608 
MR. H. F. BLANFORD ON THE WINDS OF NORTHERN INDIA. 
Table, showing the mean density of the air-columns between three hill-stations and 
two reference-stations on the plains in each month of the year. 
1. Darjeeling and Goalpara. 
Goalpara, 
K 
Darjeeling, 
b„. 
Difference, 
b-b 2 . 
Mean 
density 
of air, 
d r 
+\ 
e i+ e 2 
Mean 
density 
of air, 
d. 
2 
2 * 
January 
29-603 
23-344 
6-259- 
0-837 
26*473 
53-6 
0-325 
0-844 
February 
•522 
•322 
•200 
•829 
•422 
55-9 
•336 
•838 
March 
•442 
•311 
•131 
•820 
•376 
61-7 
•369 
•827 
April 
•380 
•326 
•054 
•809 
•353 
66-6 
•485 
•817 
May 
•286 
•273 
•013 
•804 
•279 
69-2 
•608 
•809 
June 
•206 
•228 
5-978 
•799 
•217 
71 
•713 
•803 
July 
•195 
*224 
•971 
•798 
•209 
72-4 
•732 
•799 
August 
•247 
•261 
•986 
•800 
•254 
72-6 
•749 
•801 
September 
•328 
•322 
6-006 
•803 
•325 
71 
•700 
•807 
October 
•426 
•391 
•035 
•807 
•408 
68*1 
•582 
•815 
November 
•567 
•424 
•143 
•821 
•495 
60-6 
•615 
•829 
December 
•621 
•409 
•212 
•830 
•515 
54-4 
•336 
•843 
Year 
29*402 
23-320 
6-082 
0-813 
26-360 
64*6 
•546 
•819 
Range 
0-426 
0*202 
0*288 
0-039 
0-306 
19-0 
•424 
•045 
the conclusions subsequently drawn from the physical analysis of such a column, on this assumption, are 
therefore invalid. Some weight must doubtless he given to both these objections ; hut I believe nevertheless 
that the conclusions I have drawn are substantially trustworthy. Leaving for a moment the question of the 
obliquity of the column, a question which affects both computations of the density, I would point out that the 
errors of the values of d 2 , in so far as they are due to an irregular distribution of temperature and vapour, are 
shown by their difference from those of d v since these latter are computed from independent data ; and although 
this difference is considerable in some months, it is least so in January and July (or June at Shillong), when 
the exchange of air between the upper and lower strata of the atmosphere is most active, and when therefore 
both temperature and vapour must he least irregularly distributed. The question remains how far can d 1 he 
accepted as representing the mean density of a vertical column, (dj gives the mean density of a column of 
air (7i 2 — \) feet in length/which is equilibrated by a column of mercury at 32° F attr. ., represented by (6 1 — b 2 ). 
The error, if any, must lie in the assumption here made, that b 2 , the pressure at a somewhat distant hill-station, 
really represents the pressure at height \) vertically over the station when the pressure is b v and such 
error would arise were there a persistent difference of pressures between the two places in the same horizontal 
plane. Now any such difference would tend to produce wind-currents from one place towards the other, 
and this tendency should be shown by the wind-registers. There are such currents in July, from Goalpara 
towards Darjeeling, and from Roorkee towards Simla, indicating therefore the existence of a somewhat greater 
pressure over the former stations than at the latter in the same horizontal plane. But from the annual 
range of pressure at the hill and plains’ stations given in the Table above, it appears that this difference of 
pressure at the height of 7000 feet is probably not more than half as great as at the ground surface, 
where the barometric gradient does not exceed 0-1 inch in 800 miles, or 0-012 in 100 miles (see page 604). 
Take the probable error of b 2 , then, as -006. Then increasing the value of b 2 by that amount, the value of d T 
will be diminished by 0-001, and that of d 2 increased by half as much. The error of the results in the month 
of July cannot therefore much exceed this. In January the mean movement of the air is so low that the errors 
of the assumption similarly estimated must be quite insignificant. 
I should expect that the differences of d 2 and d 1 are principally due to the real mean temperature of the air- 
column in most months being somewhat higher than the assumed temperature A certain amount of 
error must arise also from the fact that the adopted values of vapour-tension are probably in all cases lower 
than the real tensions ; but the effect of this cannot be very great. 
