Meteorology

The area dealt with in the present chapter includes the whole of India proper, with Burma and Baluchistan. It is not a self-contained area, as its meteorology depends very largely upon the oceanic area to the south and also, to a slighter extent, upon the regions to the north and north-west, more especially Central Asia and the Persian area or plateau. The object of this chapter is not only to give a statement of the larger seasonal weather changes, but also to indicate, so far as at present possible, their relation to each other, and their coordination to meteorological conditions and actions in the neighbouring areas of land and sea.

India probably presents a greater variety of meteorological conditions, actions, and features than any area of similar size in the world. The normal annual rainfall varies from 460 inches at Cherrapunji in the Assam hills, and from between 300 and 400 inches (probably) at suitably exposed positions on the crests of the Western Ghāts and the Arakan and Tenasserim hills, to less than three inches in Upper Sind. The largest rainfall actually measured in India in one year was 905 inches, at Cherrapunji in 1861, while at stations in Upper Sind it has been nil. A rainfall exceeding 25 inches within twenty-four hours is of occasional occurrence, and falls exceeding 15 inches are comparatively frequent. At one period of the year parts of India are deluged with rain; at another persistent dry weather with clear skies prevails for weeks or months. During the rains the air is almost supersaturated with moisture in some of the coast districts and in the hills, while in the hottest weather it is occasionally so dry in the interior that the methods employed for calculating humidity in Europe have given negative and hence impossible results. The coasts are occasionally visited by cyclones fiercer and more concentrated than have probably ever occurred in Europe. These bring up storm-waves that sweep over the low coast lands of Lower Bengal or the deltas of the Mahānadī, Godāvari and Kistna, destroying the crops, and drowning the inhabitants by tens of thousands. In one season of the year India is the scene of the most wonderful and rapid growth of vegetation; in another period the same tract becomes a dreary brown sunburnt waste, with dust-laden skies and a heated atmosphere that is almost unendurable even by the natives of the country. The transition from the latter to the former phase over the greater part of the interior often occurs in a few days. In one year the rains may be so distributed as to cause a severe and extensive famine over several Provinces, necessitating for months afterwards the continuous relief of millions of the population; in another the meteorological conditions may be so favourable that the crops far more than suffice for the normal food demand.

India again presents a noteworthy combination of tropical and temperate region conditions. Tropical heat, heavy and frequent rain, and fierce cyclones are prevalent at one period of the year; while moderate temperature and rain, with shallow extensive storms—conditions resembling those of South-eastern Europe—obtain at another.

In the third place India is par excellence the area in which the contrast of what are termed monsoon phases or conditions is exhibited most strongly over a large area. These conditions are the prevalence of dry land winds, with little cloud and rain, during one half of the year, and of winds of oceanic origin, with high humidity, much cloud and frequent rain, during the other half.

The work of meteorological observation was begun by the East India Company at the end of the eighteenth century. Observatories of the first rank were established at Madras in 1796, at Simla in 1840, at Bombay in 1841, and on the Dodabetta Peak (Nilgiris) in 1847. The observatories at Simla and Dodabetta were closed after a few years’ work, but those of Bombay and Madras have been continued until the present time and have furnished most valuable series of observations. Observations (chiefly of temperature and rainfall) were also taken at many hospitals, but in many cases these were carelessly recorded and are of little scientific value. The progress of meteorological inquiry in England, and its utilization for the purposes of storm warning after the Crimean War, suggested the commencement of systematic meteorological observation in India. Isolated and independent Meteorological Departments were started by four of the Provincial Governments—in Bengal in 1865, in the United Provinces in 1864, in the Punjab in 1865, and in the Central Provinces in 1868. The Bengal Meteorological Department was initiated chiefly for the purpose of conveying warnings to the port of Calcutta; the other three for supplying information to the local medical authorities, in the hope that it might enable them to trace out the relations between weather and disease. These local departments were of some service in collecting meteorological data, but they were found to be of little use for the investigation of the larger phases and changes of weather in India. The Government of India accordingly decided to imperialize the system, and sanctioned the necessary arrangements for the extension of the work of observation to the whole of India, for the adoption of uniform methods of observation, and for the systematic discussion of the observations as a whole. The Imperial Department thus formed was placed under the control of a scientific officer with full powers to carry out the sanctioned reforms. This change was effected in 1874-5, and the present system has thus been in operation for thirty years. At the end of 1902 the following observatories were in existence:—

A magnetic and meteorological observatory at Colāba, Bombay (of the first rank).

A solar physics, magnetic, and meteorological observatory at Kodaikānal, on the Palni hills in Southern India (of the first rank).

An astronomical and meteorological observatory at Madras (of the first rank).

A central meteorological observatory (of the first rank) at Calcutta (Alipore), where all instruments are tested before issue and their corrections to the India standards determined.

231 meteorological observatories (of the second and third rank) maintained by the Government of India or by Native States working in co-operation with the Government. Twenty-three of these are mountain observatories situated at elevations exceeding 5,000 feet, the highest being those at Leh in Kashmir (elevation 11,503 feet), at Kyelang (10,087), and the Chaur Peak (11,200) in the Punjab.

2,390 rain-gauge stations, recording and reporting rainfall only.

The gazetted staff of the Department includes—in addition to its head, who is Director-General of Indian observatories—the Directors of the Kodaikanal and Colāba observatories and the First Assistant Meteorological Reporter to the Government of India (all of whom are whole-time officers); the Provincial Meteorological Reporters at Allahābād, Bombay, Calcutta, and Madras, of whom the three latter issue Provincial Daily Weather Reports; and an Assistant Meteorological Reporter in charge of the Alipore central observatory. The five officers last mentioned are half-time officials, who hold other appointments in the Educational or Telegraph departments¹.

For many years the Indian region, including India proper, Burma, the Arabian Sea, and the Bay of Bengal, was considered as an independent meteorological area, in which the weather was supposed to be determined chiefly, if not solely, by the conditions within that area. It was assumed that India was protected on the north by the lofty barrier of the Himalayas, and on the west by the moderately high range of the Sulaiman, from the cold winds coming from northern regions, and that it was only exposed to the influence of equatorial sea currents. The presence of this northern mountain barrier does undoubtedly exercise a very considerable influence on the meteorology of India and more especially of the Indo-Gangetic plain; for a comparison of the temperature data of Northern India with those of the south and centre of the United States in the same latitudes indicates that the intervention of the Himalayas increases the temperature of the Indo-Gangetic plain from 3° to 5° above what it would have been if a low-level plain had extended northwards to the Arctic regions. Nevertheless, as will be seen later on, meteorological conditions in India generally are very largely determined by outside influences.

The physiographical and geographical features of India are of great importance, in so far as they modify more or less considerably the lower air movement, and hence the distribution of temperature, pressure, humidity, and rainfall; and it is necessary to bear them in mind in any scientific discussion of the meteorological conditions and actions of that country. India is the middlemost of three great Asiatic peninsulas which project southwards into the Indian Ocean, and which are more or less dependent on that ocean for their broader meteorological features. It consists of a peninsula proper (to the south of latitude 22° N. or the Tropic of Cancer), and of a broad low alluvial plain the axis of which runs east and west.

The Peninsula is of comparatively low elevation and has a backbone of hills, near the west coast, from which the land slopes slowly eastwards². To the north of the Peninsula is the low plateau of Central India, gradually falling to the extensive Indo-Gangetic plain, which nowhere rises, except in the immediate neighbourhood of the hills, above 800 feet. To the north of this extensive plain is the lofty continuous barrier of the Himalayan mass, the central axial range of which averages over 20,000 feet in elevation. Farther north is the elevated Tibetan plateau, an extensive area 2,000 miles in length from east to west and 200 to 500 miles in breadth from north to south, averaging over 10,000 feet in altitude. The continent thence falls northwards by a succession of slopes to the Arctic Ocean, to which it presents a vast low unbroken plain similar in general character to that of the corresponding plain in North America. Any general air movement on the Tibetan plateau would, if it extended into the plains of Northern India, have to rise 10,000 feet and then be precipitated in cascade form over the Indian hills. There is no evidence of any such general movement in the lower atmospheric strata. Actual observations indicate that the air motion over the Western Himalayas is mainly an interchange between the hills and plains due to local actions and conditions, and is not the continuation of general air movements over Central Asia, or the Indian Ocean and its two arms, the Arabian Sea and the Bay of Bengal. Any such general air movement is confined to the middle and higher atmospheric strata.

Again, the north-west frontier of India proper consists of a series of hill ranges running north and south, forming the escarpment of a plateau stretching westwards for at least 1,000 miles into Western Persia. The average elevation of these ranges north of Jacobābād is at least 6,000 feet, and of the greater part of the plateau to the west 4,000 to 5,000 feet. The lower air movements upon that plateau are not at any period of the year continued over the frontier ranges into Northern India. There is a drift down the passes, but it is insignificant. Hence the important conclusion that India is, in its lower air movements, cut off towards the north, north-east, and north-west. Northern India is at one season the goal of the sea winds that blow in the lower strata, and at another season the source of the land winds that blow from it to the neighbouring seas; but the origin of these movements is not to be found directly in the heating and cooling of the large central area of Central Asia, as has been stated by some meteorologists.

It will, however, be shown later on that the cold weather rainfall in India is largely dependent upon storms which originate in the higher atmosphere to the north-west of India, and that the distribution of the monsoon rains is in part dependent upon conditions in the distant Indian Ocean, and in part upon conditions in East Africa, the Persian area (probably), and perhaps to a slight extent Central Asia.

The primary fact in the meteorology of India is the alternation of seasons known as the south-west and north-east monsoons. If the Indian Ocean extended northwards over the whole or greater part of Asia, the air movement, and meteorological conditions generally, over the area between 35° S. and 35° N. lat. would be similar to those obtaining in the corresponding belts of the Atlantic and Pacific Oceans. Northeast winds would blow steadily and permanently over the northern, and south-east winds over the southern half. Between these two systems of winds there would be a narrow belt near the equator, which would move slightly northwards or southwards during the year, and in which variable winds, with frequent squalls and showery weather, would prevail. These wind systems in the Pacific and Atlantic Oceans are due to the large and permanent differences of temperature between the tropical and polar regions. Over the belt of high temperature in these oceans near the equator there is a more or less continuous ascensional movement, and an outflow in the upper atmosphere northwards and southwards with an indraught from north and south in the lower strata. Pressure is lowest in the area of squally weather and variable winds near the equator, and increases northwards and southwards to about 35° or 40° N. and S. lat., near which latitudes high-pressure belts or narrow elongated areas lie over these oceans. These pressure features are as permanent as the trade winds.

The presence of the Asiatic continent, extending from the neighbourhood of the equator northwards, modifies the air movement and pressure conditions very largely in the Indian Ocean and seas, and converts, in the northern portion, a permanent into a periodic, or monsoon, air movement. During one period of the year Central and Southern Asia is much cooler (from 15° to 25° in the winter months), and during the remainder of the year its temperature is considerably higher (probably from 10° to 15° in the summer months), than the Pacific and Atlantic Oceans in the same latitudes. These are the fundamental facts of temperature that underlie the explanation of the monsoons in India.

The north-east monsoon is a season of winds of continental origin, and thus of great dryness except where they have passed over some extent of sea, and hence in India of clear or lightly-clouded skies and of light occasional rain, due either to feeble depressions or to local disturbances. The south-west monsoon, on the other hand, is a season of winds of oceanic origin, of high humidity, and of frequent and heavy rain over nearly the whole area. The average annual rainfall of India, as determined from the returns of about 2,000 stations, is 45 inches, of which nearly 90 per cent. falls during the south-west monsoon. From this point of view the terms dry and wet monsoon seem to be more appropriate, as suggesting the most prominent and important feature of the seasons in the land area of India. The dry monsoon, which lasts from about the middle of December to the end of May, may be divided into two periods, one of comparatively low, and the other of increasing and high, temperature. The former, the cold weather of Northern India, includes the months of January and February, and the latter extends from March to May. The wet monsoon may also be subdivided into two periods. The first, from June to September, is the monsoon proper, when general and heavy rain falls occasionally or frequently in nearly all parts of India. The second is the retreating monsoon period, when the area of rainfall contracts southwards, and the intensity and volume of the fall diminish, until the humid currents pass out of the Indian land and sea areas and withdraw to the equatorial belt: this period extends from October to December.

The north-east monsoon air movement is fully established over the Indian land and sea areas by the beginning of January, when temperature is lowest in the Asiatic continent. A belt of high pressure and anti-cyclonic conditions then stretches from the West Mediterranean to Central Asia and North-east China. This corresponds to the belts of high pressure in about latitude 30° N. in the Pacific and Atlantic Oceans, but lies farther north. Pressure decreases southwards to a belt a little to the south of the equator, and thence increases again to a broad belt stretching from South-west Australia across the South Indian Ocean to the Cape. The northern belt of high pressure appears to separate completely the region traversed by the storms of Northern Europe (most of which enter from the North Atlantic) from the areas of the local storms of the Mediterranean Sea, Persia, and India, and limits to the north what may be termed the Indian monsoon region. The pressure conditions evidently accompany and determine northerly winds (deflected to north-east in consequence of the earth’s rotation) in the Indian seas, and south-easterly winds in the centre of the Indian Ocean. The winds in the land area are modified by the trend and elevation of the mountain ranges, and by the direction of the mountain valleys. Temperature begins to rise in general in the month of February over the Asiatic continent, but the changes are usually small until the end of that month. Hence the normal air movement and pressure conditions are practically the same during these two months over the whole area between lat. 40° S. and 35° N., and the air movement in that large area virtually consists of two independent circulations. The rough sketch below shows the general character of the pressure conditions and air movements in these areas.

Fig. 1. Sketch of air movement in India during cold weather period.

The meteorology of India during this period is only concerned with the actions and changes in the northern circulation. The Central Asia high pressure or anticyclone, although intense, is ineffective so far as the lower air movement in India is concerned, as it not only lies over very elevated ground but is shut off from India by the Himalayan barrier range.

Clear skies, fine weather, low humidity, large diurnal range of temperature, and light land winds are the characteristic features of the weather in India during this period. These conditions are as a rule initiated in the Punjab in the beginning of October, extend slowly eastwards and southwards, and finally prevail over the whole Indian land and sea area to the north of lat. 8° N. before the end of December. The months of November and December are, on the whole, the pleasantest of the year in Northern India. The air movement in Northern and Central India is from the west down the Gangetic plain, recurring across Bengal through north-west and north, and in the Bay of Bengal, where it is from north-east to east. This latter movement is continued across the Peninsula from the east, and passes out into the Arabian Sea. The west coast districts from Gujarat southwards and the adjacent sea area are, however, protected from this general movement by the Western Ghāts, and hence light to moderate land and sea breezes obtain in a well-marked form in these districts. The most noteworthy feature of the air movement in India at this time is that it is from practically opposite directions in Northern India and in the Peninsula, viz. from the west in the former and from the east in the latter area. The intermediate belt, which includes the greater part of Khāndesh, Berār, and the southern half of the Central Provinces, is, as might be expected under these conditions, a region of light unsteady variable winds. It is also the seat of the largest deviations from the normal weather conditions (more especially of temperature), to which attention will be directed later.

The most interesting feature of the northern wind system at this time is the return of the upper air current over India from the south (deflected to the south-west and west-southwest by the earth’s rotation). It is the continuation of the ascensional movement of the humid currents in the equatorial belt, which have previously passed over a large extent of sea surface, but which, by the action of ascensional movement, have been deprived of a considerable portion of their aqueous vapour in that belt. This current is not fully established until the end of December, by which time the circulation of the south-west monsoon has been fully replaced by that of the north-east monsoon.

These facts explain to some extent the character of the storms and disturbances of the period in Southern Asia. A succession of shallow storms passes eastwards across the Persian area and Northern India during this period. These storms are not the continuation of European disturbances, and are land-formed. The precipitation accompanying them is small in amount over the greater part of the area they cover, but is very large on the higher elevations of the mountain ranges to the north, including the Hindu Kush and Himalayas. A unique feature of these storms is the formation of brief subsidiary or secondary depressions in the Punjab, accompanied by moderate to heavy rain in the plains, and by violent gales and heavy snowstorms in the middle and higher Western Himalayas.

These storms are attended with remarkable temperature effects. Temperature usually rises very considerably in front of them, owing in part to the presence of cloud which diminishes largely the fall of temperature by night, and in part to the shift of winds to easterly and southerly directions. In the rear of the storms unusually dry clear weather obtains as a rule, with stronger and cooler westerly winds than usual. This change is, of course, emphasized by the previous rise of temperature over the same area when it was in the advancing quadrant of the storm. Temperature occasionally falls as much as 20° or even 30° in forty-eight hours in Baluchistan and Northern India owing to the passage of these warm and cold waves. It should, however, be noted that these temperature changes are more directly related to the extent of snowfall in the neighbouring mountain and plateau areas than to the intensity of the storms.

The few cloud measurements that have been taken at Allahābād and Simla during these storms establish that the cloud, and hence the field of condensation, in these storms is at an elevation of 15,000 to 20,000 feet. The distribution of the precipitation, and the occurrence of stormy winds and weather, not in the plains, but in the mountain districts of Northern India, indicate clearly that the storms are disturbances in the upper humid, and not in the lower dry land, current. The great majority of them originate either over the Syrian desert area or over the Persian plateau, and usually begin to affect Baluchistan and North-western India in December. They are, in their full development, a phenomenon of north-east monsoon or cold-weather conditions only, and are, compared with the storms which visit Northern Europe at this time, very feeble disturbances. The barometric depression in the primaries rarely exceeds two-tenths, and in the secondaries five-tenths, of an inch. The number and character of these storms vary very largely from year to year. The conditions determining these variations have not yet been fully investigated; they appear to have no relation to the local conditions in the lower atmosphere over India.

In Northern and Central India periods of bright sunny weather, with clear or lightly clouded skies, alternate with periods of cloudy showery weather accompanying the passage of the shallow extensive cyclonic or cold-weather storms of the period; and light to moderate showers fall in the plains of Northern India during the advance or passage of the storms. In Rājputāna, Central India, and the Central Provinces the rain is much less frequent and more local in occurrence, and usually accompanies thunderstorms. Occasionally severe hailstorms occur during the passage of these cold-weather storms across Rājputāna and Central India, and do much damage to the poppy and other crops. The influence of these storms rarely extends to the south of the hill ranges lying across the head of the Peninsula. On the other hand, the north-easterly to easterly winds which blow across the Coromandel coast districts are more or less largely charged with vapour, due to their passage over a considerable extent of sea area in the Bay, with the result that some cloud prevails in Southern India during this period, especially over and near the hills. The intermediate belt, including the Konkan and the Deccan, has less cloud than any other part of India at this time. The cool north-west winds in the rear of the storms advancing from Persia and passing over Northern India occasionally extend across the north-east of the Arabian Sea, and reduce temperature considerably as far south as Goa and Kārwār.

During this season the rainfall is very small in amount over Lower Burma, and practically nil over the greater part of the Peninsula. The rainfall in Northern India is greatest in the submontane districts and decreases in amount with increase of distance from the hills. The chief feature of the precipitation is the heavy snowfall in the Western Himalayas and the higher mountain ranges of Afghanistan and Baluchistan. It increases rapidly with elevation and probably exceeds on the average 30 feet at 15,000 feet of altitude, and varies between that amount and at least 60 to 100 feet over the higher ranges.

Fine weather prevails generally in Southern India at this period; but occasional showers, frequently accompanying thunderstorms, are received in the coast districts, and on the Palni, Nilgiri, and Anamalai hills, usually immediately after the breaking up or disappearance of the more vigorous cyclonic storms of the period in Northern India, when strong north-east winds flow down the Bay and are continued as easterly winds across the Coromandel coast. This is, it may be pointed out, one of the many examples of a marked contrast of conditions between northern and tropical India. Higher temperature or cloudy weather in one area often accompanies lower temperature or weather finer than usual in the other.

It has been pointed out that there is a debatable tract between the areas of westerly winds in Northern and Central India and of easterly winds in the Deccan. During the period from November to February the rainfall in this tract, which includes Khandesh, Berar, and the southern half of the Central Provinces, is usually very scanty. In normal years heavy fogs or mists occur nightly, contributing largely to the progress of the great wheat crops of these districts; but in years when the cold-weather rains are late and scanty, temperature rises very largely above the normal in this area and fogs are then comparatively rare.

The second half of the dry season includes the months of March, April, and May. During this period increase of temperature in the northern region, more especially in the Indian land area, accompanies a fairly continuous decrease of pressure; and decrease of temperature in the southern region, including the Indian Ocean and adjacent land areas of Africa and Australia, causes a slight but continuous intensification of the southern anticyclone referred to on page 110. These pressure changes accompany corresponding changes in the two air movements. The south-east trades movement increases in intensity and volume, and the northeast trades movement not only decreases, but is to some extent broken up, by the establishment of local circulations in India, and perhaps in Arabia and East Africa. An important feature in connexion with these changes is a slow transfer northwards of the equatorial low-pressure belt. This northward advance is probably due chiefly to the strengthening of the southern, and in slight part to the decrease of the northern, circulation. It is noteworthy that during this period the area of general rainfall due to the south-east trades is transferred northwards from South-east Africa to the Equatorial Lakes region of that continent.

Meteorology of the second half of the dry or north-east monsoon.

The conditions existing during the first half of the dry monsoon season begin to be modified over the whole land and sea area of India in the latter half of March, owing to rapid increase of temperature. North-east winds continue in the Arabian Sea and Bay of Bengal, but fall off in strength, and in April and May light winds obtain in the centre and south of the two seas, more especially in the Arabian Sea. Very large and important changes in the pressure, temperature, and humidity conditions are initiated in the interior of India during these two months. Temperature increases rapidly and steadily³, and pressure diminishes pari passu, over the heated land area, owing to the transfer of air to neighbouring cooler regions, more especially the sea areas to the south. India thus becomes a low-pressure area relative to the adjacent seas. The depression is very slight at first, but increases in intensity with the advance of the season. The most important local features in this general depression are the formation and development of a deepish low-pressure focus or sink in the hottest area, including Upper Sind and the adjacent districts of the Punjab and Rajputana, and the formation of a shallow depression in Chotā Nāgpur. There is thus increasing local indraught from the adjacent seas across the Bengal and Bombay coasts, which is, however, only clearly marked over comparatively narrow belts round the coasts landwards and seawards. These belts slowly increase in width with the season; but in the Arabian Sea the belt, even in May, probably does not at the utmost extend more than 300 miles seawards, thus leaving a large area of light northerly or variable winds in the centre and south of that area. In the Bay of Bengal, on the contrary, the local sea winds extend in May over the greater part or the whole of its area, in consequence chiefly of its comparatively small extent. These changes convert the air circulation in India and the adjacent seas from a general into a local movement, marked by strong hot winds down the river valleys in Northern India as well as by the strong sea winds above mentioned. It is no longer part of the previously established north-east trades, nor is it, as some have maintained, the commencement of the south-west monsoon. The period is undoubtedly a transition from the general air movement of the north-east to that of the south-west monsoon, but the final transformation occurs under conditions, and in the manner, explained later.

In normal years, India is not disturbed during the hot weather months (March to May) by storms appearing in Central or Western Asia⁴, for this is the only period of the year, it should be remembered, in which India is, to a large extent, an independent meteorological area. It is in consequence of this isolation that the storms of this period are local in character. They are due to various large actions and contrasts of conditions accompanying high temperature, large diurnal range of temperature, and great contrasts of humidity. These storms all belong to a class which may be termed hot-weather storms. They include the dust-storms of the dry heated interior⁵ (the Punjab, Sind, Rajputana, and the Gangetic plain), and thunderstorms in the regions in which there is more or less interaction between the damp sea winds and the dry winds from the interior, viz. Assam, Bengal, Burma, the west coast districts, and the Deccan. Hailstorms are more common in Assam than in any other part of India. They also occur frequently in Central India, and occasionally in Bombay, Rajputana, the United Provinces, Bengal, the Central Provinces, and Sind. They are of very occasional occurrence in Upper Burma, Hyderabad (Deccan), and Kathiawar, and rarely, if ever, occur in Mysore, South Madras, Malabar, the Konkan, and Lower Burma. The interaction of the dry and damp winds is, in Bengal and Assam, supplemented by the action of the hills in giving rise to vigorous forced ascent, and these thunderstorms, or ’nor-westers’ as they are locally called, are of frequent occurrence, and occasionally of great violence and intensity. Sometimes they develop into tornadoes, the most intense form of small revolving whirls. These are rarely more than a few hundred yards in diameter, but advance rapidly, overturning and destroying houses and trees, and causing great destruction of life and property. Their path of destructive violence is seldom more than 5 to 10 miles in length. They are comparable in violence and intensity with the tornadoes of certain districts of the United States. The rainfall due to these hot-weather storms is very important in North-eastern India, especially in Assam, where it is of great service for the tea crop.

These storms, which are peculiar to the hot weather, are evidently due to the vigorous convective air movements of the period, and usually occur in the afternoon hours when these movements are strongest. Their frequency and strength increase, as might be expected, with the advance of the season, and with the increasing intensity of the thermal conditions and actions. This is very clearly shown by the following data for the rainfall of the period in Bengal and Assam:—

The amount of aqueous vapour in the air increases slowly, but fairly steadily, during this period, owing to the increasing influence and extension into the interior of the sea winds. This increase is most marked in the coast districts. In the interior the humidity percentage decreases considerably in consequence of the rapid increase of temperature, and the air is as a rule exceedingly dry in the afternoon hours during the months of April and May.

We have next to deal with the south-west monsoon. In the first or cold-weather period the weather in India is mainly determined by continental, and in the second or hot-weather period by local, conditions. During the south-west monsoon the weather is determined chiefly by oceanic conditions, which we have now to consider.

The first portion of the wet season, from June to September, witnesses the extension of the south-west monsoon currents over the whole of India and their persistence during these months as rain-giving currents. This is the season of the south-west monsoon proper. It is the most important season of the year, as it is upon the amount and distribution of the rainfall of this period that the prosperity of at least five-sixths of India mainly depends.

Accompanying the rapid increase of temperature in the Asiatic land area in May, pressure generally decreases, so that at the end of May the Asian winter anticyclone has been replaced by a deepish low-pressure area, the axis of which stretches from the Sūdān to Sind and Western Rājputāna and thence to Bengal, and is hence farther south than the median line of the winter anticyclone. A graphical representation of the chief pressure conditions in India and the Indian Ocean in May is given below. As compared with that given on p. 111, it indicates the replacement of a stable by an unstable system. Finally by a spasmodic and sudden burst, due chiefly to the momentum of the current and the diminution of resistance, and hence to vis a tergo and not to indraught towards the low-pressure area in India, the equatorial belt closes up, and the south-east circulation extends northwards across the equator into the two areas of the Arabian Sea and the Bay of Bengal, whence it advances more or less quickly into the neighbouring land areas of Burma, India, and Abyssinia.

Fig. 2.

This advance in many respects resembles, on a large scale, the extension of the south-east trades across the equator into the Indian sea and land areas. A characteristic feature of the advance is that often a cyclonic storm forms in this area. When it occurs the most characteristic feature of the advance is the formation of a cyclonic storm in the areas. The storms in the Arabian Sea usually advance parallel to the coast until they reach the latitude of Bombay, when they generally recurve to the west and march to the Arabian coast, where they break up. Occasionally they form farther west and advance past Sokotra into the Gulf of Aden. These storms are on the whole the most dangerous that are encountered by mariners in the Arabian Sea. The most disastrous in recent years was the Aden cyclone of May 30 to June 3, 1885, in which the English SS. Speke Hall, the French man-of-war Renard, the German corvette Augusta, and numerous native craft were lost with all hands. The storms in the Bay of Bengal, unlike those of the Arabian Sea, advance landward and usually strike the Bengal or Orissa coast and thence march in directions between north and west, giving more or less general rain⁶. If the humid currents are strong they quickly spread over the whole of India, and the transformation from the hot weather to damp cloudy rainy weather is effected in a very short period. During the next three or four months the horizontal air movement over the Indian Ocean and seas into the adjacent land areas is converted over India into an irregular ascensional movement accompanying rainfall, the actions connected with which not merely maintain the low pressure established by the high and increasing temperature in May, but intensify it to some extent in June and July. The figure on the opposite page represents the lower gradients and air circulation in June, July, and August over the area in question.

When this current is established over the Indian seas and adjacent land areas, a continuous air movement extends over the Indian Ocean, the Indian seas, and adjacent land areas between latitudes 30° S. and 30° N., the southern half of which is the south-east trades and the northern half the south-west monsoon current. The strength of this movement differs very considerably in different parts of its area of extension, owing to various causes. The most important fact is that it is a continuous horizontal air movement in the lower atmosphere that occurs mainly over an extensive tropical oceanic area, in the greater part of which it is passing from cooler to warmer areas and in which weather is generally fine with clear or lightly clouded skies. Hence when it enters the Indian seas and approaches the coasts of Southern Asia the air is highly charged with aqueous vapour.

The south-east trades, previous to their extension across the equator, are largely directed to the African coast, and give more or less general rain, during the dry monsoon period of India, in South and East Africa. During April and May the rainfall occurs most largely in the Equatorial Lakes area and the south-east.

Fig. 3.

Adjacent districts. This rainfall produces the first seasonal rise of the Nile in May and June. The extension of the south-east trades as south-west winds over the Arabian Sea is followed by a diversion of a part of the monsoon current to Abyssinia and the Sūdān. The rainfall in the mountain districts of Abyssinia is similar in period and character to that of Northern India. Sir W. Willcocks estimates that an average of 60 to 70 inches falls over that area from June to September. This gives rise to a much higher and greater flood in the Nile than does the rainfall in Central Africa. The Nile floods hence reach their culmination in September.

Arabia, so far as is known, shares only slightly in the Arabian Sea monsoon current, which is chiefly deflected eastwards over the centre and hence parallel to the Arabian coast. Some of the higher regions in Southern Arabia, more especially Yemen, receive moderate rain during the period. Persia, Afghanistan, and Baluchistan are practically outside the sphere of the monsoon currents.

The Arabian Sea monsoon current is mainly determined to India, across the Bombay coast. The Bombay current prevails exclusively over the Peninsula, Central India, Rajputana, and Northern Bombay. Another portion of the monsoon current passes up the Bay of Bengal, and part of it enters India proper, giving rain to Assam, Bengal, and the Gangetic plain. The volume of the Bombay current is probably three times as great as that of the Bengal current. The remaining portion of the current in the Bay passes from the Andaman Sea across the Lower Burma coast and up the river valley of the Irrawaddy, giving frequent rain to the whole of Burma. Siam and South China receive most of their rainfall during the same period; but this, it would seem, is due chiefly to an influx from the China Sea and the Gulf of Siam, and not from the Indian seas.

It will thus be seen that the south-east trades, after they have crossed the equator, pass in part northwards up the Arabian Sea and in part up the Bay of Bengal, dividing into two semi-independent movements. Each of these again gives rise to two separate movements, one to the west and the other to the east. Abyssinia and Burma are each served by one of these four subdivisions, and India proper by two.

The following paragraphs describe the action of the Indian monsoon currents in greater detail. The current in the south of the Bay of Bengal is from west-south-west to south-west, and is therefore at this stage largely directed towards Burma and Tenasserim; and probably at least one-third of its volume proceeds to that area, and is either forced across the Tenasserim hills or passes up the Irrawaddy valley. The remaining portion advances up the Bay and is deflected westwards by the action of the Arakan hills, with the result that at the head of the Bay the mean direction of the air movement is from south-east and south. The advancing current chiefly crosses the coast between Chittagong and Purī, and comes almost immediately under the influence of the Assam hills and the Eastern Himalayas. That portion of the current which passes into the cul de sac formed by the Assam and Chittagong hills is forced upwards, and gives excessive rain (the heaviest probably in the world) to the southern face of the Assam hills. The remainder of the current is deflected westwards, owing largely to the action of the lofty barrier of the Himalayas, to the lower ranges of which it gives almost daily rain along its whole length from Sikkim to Kashmir. The field of this portion of the monsoon current is therefore Assam, Bengal, and the greater part of the Indo-Gangetic plain. It is more effective as a rain-distributing current than the Bombay current, as it is not directly impelled against and across a line of hills, and forced by ascent to part with the greater part of its moisture before reaching the interior, but advances up a broad river plain, the elevation of which increases very slowly with distance from the sea.

The greater portion of the Arabian Sea current on reaching the Bombay coast is directed from west-south-west to west against the west coast districts. Over a large portion of that area it meets with an almost continuous hill range, from 3,000 to 6,000 or 7,000 feet in elevation and rising abruptly from the coast, and is forced across in part by the momentum of the whole current. This forced ascent cools down the air considerably, and gives rise to frequent and very heavy rain upon these hills. Heavy rain, too, falls in the low coast districts, the total averaging roughly 100 inches, nearly the whole within about four months. The current, after depositing the greater portion of its aqueous vapour in the coast districts and the western face of the Ghāts, advances across the Peninsula, giving strong winds but only occasional rain to the Deccan, and passes out into the Bay of Bengal, where it mixes with the Bay current, giving rise to occasional squalls.

The northern portion of the west coast current, which crosses the Kāthiāwār, Cutch, and Sind coasts, passes over the sandy plains of Western Rājputāna, and gives little rain, except in the coast districts, until it reaches the Arāvalli hill range. It thence passes on north-eastwards and northwards, being deflected from the hottest and driest area (the Sind permanent low-pressure area of the period) by the action of the earth’s rotation, and reaches the Eastern Punjab, where it intermingles with the current from the Bay and is in part deflected westwards, and in part forced upwards by the Himalayan range, where the two currents combine to give moderate to heavy rain in the Eastern Punjab, Eastern Rājputāna, and the Western Himalayas.

An important feature of the period, not yet noticed, is the

debatable area in the south of the Indo-Gangetic plain between the independent fields of the two currents. It usually stretches from Sirsa, through Agra, Allahabad, and Hazaribagh, into Orissa, and may be termed the monsoon trough of low pressure. It is characterized by lower pressure, and by the prevalence of much more variable and unsteady but stronger winds, than the areas immediately to the north or south. A very marked tendency exists for cyclonic storms forming in the north of the Bay during the period to advance along this trough. It is mainly due to this that the rainfall is locally heavy in the rice-growing districts of the Central Provinces.

The normal dates on which the monsoon currents advance into the larger Provinces of India are given below:—

The dates do not, as a rule, vary much in the coast districts, but range over a considerable period in North-western India. The monsoon is usually established over the whole Indian area by the end of June. July and August are the months of greatest extension of the currents to the northern limits of the Punjab, and of heaviest and most general rain. The strength of the currents, and the accompanying rainfall, begin to diminish in the first or second week of September and decrease rapidly during the latter half of the month.

The period of about three months during which the monsoon usually obtains its full strength is, however, not one of continuous rain in any part of India. More or less general rain is received during periods varying very considerably in length, with intervals of fine weather and passing clouds in the interior and showery weather in the coast districts. In other words, bursts of general rain alternate with breaks, partial or general as the case may be. This is due to what may be described as the self-feeding actions accompanying rainfall. Some ascensional movement is requisite to cool the air in order to give rise to condensation and rain. The process of condensation releases energy, which strengthens the ascensional movement and hence up to a certain point increases the rainfall. It however drains and partially exhausts the atmosphere of aqueous vapour. The rainfall then diminishes in amount, and pari passu the ascensional movement decreases, until the rainfall ceases or has become local in character and distribution¹. A short period of lighter winds of indraught follows, with a slow accumulation of energy, in the humid currents over the south and centre of the Indian seas. This is followed by a repetition on a small scale of the actions accompanying the first great advance of the monsoon. In other words, the strong humid winds in the Bay tend to press forward by their momentum, and squally rainy weather sets in over the front of the current. A cyclonic storm forms, and the storm, with the strong humid winds in its rear, advances landwards and gives another burst of heavy and more or less general rain. The pulsatory character of these actions and of the rain precipitation is one of the important features of the monsoon period.

The majority of the storms which form in the Bay of Bengal during this period rarely develop so far as to have an inner calm central area, and only about one storm in four is of sufficient intensity to give winds of force 10 (gale) or upwards². They are chiefly remarkable for the rapidity with which they form and begin to move landwards, for the heavy rain they distribute to the belt they traverse, and for the general moderate rain that follows during a few days over the greater part of Northern India. They are thus of great importance in the economy of the season, as they not only modify considerably the rainfall distribution, but are apparently an essential element in the dynamics of the current for the production of general rainfall. In some parts of the interior, rainfall during the wet monsoon occurs only in connexion with these cyclonic storms. The great majority of these storms advance far into the interior, into Rajputana or the Punjab. In fact their vitality is one of their most characteristic features. Their tracks depend largely upon the distribution of pressure at the time of their advance. They almost invariably march in a west-north-westerly direction, except at the commencement and near the end of the period, when they occasionally advance northwards into Bengal. The great majority form in the north of the Bay, and occasionally, under special conditions, over South Bengal, where their origin is associated with concentrated heavy rain.

The following table shows the general distribution of rainfall month by month from May to October. These are the averages of a large number of stations not equally distributed, values which are only approximate but sufficiently exact for the purpose intended.

An important feature of the rains is the marked tendency to uniformity of temperature, cloud, and humidity conditions over by far the greater part of India. This will be seen from the following data for twelve representative stations in India for the month of July:—

As the south-west monsoon currents are the direct continuation of the south-east trades, it is a priori probable that they will more or less reproduce any large variation in the strength and character of the latter. The data for estimating the intensity, volume, or humidity of these two great currents are, however, too few and uncertain to be of any value for detailed comparison except as regards the accompanying rainfall. The experience of the past ten years has shown that there is from this standpoint a close connexion. The period 1895–1902 was one of almost continuously deficient rainfall in India, and of severe droughts in 1896 and 1899. The same period was notoriously one of scanty rain and droughts in Australia, and of severe droughts in different parts of South and East Africa. Mauritius suffered from drought in the years 1896–1900, or immediately subsequent to the scanty rainfall in India that brought on the famines of 1897 and 1900. It is not possible to make a more exact comparison at present; but the evidence is sufficiently strong to justify the conclusion that deficient rainfall in the south-east trades region in the period November to April is preceded or followed by a weaker south-west monsoon than usual, more especially in the Arabian Sea branch of the monsoon currents. The rains are in such years late in their establishment over Western and Northern India, and the Bombay current retreats earlier than usual and gives considerably less rain than the normal over the whole area dependent on it. In such years Abyssinia usually suffers equally with Western India, as it is now fully established that years of drought in Western or North-western India are almost invariably years of low Nile flood. The relation is further confirmed by the fact that years of heavier rain than usual in Western India are also years of high Nile flood. It is also found, as might be anticipated, that the precipitation in these areas due to the humid south-west monsoon currents is not invariably distributed in the same proportion. Local conditions in one area may be more favourable than usual and its share is hence increased. There are no data available for Abyssinia, and the variations in the height of the Nile floods are not sufficient evidence, as they are probably due in part only to the Abyssinian rainfall. There is, however, marked opposition between the rainfall in Burma (and to a less extent in Bengal and Assam) and the rainfall in North-western and Western India.

The table on the next page, giving the annual variation of the rainfall in different areas from 1878 to 1902 as compared with the strength of the Nile flood, illustrates the preceding remarks.

India shows, moreover, large local variations from the normal distribution of the rainfall. When such is the case, it is found that the current and rainfall are determined largely by the abnormal pressure conditions set up during the hot weather and hence antecedent to the arrival of the rains. The currents are directed more largely than usual towards any area over which there has been for some time previously a local deficiency, and less largely to areas in which there is from any cause a local excess, of pressure. These conditions cause a variation of rainfall from the normal in the earlier stages of the monsoon, viz. relative excess in one area and relative defect in another. This special rainfall distribution tends to maintain the pressure and other conditions which determined the initial set of the currents and consequent rainfall, and hence the areas of increased rainfall continue to be areas also of deficient pressure. Such persistence of abnormal pressure conditions is one of the most striking features of the meteorology of India, and it is mainly for this reason that the study of pressure anomalies in India must be the basis for any successful system of long period or seasonal forecasts from meteorological data. The continuity of pressure anomalies is sometimes persistent throughout the whole monsoon period, but occasionally interfering actions occur, such as snowfall in June or July in the Western Himalayas.

One of the more potent influences in modifying the pressure conditions antecedent to the advent of the monsoon is late and prolonged snowfall in the Himalayan regions. Mr. Blanford first fully investigated the question of the strength of the monsoon and distribution of the monsoon rainfall so far as it was modified by Himalayan snowfall, and upon an incorrect idea of his conclusions has been based what newspapers term the snowfall theory of the variations of the monsoon rainfall. According to this theory, snowfall is either the predominant or the only important factor, scanty snowfall preceding a strong, and excessive and prolonged snowfall a more or less deficient, monsoon. This snowfall theory is, however, not a theory of the Meteorological Department. All that Mr. Blanford established was that, in certain cases which he investigated, heavy and late snowfall in the Western Himalayas was followed by a late and scanty monsoon, more especially on the Bombay side, and that the relation held so closely in these cases that it might be used as an empirical guide for forecasting droughts. The relation was employed successfully in the last period of excessive and prolonged snow in the Himalayas to forecast drought in Central Rajputana and adjacent districts in 1891.

The preceding remarks show that the distribution of the monsoon rainfall in India is not a simple but a very complex problem. It depends upon a large number of factors, every one of which must be taken into consideration if a rational explanation of the variations from year to year is to be given; and in attempting to forecast future conditions not only must each large factor be taken into account, but proper weight or value must be given to each factor. It should also be remembered that rainfall is, as a meteorological factor or element, quite different from pressure, temperature, and air movement. It is a discontinuous element, the only discontinuous element of importance at the present stage of meteorological discussion. It is, however, found that the relative or local distribution of any one south-west monsoon rainfall is similar in character throughout the whole period in about four years out of five. The application of statistical methods based on this general fact or principle (which may be termed the persistence of weather or meteorological conditions) will, it is hoped, continue to give more and more exact and numerous relations, all of which will be of use for seasonal forecasting in India.

The following are the most important variations of the period and distribution of the monsoon rains from the normal which may occur:—

(1) The commencement of the rains may be considerably delayed over the whole or a large part of India. This is, on the whole, most frequent in Northern Bombay and North-western India. In 1877 the commencement of the rains on the Konkan coast was a fortnight later than the normal, and in 1896 a week later. The delay was most marked in the year 1900, when it amounted to three weeks in the Konkan, Kathiawar, and the Central Provinces, and to nearly a month in Central India and Rajputana.

(2) There may be a prolonged break lasting over the greater part, or the whole, of either July or August (most usually August). Thus, in 1880 there was a break from August 14 to September 3 over the whole of North-western and Western India. In 1883 there was a break lasting nearly six weeks, from July 19 to the fourth week of August, in North-western and Central India.

(3) The rains may terminate considerably earlier than usual. The economic effect of this is occasionally very serious. The absence of rain may cause the chief crops, e.g. rice, to dry up, with the result that the harvest may fail either partially or completely. In 1874 the rains ceased in Bihar early in September. In 1883 they terminated about a month before their normal date over the greater part of Bengal. Similarly in 1896 they ceased four weeks earlier than usual in North-eastern India and six weeks before their normal date in the Central Provinces. In 1899 they terminated four weeks, and in 1902 more than a fortnight, earlier than usual in the Central Provinces.

(4) The rains may, as above observed, be directed more largely than usual throughout the whole or great part of the monsoon period to a given area and less than usual to another. This is the most common variation, examples occurring every year. For example, Assam, Eastern Bengal, and Burma almost invariably have heavier rain than usual when the fall is very deficient in the east of the Punjab and the Gangetic plain.

The following chart shows clearly the normal distribution of rainfall of the period:—

CHART SHOWING THE NORMAL RAINFALL IN INCHES IN THE PLAINS OF INDIA, i.e. AT STATIONS BELOW 3,000 FEET IN ELEVATION, DURING THE PERIOD JUNE TO SEPTEMBER

Fig. 4.

The second half of the wet season forms a transition period leading up to the establishment of the conditions of the dry season. During this period the area of rainfall contracts and the monsoon extends southwards and diminishes in intensity, but is determined by the same winds or air currents, and by the same general conditions, as in the preceding period, June to September. The transition from the wet to the dry season commences in the latter part of September and is usually not completed until the third week of December. It is first exhibited in Northern India, and frequently follows upon a final burst of rain and thunderstorms accompanying the march of a cyclonic storm from the head of the Bay. The skies clear with the dispersion of the storm, and light and dry westerly airs replace the damp winds which previously prevailed. The night temperature decreases considerably with the increased dryness of the air, and the day temperature increases slightly on account of the absence of cloud. Hence the mean daily temperature is at first almost unchanged, but the diurnal range is largely increased.

The Northern India humid current retreats down the Gangetic plain and across Central India by a series of intermittent actions. As a rule the lower air current of the wet monsoon withdraws from the Punjab in the third week of September, from the western half of the United Provinces in the last week of September, from the eastern half of the United Provinces and Bihar in the first or second week of October, and from Bengal and Upper and Central Burma in the third or fourth week of that month. The dry land westerly winds during this period extend eastwards down the Gangetic plain, and at the same time increase slightly in intensity, and are usually established over the north of the Bay, and the whole of Northern and Central India, before the end of October. While these changes are in progress in Northern India and the Bay area, similar alterations occur in Western India and the Arabian Sea. The south-west humid current usually obtains at the beginning of September over the whole of that area, but decreases in strength and probably also in elevation and volume. It begins to retreat from the head of the Arabian Sea, Northern Bombay, and Rajputana in the second or third week of September. This process, as in the corresponding retreat of the Northern India or Bay current, occurs by a series of intermittent actions. The rains usually cease in Central India at the end of September, and in Berar and the Central Provinces in the second or third week of October. The further retreat of the currents in both sea areas is marked by novel features, more especially in the case of the Bay of Bengal. These features are most easily explained by reference to the pressure changes.

During the south-west monsoon proper pressure is lowest in a belt across the Persian Gulf, the north of the Arabian Sea, Baluchistan, Sind, Eastern Rajputana, and the southern districts of the United Provinces. As the air temperature diminishes, at first slowly and then rapidly, until the beginning or middle of December, the air contracts over the cooling area and there is also a flux in the higher regions from the areas to the south, i.e. the Indian Ocean. This double action causes a continuous increase of pressure over the Indian land area, largest in amount where temperature decreases most rapidly and pressure was previously lowest, i.e. in North-western India. The low-pressure conditions previously prevailing in India are thus almost obliterated by October, when pressure is remarkably uniform. The only important feature during the remainder of the period is a residual shallow depression stretching eastwards and westwards across the Bay and Arabian Sea, with slightly higher pressure in the Deccan and South Indian areas. This depression is slowly transferred southwards during the period, and lies over the centre of the Bay at the beginning of November and over the south of the Bay at the beginning of December. The humid current of the retreating south-west monsoon continues over the area to the south of the belt and varies very largely in strength. When it is vigorous, it recurves round the eastern portion of the low-pressure belt and extends westwards in the area immediately to the north of it. At the same time, and as part of the general actions, a cyclonic storm forms and passes westwards, carrying the humid currents of the retreating monsoon with it and giving heavy rain. In the intervals between these actions the westerly winds in the south of the Bay are light in force. Light variable winds with fine weather prevail in the centre of the Bay, and light to moderate north-easterly winds in the north and north-west of the Bay. Periods of fine dry weather thus alternate with periods of showery and disturbed weather in the east coast districts of the Peninsula.

The monsoon current recurves at the head of the Bay and in Bengal, in the same general manner (i.e. through north and north-west to west) as during the south-west monsoon proper. The recurvature is now, however, due not to the obstructive action of mountain ranges, but to special pressure conditions in the Bay. The retreating south-west monsoon gives occasional or frequent general rain during this period to the Madras coast. In October and the first half of November the precipitation occurs chiefly in the North Coromandel and Circars districts, and in the second half of November and December chiefly or solely in the South Coromandel districts. In the second or third week of December the belt of low pressure usually passes out of the Bay limits into the equatorial belt, where it forms a permanent feature of the meteorology of the Indian Ocean during the next five months. The heaviest and most general rainfall during this period accompanies the passage of the cyclonic storms of the period, which almost invariably advance westwards or north-westwards to the Madras coast. The Bay islands and the Malayan peninsula receive occasional rain during this period, but to a less extent than the Coromandel coast.

Similar conditions probably obtain in the Arabian Sea. The humid current retreats slowly over the southern half of that area in October and November, and is in slight part directed to the Malabar coast, to which it gives occasional rain. It is probable that a belt of slightly lower pressure than elsewhere demarcates the northern extension of the current, which recurves to west and determines more or less general rain from lat. 16° N. southwards to the eastern coast of Africa, and also in the Equatorial Lakes region, the area of rainfall passing slowly southwards with the advance of the season. Cyclonic storms occur much less frequently in the Arabian Sea than in the Bay of Bengal during this period. They rarely form in that sea, but advance into it from the Peninsula and Bay area.

The humid current in the Bay continues to exhibit the same intermittent or pulsatory character as in the wet season. Periods of rain and strong winds alternate with periods of dry weather and light winds. Each period of general and heavy rain is as a rule initiated by the advance from the Bay of a cyclonic storm, which gives concentrated rainfall over long narrowish belts of country. These bursts of rain occur at longer intervals than in July and August, owing to the decreasing intensity of the humid monsoon current. It is, however, interesting, as an example of the opposition which frequently characterizes meteorological actions, that the antecedent conditions of uniform pressure and temperature and light variable winds during the intervals of fine weather between the rain periods favour the slow incubation of storms in the centre of the Bay, with the result that they are occasionally very intense and violent in character. The conditions are, on the whole, favourable for the formation of the most severe cyclones in October, when the humid currents that provide the motive power are still of moderate strength. These October cyclones are examples of the most intense tropical storms. They differ in several respects from the cyclonic storms of temperate regions. They are usually of small extent, occasionally not exceeding 200 miles in diameter. Hence the shifts of wind accompanying them are very rapid and dangerous to vessels, and October is the most critical period of navigation in the Bay of Bengal. The pressure gradients are very steep and the winds of hurricane intensity in the inner storm area. The precipitation is excessive, the rain being commonly described as ‘falling in torrents.’ The most characteristic feature in the worst storms is an inner central area of calms or light variable winds, occasionally 10 to 20 or even 30 miles in diameter, which is termed by sailors ’the eye of the storm.’ The transition from the calm area to the belt of hurricane winds is usually exceedingly rapid. Another characteristic feature of these cyclones is the piling up of a mass of water in the inner storm area and area of lowest pressure. This advances with the storm and strikes the coast as a ‘storm-wave.’ The effect of this in flooding the coast districts depends largely upon the phase of the ordinary tidal wave at the time when the storm-wave strikes the coast. If the storm-wave strikes the coast about high water or shortly after it may produce the most disastrous results, flooding low coast districts in a few minutes to the depth of ten, twenty, or even thirty feet above tidal high-water level. In such an inundation, caused by a storm-wave which spread up the Hooghly in 1737, 300,000 people are said to have perished, but the number is probably exaggerated. The storm-wave accompanying the Calcutta cyclone of October, 1864, drowned 50,000 people and caused immense destruction of shipping. That accompanying the Backergunge cyclone of 1876 was one of the most disastrous on record: probably over 100,000 people were drowned in less than half an hour in the islands and low districts at the mouth of the river Meghna, while an equal number died from epidemics of fever, cholera, and other diseases which almost invariably follow a storm-wave. The total height of the wave was in some districts as much as 30 to 40 feet. Cyclones of the most dangerous type are fortunately rare, not more than one, on the average, occurring in five years.

A moderate to rapid increase of pressure in October following the last burst of rain initiates cold-weather conditions, with fine dry weather, almost continuous cloudless skies, and light variable winds in Upper India, the monsoon current being, during the first fortnight or three weeks of that month, determined chiefly to North-eastern India and Burma, which areas usually receive moderate to heavy rain. The cessation of the rainfall in that area is followed by a rapid local rise of pressure in Upper and Central Burma, Assam, and Bengal, and a general decrease of temperature and increase of pressure continue during the next two months. These changes are greatest in Upper India, and are small in amount in the extreme south of the Bay and in Southern India. Gradients for northerly winds increase more or less steadily throughout the period in the Indian area and the air movement increases in intensity. North-westerly and westerly winds obtain in the Gangetic plain, are continued as north-easterly winds in the north of the Bay, and extend southwards over the centre and south of the Bay as the season advances. A narrow belt intervenes in which winds are irregular and unsteady, ranging between south and east, while further south-south-west humid winds continue but decrease in general strength and volume.

The most important feature of the air movement of the whole period is thus the slow and gradual withdrawal of the south-west humid winds from the Indian area, and the gradual extension of the winds of the dry season from Upper India eastwards and southwards over the whole land and sea area. This change is usually completed before the end of December, when true north-east monsoon winds are established over the whole Indian region. The season above discussed is thus a transitional period of considerable length, and is in this respect unlike the transitional period at the commencement of the south-west monsoon. That monsoon is usually established rapidly over the Indian area, and extends over the greater part of the whole of the interior in the course of a week or a fortnight. The advancing current is a vigorous movement, while the retreating current, and the current which gradually displaces it, are both feeble.

The rainfall due to the decaying and retreating current is not only much smaller in amount than during the previous three months, but also much more irregular. The whole area of North-western and Central India and Northern Bombay receives practically no rain during this period. The Punjab and the adjacent districts receive a small amount, chiefly due to one or more cold-weather storms in the last fortnight of December. North-eastern India and Upper Burma obtain light to moderate rain, chiefly in October, and Lower Burma moderate to heavy rain in October and the first half of November. On the other hand the Madras coast districts receive their chief rain supply during this period. The amount increases southwards from Ganjam to Negapatam, but decreases very rapidly in amount from the coast to the interior.

The important variations to which the weather is subject in this period are as follows:—

A. The number and character of the storms which form in the Bay, and hence also the distribution of the rainfall accompanying the retreat of the south-west monsoon humid current, vary very largely from year to year. In some years, as for example in 1896, no storm of importance occurs, while in others two to four severe storms or cyclones may be experienced. The years 1876, 1886, and 1891 were remarkable for the number of fierce and extensive cyclones which occurred during this period.

B. The character of the large general and local pressure changes which accompany the gradual withdrawal of the monsoon current in India, and which appear in part to determine and in part to be determined by rainfall, varies largely from year to year. Thus an abnormally large increase of pressure in October over North-western and Central India is almost invariably accompanied by absence of rain in the Deccan. Similarly, a delay in the establishment of the high-pressure conditions in North-eastern India and Burma accompanies, and in part at least determines, a prolongation of rainfall in that area to the exclusion of the Peninsula. The late rains are then not only delayed in that area, but are scanty and more irregular than usual.

The most important variations from the normal of the rainfall distribution of this period may be summarized as follows:—

(1) Prolonged rainfall in Burma, Tenasserim, and the Bay islands, and consequent deficient rainfall in Madras and the Deccan.

(2) Early termination of the rains in Burma, accompanying increased rainfall in the Peninsula.

(3) Deficient and scanty rainfall in the Deccan and Northern Madras, accompanying normal or increased rain in Southern India.

(4) Early termination of the rains in both the Burma and Madras areas, due to more general actions and conditions than the local conditions in the Indian area; these have not yet been fully investigated.

(5) Early commencement of the cold-weather rains in Northern India, almost invariably preceding prolonged stormy weather with much snow in the northern mountain ranges.

II. Special Remarks on Rainfall in India, and Droughts

The air is always more or less charged with aqueous vapour. The amount is measured (in English measures) absolutely by the weight in grains per cubic foot, and varies greatly with the season and with the elevation and position. In the Bombay and Bengal coast districts, during the height of the wet monsoon, it is about 10 to 12 grains per cubic foot; whereas in the interior of India during the driest periods it is not more than four grains per cubic foot, and occasionally even less. ‘Relative humidity’ is humidity in relation to temperature, and indicates how far the air departs from saturation with respect to aqueous vapour.

Unsaturated air can always be cooled down until the space is saturated and further cooling gives rise to condensation. There are many complex problems connected with the processes of condensation and rainfall, but the main actions are simple. Ascending air always cools down at a rate depending upon the amount of aqueous vapour it contains and whether condensation is in progress. If the air be not saturated, it cools down at a rate of about 1° F. per 200 feet; but as soon as it reaches the saturation stage, followed by rain, condensation commences and continues so long as rapid ascensional movement continues. The heat given out in this process acts mechanically on the atmosphere and tends to increase the ascensional movement. The rising air cools down, but at a much slower rate than previous to condensation. In this stage the rate of cooling may be as low as 1° F. per 500 feet. Uptake or ascensional movement is thus a very powerful means of producing the cooling necessary to condensation and rainfall. There are other processes, as for example the mixture of cool and warm masses of air, but they are probably of comparatively rare occurrence and seldom give rise to heavy general rain. Hence it may be assumed that rainfall in a humid current is chiefly, if not almost entirely, due to upward movement, which may be initiated by various actions. Whatever the initial cause, the continuance of the movement is in large part strengthened and maintained by the energy set free by the condensation of the aqueous vapour, and converted by physical processes into the energy of motion of the atmosphere affected.

The initial ascensional movement may, in part at least, be due to the ordinary diurnal changes in progress caused by the heating of the land surface during the day. When the ascensional movement is very large, as in the hot weather, it is possible to get this action even although the column of ascending air may be very dry near the earth’s surface. This is shown by the occasional occurrence of showers of rain or hail in the hottest and driest districts. There is hence a marked tendency, especially in the hot-weather period, to the occurrence of rain at the same time of the day.

The uptake or ascensional movement may also be due to the forced ascent of a humid current against and across a range of hills, initiated by the momentum of the moving mass in the rear and further accentuated by the rainfall to which it gives rise. Rainfall is therefore usually heavy on and near any range of hills facing a humid current, but diminishes very rapidly on passing beyond the crest of the hill range, when the current not only ceases to rise but may descend slightly, and has, during the process of rising over the mountain ridge, been deprived of a considerable portion of its moisture. There is in such cases a very marked contrast between the rainfall on the two sides of hill ranges across which humid currents pass and the adjacent plain districts.

Ascensional motion also occurs on a large scale in cyclonic storms, and hence heavy rainfall is the most prominent feature of these storms in India. The rainfall is usually greatest in the advancing quadrant of these storms. The heaviest downpours frequently occur in the interior of India and not in the coast districts, and more especially in the neighbourhood of the hills athwart the track of a storm. Falls ranging between 20 and 35 inches in twenty-four hours occasionally occur under these conditions. Such was the heaviest fall on record in the plains of India: namely, 35.38 inches at Purnea, in Bihar, on September 13, 1879.

The rainfall of the ascensional motion may, lastly, be due to the increased resistance to the horizontal movement experienced by a humid current crossing a coast and advancing landwards, as, for example, across Bengal and up the Gangetic plain. This is undoubtedly one action determining a considerable portion of the monsoon rainfall.

The above are the chief actions producing rainfall in the humid south-west current. Other minor actions might be enumerated; but it is sufficient to refer to Mr. Blanford’s monograph on The Rainfall of India for a full statement and a fairly complete explanation of the causes co-operating to determine the rainfall in each district of India, and also to some extent the variations from year to year.

A brief account of the character and distribution of the rainfall of each of the seasons has been given under the meteorology of that season. The following gives a statement of the mean or normal rainfall in India, derived from the average data of about 2,000 stations distributed somewhat unequally over the whole of the country:—

Cold weather (January-February) . . .

Of the mean annual rainfall 12 per cent. occurs during the dry, and as much as 88 per cent. during the wet, season. The ratios differ very considerably in different parts of India, varying for the dry season from 3 per cent. in Bombay to 19 per cent. in Bengal, 21 per cent. in the Punjab, and 30 per cent. in Assam. These figures show the importance of the hot-weather or spring rainfall in Assam, and of the cold-weather rainfall in the Punjab. Appendix A (p. 153) gives data for the meteorological divisions lately adopted in the India Daily Weather Report.

The cold-weather rainfall is due almost entirely to disturbance and condensation in an upper current, probably over 15,000 feet in elevation. Hence its distribution is very slightly modified by the configuration of the country over which the current passes, except over and near the only mountain range sufficiently elevated to give rise to forced ascent on the large scale, viz. the Himālayas. The precipitation is large over the interior elevated ranges, and moderate over the outer ranges, of the Western Himālayas. It diminishes slightly from the sub-montane districts southwards, and from the Punjab eastwards to Bihār, but thence increases again to Assam. It is very irregular in its occurrence, but the normal amount differs very slightly over the plains of Northern and Central India, varying from half an inch to one inch, except in the Punjab and Assam, where it averages two inches.

As already pointed out, the precipitation of the cold-weather period occurs during the passage of shallow extensive depressions across Northern India from west to east. These storms are almost the only source of rain in Persia, Afghanistan, and Baluchistan, and to a less extent on the higher and interior ranges of the Western Himālayas. The cold-weather rainfall is small in absolute amount in Northern and Central India, but is nevertheless of great economic importance over the larger part of that area, as it is upon this rainfall that the wheat and other cold-weather crops of the non-irrigated districts in Northern India depend. In the districts to which irrigation extends the crops are practically independent of the rainfall; but over the unirrigated districts, including the greater part of Rājputāna, Sind, Central India, and parts of the Punjab and the United Provinces, such cultivation as there is largely depends upon the amount and time distribution of this limited rainfall.

The distribution of the hot-weather rainfall presents much greater contrasts than that of the cold weather. It is due, as previously stated, to large convective movements over the interior plains raising air masses to a sufficient height to produce condensation, and to forced ascent of local sea winds across hill ranges in Assam, Bengal, Orissa, and Malabar, supplemented more or less largely by convective air movements due to the intense thermal actions of the period. The rainfall often accompanies dust-storms in the interior, and is small in amount, averaging only about an inch, over Bombay, Berar, the Central Provinces, Central India, Rajputana, and the Gangetic plain. It accompanies thunderstorms (occasionally of extraordinary violence and intensity) in Assam, the greater part of Bengal, Arakan, and Lower and Upper Burma, and is largest in amount in Assam, where it averages 31 inches. Economically it is of great importance for the tea crop of Assam, and in Bengal, where it favours an early spring crop of rice. Over the greater part of the interior it is of little agricultural value, and the violent winds and heavy rain accompanying the storms occasionally do great damage.

It is upon the south-west monsoon rainfall that the prosperity of India chiefly depends. As already stated, nearly 90 per cent. of the whole rainfall of the year falls on the average during this part of the year, being due to the invasion of India by oceanic currents of great volume and elevation from the Arabian Sea and the Bay of Bengal. The goal of these currents from June to September in India is a persistent low-pressure area in Upper Sind and the neighbouring districts of Rajputana and the Punjab. The transformation effected by the invasion of these currents is comparatively slight in the regions of frequent thunderstorms in April and May: that is, in Lower Burma, Bengal, Assam, Orissa, and the Madras coast. Over the remainder of India, and more especially in the interior of the Peninsula and in Northern and Central India, it is very pronounced. Previous to its advance strong land winds, almost unbearably high temperature, excessive dryness of the air, and a dust-laden atmosphere are the chief features. Afterwards strong to moderate sea winds, moderate temperature, great humidity, much cloud, frequent rain, and a comparatively clear atmosphere are prevalent. From June to September the two currents give rain chiefly to Burma, Northern and Central India, and the northern half of the Peninsula, while over the southern half of the Peninsula rain occurs very irregularly (chiefly during breaks of the rains in Northern India) and is scanty except on the Malabar coast. The distribution of the rainfall during this period (June to September) is determined chiefly by ascensional movement due to (1) forced ascent across hills, (2) convective ascent in cyclonic storms, and (3) irregular ascensional movement caused by resistance to horizontal air movement, or due to actions accompanying the heating of the atmosphere during the day hours over the drier parts of the interior.

The action of forced ascent across or against hill ranges gives heavy rain in (a) the west coast districts of Malabar, the Konkan, and Gujarat, (b) Tenasserim and Arakan, (c) the Assam hills and Sylhet, and (d) the Lower Himalayas and submontane districts, more especially the eastern half. Actual data show that in the Western Ghāts the rainfall is on the average at least three times as great on the crest of the hills as at sea-level. The average rainfall of the period is roughly 100 inches in the coast districts of the Konkan, and thence increases up to an average of 300 inches at stations situated on the crest of the hills facing the sea at an elevation of 3,000 to 4,000 feet. A similar law of distribution almost certainly holds in the case of the Tenasserim, Arakan, and Assam hills. Sylhet at the foot of the Assam hills has an average total of 157 inches, while Cherrapunji in the Assam hills, at an elevation of about 4,000 feet, has an average of 458 inches. The inverse action to that of forced ascent is found in the districts on the landward side of ranges of hills which face the monsoon currents. The current, which has been deprived during ascent of a large portion of its aqueous vapour, after it passes over such a range, either moves horizontally or descends slightly, and is then, unless some other special actions come in, virtually a non-rain-giving current. Hence the Bombay and Madras Deccan Districts of Sholapur, Bijapur, Ahmadnagar, Poona, and Bellary are very dry areas and specially liable to drought. Similarly, a portion of Central and Upper Burma, including the Districts of Magwe, Minbu, Myingyan, Kyaukse, and Sagaing, is a very dry and precarious area, with a scanty rainfall averaging not more than 30 inches for the year. The Himalayas are not exposed to the full burst of the annual monsoon currents. The Bay current advances towards them, but is deflected, and though there is much ascensional movement, it is not forced ascent accompanying movement across the line of hills. Hence the distribution of the rainfall does not follow exactly the same law as in the Western Ghāts or the Assam hills. The rainfall is, however, considerable over the Lower Himalayan ranges up to levels of 8,000 to 10,000 feet, but is slight on the higher elevations or in the interior valleys. The rainfall again decreases considerably southwards in the submontane districts, and over the Gangetic plain it is determined partly by ordinary ascensional movement and partly by cyclonic storms.

A considerable proportion of the wet monsoon rainfall over the greater part of India is due to the ascensional movement accompanying cyclonic storms. On the average eight storms of moderate to considerable intensity pass from the Bay of Bengal into India between June and September. They travel comparatively slowly and frequently traverse the whole breadth of Northern India. Their characteristics, and the accompanying rainfall, have already been noticed on page 125.

After the Bay current has withdrawn from Northern India and Upper Burma, and is recurring over the Bay, it is, as already stated, directed to the Peninsula. It is now, however, comparatively feeble, and probably its elevation as well as its volume diminishes and it becomes less effective as a rain-giving current. Cyclonic storms occur at longer intervals, but continue to give large amounts of rain to the areas they pass over. Their path is, usually in the latter part of October and almost invariably in November and December, westward to the Circars and Coromandel coasts. The rainfall in the Deccan and Southern Madras coast districts during this period occurs almost entirely in connexion with these storms, and is hence extremely irregular in its distribution. The rainfall is heaviest in a narrow belt of the Coromandel coast, where it ranges between 20 and 30 inches, and decreases rapidly on proceeding into the interior. It is occasionally very heavy in the Nilgiris, when the storms advance into the interior of Southern India and break up against the high hill ranges.

A characteristic feature of Indian meteorology is the excessive downpours which occasionally occur, chiefly during the passage of the cyclonic storms of the south-west monsoon, and also in the Western Ghāts and Lower Himālayas during a strong influx of monsoon winds. The fall at Purnea in Bihār of 35.38 inches in one day, in September, 1879, has already been mentioned (p. 139). Other phenomenal downpours, exceeding in each case 24 inches within the twenty-four hours, have been:—

Another noticeable feature is the very heavy downpours which occur in very short periods, chiefly accompanying thunderstorms. The most striking examples have occurred at Calcutta and Nāgpur, at each of which places slightly over an inch has fallen within ten minutes, equivalent to a rate of nearly seven inches per hour.

A most important feature in determining the economical value of rainfall in any given district is its variability. This can be estimated in various ways. Mr. Blanford, in his monograph on The Rainfall of India, has investigated the variability in the twenty-two rainfall divisions which he adopted for the seasonal returns submitted regularly to the Secretary of State for India by the Meteorological Department. According to this method (based on the law of errors and method of least squares) the variation from the normal rainfall for which the probability is $\frac{1}{2}$ (i.e. the chances are 1 to 1) is determined, and then expressed as a percentage of the normal rainfall. Some of the results of Mr. Blanford’s calculations will be given on the opposite page. This is undoubtedly the most scientific method of determining estimates showing the probable range of variation of rainfall. It is, however, too elaborate for general use over a very large area like India. The simplest method is to adopt as the measure of variability the percentage ratio of the difference between the highest and lowest actuals for the period of observation to the normal or average rainfall of the period. The largest percentages of variability as thus calculated are those of the areas named below:—

The most noteworthy feature is the very great variability of the rainfall in Gujarat. This is due to the fact that it is liable not only to very scanty rain in dry years, but to excessive downpours from cyclonic storms in years of ordinary or strong monsoon conditions.

The following table gives Blanford’s estimates of variability for thirteen areas:—

When the rainfall in any area is too scanty for the staple crops of that area, and a partial or complete failure of the harvest ensues, the season is one of drought which may, if severe and long continued, produce famine¹. A large deficiency of rain in an area of heavy rainfall, as for example Arakan, East Bengal, &c., rarely, if ever, affects the staple crops to an extent sufficient to justify the use of the term ‘drought.’ Droughts chiefly occur in the interior districts where the normal rainfall, when properly distributed, is sufficient for the crops, though not ample or abundant. The table on the preceding page gives a list of the larger areas liable to drought in the order of the variability of the rainfall as determined by Mr. Blanford (vide The Rainfall of India, p. 129).

Mr. Blanford, from examination of the data, inferred that droughts leading to famine were especially likely to occur in all districts in which the variability, as measured by the process described on p. 125 of his memoir, is twelve or upwards.

Scanty and ineffective rain in the wet monsoon in any area is due to one or more of the causes referred to on p. 130. The rainfall of the first half of that monsoon is much less valuable and effective, on the whole, for agricultural purposes than the rainfall of the second half. It is essential that there should be occasional to frequent rain in August and September, and occasional rain in October, in order to bring the chief food crops to perfection. An abrupt termination of the rains in August may cause the whole of the rice crop in a Province to fail to come to maturity.

In North-western and Central India not only may the rains of the south-west monsoon fail more or less completely, but also the preceding or following cold-weather rains, and failure of the first is, in about two cases out of three, accompanied by failure of the second. This double drought entails loss of the harvest of both seasons and is most disastrous. The most severe droughts in the Punjab and the Gangetic plain have been of this kind. Finally, a severe drought leading to famine may occur after a series of bad years, owing to excessive or deficient rain, either of which may diminish the crop returns for several years by large percentage amounts. This was the case in the Central Provinces in 1896.

II. Special Remarks on Indian Temperatures

Temperature is perhaps next to rainfall the most important feature of meteorological observation in India from the economic standpoint. Temperature in relation to the amount of aqueous vapour present in the air and the rainfall determines the character and abundance of the staple crops. The temperature of the air in India is carefully measured under, so far as is possible, approximately similar conditions. The thermometers are exposed in open sheds, through which the air freely circulates, and are protected from the direct influence of the sun by a thick roof or thatch of straw. The bulbs of the instruments are placed as nearly as possible at the same standard height above the ground, viz. four feet.

The sun is practically the only hot body which materially modifies and determines the temperature of the earth’s atmosphere. The solar radiation is absorbed in part during its passage through the earth’s atmosphere and in part by the earth’s surface, which again gives up a portion of its heat to the adjacent air.

The average temperature of the ground surface in India is, at the hottest time of the day in the cold weather, from 10° to 20° above the temperature of the air four feet high. The difference increases until the months of April and May, when the excess is usually as much as 40° and sometimes 45° or 50°. The difference falls off rapidly during the rains, and in August is as small as in the height of the cold weather. At Jaipur the ground has been found to be cooler than the adjacent air four feet above for sixteen hours out of the twenty-four in December and for twelve hours in May.

The air in contact with the earth’s surface is largely and rapidly heated during the day time, and is then carried upwards by virtue of the convective movements thus initiated. These movements tend to distribute the heat taken in by the air at the earth’s surface throughout the superjacent mass, and give rise to changes of temperature of great importance.

The temperature of the air at any fixed position or place is in a constant state of flux or change due to a variety of actions, more especially (a) the absorption of heat or solar energy either directly or indirectly, (b) the radiation of heat to the earth, clouds, or space, (c) the vertical or horizontal movement of the air in the lower strata. The most effective cause of rapid temperature changes is the displacement of air of one character by that of another, such as almost invariably occurs in the alternation of land and sea breezes, or accompanies hailstorms, thunderstorms, cold-weather cyclonic storms, &c. These changes are usually very irregular in character. The temperature of the air varies (1) regularly in consequence of the regular variations in the causes or actions modifying it, the most important of which is that of the sun’s thermal action as modified by elevation, and (2) irregularly owing to the actions enumerated above.

The temperature of the air is increased by the absorption of solar radiation or radiant energy directly or indirectly during the day hours, while it is affected during the whole diurnal period by radiation to clouds, space, &c. The result of these two actions is that in the lower atmosphere the temperature of the air is lowest very shortly before sunrise, and increases until two or three hours after noon, after which it decreases until next morning before sunrise. This fairly regular variation, with a maximum and minimum, which divides the day into unequal periods of about eight hours and sixteen hours on the average of the year, is the diurnal variation.

The difference between the maximum and minimum temperatures of any twenty-four hours period, as for example the ordinary civil day, is the ‘diurnal range.’ The diurnal range depends chiefly upon the amount and distribution of cloud and also upon the humidity. The following summarizes the chief features of the diurnal range:—

(1) It is much smaller in the wet than in the dry season, the difference being most marked at stations in the interior.

(2) It is smallest in amount at the coast stations, more especially at those on the Konkan and Malabar coasts, where it averages 10°, and increases on proceeding into the interior to a maximum of 30° on the mean of the year in the Punjab and Upper Sind.

(3) It has, over the greater part of Northern and Central India, two maximum and minimum values. It is usually absolutely greatest in November, the most serene month of the year; falls slightly up to January or February; increases again up to April and May; and falls to the absolute minimum in July or August, the height of the rains.

Again, during one part of the year, from January to May or June, the increase of temperature by the solar action is greater than the decrease or loss by radiation and other actions, and hence temperature rises more or less steadily during this period of increasing elevation of the sun. During the remainder of the year the balance is the other way and temperature steadily decreases from June or July to December. This regular increase during one half of the year, followed by decrease during the other half, is known as the annual variation. The annual variation is estimated in various ways. The most usual methods are (a) as the difference between the highest and lowest monthly means, (b) as the difference between the highest and lowest daily means, (c) as the difference between the highest normal daily maximum and lowest normal daily minimum temperatures during the year, and (d) as the mean of the annual absolute range of temperature: that is, of the differences between the highest maximum and lowest minimum observed in each year. The data establish that the annual variation (calculated by any of these methods) is small in amount at the Bay islands and in Malabar. It increases rather rapidly northwards in proceeding along the east and west coasts of India. It is twice as great at Bombay and Rangoon as in Malabar, three times as great at Madras and Chittagong, and over five times as great at Karachi. It is from eight to ten times as great at the stations in the North Deccan and Northern and Central India, and is absolutely greatest in the most inland and the driest districts, including Upper Sind and the Punjab.

The highest day temperatures are usually observed in Northern India in the month of May. The hottest area is undoubtedly that part of Upper Sind known as the Pat desert, of which Jacobabād is imperfectly representative, and after that, the adjacent tracts, including the South-western Punjab and North-western Rajputāna. In seventeen out of the last twenty years the highest day temperatures, ranging from 117° to 126°, were recorded at Jacobabād. As there is much irrigation in and near this town, which is more or less sheltered by trees, it is almost certain that at Sibi and other places in the driest parts of the Pat desert the maximum temperature probably reaches 130° in the hottest periods. Maximum temperatures exceeding 120° have been recorded at the stations named below:—

Jacobābād. Jodhpur.

Pachbhadrā. Sirsa.

Hyderābād. Multān.

Montgomery. Khushāb.

Dera Ismail Khān. Lahore.

Agra