RELATION TO THE DISTRIBUTION OF BAROMETRIC PRESSURE. 
355 
A similar relation obtains 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 November ; 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 d 'priori probable, there must be a 
greater pressure 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 find the distribution of pressure over India at a considerable height in 
the atmosphere, say 10,000 feet. For this purpose I have selected some 40 stations 
out of the large number in India and Ceylon for which 10-year averages of pressure 
are published in the Meteorological 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, fourteen stations situated at levels below 1500 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 Dodabetta, the published barometric 
heights had first of all to be corrected for the temperature of the mercury. The 
elevations given have been found by levelling to Great Trigonometrical Survey stations, 
or other datum levels, in every case except Pithoragarh and Shillong. The elevation 
of the last-mentioned has been found by comparing barometric observations 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 determination cannot amount to more than two or three feet. 
The elevation of Pithoragarh has been computed by comparing, month by month, the 
barometric observations 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, when 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 too high, and that of Mercara 
135 feet too low. These elevations have been deduced trigonometrically or by spirit- 
levelling from points fixed in the early days of the Survey, when the uncertainties of 
terrestrial refraction were not sufficiently allowed for. 
The observations made at Simla have also been rejected, as there is some uncertainty 
about the true elevation of the barometer. 
2 z 2 
