29 
Ten years, 1841-50, Rainfall, 24-835, Days 159. 
Twenty years, 1841-60, „ 24 05, ,, 154. 
Thirty „ 1841-70, „ 24 01, „ 158. 
Forty ., 1841-70, 1881-90, „ 24*25, „ 167- 
Fifty „ 1831-70, 1881-90, „ 24-20 
Sixty-five,, 1811-24,31-70,81-91, „ 24-08 
Here, by very simple doctoring of the results (omission 
of the seventies), we get a variation for averages from twenty 
to sixty-five years of less than a quarter of an inch. But 
persons do not always begin with the first year of a decade. 
Suppose the two following were to set up as authorities:— 
A, for the 12 years, 1850-1861, rrean of 21-73 inches, and 148 days. 
B, „ 11 „ 1872-1882, „ 28-78 „ and 195 „ 
Apparently at least 100 years are needed to approaoh 
within 1 per cent, of the actual value of the rainfall. For we 
may assume that the error for the fifty years’ mean does not 
exceed one-fifth of the variations of the ten-year means, that is, 
it is slightly under half an inch on 25 inches, which is 2 per cent 
To assert that a rainfall is permanently increasing or 
decreasing over a given region is obviously very unsafe, if we 
notice the conclusions which would have been drawn by two 
observers for twenty years, one beginning in 1860, the other 
in 1870. 
An interesting example may be found in the Palestine 
Exploration reports. The rainfall for 32 years averages 
25*23 inches 
From 1861-76 it averages 22 26 inches. 
From 1877-92 „ 2820 
This gives an excess upon the previous period of 5*94 inches. 
Hence, says Mr. Gflaisher in his comments (for it is no less an 
authority who is here oaught napping) there is “an evident 
increase of the fall of rain in the latter years of the series,” which 
is “ a somewhat remarkable point brought out in the discussion.” 
But suppose we take a similar period for York, starting with 
the second half of the present century. 
The mean rainfall, 1851-82 is 25 - 45 inches. 
„ 1851-66 is 23 02 „ 
„ 1867-82 is 27 87 „ 
This gives an excess upon the previous period of 4 85 inches. 
This is, indeed, not quite so large; but the values correspond 
strikingly, and the shorter slightly. 
The same caution is obviously needed with reference to the 
supposed changes in American prairie land from the effect of 
