158 



S. A. mil— TonmiJjcs and TTaih/orms of April 



[No. 2, 



ly winds ovor the submontane belt were dying out. Next morning, as 

 above stated, westerly winds reasserted tbemselves riglit up to the 

 mountains of Nepal. 



It is clear from the study of these four charts that there was no- 

 thing in the general distribution of pressure at sea-level, even just 

 before or during the progress of the tornadoes, to account for their 

 formation. The general features of this distribution on the first three 

 charts are identical, yet violent storms occurred in several places on 

 the 30th Apiil and 1st May, whilst there was nothing of the kind on 

 the 29th April, if wo except a small hailstorm which came on late in the 

 evening ovor the Siwalik hills and Roorkee. 



In his Tornado Studies for 1884 (Professional Papers of the Signal 

 Service No. XVIJ, Finlay has found that tornado tracks in the United 

 States lie almost invariably to the S. E. of a region of low pressure, that 

 is to say, they lie on the side of the depression covei-ed by warm moist 

 winds from the Gulf of Mexico. The analogous position for the upper 

 Gangetic plain would be tho N. E. side of the depression ovor which 

 moist winds from the Bay of Bengal blow, and this is where the storms 

 under discussion actually occurred But while the American tornadoes 

 in all cases travel almost parrallcl to tho isobars and in very nearly the 

 same direction as the S. W. winds proper to the octant in which they 

 are found, four at least of the five storms here described, and very 

 probably the fifth also, travelled against the baric gradient and against 

 the wind previously existing at the level of the plain. It is extremely 

 likely therefore that the conditions of pressure which produced these 

 tornadoes were not those existing at sea-level (for these, in a region 

 including high mountains, are to a great extent hypothetical merely), 

 but rather those obtaining at some definite higher' plane in the atmo- 

 sphere, probably, as pointed out by Mr. Archibald in a review of a 

 French work on the experimental production of such disturbances on a 

 small scale {Nature, Vol. XXXVIII, p. 104), at the place where cloud for- 

 mation begins. If once a toi-nado is commenced at this level, it may be 

 maintained for an indefinite time by tho energy converted from the poten- 

 tial to the kinetic form in tho condensation of vapour ; and the movement 

 may bo rapidly propagated downwards by means of the viscosity of the 

 air, and by indraught from below towards the partial vacuum in the 

 vortex. 



Now, though the situation of tho Rohilkhaud plain and Upper Doab 

 at tho foot of the Himalayas does not favour a proper undorstanding of 

 tho distribution of pressure at sca-lcvol, it has ho advantage of enabling 

 us, by observations made at adjacent plain and hill stations, to get a 

 very fair idea of the vertical distribution of temperature up to a liciglil 



