54 
Probable ap¬ 
proximation to 
actual mean. 
Remarkable 
wetness of the 
Seventies. 
Extremes. 
Fallacy of 
conclusions on 
limited data il¬ 
lustrated. 
Further illus¬ 
trations of the 
excess in the 
“ Seventies.” 
Again, assuming for the 50 years a limit of error J the 
variation of these decades from the mean, we get a Rainfall 
of 25’005 dy 0'477. The evidence, however, points very 
strongly to this error being negative. For the increase in 
Rainfall, and Days wdth Rain, during the last two decades, is 
most striking, hut especially the increase of Rain in the 
Seventies. This raised the 30 years’ mean by an inch for 
that of 40 years. The earlier returns, moreover, confirm the 
exceptional character of that decade. 
Thus the mean fall from 1831—90 is 24*83. 
„ „ 1811—24 and 1831—91 is 24 606; 
this over a period of 75 years*. 
As the Yearly Rainfall has so much wider a range than either 
Mean Pressure, or Temperature, the close approximation of all 
hut one decade as to total fall is, perhaps, more unexpected than 
the extraordinary divergence of this exception. However, a 
little further selection results in even more striking extremes :— 
Mean of the 12 years, 1850—61 :—21*73 inches ; 148’4 days. 
,, 11 ,, 1872—82:—28 78 inches ; 194*6 days, 
or ^ excess. Everything points to the danger of reckoning rain 
averages upon too small a basis. Evidently at least 100 j^ears 
are required for even an approach to only one per cent, of error. 
One often hears it asserted, also, that the rainfall is increasing 
or decreasing. The latter, for instance, has been frequently 
assumed of the TJ.S. Prairie regions. But the first 30 years of 
our returns would have led to the same result. Still more 
would it have been so had any one begun observations in the 
early Seventies. 
The exceptional character of the Seventies can he shown in 
other ways; it is iudicated by the following table :— 
* See also conclusion. 
