goes a long way in my opinion to confirm the correctness 

 of this portion of Mr. Symons' diagram. 



It will be seen that in the year 1815 they were, in Eng- 

 land, just at the termination of the second severe drought, 

 shown on the diagram, which commenced in 1795, during 

 which the loss of rain below the mean amounted to more 

 than double the quantity mentioned by Mr. Glaisher as 

 having been lost in the period from 1854 to 1858, and 

 although something less than two-thirds of this loss had 

 been made up by an accumulated rise in the rainfall to the 

 yt-ar 1831, this rise was not maintained, as it will be seen 

 that rains below the average were experienced in excess 

 of those above it for twenty years up to the year 1851. 

 One year of high rainfall then followed in 1852 (which year 

 corresponds with that of 1909) to be followed by a long 

 period of drought in which the accumulated loss of rain 

 (luring nineteen years to 1871 pulled the curve down to its 

 lowest point, there having been only six years above the 

 average in the whole period. 



Another important feature is conspicuous in the diagram, 

 namely, the period between the termination of the two 

 droughts in 1814 and 1871 is exactly 57 years. 



With this evidence before them, and the fact that critical 

 points of change are discernible in 1738, 1757, 1776, 1795, 

 1814, 1833, 1852, 1871, 1890, aud 1909, the intervals between 

 which are nineteen years, it is difficult to understand how 

 meteorologists in England have not, up to the present time, 

 noticed this peculiarity. If the residual mass curves had 

 been worked up the phenomenon of secular variation would 

 have been discovered long ago. 



Mr. Symons, when referring to his diagram at the dis- 

 cussion on Sir Alexander Binnie's paper in 1892,' said: — 



"I have exhibited the diagram in order that the members 



might form a rough idea of the amount of oscillation to which the 



annual total rainfall was liable. I will not enter into the question 



1 Proceeding In-t. <'.L\. Vol. cix, p. 139. 



