360 LOCK : STUDIES IN PLANT BREEDING 
A method which deserves particular emphasis is one which has 
yielded valuable results in the hands of De Vries, and consists in 
deducing the character of a particular plant—whether originating in 
a cross or otherwise—from a study of its offspring by self-fertilization. 
For this purpose all the seeds of the plant under examination should 
if possible be sown ; but in cases where these are too numerous, at 
least a considerable sample. The nature, or Mendelian constitution, 
of any particular cross-bred plant can in this way be deduced, some¬ 
times from the examination of one, or as a general rule of at the most 
three generations. This method is particularly applicable in a 
climate under which as many as three generations can be grown in 
a single year. 
In addition to this work on crossing, the writer also attempted 
some experiments in definite selection on lines similar to those 
recently described by Johannsen ; one of the characters examined 
being the size of individual pea seeds. These attempts were given 
up however owing to the apparent impossibility of arriving at 
anything approaching to uniform conditions between plant and 
plant, or between generation and generation. The latter difficulty 
was perhaps the more serious of the two, owing to differences in 
the rainfall and other conditions of climate at different times of 
year. In order to carry out such an experiment properly, only one 
generation must be grown each year, and further, the sowing should 
take place each year at a corresponding season. But the 
differences apparently caused by irregularities of nutrition in a 
potentially uniform batch of seedlings were also sometimes astonish¬ 
ing ; and it was found most difficult to obtain an even preparation 
of soil in the seed beds by native assistance, even under the closest 
personal supervision. This difficulty would appear to be a funda¬ 
mental one in all exact experiments of this kind, and renders such 
work of little value unless carried out on a very large scale for many 
generations, in order that accidental differences of soil, &c., may 
become equalized. Moreover the crop obtained from uncrossed 
strains of peas was very small at the best of times, and when the 
conditions w T ere unfavourable it became almost nothing. In such 
circumstances it was thought better to confine the attention to 
