88 WHEAT PRODUCTION IN NEW ZEALAND 
But, besides the error ascribable to variation in soil 
in a single season, there is a still larger probability of 
error due to variations between season and season. This 
fact is, however, so generally recognised that to enlarge 
upon it is unnecessary, except to point out that compara- 
tive trials carried out during successive seasons must be 
carried out on the same class of land if any conclusion 
is to be arrived at. It is obvious that a trial made on 
heavy soil in a dry season and on light soil in a wet 
season will tend to give the same result, and the experi- 
menter might say, ‘‘The result is reliable, for I have 
proved it on two different classes of land during two 
different seasons.’’ In reality, the result might be mis- 
leading as regards both heavy and light land in a normal 
season and certainly is inconclusive for heavy land in 
wet seasons, and light land in dry ones. The prime 
necessity of carrying out a series of experiments during 
successive seasons only on lands of the same type seems 
to be the weightiest argument in favour of a soil survey. 
5. Details of Experimental Work. 
Having described the principles of single ear selection, 
and the necessity of replication of experiments in order 
to arrive at reliable results, the details of the method of 
experiment may now be described. It was decided to com- 
mence work on the varieties commonly grown in the 
country, on the assumption that these were best suited to 
it, as having stood the test of half a century of culti- 
vation here while other varieties had been tried and 
rejected. The varieties dealt with, were, therefore, Hun- 
ter’s or Red Chaff, White Straw Tuscan, Solid Straw 
Tuscan, Velvet, and Pearl. The experience with Hun- 
ter’s will be described, as it has produced the most 
noticeable results. 
Some hundreds of heads differing in appearance were 
selected from a normal crop, and these were reduced by 
