when the mean summer position of the circumpolar vortex is displaced southward, causing colder weather in Ice- land, the belt of subtropical anticyclonic high pressure systems, typically associated with dry weather, does not extend as far north of the equator as it otherwise would; consequently, the moisture-laden summer monsoons blowing out of the Indian Ocean are weaker and penetrate less deeply into the subconti- nent. This particularly affects its arid northwestern regions. Although temperature measurements per se in Iceland do not go back very far in time, they have been recorded long enough for one to know, with some precision, their relationship to the southerly drift of icebergs, which has been recorded since c. A.D. 1000. Thus, temperatures inferred from the latter data can be used in the regression equation to generate a set of average rainfall figures for northwestern India, by decades, since that ¹ Sambhar, at 27°ree;N 75°ree;E; Lunkaransar, at 28°ree;30'N 73°ree;45'E; and Didwana, at 27°ree;20'N 74°ree;35'E. ² Recent research in Southeast Asia suggests even earlier begin- nings of agriculture or, more precisely, plant "tending" in Thailand; however, the domesticates were legumes, tuberous root crops, and other non-grasses, and the system of cultivation was not closely re- lated to that of South Asia and the Middle East. But even in the latter areas, grain cultivation does not necessarily imply sedentary settlement. Seeds could have been planted in favorable localities, left untended as the "cultivators" went about other nonsedentary pursuits (hunting or pastoralism or both), and harvested at the proper season with little more than a brief sojourn at the place of planting. ³ We have taken the graph essentially as presented by Bryson, whose timeline for the beginning of the Indus civilization makes it somewhat older than would our discussion and maps in section II. date. This brief sketch explains the data presented in figure 1.8 below.

Finally, with regard to the past century or so, opinions differ. Bryson (1975) perceives a slight recent worldwide warming trend over a period of several decades before the mid-1940s, with moister conditions in South Asia during that period, fol- lowed by a sharper cooling trend with progressively greater incidence of drought. However, in a careful analysis of data from widely scattered stations in India, some of it extending back more than a hundred years, Rao and Jagannath (1963) fail to discover any statistically significant recent rainfall trends.

Sources of Data for Plate I.C.1

General References

The Gazetteer of India (1965–); Kabul Times (1967).

Government Documents

Ceylon, Department of Census and Statistics (1950?–); U.S. Weather Bureau (1967).

Atlases

India, Meteorological Department (1906); India (Republic), National Atlas Organisation (1957); Nepal, Ministry of In- formation and Broadcasting (1966); Oxford school atlas for Pakistan (1959).

Other Books

K. S. U. Ahmad (1972); D. B. Carter (1954); C. A. Fisher (1964); H. Gaussen et al. (1967); J. Humlum (1959); L. Labroue et al. (1965); H. Ludwig (1953).

Selected Studies of Climatic Change

Books and Articles

S. M. Ali (1941); R. A. Bryson (1975); R. A. Bryson and D. A. Barreis (1967); K. W. Butzer (1971); J. E. Chappell (1970); H. De Terra and T. T. Paterson (1939); International Symposium on World Climate (1966?); R. L. Raikes and R. H. Dyson (1961); K. N. Rao and P. Jagannathan (1963); S. K. Seth (1963); G. Singh (1970, listed under Unpublished Works); G. Singh et al. (1972).

Oral Presentations

Reid Bryson and Waltraud Brinkman, "Historical climatology of Northwest India and its significance: Past patterns and future prospects" Part 1. "The last ten millennia;" Part 2. "The last one thousand years." Fourth Wisconsin Conference on South Asia, Madison, Wis., 7 November 1975.

I.C.2. Forested Areas and Natural Vegetation Types

Closely reflecting the pattern of climate is that of natural vegetation. It must be noted, however, that in few parts of South Asia does one encounter any considerable stands of forests that have not been significantly altered by human inter- vention. Hence the "natural" vegetation indicated on our map and the vegetation actually encountered in the field may differ substantially. The forests have been changed by man not only in type, but even more drastically in area covered. In the Republic of India today, only about one-fourth of the total surface is under forest, including much low scrubby growth of very limited use. The zonation of natural vegetation belts, however, would surely have shifted over time in phase with climatic alterations, the periodicity of which has been dis- cussed and partially illustrated in relation to plate I.C.1. The remaining forested areas, to the extent that they can be shown at the scale of our map, are portrayed on plate I.C.2. Further, wherever possible, woodlands are classified by density of forest cover.

Generally speaking, the types of forest and other wild vege- tation currently found in South Asia, particularly in areas long occupied by man, suggest climatic conditions drier than those actually encountered. This is so because repeated forest