GENERAL CONDITIONS IN NEW ZEALAND 49 
[245.65 
380 
= "8.19 
= 2.85 
_ 158.20 
00-EFFICIENT OF CORRELATION =—gyxenexd te 
= +.602 
1 —(.602)2 x .6745 
v30 
=.0786 
STANDARD DEVIATION OF RAINFALL 
PROBABLE ERROR = 
and rainfall for that state, and @ priort for the whole 
of Australia. Fluctuations in rainfall are invariably 
followed by fluctuations of a similar nature in yield. 
The co-efficient of correlation between these two series 
is + -602 and the probable error only -08, so that 
correlation may be regarded as a practical certainty, 
according to Professor Bowley, who takes as one test 
of correlation a probable error of less than one sixth of 
the co-efficient of correlation. 
From the graphs for New Zealand and Australia given 
here many interesting conclusions could be drawn, but 
we must content ourselves with a brief summary of 
results.* 
The correlation between yield and rainfall in Aus- 
tralia is more vital than it is in New Zealand, because 
Australian rainfall is so small that fluctuations either 
way greatly affect the productivity of the soil. In New 
Zealand, on the contrary, it is only in very exceptional 
circumstances that rainfall is so low as to interfere 
greatly with productivity, while it is generally of such 
an amount that an increase does not add much to the 
yield of produce. 
This conclusion adds further to the contention that 
from the point of view of rainfall, New Zealand enjoys 
a differential advantage in wheat production. 
*Statistics from Lincoln College, Canterbury, and Roseworthy 
College, South Australia, are given in Table XII., and they 
support the general conclusions arrived at here. 
