]\gt ; [ 330- ] XII .
Some Anomalies in Winds of Northern India , their Rclation to the Distribution of rometric P S. A. HILL , B. , Meteorologicad Raeporter to , North-Western Provl nces and Ondh .
Communicated by H. F. BLANFORD , F.R.S. , Meteorological Reporter to the Government of India .
January 3 , \mdash ; Read January 20 , 1887 .
IN the 13 years which have elapsed since Mr. BLANFORD published his paper on the Winds of Northern India , 'very great additions have been made to our knowledge of the meteorology of the country .
The carefully organised system of observations , commenced in BeIIgal and the North-Western Provinces , has been extended to include the whole of India , and placed under the direction of Mr. BLANFORD himself , aided by local officers in all the larger provinces .
Verified instruments have been supplied to all the stations , and the elevations of these above sea-level have been determined by connecting them with the lmes of spirit-levelling , carried inland from the coast , in various directions , by the officers of the Great Trigonometrical Survey ; or , where this was impracticable , by spirit- ] evelling to some of the trigonometrical stations of the Survey .
In this way , trustworthy and inter- comparable series of barometric observations , extending over ten years or more , have been obtained for all the more important stations .
At the same time , the diurnal variations of the barometer at certain selected stations have been determined by longcontinued series of hourly observations , with the object of enabling us to reduce tho readings made in the ordinary way usually to true daily means .
Simultaneously with the collection of this immense quantity of accurate and reliable barometric data , observations have been made of temperature , humidity , cloud , wind , and rain .
Latterly also barometric and wind chalts of the Bay of Bengal have been prepared from observations made on board ships navigating those waters .
During these 13 years , the winds p1evailing over the Indian continent and the Bay of Bengal , and their relations to the distribution of pressure at sea-level , have * ' The Winds of Northeln India by HENRY F. BLANI , F.G.S. , Phil. Trams vol. 164 , 1874 , Part 16 .
11 .
87 RELATION TO THE DISTRIBUTION OF BAROMETRIC SURE .
Part Wind ?
cction The hot winds of the Indian plains have been often described by travellers and writers on Meteorology , but the exiqtence of such a condition as that mentioned above , viz. , their ] directly against the baric gradient , has been doubted by competent authorities , or at all events suspected to be only an erroneous deduction from a comparison of monthly mean pressures with the resultant wind directions .
The daily weather telegraphic reports of the last few years , however , leave no room to doubt that , almost every year , there are several days in April , May , or June when this primd facie impossible condition obtains over a large extent of country at 10 A M. , while , from the known character of the diurnal variations of pressure and wind direction , we may safely infer that the condition is much more frequent and more distinctly marked in the early hours of the afternoon , the diurnal fall of the barometer increasing slightly as we advance inland from Bengal , and the wind direction steadiest from the north-west about 2 The following examples , selected from the 10 ] egraphic reports , will prove the existence of this paradoxical condition of pheric circulation .
stations are enumerated in order from west to east , the general direction of the lines them being about W.N.W. to E.S.E. The pressures are reduced to sea-level .
TABLE I.\mdash ; Instances of Wind blowing in opposition to Baric Gradient .
May 17 , 1882 .
May 9 , ISS3 .
MueCeruntoyw .
sBetr..icpressute * Indian Meteorological Memoirs , ' vol. 1 , page 353 , and Plate XXX .
MR. S. A. HILL ON THE WINDS NORTHERN INDIA , AND THEIR May 29 , 1884 .
\mdash ; \mdash ; Station .
Barometric pressure .
May 29 , 1884 .
\mdash ; \mdash ; Station .
Barometric pressure .
May 29 , 1884 .
\mdash ; \mdash ; Station .
Barometric pressure .
May 29 , 1884 .
\mdash ; \mdash ; Station .
Barometric pressure .
May 29 , 1884 .
\mdash ; \mdash ; Station .
Barometric pressure .
1nches .
29.466 29.485 Inches .
Roorkee 29.466 Meerut 29.485 Bareilly .
29 .
Lur , know .
Allahabad 29.520 Inches .
Roorkee Meorut 29.485 Bal.eilly .
Lur , know .
Allahabad 29.520 Inches .
Roorkee Meorut 29.485 Bal.eilly .
Lur , know .
Allahabad 29.520 Inches .
Roorkee Meorut 29.485 Bal.eilly .
Lur , know .
Allahabad 29.520 Inches .
Roorkee Meorut 29.485 Bal.eilly .
Lur , know .
Allahabad 29.520 Inches .
Roorkee Meorut 29.485 Bal.eilly .
Lur , know .
Allahabad 29.520 Inches .
Roorkee Meorut 29.485 Bal.eilly .
Lur , know .
Allahabad 29.520 Inches .
Roorkee Meorut 29.485 Bal.eilly .
Lur , know .
Allahabad 29.520 Inches .
Roorkee Meorut 29.485 Bal.eilly .
Lur , know .
Allahabad 29.520 Inches .
Roorkee Meorut 29.485 Bal.eilly .
Lur , know .
Allahabad 29.520 Inches .
Roorkee Meorut 29.485 Bal.eilly .
Lur , know .
Allahabad 29.520 Inches .
Roorkee Meorut 29.485 Bal.eilly .
Lur , know .
Allahabad 29.520 Inches .
Roorkee Meorut 29.485 Bal.eilly .
Lur , know .
Allahabad 29.520 Inches .
Roorkee Meorut 29.485 Bal.eilly .
Lur , know .
Allahabad 29.520 Inches .
Roorkee Meorut 29.485 Bal.eilly .
Lur , know .
Allahabad 29.520 Inches .
Roorkee Meorut 29.485 Bal.eilly .
Lur , know .
Allahabad 29.520 Inches .
Roorkee Meorut 29.485 Bal.eilly .
Lur , know .
Allahabad 29.520 \mdash ; Slation .
Roorkec Meerut Bareilly . . . . . .
I-ucknow .
abad Wind direction .
Roorkee Meorut Bal.eilly .
Lur , know .
Allahabad Roorkee Meorut Bal.eilly .
Lur , know .
Allahabad Roorkee Meorut Bal.eilly .
Lur , know .
Allahabad Roorkee Meorut Bal.eilly .
Lur , know .
Allahabad Roorkee Meorut Bal.eilly .
Lur , know .
Allahabad Roorkee Meorut Bal.eilly .
Lur , know .
Allahabad Roorkee Meorut Bal.eilly .
Lur , know .
Allahabad Roorkee Meorut Bal.eilly .
Lur , know .
Allahabad Roorkee Meorut Bal.eilly .
Lur , know .
Allahabad Roorkee Meorut Bal.eilly .
Lur , know .
Allahabad Roorkee Calm .
Meorut 29 .
Bal.eilly .
Lur , know .
W. Allahabad NW. May 24 , 1885 .
May 27 , 1885 .
May 20 , 1886 .
Wind directi onlmsw .
RELATION TO THE DISTRIBUTION OF BAROMETRIC PRESSURE .
These instances abundantly prove the frequent occurrence of the condition described \mdash ; the wind blowing from a place of low pressure to one where the pressure is slightly higher ; but it may be objected that the places enumerated probably happen to lie nearly on an isobaric line , while the wind blows almost parallel to it , so that , if the slight disturbances of the true win direction due to local causes were eliminated , there would be no anomaly to explain .
The answer to this is that , at the season under consideration , the stations enumerated lie near the middle of a very large area of uniform low pressure ( see Plate 19 ) , while the winds are the strongest and steadiest of the year .
It may be added , also , that near the equator the wind does not blow parallel to the isobars , even over the sea , but makes an angle of two or three with them ; whilst at inland stations , where the friction coefficient is greater , the angle must be still greater , as FERREL has pointed out in his works on the movements of the atmosphere .
The charts prepared from the daily telegraphic weather reports for India show an average angle of like between the wind directions and the isobars .
The characteristic winds of the hot season , therefore , do really blow sometimes in opposition to the pressure gradient , and the cause which sets in motion must be sought somewhere else than in differences of pressure at sea-level , or rather at the level of the plain over which they blow .
Since , in these winds , the air-particles are not urged from west to east by any considerab]e difference of pressure at the ground-level , whilst they are retarded by friction , and sometimes even by an increasing pressure , as they go eastwards , their velocity must be gradually diminished as they approach the Bay of Bengal .
This fact is familiarly known to residents on the plains of the Ganges .
The " " hot winds which are excessively dry as well as hot\mdash ; and which , therefore , serve very effectually to refrigerate the interiors of dwelling-houses by blowing through wet tatties , or screens of fragrant grass , placed in the door and window openings on the windward side\mdash ; seldom extend as far eastwards as Monghyr , in Behar , their extinction takes place very gradually .
It is a matter of extreme difficulty to obtain truly comparable observations of wind velocity , owing in part to slight differences in the anemometers , but chiefly to the impossibility of ensuring umiform conditions of exposure ; but the following sets of stations , where the surroundings are as nearly alike as possible , show clearly a gradually diminishing velocity as we go eastwards during the hot weather , while , at other seasons there is , in some cases , a tendency for the velocity to increase as we approach the sea .
The are taken from the Report on the Meteorology of India for 1884 .
They represent the averages of ) years .
a short time at the hottest hours of the day westerly winds are felt in Bengal , and even in Assam , but they do not possess the dryness and gusty character of the winds under discussion , though when they blow are probably due to similar causes .
RELATION TO THE DISTRIBUTION OF TRIC PRESSURr .
the drier months of the year there is a decided diminution of vapour pressure at the hottest hours of the day , when west wind similar to , or identical with , the " " hot winds are blowing .
The following hourly values of the pressure of vapour at Allahabad will illustrate this .
They are taken , with some corrections derived from subsequent observations , from a paper on the " " Results of Meteorological Observations at Allahabad published in 1881 in the first volume of the 'Indian Meteorological Memoirs .
' ITI.\mdash ; Diurnal Variation of the Pressure of Vapour at Allahabad .
Month .
20 JanuaryMayJuly .
' October It will be seen that , whereas , in the month of July , the quantity of vapour in the air increases , with slight irregularities , as the temperature rises from 6 .
up to 2 , and diminishes again steadily the temperature falls after 2 , this regular march is interrupted in the other months by a fall in the middle of the day .
The fall is very slight in January , but considerable in October , making the afternoon minimum in that month slightly lower than that of the early morning .
In however , and in the other dry hot months , the afternoon fall so completely masks the variation due to the diurnal changes of temperature , as seen in July , that the morning minimum is altogether ted .
Whence comes this excessive dryness in the afternoons , which is not merely a decrease of relative humidity , due to rise of temperature , but a diminution of the absolute quantity of vapour in the air near the ground , from an average pressure of 621 inch in May , at 8 o'clock in the morning , to 515 inch at 4 in the afternoon It may be caused by the wind coming from drier regions to the north-westward ; but this can hardly be the true cause , for in May there is very little decrease in the mean pressure of vapour as we go north-westwards over the Gangetic plain , and there is still less in April .
For example , we have for three selected stations on the plain of the GaIjges , arranged in order from SE .
to NW. , the following values for the mean pressure of vapour . .
RELATION TO THE DISTRIBUTION OF BAROMETRIC PRESSURE .
and to the value of gravity at latitude .
The isobars have been extended across the Bay of Bengal by means of observations extracted from ships ' logs , recently reduced and tabulated in the Bengal Meteorological Office .
The arts given in Plate 19 are , therefore , identical with those for the same months attached to Mr. BLANFORD'S paper the Rainfall of India , *except that the observations of one or two stations which seemed to me of validity have been rejected .
On the chart for January , the resultant wind directions , on the whole , conform to the distribution of pressure in accordance with BuYS BALLOT'S , but there are some exceptions .
For example , at Neemuch , in Southern Rajputana , south-westerly winds prevail where the law would indicate an excess of easterly components .
The chief exception is in Northern Bengal , where the winds should be eastel .
lie or south-easterly according to the law , whereas there is a decided prevalence of north-westerly winds as far as Chittagong .
Elsewhere , except on the west coast , where the sea-breezes prevent observations , made only in the day time , from giving the true resultant direction , the prevailing wind agrees closely with that given by the law ; when it is borne in mind that , near the Equator , the angle bet , ween the wind direction and the isobars is larger than in higher latitudes .
In May , while there is close co'nformity to the law in Southern India , and near the coast generally , there is no apparent relation whatever between distribution and wind direction over the very large area north of the parallel of and west of the meridian of .
The greater portion of this is enclosed by the very peculiarly curved isobar of , within which are two distinct areas where the pressure is below .
One of these occupies the upper valley of the Mahanadi , in the eastern part of the Central Provinces , and the other and larger is in Upper Sindh and the Bikanir desert .
Over these areas bounded by closed curves , and across the axis of low pressure them , as well as across the isobar of which runs along the southern face of the Himalaya , the westerly winds blow without interruption as far as Behar and Chutia Nagpur , and , at a higher level , as far as Darjiling ; whilst the easterly winds , which should prevail , according to the law , at all stations north of the low pressure axis , are only felt along the base of the Himalaya up to Gorakhpur .
These west winds have a southerly component in the Indus Valley and Western Rajputana , as well as at most of the HimaJayan stations of 6000 or 7000 feet elevation , whilst , over the rest of the region , they usually a northerly component ; but a glance at the chart for May will show that they are probably all stream lines of one continuous atmospheric flow .
The chart for July shows that the pressure gradients over the whole of India and the Bay are much steeper than in the other months , while the usual relation between the direction of the wind and the isobars holds good over nearly every part of the * ' Indian Meteorological Memoirs , ' vol. 3 .
At most stations the observations are made only at 10 .
and 4 RELATION TO THE DISTRIBUTION OF BAROMETRIC PRESSUI tion of pressure , at levels , be different from that observed near the ground surface , we may , on KoPPEN'S theory , expect to find in January a greater departure of the actual wind direction , from that given by the usual law , we in July .
On the Upper Gangetic plains , and in the Punjab , the rapge of temperature in January is quite as high as in Bengal , though the are somewhat more cloudy .
We do not there , however , find departures from BUYS BALLOT'S law to occur so fi.equently in this month ; probably because the disturbances of thermal causing convection currents , are less intense .
For this there are two reasons .
In the first place , whilst the diurnal range of the thermometer is but little greater on the plains than it is in , bhe range at the nearest hill stations is considerably greater than at Darjiling , so that the difference between and low stations , at the hottest hour of the day , is not so considerable in North-western India as it is for the same difference of altitnde in .
Again , the temperature probably diminishes less rapidly with height at the North-western stations during January than it does in Bengal .
these two causes together , effect is that the rate of upward decrease of temperature in the North-west is not by any means so rapid at the hottest time of the day as it is in Bengal .
In the following Table , three pairs of stations are compared , so as to show the diurnal variation of the rate of temperature decrement with height during the month of January .
TABLE V. Temperature Decrements at different times of the day during January .
Difference of Mean Rate per 1000 feet Difference of Maxinla Rate per 1000 feet Difference of Minima Rate per 1000 feet Roorkce .
Murrec The vertical decrement of temperature in North Bengal at the hottest time of the day is , therefore , 43 per cent. more rapid than at north-western stations , namely , F. per 1000 feet , instead of or , according to the Table .
These of decrement are not those which actually obtain in the free atmosphere , since they include a certain variation with latitude ; and , moreover , there is reason to believe that , on mountain slopes , the decrease is considerably less rapid than that given by balloon ascents .
The ratio between rates for the two regions compared is , how* This column is derived from ' Indian Vado Iecum , ' .
The are from my paper on the Temperature of North-Western India , ' Indian Metcorologicul M ' vol. 2 .
of the winds from their theoretical directions , in the two regions compared , may be explicable by KoPPEN'S hypothesis .
The following considerations will enable us to discuss this point a little more fully .
At page 236 of his valuable work on the Meteorology of the Bombay Presidency , Mr. C. CHAkJBEBS , F.R.@ .
, has shown that rapid convective movements of the atmosphere , of the kind he calls " " topsy-turvy , \ldquo ; probably extend in the day time up to the level at which the ternperature is the same as that given by the law of convective equilibrium , which may be proved from thermodynamical nciples to be , for dry air , decrease of 1o F. in 183 feet .
There is no reason , however , to believe that all interchange between successive layers ceases at this level , for the molecular theory of gases indicates that the ultimate distribution of temperature would be uniform in a vertical column of the atmosphere free from all disturbing causes .
It is difficult to estimate , with any exactitude , the height to which such rapid convective movements here extend , for we do not know the rate of temperature decrement in the air ovel the plains India ; all our observations at high levels having been mountain stations , where the temperature , at least in the day time , is higher than it would be in the free atmosphere .
The results of GLAISHER'S balloon observations may help us , however , to obtain some idea of this height .
When slightly smoothed , they may be thrown into the form of Table VI .
, which gives a rapid decrement of perature in the first 2000 feet , especially with clear sky .
TABLE VI.\mdash ; Rates of Temperature Decrement from GLAISHER'S Observations :\mdash ; This Table is represented graphically on Plate 21 , which also shows the line of uniform decrement under the law of adiabatic convection .
It will be seen that , RELATION TO THE DIbTRIBUTION OF TRIC PRESSURE .
according to the principles above discussed , the ) -turvy movements extend , when the sky is clear , to an average height ( in summer in ) of about 2100 feet , whilst in cloudy weather they can hardly take place at all .
In India there is , in all probability , a still more rapid decrement of temperature near the ground , when the sky is free from cloud .
From four years ' made at Alipore , near Calcutta , * I find that at the hottest time of the day , about 2 .
during the months of greatest serenity , the temperature decreases F. between 4 and 40 feet bove the ground , or at the rate of per 1000 feet .
This we may take to be the most probable initial rate of decrease for clear weather .
In the cloudy weather of July the rate is only per 1000 feet , or between 4 and 40 feet .
Both ) are very much greater than the initial rates given by GLAJ Sions for ; which , by drawing tangents at the starting point of the curves on Plate 21 , or the formulae representing them , are found to be and ) 1000 feet respectively ; and we may infer that convective action is consequently much more energetic in India .
Whilst observations made at mountain stations thus give rates of decrement which are probably different from those obtaining in the free atmosphere , at all even ts for the layers of the atmosphere nearest the ground , we still , however , get an idea of the intensities of diurnal convection currents over different parts of the Indian plain by a comparison like made in Table V. On the Central Indian plateau , and in South Rajputana ) Sindh , where there is usually but little cloud in January , and the , of { emperature is very reat , such convection movements doubtless take place much more frequently , and extend to a greater height than they do in the Punjab and North-Western Proyinces .
To the interchange thus effected may probably be attributed the anomalous wind direction erved at Neemuch .
During the hot weather , vertical convection currents are extremely active oyer Northern and Central India , as testified by the constant occurrence of dust-whirls , familiarly known as " " devils as well as by the frequent piling up of cumulus clouds in the afternoons and the occasional occurrence of thunderstorms .
Another evidence of such action is the gradual charging of the air with dust up to a height of S000 feet or more ; so that , just before the rains set in , it may be likened ) to muddy water or pea-soup .
At this season , the mean temperature of the 24 hours decreases rapidly on ascending ; the mean rates on mountain slopes , when variations in latitude and longitude are eliminated : being the owing in the month of * ' Indian Meteorological Memoirs , ' vol. 2 , et seq. See ' Indian Meteorological Memoirs , ' vol. 2 , pagcs 182 to Sea-lcvel to 1000 feet allowance is made for the decrease of diurnal range on ascending , these rates become , for the Such very rapid rates for Central India are partly confirmed by observations made during the present year ( 1886 ) in a tower at Allahabad , which for April and May give initial rates of decrement of and per thousand feet respectively at the time of diurnal maximum .
According to the law of decrement deduced the last Table , the -turvy movements described by CHAMBERS would , at the time of the day , ascend the slopes of the North-Western Himalaya to an elevation of some 2600 feet ; while , over the mountain tops of Central India and Rajputana , they would rise to aboye 11,000 feet , if the law of temperature decrement held good .
The heights thus computed , of course , have no pretension to exactness ; in fact , the former is undoubtedly too low , for the temperature decreases much less rapidly on a broad mountain zone heated by the sun it does in the free atmosphere .
They suffice , however , to show that about midday , in the hot season , convective currents are very active to an altitude of several thousand feet over the drier part of Northern Central India .
It is probable , therefore .
that the wind direction anomalies , which are so striking at this season , ) be explained by the descent , from a considerable , of atmospheric strata which retain , for a time , the velocity acquired by them under ) pressure differences at the level from which they descend .
temperature of the HimaJayan slope may be represented by the formula , while according to the ultimate law of convection being the sealevel temperature and the expressed in thousands of feet .
Equating these , we get feet .
For entral India and Rajputana , the formula best expressing the results in the Table is , from which we find the height , where the temperature would be the as that given by the of couvective equilibrium , to be 11,330 feet .
RELATION TO THL OF TRIC P In Northern Bengal and Assam , at season , the ) of maximum temperature with height is so slow in the first three or four feet , no rapid convectiye action during hine can take place .
The 286 feet , and Darbhanga , feet ; at , 4794 feet , and Kathmandu , 4388 feet ; and at Darjiling , 7421 feet , give a combination from which variations in latitude and longitude are almost completely eliminated , and from the maximum temperatures of these places in May we get the rates of ement : Sea-level to 1000 feet 1000 , , 2000 , , . . . . . . . . .
3.3 . . . . .
3.6 These are so much slower than those for North-Western and Central India , that there is , comparatively speaking , no disturbance of the kind aboye discussed , and accordingly we find that the anomalous hot winds do not usually penetrate into Bengal , or become greatly modified in character by mixing with the lower atmospheric strata , if they do .
During the earlier part of October , south-west monsoon conditions over Bengal , and the wind directions are those given by the local pressure dients ; but , towards the end of the month , convection currents , induced by high of temperature , become active , the day winds at the same time backing from east to north and north-west , thus causing the resultant for the month to be almost due north .
In Central and North-Western India , where this month is characterised by clear skies and a large temperature range , there is ) a good of convective action , as also be inferred from the afterno of in this month ( see Table III In this region , the mean temperature in October decreases with ) above sealevel at the following rates:\mdash ; Sea-level to 1000 feet , 3.69 , , 3 .
When these are corrected for the diminution of daily is extremely rapid\mdash ; the decrease , for example , between fact the putana plain and 4000 feet ( Mount Abu ) being from about to \mdash ; the figures become for the daily maxima:\mdash ; *Formerly 6941 fcct , but the ) moved to ncw sitc in 1882 .
MR. S. .
HILL ON ] ] WINDS OF NORTHERN INDIA , AND THEIR Sea-level to 1000 In this legion , therefore , we should expect the convective action to be even more energetic after the close of the rainy season than in the dry hot weather , though the rate of decrement falls off so rapidly that probably the action does not extend to so great a height .
If , therefore , we find , as we do , that the wind directions in October conform more closely than in to the normal directions inferred from the distribution of pressure at sea-level , it is probably because there is less difference in the distribution of pressure at low and high levels in October than in May .
This will be } in the next section .
.\mdash ; The Velocity Anomaly .
In no part of the world is the diurnal variation of the wind velocity better marked than on the plains of Northern India .
In the rainy season there is more or less wind both and day , though the velocity at night is , as a rule , considerably less than in the day time ; but during the dry season , from October to May , the nights are almost always perfectly , the only occasions when there is any wind at night occurring during the showery weather sometimes observed in January or February , and in occasional evening dust storms towards the end of the dry hot weather .
At night , as the observations made at Alipore prove , the temperature increases for some distance with height above the ground , and under ordinary circumstances , therefore , there can be no convective interchange between upper and lower atmospheric strata .
The diagrams on Plate 29 of the second volume of the ' Indian Meteorological Memoirs ' show that from about 6.30 .
to 8.15 A. , on the average of the year , temperature increases with .
Between these hours , therefore , the wind movement is that due to the local pressure gradients at sea-level , whilst in the day-time , when convection takes place , the velocity is greater , owing to the descent of air from regions where the retardation by friction is much less than near the ground .
This diurnal variation may be illustrated by the results of the Calcutta anemograms published at p. 23 of the first volume of the 'Indian Meteorological Memoirs , ' and by three-hourly readings of a common anemometer made at Agra on four days each month for eight years , both of which are given in Table VII .
For the last few years , valuable anemographic traces * Another reason is that this rapid diminution of temperature in October obtains only during a short time each day .
The temperature on the plains rises and falls so rapidly that , as will be shown in the next section , the mean rate of decrement during the six hours from 10 .
to 4 P.ll .
is only about half as great as that at the hottest time of the day .
RELATION TO BUTION OF TRIC p ] have been obtained three stations in ndia -Jeyl ) , oorkee , and Lucknow\mdash ; but they have not been discussed yet .
TABLE VARIATION OF WIND VELOCITY .
The figures in the Table represent annual means .
They prove that the velocity attains its maximum about the hottest time of the day , and its minimum about midnight at Agra and 4 .
at Calcutta : the exact instants of the points , found by taking diffel.ences , being 3 .
and 12 .
at Calcutta , and .
and 13 .
at Agra .
The anemograms for several years at Karachi have been discus.sed by Mr. .
CHAnIBLRS and published in vol. 1 .
of the ' Meteorological Memoirs , ' but , as station is on the sea-coast , where there is little frictional retardation , the hourly means do not exhibit anything like such a large variation in proportion to the mean velocity , which , at that station , is nearly 17 miles per hour .
The extrell ) hourly distances traversed are miles between .
and , and milcs between .
and At all the stations in North In dia , for which the hourly variations of the wind resultants have been a double diurnal oscillation , related to that of barometric pressure , has been more or less distinctly observed ; but the amplitude of this double oscillation , which be taken to depend the diurnal variation of pressure differences near sea-level , is extremely small in comparison with that of the single oscillation due to diurnal heating of the ground and consequent convect ve action .
We have already seen tJlat , in the month of , the pressure differences ' the upper Gangetic pl ins are often eyanescent ; and it will be shown below that , this region , the differences of the mean pressures are very small at all times of the year .
It must frequently happen , therefore , that any east and west pressure gradient , which may exist , is due solely to the difference in phase of the diurnal variation owlng to difference of itude ; the epochs of maximum and minimum , the range of this variation being very nearly constant ] many degrees of longitude .
Thus , taking two points 15 degrees the pal'allel of habad , we should find * 'Inclian feteorological Memoirs , ' vol. 1 ( Papers 1 , 9 , and 10 ) .
{ .
S. A. HILL ON THE WINDS NORTHERN INDIA , AND following pressure differences on days in May he mean pressures at sea-level alike : \mdash ; VIII .
sure Differences in May dependent on Diurnal Variation only .
Hour Iour Iour Difference in Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour Difference in Jour 12 to 13 , , 14 , , 0248 15 , , 16 , , 17 , , These diHel.ences , which are taken the ' Indian Meteorological Memoirs , ' vol. 1 , page , would produce N.W. winds , indicated by the positive sign , from about .
to 17 , and from 22 .
to 2 , the remaining hours having gradients for S.E. winds , indicated by the negative sign .
Now , from the paper last referred to , we learn that , the hot season , the rebultant wind at Allahabad blows at all hours from points between S. W. and N. W. , the double oscillation suggested by the Table being combined with and masked by a prevalent N.W. direction , though it is plainly indicated in the diagram formed byjoining the ends of he lines representing the resultants for the several hours .
The differences in the Table , in fact , are so small , not exceeding0275-inch in about 813 geographical lililes , or 002-inch per degree of the earth 's surface , that near the ground , where the friction coefficient is , the winds due to them would be so light as to be almost imperceptible ; and , accordingly , we find that the double gyration of the wind vane , which would be produced by this cause , is almost entirely hidden by the influence of the more powerful currents of the higher atmosphere which descend in the middle of the day .
The N.W. winds due to the diurnal inequality of pressure are , however , probably more than twice as strong in the afternoons as in the early mornings , the maximum pressure differences being in the ratio of to 112 , so that this cause combines with cvective interch nge between the lower higher strata to make the velocity greatest about the hottest time of the day .
On comparing the mean pressures for each month at several stations on the Upper Gangetic plains , we shall find , as been stated in the introductory paragraphs , that their differences have no apparent relation to the mean wind .
but the latter has , in the annual as in the diurnal period , a very distinct relation to the temperature .
In the North-Western Provinces and Oudh there are four stations , for which we possess registers extending , with few interruptions , over nearly 18 years .
When the monthly mean pressures of these are reduced to sea-level and the value of gravity at latitude , we get the figures given in Table IX .
, the differences of which from place to place are strikingly small .
RELATION TO THE DISTRIBUTION OF BAROMETRIC PRESSURE .
TABLE IX.\mdash ; Mean Pressures at Places in the North-Western Provinces and Oudh , reduced to Sea-level and Latitude If we take the central station of the group\mdash ; Lucknow\mdash ; and measure the gradient from it towards each of the others without reference to the direction of the slope , we get the mean and extreme gradients per 60 geographical miles , which are compared in X. with the mean wind velocity at the same four stations , and with the mean temperature and the diurnal temperature range .
TABLE X.\mdash ; Wind Velocity in the North-Western rovinces compared with sure Gradient and Temperature .
This Table fully justifies the statement above made as far as the monthly means concerned .
The smallest values of the gradient , both mean and extreme , are found in September , when the wind velocity does not differ much from the .
The mean gradient is steepest in May and November , the one month having a wind velocity , and the other the lowest of the year , while the maximum value of the steepest gradient is also in Noyember , the month of least wind lnovement .
The gcstion , CCLXXXVII .
.3 MR. S. A. ON THE WINDS OF ORTHERN INDIA , AND THEIR however , arises that the pressure differences of any two months may have the same small mean value , while in one month the barometer stood at a uniform height all over the country , and in the other there may have been a series of more or less violent nces c strong winds ; in short , it is useless to compare mean values instead of actual observations .
In North India , however , no feature of the meteorology is more remarkable than the simultalleity of rometric movements oyer alarge extent of country ; all usual fluctuations , which are very numerous , though of small range , even in the most settled weather , occurring with the most absolute uniformity over an extent of country often larger than France , Germany , and Austria .
Small disturbances , of the nature of cyclonic storms , do occur occasionally , more especially during the rainy season and in the cold weather , during the month of January or about the end of December ; but , while they , doubtless , make the mean wind velocity , for the mo1lths mentioned , somewhat greater than it would otherwise be , they do not perceptibly interfere with the regular marct ] of its annual variation .
Table X. also shows , contrary to what might be expected from the part of this Paper , that the velocity is much more distinctly related to the actual temperature of each month than to the daily range of the thermometer .
The explanation of this probably is that in the interior of Northern India the daily range at all seasons is sufficient to set up convection currents , while the velocity , of the winds which descend from a considerable elevation over the plains is dependent upon the pressure gradients prevailing at high levels .
Now these gradients are doubtless subject to a distinct annual range , depending on the temperature ; for , if the plains be heated more than usual relatively to the Himalayan slope , the result will be to raise the planes of equal pressure and make them incline towards the mountains , thus giving rise to westerly upper currents , the velocity of which will , coeteri.paribus , be proportional to the temperature difference between plains and mountains .
Now , if we compare the mean temperature for stations on the plains of the NorthWestern Provinces and Oudh , given in Table X. , with that of the mountain slope at height of about 5700 feet , as computed from the observations of Chakrata , Ranikhet , Pithoragarh , and Kathmandu , we find the following differences , which are subject to a very distinct annual variation , nearly coincident in phase with that of wind velocity the plains RELATION TO THE DISTRIBUTION OF BAROMETRIC PRESSURE .
355 A similar relation obbains to the greatest elevation at which observations have been made on the mountains .
The temperature difference , between Leh ( 11,503 feet ) and the Punjab plains , is greatest in May and June and least in Noyember ; so that , if we may assume the temperature decrement in the free atmosphere over the plains to be less variable than on the mountain slope , as seems priori , there must be a greater presstlre gradient for westerly winds at high levels in May and June than in November .
The proper way to verify this and various other deductions , which have been made from the hypothesis that convective interchange is the cause of the observed wind anomalies , is to the distribution of pressure over India at a considerable height in the , say 10,000 feet .
For this purpose I have selected some 40 tions out of the number in India and Ceylon for which 10-year of pressure are published in the Report for 1884 , and have tried to reduce their mean pressures for January , May , July , and October to the proposed elevation .
The majority of stations selected are those on mountain tops or high plateaux , but , in order to complete the maps on Plate 20 , fourteer ] stations situated at levels below feet have had to be added .
Table XI .
gives the stations , with their elevations and the mean pressures of the months selected .
Two of the stations mentioned , Dodabetta and Shillong , are not now on the list of meteorological observatories , but their pressures and temperatures have been taken from old reports .
In the case of , the published barometric heights had first of all to be corrected for the temI ) e , rature of the mercury .
The elevations given have been found by levelling to Great Survey stations , or other datum leyels , in every case except Pithoragarh and Shillong .
The elevation of the last-mentioned has been found by comparing barometric obscryations at the station with those made simultaneously at Darjiling and Silchar , which places are on opposite sides of Shillong , and are respectively considerably higher and lower than that station .
The error in the ermination cannot amount to more than two or three feet .
The elevation of has been computed by comparing , month by month , the barometric obseryations of four years at that station with those of Ranikhet and Kathmandu , and taking the mean of the results .
The error of this mean must be very small .
The pressure observations of two stations in Southern India , Wellington and Mercara , have had to be rejected , as , reduced to a common level , they did not agree with those of neighbouring stations , such as Coimbatore , Dodabetta , and Bangalore , while the latter agreed very fairly with one another .
The assigned elevation of Wellington appears to be about 200 feet boo high , and that of Mercal.a feet too low .
These have been deduced trigonometricaJly or by spiritlevelling from points fixed in the early days of the Suryey , when the uncertainties of terrestrial refyaction were not sufficiently allowed for .
The observations made at Simia have also been rejected , as there is some uncertainty about the true elevation of the barometer .
TABLE XT.\mdash ; Observed Pressures KarachiDiamond IeshawarAjmereMount AuRanikh.ibag .
Pachmarh .
PoonaNewara EaChik 1mDaccaSecun , oreSta.tion .
In the formula connecting barometric pressu important variable is the temperature of the ai to find out the cause of the anomalous day time when these wiuds blow , is that which a height of 10,000 feeb , which is now atten ) this is obtained with sufficient accuracy by com that of the usual hours of observation , not been at these hours , as at the Cey RELATION TO THE DISTRIBUTION OF BAROMETRTC mean a correction derived from some similarly ated station which is not too far off , bhis correction being made proportional to the daily range .
These day temperatures are given in the 2nd , 3rd , 4th , and 5th columns of TABLE XII .
\mdash ; Temperatures used in constructing Table XIV .
* The temperature of this station appears to be about too low in comparison with Coimbatore , near which it lies .
Coimbatore , is , however , in tho dry region to the lee of the mountains , and Dodabetta is compared with the west coast stations it appears warm in comparison .
Thus mean day temperatures of Cochin Mangalore when reduced to 10,000 feet January .
May .
July .
MR. S. A. HILL ON THE WINDS NORTHERN INDIA , AND THEIR The pressures given in ble XI .
do need any correction for daily range , in every part of India the height of the baro1neter between 10 and 16 hours not differ by more than a few thous ndths of an inch from the of the twenty-four hours , and it will be seen immediately that , with the unavoidable uncertainty there is as to temperature , it is unnecessary to take such small differences into account .
It has already been pointed oui , in ssing the probable height to which convection currents extend at the hottest time of the day , that nothing accurate is known rding the decrease of the day temperatures with height above the sea in the free atmosphere .
But , by the differences between the day temperatures given in Table XII .
for stations in North-Westerll and Central India and the sealevel values of the mean temperatures of the same places , published in vol. 2 of the Memoirs , ' pages 137-141 , and combining the results into groups for stations below 2000 feet , between 2000 and 5000 feet , between 5000 and 8000 feet , and above 8000 feet , we construct a set of curves giving the variations of temperature on mountain slopes , unaffected by changes of latitude or longitude .
By graphic interpolation we can , then , get from these the rates of decrement .
The results for the four months in the Table are shown in Table XIII .
They exhibit a considerable diversity of character , and a tendency in the summer to an crease of temperature with height at the greatest elevations , owing to the greater diurnal range of the thermometer at Leh than at the stations on the outer ranges about 7000 feet elevation , whilst it is practically certain that there is no such tendency in the free atmosphere .
TABLE XIII.\mdash ; Vertical Decrement of Day Temperature on the Mountains of Central and North-Western India .
total decrement between level and 10,000 feet is less in every month than found GLAISHER in the free atmosphere over England and France , a , s was to RELATION TO THE DISTRIBUTION BAROMLTRIC PRESSURL .
be expected .
The difference is not very great in the summer months , but amounts to or more in January .
At this season , however , the decrement in .
is , as we have seen , much more rapid , and such appears also to be the case in Southern when we compare the ture of Dodabetta with of Coimbatore and the nearest stations on the west coast .
For the whole count ry , therefore , after due consideration , I have decided to adopt the rates of decrement found GLAISHER , and given bove in Table VI .
These include an allowance for di1ninishing diurnal range , since the observatiens were made in the day time only .
Though considerably more rapid for some nths than those in the last Table , they are , perhaps , too slow on the whole , the decrement in India being probably greater near the ground than in England , where ] ation is not so powerful .
The last four columns of .
give the probable temperatures at 10,000 feet , deduced by means of Table VI .
, the rate for cloudy skies being used when the average cloud proportion for the month amounts to half the expanse or more , and the rate for clear skies in the other nlonths .
The formula adopted for the barometric reductions is the simple one:\mdash ; With the values of and in Table XII .
, and the mean value of , about 21 inches , it can be shown , by differentiating , that for a difference of elevat , ion of 10,000 feet , an error of 1o in estimating would give an error in the resultin value of lying between and 010 inch .
As the margin of possible in many of the values of may amount to several degrees , there is evidently nothing to be by using a more complicated fonnula , in which the variations of gravity and of density , owing to the presence of more or less water vapour , are taken into account .
The computed results , which have been reduced to the standard value of gravity , given in Table XIV .
Considering the uncertainty of the adopted rates of decrease of temperature , and the widely difierent altitudes of the bose stations , these results are remarkably ccnsistent .
Even apparent exceptions to this consistency , such as the low pressure over Dodabetta compared with the hbouring station Coimbatore , or of Chikalda compared to Poona , or Mount Abu to Neemuch , serve to confirm the probable correctness of the rates of temperature decrement aclopted ; for these are instarlces of the mid-day distribution of pressure which gives rise to the diurnal mountain winds , observed in all the warmer regions of the world .
TABLE .\mdash ; Mid-day Pressures at 10,000 Feet ( corrected to Gravity of Latitude ) .
Vizag pingKandy 2arhC iatore .
Madras 2Diamond.sagar 2Dacca 2Bhuj 2 rfhe pressures given in this Table are charted on Plate 20 , which is headed " " Mid-day Pressures at 10,000 Feet These charts deserve some study in detail .
In January , the region of lowest pressure , mded by the isobar of inches , covers a considerable part of Afghanistan and the , and extends through into Ladakh and Thibet .
The north-east corner of the PmJjab and the region of the Hindu Kush have probably a slightly higher pressure .
The isobar of RELATION TO THE DISTRIBUTION OF PRESSU R inches sends a long loop down the valley of the Ganges to the north of Delhi .
The next two isobars are very similarly curved , but they extend further to the east over known regions , and are seen to curve southward oyer } hcad of the Bay .
The others probably form closed curves of a triangular form round a centre lying in the western near the town of Sholapur .
The system of winds , pertaining to this distribution of pressure at a where the friction coefficient is very small , would be S.W. over the coast , indh , and Rajputana , S. or S.W. at most of the ayan stations , N. or .
at Murree , S.E. along the foot of the Central , N.W. in Bengal , W. over Central India and the Gangetic plains , over neal .
lie the whole of the hIadras Presidency , or S.E. in Travancore and Cochin .
Now in almost every ce where the observed prevailing wind direction for this month is inconsistent with the distribution of pressure at sea-level it is in accordcmce with this upper system of wind A possible solution of another question of reat interest is ested by the chart .
The cause of the disturbances produce the winter rainfall of the Punjab , NorthWestern Provinces , and jputana , and the snows of the North-west Himalaya , is not yet fully derstood .
Mr. F. CHAMBERS has suggested that these precipitations may be due to disturbances of the same nature as the winter storms of latitudes entering In dia from the west , while Mr. BLANFo 1 attributes their to the disturbance of atmospheric equilibrium by the diffusion and condensation of water vapour formed locally .
It is difficult to understand ) a cyclonic , if it be of the , same nature as those formed in the Bay of Beng.al at the turn of the seasoI ) can cross a mountainous country like Afghanistan ; but that irling stonms do cross the Rocky Mountain region is a common experience in America .
Without presuming to dogmatise , therefore , about a subjec4 which I have not sufficiently studied , I would suggest that it is possible such storms be formed in more than one way , and that some of them at least may travel for ances if they do not originate in a region of relatively low pressure in some of the atmosphere .
Such whirling storms , crossing the Indus in the month of .
] might travel along the axis of lowest pressure directly into Kashmir , or they might pass along the -shaped depression in the Valley of the Ganges as ffi as Behar ; but , owing to the higher pressure in West Bengal and Nepal , could seldom to Assam , except by the roundabout way of Thibet , in which case their chance of survival after twice crossing the Himalayas would be infinitesimal .
Now these possible paths suggested by the form of the isobars are precisely those most fi.equently taken by the disturbances which bring the winter rains .
The chart for May shows the distribution of pressure at 10,000 feet elevation during this month is extremely sinll ) if we lcave out of consideration local ' Nature , ' vol. 23 , page 400 .
'Journal of the Society of Bengal , ' vol. 52 , Part 2 , and 'Indian Metcorological ' vol. 3 , pa in the Satpuras , such as cause the diurnal moulltain winds above referred to .
The pressure in the Deccan is much enlarged , and the centre shifted to ound this the isobars curye in a nearly triangular form , following closely the outline of the country as bounded by the sea , the Hala and Suleiman ranl o and the ; the most tant exception to this rule being a recurv tion of the lines of equal pressure round a low-pressure area extending obliquely across Bengal from Balasore to Assam .
The wind system , induced by this distribution of , would be S. or S.W. in the valley of the Indus and utana , W. or N.W. over the rest of Northern India , as far east as Lower Bengal and as far south as .
in Eastern Bengal and Burma , N.E. along the eastern side of the Peninstl ] from Orissa to , and S.E. or S. on the Malabar coast .
That is to say , over the interior of Northern and Central India the upper currents in this month have exactly the same direction as the anomalous hot winds which we have reason to suppose caused by their descent under the convective action set up by the daily passage of the sun .
The steepest gradient on the map is that to the of the Orissa coast .
Disturbances originating at the head of the Bay would therefore , according to the usual experience that the line of least resistance is that of lowest pressure , experience much difficulty in proceeding westwards , but would meet with less resistance in the northward direction oyer Bengal .
As a matter of fact , the majority of such storms as are formed in this region during May do proceed northwards , and those which turn westwards across the plain of Orissa do not survive to reach the Central Provinces .
Mr. LIOT aibutes their rapid break-up to the comparatively small height to which they extend in the atmosphere , and the resistance they experience from the hills west of ; but , while this may be one reason for their disappearance , the rapidly pressure in a westward direction at great elevations is doubtless another .
The low-pressure area extending obliquely across Bengal and Assam is the scene of frequent local storms , known as " " Nor'-Westers , \ldquo ; in the hot season , and the area of maximum " " spring storm rainfall\ldquo ; on Mr. BLANFORD'S charts coincides very closely with it .
At page 105 of the work cited , gives a clear and interesting description of these storms and their origin , which he attributes to the mixing , by convective action , of the southerly sea winds with the north-west winds , which blow over the Chutia plateau , and continue their course eastwards over the low plains of Bengal at nearly the same level .
The line of maximum rainfall in May , according to Mr. Table , passes through or near the following laces , the rainfall gradually as the distance from the increases and the Khasia Hills are Accounl of S.W. Monsoon Storms generated in the Bay of Bengal during the years 1877 to 1881 by J. ELIOT , ' Indiall Meteorological Memoirs , ' vol. 2 .
" " The Rainfall of India ' Indian Memoirs , ' vol. 3 .
RELATION TO TH DTSTRIBUTION OF BAROMETRIc PRESSURL .
Balasore inches Midnapore Krishnaghal .
Jessore aridpur . . .
8.33 Dacca . . .
Mymensingh 11.74 21.64 Cherra Punji 51.46 The line joining these places coincides almost exactly with the axis of lowest pressure at 10,000 feet on the chart for May .
July the triangular area of high pressure in the Deccan has retreated to tho south-east of the Peninsula , and occupies those parts of the Carnatic where the rainfall at this time of the year is very light .
Along the foot of the Himalayas there is a zone of relatively high pressure , inches , or htly less , which widens out in the west , so as to include the western Punjab and Afghanistan .
Parallel to this , and south of it , from the head of the Bay of Bengal to Sindh , extends a region of lower pressure , below inches ; which is , however , terrupted along the line of the Aravalis by a belt of higher pressure , exceeding inches in the vicinity of Ajmere .
As regards the country south of the parallel of , there can , therefore , be but little difference in direction between the winds prevailing at high and low levels , and , except for the diminution of frictional retardation on ascef ) ding , the upper currents have probably no greater velocity than those at the ground surface , since the gradients of 10,000 feet are no steeper than those at sea-level .
In Bengal the winds at this elevation should be easterly or ] , while those deduced from the distribution of pressure at sea-level by Buys BALLOT'S Law would be southerly .
The prevailing winds are south-easterly , or in Cachar north-easterly , the average angle between the wind and the isobars .
about , an effect which is possibly due in part to convective interchange ; though , as we have already seen , the activity of such interchange at this season is not great .
The northerly winds at Multan , Jeypore , and Agra , to which attention was directed in the first part of this Paper , evidently the of the high pressure in the upper strata over the vali region respectively , these stations being all in the drier part of North-western India , where clear skies and a high temperature , with their accompaniment of active convection currents , are not infrequent even in July .
The extensive belt of low pressure , at high altitudes , stretching across the Central Provinces and the head of the Bay , is doubtless connected with the fact , brought prominently to notice in Mr. ELIOT'S paper last referred to , ) cyclonic storms formed during the prevalence of the S.W. monsoon in land across the Orissa coast , meeting with little obstruction from of hills they have to .36 .
S. ON OF NORTHEiRN [ NDIA , AND often .
right across India to Sindh , the Rajputana desert , or ester)Pnnj ; while those the autumn transition periods , which either do not to so great a height or meet with increasing pressures at tions as they go , very seldom succeed in .
from to tlJe Central inces .
From the charts attached to Mr. ELIOT'S paper it ) ppears that , during the five years , inclusive , 30 cyclonic storms crossed the coast near the head of the Bay in the months of June , July , August , and September .
Of these , 23 proceeded westwards at first , and the seven went northward into .
Onlv five storms occurred in the month of July , ] of which crossed the coast in the westward direction .
The majority of such storms either proceed almost due west towards Sindh , or turn off northwards to the valley of the Ganges , doubtless owing to the obstruction presented by the -pressure region in Eastern Rajputana .
Of the 23 above enunlerated , off the Ganges Valley , 2 succeeded in the Aravali range near Ajmere and ultimately reached the Indus Valley , and the remainder either proceeded the Narbada Valley to Gujerat and Sindh , or died out during the trans-continental .
To the frequent passage of such storms over the Central Provinces and Malwa is doubtless to be attributed the fact of these ions having a vier and less variable rainfall than any other part of the interior of the Indian Peninsula .
In October , the highest sure is over Rajputana and Malwa , and the lowest in Arakan nd Upper Burma , and in the Trans-Himalayan regions of Ladakh and Thibet .
The general appearance of the isobars is that of a system of more or less wavy loops surrounding the centre in Rajputana , and extending to an unknown distance westwards over the Arabian Sea .
The system of winds due to this distribution of pressure must be E. or S.E. in the extreme south of the Peninsula , S.W. in the Indus as well as in Lower Burma and Tenasserim , W. in the Punjab , N.W. in the Valley , N. over the Bengal delta , and N.E. over the rest of the country .
Except in and on the western side of the Indus Valley , this is not dissimilar to the system due to the pressure distribution at sea-level , and accordingly , we find it is only in these parts of the counbry that there are any marked anomalies in the ection of the wind observed on the plains .
In collfirmation of the general direction of the circulation of the atmosphere at 10,000 feet elevation , we have the evidence of the winds at the hill-stations in the ayas and in Southern India .
On the whole , though these stations , with the exception of Leh , are only at elevations between 6,000 and 7,000 feet , the of the actual wind directions with those given by the distribution of * Tnstances have been known to occur in which such storms , after crossing the Orissa coast , gave little or no indication of their existence on the charts of ] evel isobars , but after a day or twq ared in the Narbada valley .
This seems to provs that , unlike the cylones of October , they are -level ) RELATION TO THE UTIO OF pressure at 10,000 feet is very fairlv may be seen by ) the wind directions on with the isobars on Plate 20 .
For greater elevations I have only the eyidence of the nephescope , by means of the directions of movement of the cirrus clouds been observeCl for eight observations were recorded and originally tabulated to points , but in them here I have reduced the number of points to S. The total number of obselvations , during the eight years , inclusive , with the resultants computed by formula , are given in TABLE .\mdash ; Ob.served Movements of Cirrus Clouds at the years 1878-8 February yJune Month .
August September October November .
December .
resultants for January , May , and October are consistent with the distribution of pressure at 10,000 feet , but that for July has too large a southerly component .
The very small percentages of steadiness for the three rainy months of lie , and September , however , indicate that the pressure gradients of the monsoon are evanescent at the level of the cirrus clouds .
As regards the country north of the parallel of and west of the meridian of E. , the distribution of pressure in October is strikingly similar to which obtains in May , but the gradients cure much less steep .
Therefore , although on account of the large diurnal of temperature , convective may be quite as active in October as in May , the winds which blow in the afterno are not nearly so strong .
In Table XVI .
, the maximum gradient Lucknow , at 10,000 feet above sea-level , is compared with the gradient at sea-level with the mean wind velocity at North-Western Proyinces stations , for each of the four months under review .
MR. S. A. HILL ON ffIE$ WINDS OF NORTHERN INDIA , AND THEIR TABLE XVI.\mdash ; Comparison of Wind Velocities with Pressure Gradients over the Gangetic Plains . . .
It seems probable from this Table that the high velocity for July may be attributed , partly at least , to the higher pressure gradient at sea-level in that month than in any of the others , and that the lower velocity in October than in May is the result of the lower gradient prevailing at high levels , The relation between the wind velocity and the dients at and low levels may be.investigated as follows:\mdash ; If be the velocity of the wind near sea-level in miles per hour , and that prevailing in a higher stratum where the gradients are the sam as at 10,000 feet in the Table , and if be the number of hours during which the wind from this high level descends to the earth 's surface , then the total distance traversed daily will be .
Should the wind not descend from a leyel where the gradients are so high as those given in Table XVI , , will not the true time during which the upper cLlrrent descends , but will be longer or shorter in inverse proportion to the dient .
From 's fornlula for the relation between the wind velocity and barometric gradient , given in his work on the Moyements of the Atmosphere , we get , on reducing to English measure , and taking the unit of distance to be one degree of the earth 's surface , ; or , aking equal to the mean Iatitud.e of the Gangetic ( about ) , and neglecting the angular velocity , we get .
In arriving at this result , however , frictional retardation has been neglected ; , if the ratio between the actual and the theoretical velocity for a given gradient be called , we may put for the actual velocity B. Substituting for and in the previous formula their values thus expressed in terms of the barometric adients for each of the four typical months , and assuming the probab ] values since the equations are indeterminate , we get the following results :\mdash ; Month .
January May 5.4 July ] Octobel 3.3 TABLE XVI .
ison of Wind Velocities with Pressure Gradients over the Gangetic Plains . . .
It seems probable from this that the high velocity for Jnly may be attributed , partly at least , to the higher pressure gradient at sea-level in that month han in any of the others , and that the lower velocity in October than in May is the result of the lower gradient prevailing at high levels , The relation between the wind and the dients at and low levels may be.investigated as follows:\mdash ; If be the velocity of the wind near sea-level in miles per hour , and that preyailing a higher stratum where the are the as at 10,000 feet in the Table , and if be the number of hours during which the wind from this high level descends to the earth 's surface , then the total distance traversed daily will be .
Should the wind not descend from a leyel where the gradients are so high as those given in Table XVI , , will not the true during which the upper cLlrrent descends , but will be longer or shorter in inyerse proportion to the dient .
From 's fornlula for the relation between the wind velocity and barometric dient , given in his work on the Moyements of ths Atmosphere , we get , on reducing to English measure , and taking the unit of distance to be one degree of the earth 's surface , ; or , aking equal to the mean Iatitud.e of the Gangetic ( about ) , aIJd neglecting the angular velocity , ws get .
In arriving at this result , however , frictional retardation has been neglected ; , if the ratio between the actual and the theoretical yelocity for a given gradient be called , we may put for the actual yelocity B. Substituting for and in the previous formula their values thus expressed in terms of the barometric adients for each of the four typical months , and assuming the probab ] values since the equations indeterminate , we get the results :\mdash ; Month .
May 5.4 Jnly ] Octobel 3.3 TABLE XVI .
ison of Wind Velocities with Pressure Gradients over the Gangetic Plains . . .
It seems probable from this that the high velocity for Jnly may be attributed , partly at least , to the higher pressure gradient at sea-level in that month han in any of the others , and that the lower velocity in October than in May is the result of the lower gradient prevailing at high levels , The relation between the wind and the dients at and low levels may be.investigated as follows:\mdash ; If be the velocity of the wind near sea-level in miles per hour , and that preyailing a higher stratum where the are the as at 10,000 feet in the Table , and if be the number of hours during which the wind from this high level descends to the earth 's surface , then the total distance traversed daily will be .
Should the wind not descend from a leyel where the gradients are so high as those given in Table XVI , , will not the true during which the upper cLlrrent descends , but will be longer or shorter in inyerse proportion to the dient .
From 's fornlula for the relation between the wind velocity and barometric dient , given in his work on the Moyements of ths Atmosphere , we get , on reducing to English measure , and taking the unit of distance to be one degree of the earth 's surface , ; or , aking equal to the mean Iatitud.e of the Gangetic ( about ) , aIJd neglecting the angular velocity , ws get .
In arriving at this result , however , frictional retardation has been neglected ; , if the ratio between the actual and the theoretical yelocity for a given gradient be called , we may put for the actual yelocity B. Substituting for and in the previous formula their values thus expressed in terms of the barometric adients for each of the four typical months , and assuming the probab ] values since the equations indeterminate , we get the results :\mdash ; Month .
May 5.4 Jnly ] Octobel 3.3 TABLE XVI .
ison of Wind Velocities with Pressure Gradients over the Gangetic Plains . . .
It seems probable from this that the high velocity for Jnly may be attributed , partly at least , to the higher pressure gradient at sea-level in that month han in any of the others , and that the lower velocity in October than in May is the result of the lower gradient prevailing at high levels , The relation between the wind and the dients at and low levels may be.investigated as follows:\mdash ; If be the velocity of the wind near sea-level in miles per hour , and that preyailing a higher stratum where the are the as at 10,000 feet in the Table , and if be the number of hours during which the wind from this high level descends to the earth 's surface , then the total distance traversed daily will be .
Should the wind not descend from a leyel where the gradients are so high as those given in Table XVI , , will not the true during which the upper cLlrrent descends , but will be longer or shorter in inyerse proportion to the dient .
From 's fornlula for the relation between the wind velocity and barometric dient , given in his work on the Moyements of ths Atmosphere , we get , on reducing to English measure , and taking the unit of distance to be one degree of the earth 's surface , ; or , aking equal to the mean Iatitud.e of the Gangetic ( about ) , aIJd neglecting the angular velocity , ws get .
In arriving at this result , however , frictional retardation has been neglected ; , if the ratio between the actual and the theoretical yelocity for a given gradient be called , we may put for the actual yelocity B. Substituting for and in the previous formula their values thus expressed in terms of the barometric adients for each of the four typical months , and assuming the probab ] values since the equations indeterminate , we get the results :\mdash ; Month .
May 5.4 Jnly ] Octobel 3.3 TABLE XVI .
ison of Wind Velocities with Pressure Gradients over the Gangetic Plains . . .
It seems probable from this that the high velocity for Jnly may be attributed , partly at least , to the higher pressure gradient at sea-level in that month han in any of the others , and that the lower velocity in October than in May is the result of the lower gradient prevailing at high levels , The relation between the wind and the dients at and low levels may be.investigated as follows:\mdash ; If be the velocity of the wind near sea-level in miles per hour , and that preyailing a higher stratum where the are the as at 10,000 feet in the Table , and if be the number of hours during which the wind from this high level descends to the earth 's surface , then the total distance traversed daily will be .
Should the wind not descend from a leyel where the gradients are so high as those given in Table XVI , , will not the true during which the upper cLlrrent descends , but will be longer or shorter in inyerse proportion to the dient .
From 's fornlula for the relation between the wind velocity and barometric dient , given in his work on the Moyements of ths Atmosphere , we get , on reducing to English measure , and taking the unit of distance to be one degree of the earth 's surface , ; or , aking equal to the mean Iatitud.e of the Gangetic ( about ) , aIJd neglecting the angular velocity , ws get .
In arriving at this result , however , frictional retardation has been neglected ; , if the ratio between the actual and the theoretical yelocity for a given gradient be called , we may put for the actual yelocity B. Substituting for and in the previous formula their values thus expressed in terms of the barometric adients for each of the four typical months , and assuming the probab ] values since the equations indeterminate , we get the results :\mdash ; Month .
May 5.4 Jnly ] Octobel 3.3 TABLE XVI .
ison of Wind Velocities with Pressure Gradients over the Gangetic Plains . . .
It seems probable from this that the high velocity for Jnly may be attributed , partly at least , to the higher pressure gradient at sea-level in that month han in any of the others , and that the lower velocity in October than in May is the result of the lower gradient prevailing at high levels , The relation between the wind and the dients at and low levels may be.investigated as follows:\mdash ; If be the velocity of the wind near sea-level in miles per hour , and that preyailing a higher stratum where the are the as at 10,000 feet in the Table , and if be the number of hours during which the wind from this high level descends to the earth 's surface , then the total distance traversed daily will be .
Should the wind not descend from a leyel where the gradients are so high as those given in Table XVI , , will not the true during which the upper cLlrrent descends , but will be longer or shorter in inyerse proportion to the dient .
From 's fornlula for the relation between the wind velocity and barometric dient , given in his work on the Moyements of ths Atmosphere , we get , on reducing to English measure , and taking the unit of distance to be one degree of the earth 's surface , ; or , aking equal to the mean Iatitud.e of the Gangetic ( about ) , aIJd neglecting the angular velocity , ws get .
In arriving at this result , however , frictional retardation has been neglected ; , if the ratio between the actual and the theoretical yelocity for a given gradient be called , we may put for the actual yelocity B. Substituting for and in the previous formula their values thus expressed in terms of the barometric adients for each of the four typical months , and assuming the probab ] values since the equations indeterminate , we get the results :\mdash ; Month .
May 5.4 Jnly ] Octobel 3.3 TABLE XVI .
ison of Wind Velocities with Pressure Gradients over the Gangetic Plains . . .
It seems probable from this that the high velocity for Jnly may be attributed , partly at least , to the higher pressure gradient at sea-level in that month han in any of the others , and that the lower velocity in October than in May is the result of the lower gradient prevailing at high levels , The relation between the wind and the dients at and low levels may be.investigated as follows:\mdash ; If be the velocity of the wind near sea-level in miles per hour , and that preyailing a higher stratum where the are the as at 10,000 feet in the Table , and if be the number of hours during which the wind from this high level descends to the earth 's surface , then the total distance traversed daily will be .
Should the wind not descend from a leyel where the gradients are so high as those given in Table XVI , , will not the true during which the upper cLlrrent descends , but will be longer or shorter in inyerse proportion to the dient .
From 's fornlula for the relation between the wind velocity and barometric dient , given in his work on the Moyements of ths Atmosphere , we get , on reducing to English measure , and taking the unit of distance to be one degree of the earth 's surface , ; or , aking equal to the mean Iatitud.e of the Gangetic ( about ) , aIJd neglecting the angular velocity , ws get .
In arriving at this result , however , frictional retardation has been neglected ; , if the ratio between the actual and the theoretical yelocity for a given gradient be called , we may put for the actual yelocity B. Substituting for and in the previous formula their values thus expressed in terms of the barometric adients for each of the four typical months , and assuming the probab ] values since the equations indeterminate , we get the results :\mdash ; Month .
May 5.4 Jnly ] Octobel 3.3 TABLE XVI .
ison of Wind Velocities with Pressure Gradients over the Gangetic Plains . . .
It seems probable from this that the high velocity for Jnly may be attributed , partly at least , to the higher pressure gradient at sea-level in that month han in any of the others , and that the lower velocity in October than in May is the result of the lower gradient prevailing at high levels , The relation between the wind and the dients at and low levels may be.investigated as follows:\mdash ; If be the velocity of the wind near sea-level in miles per hour , and that preyailing a higher stratum where the are the as at 10,000 feet in the Table , and if be the number of hours during which the wind from this high level descends to the earth 's surface , then the total distance traversed daily will be .
Should the wind not descend from a leyel where the gradients are so high as those given in Table XVI , , will not the true during which the upper cLlrrent descends , but will be longer or shorter in inyerse proportion to the dient .
From 's fornlula for the relation between the wind velocity and barometric dient , given in his work on the Moyements of ths Atmosphere , we get , on reducing to English measure , and taking the unit of distance to be one degree of the earth 's surface , ; or , aking equal to the mean Iatitud.e of the Gangetic ( about ) , aIJd neglecting the angular velocity , ws get .
In arriving at this result , however , frictional retardation has been neglected ; , if the ratio between the actual and the theoretical yelocity for a given gradient be called , we may put for the actual yelocity B. Substituting for and in the previous formula their values thus expressed in terms of the barometric adients for each of the four typical months , and assuming the probab ] values since the equations indeterminate , we get the results :\mdash ; Month .
May 5.4 Jnly ] Octobel 3.3 TABLE XVI .
ison of Wind Velocities with Pressure Gradients over the Gangetic Plains . . .
It seems probable from this that the high velocity for Jnly may be attributed , partly at least , to the higher pressure gradient at sea-level in that month han in any of the others , and that the lower velocity in October than in May is the result of the lower gradient prevailing at high levels , The relation between the wind and the dients at and low levels may be.investigated as follows:\mdash ; If be the velocity of the wind near sea-level in miles per hour , and that preyailing a higher stratum where the are the as at 10,000 feet in the Table , and if be the number of hours during which the wind from this high level descends to the earth 's surface , then the total distance traversed daily will be .
Should the wind not descend from a leyel where the gradients are so high as those given in Table XVI , , will not the true during which the upper cLlrrent descends , but will be longer or shorter in inyerse proportion to the dient .
From 's fornlula for the relation between the wind velocity and barometric dient , given in his work on the Moyements of ths Atmosphere , we get , on reducing to English measure , and taking the unit of distance to be one degree of the earth 's surface , ; or , aking equal to the mean Iatitud.e of the Gangetic ( about ) , aIJd neglecting the angular velocity , ws get .
In arriving at this result , however , frictional retardation has been neglected ; , if the ratio between the actual and the theoretical yelocity for a given gradient be called , we may put for the actual yelocity B. Substituting for and in the previous formula their values thus expressed in terms of the barometric adients for each of the four typical months , and assuming the probab ] values since the equations indeterminate , we get the results :\mdash ; Month .
May 5.4 Jnly ] Octobel 3.3 RELATION TO THE DISTRIBUTION OF BAROMETRIC PRESSURE .
These results , for the three dry months , are consistent with of the length of the , and the probable rates of decrease of teml ) Iture .
the ound ; but the result for July is much too large , and implies eithel that action does not extend to so great a height as in other mollths , or .
to the frequency of small cyclonic disturbances in July , the mean of the gradients actually obtaining from day to is consider bly steeper than deduced from the chart of monthly mean pressures .
Both reasons , doubtless , concur in producing the result just found .
The hypothesis of diurnal interchange between atmospheric strata lying at low and high levels , lght about by convection currents during the hours when the sun is shining , having thus been shown to account for the observed anomalies of the winds of Northern and Central India , both in direction velocity , I shall now proceed to enquire whether the same hypothesis will suffice to the effect of snowfall on the Himalaya in producing extraordinarily powerful and persistent N.W. winds over the Indian plains .
III.\mdash ; The Effect of Snow in the Region .
In the year lS77 , remarkable for the ahnost total failure of the summer rains , I pointed out , in a letter to the Government of the North-Western Provinces , the probability of a rule that when the winter rains of Northern are light those of the summer are heavy , and vice versed .
About the same time .
E. D. arriyed independently at the same empirical law .
The tion which led up to this law was continued and extended , and two years I published a paper on the subject in the ' Indian Meteorological ' vol. .
At page 209 of volume it is shown that out of a total number of 34 years of which rainfall statistics are discussed 25 tell in favour of the rule that , in the North-Western Provinces , the winter rainfall and that of the succeedin summer vary in inverse directions , and only nine are against it .
Mr. has ated the subject further , shown that if , instead of the rainfall during the sprlng months over the plains , we take the precipitation on the North-West } ( which usually , but not always , varies , , with the rainfall on the ) lains ) , and compare with the rainfall of the succeeding sun mer monsoon , the unfavourablo lnstances , with one or two exceptions , due to special tempol.al.y causes , all disappear .
The way , in which unusually heavy late snowfall on the xerclses a and weakenin influence on the summer monsoon , is doubtless by keeping down the temperate .
ur of those regions , which constitute one .
the goals towards which the monsoon winds ] such dry years , the westerly wlnds , which prevail over northern and western during April and , are reinforced , and continue to blow far into June or July ; or even , as in and 1877 , on through the months of the monsoon until September or October .
In his MR. S. A. HLLL ON WINDS OF HERN INDIA , AND THEIR earliest writings on this subject FORD seemed to incline to the notion that these were a nearly direct effect of the cold to the north of India , cause the ( apparent ) high pressure of the mountain region when observed were reduced to ] evel .
It has been pointed out in paragraph , however , that the result of such a distribution of actual pressure would be to cause easterly instead of westerly winds , in accordance with the usnal influence of the earth 's rotation .
In his last paper ( ' Proceedings of the Society , ' vol. , 1884 ) this idea that the north-west winds are a direct effect of ayan cold has been dropped , as will be ffie following quotation , aild he tes the explanation on the hypothesis of convective interchange , which I now attempt to give .
Ab page 18 he says : " " The question then presents itself , 'What is the of the dry westerly current ?
' The supposition hat the indraught from the south-west ishes more than a small portion of the stream is at once atived by the fact that , even at Karachi and Bhuj , southerly winds * do not preponderate oyer until May , and then almost inappreciably\mdash ; at jkot not before June ; and the very fact of the great dryness of the west and northwest winds militates against the idea that any considerable portion of their air-mass can be from the sea .
Neither is it derived to any considerable extent from the valleys and lower slopes of the surrounding hills .
There is no permanent drainage of air from these hill-slopes , and strong winds blowing outwards from the larger valleys , like the of Hurdwar , are local and exceptional phenomena , restricted to certain hours of the day .
At all the hill-stations of the outer North-West Himalaya , as far as the existing registers show , herly winds preponderate oyer northerly all through the year ; and , although this is probably due in some ) easure to the fact that the night winds have not itherto been registered , it suffices to show that , up to a leyel of 7000 feet , there is no steady outflow of air from the to the plains .
' ' There rernains then only the supposition that these winds are fed by the descent of air from an upper stratum , a current moving at a considerable elevation west to east .
And that this is their true explanation several facts seem to testify .
In the first , they are characteristically winds of the day time , theiIb moyement being at a } inimum ( almost or quite a cahu ) in the hours , and indeed up to 9 or 10 o'clock in the forenoon\mdash ; then increasing with temperature , and fallin again towards ; and , secondly , such observations as have been made on the decrease of temperature with e1evation show that , in the dry weather , the vertical decrement is such as is incompatible with the vertical equilibrium of an air column , being considerably more than 1o in 183 feet .
The diurnal variation of the movement is then probably to be accounted for on KoPPEN'S hypothesis , , the of the higher and lower air strata , by which do not affect the existing horizontal movement of the nosphere , so that the of the latter , after its descent , preserves for a time its original vard motion .
hypothesis of convective receives there support the racter of the diurna TO THE DISTRIBUTION OF BAROblETRIC PRESSURE .
variation curve of vapour-tension in a dry nosphere n the is the same in all parts of India .
This shows a rapid fall of the absolute humidity of the air after 8 or 9 o'clock in the forenoon , reaching its minimum about the time of greatest heat , and a more or less sudden rise before sunset , which it is difficult to account for on any other supposition than that it coincides with the tion of the convecbiye lnovement Having given thus clearly and concisely the evidence that the explanation of the diurnal hot winds is probably to be found in esis , Mr. BLANFORD does not go on to explain how the bundance or paucity of the snowfall on the mountains will affect their intensity .
From has gone before , however , it will be evident that their intensity is likely to be affected only by two causes : ( 1 ) the intensity of the convective action which brings about an between higher and lower atmospheric strata ; and ( 2 ) the pressure gradients and consequent intensity of the winds at levels .
The rmer , which depends on the rate of variation of temperature with elevation , is doubtless affected by rainfall over the plains in what is ordinarily the dry season ; for such rainfall invariably lowers the temperature at the ground surface , though , from theoretical reasons , it would appear raise it by the ]iberation of latent heat at the level where precipitation occurs , the double effect .
a decrease in the rate of vertical decrement .
The cooling eifect of occasional showers over the plains is , however , a transient phenomenon , compared with that of the snows on the mountain slopes ; hence it is probably by an increase of the baric gradient at levels , to the sinkin of the planes of equal pressure over the cold mountain zone , that the terly winds are intensified to the extent observed in dry years .
A thorough investigation of this part of the subject would demand the preparation of charts of the distribution of pressure at a high level for each of a considerable number of years , a comparison of the gradients on which , month by month , with the rainfall and the prevailing wind directionS over the , would enable us to completely verify or disprove the hypothesis .
It is to be hoped that , in future , such highlevel* charts may be epared for the monthly Illean pressures , as regularly as the charts at sea-level now published in the Reports of India .
Such charts could not fail to be instructive .
The labour of preparing such a series would , howeyer , be very great , and all that can be attempted ) is to ascertain as nearly as possible what were the high-level gradients prevailing over the Gangetic valley in the typical months of a few years of aracteristics as rainfall .
During the last eight years for which the Meteorological ) orts have been publiQhed , the of the Punjab , utana , Central India , the North-West Provinces , * The elevation , 10,000 feet , adopted in the charts attached to this paper , has been choscn only because it is expressed by a convenient round number , and because it lies bovo all mountaius in India except the snowy ranges .
It may be found in practice more convcl1icnt to adopt some other level .
These are the riations computed on the total rainfall of the year , which averages inches .
Owing co the opposition in sign of the variations for the winter and ummer months , the summer rainfall alone would give larger departures from the average , the defcct in 1877 for example , instead of 35 per cent. The totals for the separate seasons are not given in the summaries from which the above figures are taken ; but , as those of the summer nearly always vary in the same direction as the annual totals , the latter will suffice for our purpose , which for the present has reference only to direction and not to extent of variation .
Leaying out of account the two years 1878 and 1881 , for which the variations lie within the limits of possible error , and also , the barometric data for which at one of the stations given below are incomplete , there remain five years of distinctly marked positive or tive variations in the rainfall .
Table XVII .
gives the observed pressures and day temperatures , for the four typical months of these years , at three stations on the outer Himalayas , and three others in a nearly parallel line on the plateau to the south of the and Table XVIII .
are given the comuted pressures at an elevation of 10,000 feet .
In making the computations for this Table , the same rates of temperature decrement have been assumed as in constructing the Table of normal values , though there is reason to believe , as above stated , that these rates may be different for dry and showery months of the winter and spring ; but , then , we have at present no means of estimating exactly what the variations from month to month of this element of our calculations may be .
Some of the apparently anomalous results worked out below would , doubtless , disappear if rates of decrement were known and used .
to the heavy rainfall of the winter and spring months of 1877 , the total rainfall of the N.W. Punjab that year was considerably over the Near the end of the of each Annual Report on the Meteorology of India .
RELATION TO THE DISTRIBUTION OF BAROMETRIC PRESSURE .
TABLE XVII.\mdash ; Pressures and Temperatures at North Indian Stations in Certain Years .
ctober .
89.0 61.6 64.5 6.38 .3 RELATION TO OF BAROIIETRIC direction of the wind is powerfully mfluenced by the distribution of pressure at high levels , it is important , as far as the weather of the winter months is concerned , to know the variations of pressure from the normal at the highest hill-stations than at any place on the plains .
In May there are considerable fluctuations over both regions , the greatest range being shown by Ajmere and the next test by Ranikhet ; but in July variations over the mountains , which are always more or less moist and equable in temperature at this season , are very small , the nges not exceeding inch in the five years compared , while at Ajmere and Sutna they amount to 144 and 120 inch .
In October , as we should expect from the fact of the mean temperature of the month being about the average of the year , and the normal distribution of pressure very uniform at all levels , the range of the variation is less than in any of the other three , and is greatest at Hazaribagh , which , during the first half of the month , is on the border of the region where the rains still penetrate after they have ceased in Upper India , while it also lies near the track of some of the October cyclones .
The great variability of pl.essure at 10,000 feet over India and P.ajputana during the rainy season , which depends chiefly upon the still greater variability of temperature in the same region , is a good instance of the general tendency for established set of meteorological conditions to persist by the interaction of cause and effect .
An unusually high pressure in the upper atmosphere over this in the months of May and June probably produces , as we have seen , westerly and northwesterly winds ; these , being dry winds , neither cool the earth 's surface by precipitation and subsequent evaporation , nor , by the interposition of a screen of clouds , prevent its temperature from rising high in the day-time ; then , in consequence of the high day temperature in this region as compared with the , which are more or less ( .loudy and moist , the pressure at 10,000 feet or other high levels remains relatively excessive ; and again , in consequence of the high baric gradient thus conserved , the westerly winds continue to blow on , until some disturbance origmating at a distance supervenes , or , as in lS77 , until after the autumnal equinox , when the night begms to exceed the day in length , and the loss of heat by radiation to the clear sky overbalances the gain .
The mean gradients or pressure differences , measured more directly across the Gangetic plain , in each of the months compared in Table XVIII .
, are Table XIX .
, the figures representing thousandths of an inch , and positive a gradient for westerly winds , i.e. , .
that higher pressures belong to the southern tations , such being the most frequent condition . .
\mdash ; Pressure Differences over the ngetic Valley in Thousandths of Incl ] .
On aring this Table with the figures given above for the variations of rainfall , it will be seen that , in the worst years , 1877 and 1880 , the mean gradients for January were least , indicating the probable existence , for a considerable number of days in the month , of moist south-easterly currents , the snowfall , which , according to Mr. BLANFORD'S theory , was the cause of the subsequent drought .
These moist winds were observed in , 1877 , on the plains ; but in 1880 , as Mr. BLANFORD fias pointed out in the paper above quot , ed , they were confSned chiefly to the valleys and slopes of the Himalaya .
In May , the variations of the pressure ces are less evidently connected by any rule , and in July there apparently is in the mean of the three pairs of stations a nearly uniform decrease in the dients for westerly winds or an increase in those for easterly ones during the eight years from 1877 to 1884 .
It does not seem right , however , to group the Hazaribagh-Darjiling pair with the others , as the character of the ] of any season in Bengal is usually different from and often opposite to that which obtains in Upper India .
Thus , in 1883 , in which the gradients over West Bengal appeared rpther favourable for rainy winds , the defect of rainfall in that province was only 4 per cent. , as against 17 per cent. in the upper , and this small defect may possibly be explicable by other causes ; ' whilst in 1880 , which the Tadient between Hazaribagh and Dar* For example , diminished absolute humidity of tbo air , owing to the t being below the mean of whole of in was below average , and tho inch in defect .
RELATION TO THE DISTRIBUTION { BAROMETRIC SSURE .
375 jiling was still more fayourable for easterly winds , there was an actual excess of rainffil over the Bengal provinces , amounting to 10 per cent. , against a defect of 20 per cent. in Upper lndia .
The character of the Ajmere obsey.vations for accuracy , moreover , is not nearly so high as of the work of the observer , and the barometer at Ajmere has been several times changed , while there has also been a change in its position during the years compared .
Hence , every care has been taken to correct the obserfor such changes , it is possible that there may be some residual errors undetected , which , though ligible , or nearly so , in the mean of a term of years , many of which were anterior to the mentioned , may introduce some confusion in comparing the observations of single years .
For these reasons , it is probable that the best criterion of the pressure , prevailing over Bundelkhand , Rajputana , and the North-West Provinces , will be a comparison of the observations of Sutna and Ranikhet .
Thus estimated , the gradient for westerly winds was relatively high in May , July , and October , 1877 .
It was low in May and October , 1879 , and in July there was a considerable gradient for easterly winds .
In 1880 , the gradient for westerly winds was high in May and October , and there was a small gradient for easterly ones in July ; thus , on the whole , the upper winds were such as to make the rainfall less than usual .
In the dry year , 1883 , the gradients for westerly winds were greater , or for easterly ones less , than in the wet year lS84 , both in May and July ; but in July , 1883 , there was nevertheless a rather high gradient for easterly winds , contrary to what might be expected .
It was in August , 1883 , however , not in July , that the principal deficiency of rain occurred .
The verification of theory by of the deductions from observations embodied in Table XIX .
being thus less definite than is desirable , it be worth while to work out the pressure gradients between Sutna and Ranikhet for the other months of the rainy season in each of the years compared .
Table .
gives the pressures 10,000 feet for the months of June , August , and September .
TABLE .\mdash ; Pressures at 10,000 feet for the other Nlonths of the ainy Season in certain Years .
Ranikhet .
to say , for the six months May to October inclusive , which almost the whole rain of the year falls on the etic plain .
essure Differences between Sutna and Ranikhet for the Summer Half-Year .
In .
the relation between the pressure differences and the rainfall variation becomes quite clear .
The driest years , 1877 and 1880 , had the largest , for erly winds , that is to say , in those years , easterly winds at high levels over the plains of Nortlbern India were probably , or restricted to an unusually narrow belt along the Himalayas ; on the other hand , the wet years , 1884 and 1879 , had the smallest gradients , or in those years the easterly upper currents were steadiest , and reached their widest extension ; while in the moderately dry year , 1883 , there was a considerable gradient for winds , but nothing like so great as in 1877 and 1880 .
If the month of May be left out in striking the mean for the season , the relation indicated becomes still more distinct ; thus\mdash ; It appears , therefore , from this amalysis of the observations , that the theoretical uctions from KoPPEN'S convection hypothesis are fairly substantiated by experience ; that in years of unusually heavy snowfall , and consequent ] , over the North-West Himalaya the pressure gradients producing westerly winds in the upper atmosphere intensified , and that in consequence of this the easterly , which prevail as far south as the Ganges up to the level of the lower cirrus clouds in the rainy season of average years , are greatly restricted in horizontal extent , and doubtless also in height as well as in time , the result being a diminution of the rainfall , sometimes to a disastrous degree .
Though these conclusions are ntiated by the average of the observations of five or six months , they are not completely borne out by those of any single mon th , ' RELATION 1)ISTRIBUTION OF BA ) except September , and ffable X. it would appear impossible to predict in May or June what 1ill be the prevailing pressure difference later on in the year .
irregularities in the pressure differences deduced from the observations of single months may , however , as already pointed out , be due in great to the uncertainty of the assumed rate of decrease of temperature , which , while probably nearly true , or at all events giving consistent results when applied to the average of many years ' observations for any calendar , or even to the mean for five or six months of a single may be very considerably in error when ) plied to the observations of one month standing by itself On making a retrospective survey of all the evidence put forward in the preceding pages , it will , I think , be generally admitted thab the hot winds of Northern and Central cannot be ] explained by the distribution of pressure at the earth 's surface , and that KoPPEN'S hypothesis of convective interchange between upper and lower strata is probably the true explanation of them , ( 1 ) because tho vertical distribution of temperature is such that convective action must take ( 2 ) because the diurnal variation of the intensity of these winds and their characteristic dryness suggest such an for them ; and ( :3 ) because the distribution of pressure at 10,000 feet above sea-level in May is such as to winds of the observed direction .
The distribution of the upper currents as suggested by the charts also elucidates many other obscure points in the wind system of India , and the same hypotbesis of the origin of vesterly winds gives a rational mation of the law worked out by Mr. BLANFORD connecting the spring snowfall of the North-west Himalaya with the rainfall over the plains during the succeeding summer monsoon .
If these conclusions have not been all completely established , as I am by no means desirous of asserting , they have , I think , been shown to be sufficiently probable to warrant a more complete examination in detail by } determination , if possible , of more trustworthy rates of temperature decrement with than those assumed in this paper , and by the syst , ematic ) of high-level pressure charts for the whole of , month by month .
PLATES .
Plate 19 shows the isobars or lines of equal pressure at sea-level over India and the Bay of Bengal for the months of January , May , July , and October ; typical respectively of the cold , the hot , and the rainy season , and the autumn transition period .
The pressure , indicated by each line , is ented by the figures attacl ) to it , those for successive lines differing by inch .
The have all been COl'rected for variations of gl.avity with latitude .
The charts also give the prevalent wind at the more important stations , these ' resultant\ldquo ; directions having been computed by means of LAlIBERT'S formula from the observations of many years .
Plate 20 gives the isobars for a stratum of the atmosphere 10,000 feet oove sea-level .
The pressures represented are not , however , middle of the day , between 10 , and 4 observations made at 40 stationoe uding assuming the temperature to decrease at the rat a Table computed by Major ALLAN RANKINE'S modification of LAPLACE'S formula .
Plate 21 shows the probable rate of day time , through the atmosphere over a pla sJightly dflerent for clear and cloudy obseryations made during balloon ascents .
rate of temperature decrement in a mass of of heat . .
A. I ! .
MEAN PRESSURFS AT .
1887 .
A. MID-DAY PRESSURES AT 10.000 FEET

