W. F. R. Weldon 
371 
a favourable mutation has occurred to isolate the otFspring. Regression will lead 
such ottspriug to assume the mean character proper to the new species ; and when 
this has been attained selection can do notliing more of permanent effect until the 
occurrence of a new mutation. Indeed, if the individual which exhibits the 
mutation should not be a " Minusvariant," but should be at the mean of the new 
species, all the attainable improvement will have been effected at once, and no 
further step can be made without a new mutation. 
The existence of the very remarkable form of regression here postulated can 
only be proved by full evidence of the correlation between parents and offspring in 
cases which ai-e said to be due to mutation ; but such evidence is never provided 
by Professor de Vries. 
The nearest approach to an adequate account of the relation between a 
mutating individual and its offspring is given in the case of the form called 
Trifoliutn 'pratense quinquefoliiim*. In 1886 Professor de Vries found by a road- 
side several wild plants of T. pratense, bearing leaves with four or five leaflets in 
addition to normal leaves with three leaflets. Two of these plants were removed to 
his garden and cultivated. In 1889 the two plants together bore 64 leaves with 
four, and 44 with five leaflets, among a very large number of normal leaves. 
Data by which the deviation of these plants from the mean of their parental 
generation could be determined are not given. In 1889 the seed from these 
plants was collected, and in 1890 there were 100 offspring of the first generation. 
" About half" of these bore only normal leaves, and were destroyed. The remaining 
half bore some leaves with four leaflets, and some with five ; but the proportion of 
abnormal leaves is not recorded. The four best plants were saved for seed, and the 
rest destroyed. In August — September 1890, the four seed-plants had amongst 
them 69 leaves with four leaflets (64 accoi'ding to the M utationstheorie) and 44 
with five. Of the plants with some abnormal leaves which were not saved for 
seed the best twelve had amongst them 48 leaves with four and 11 with five 
leaflets. 
The whole number of leaves on yearling plants is less than the number on older 
plants, so that the four plants chosen for breeding, and some of the plants thrown 
away, had a larger propoi'tion of abnormal leaves than their parents ; but the 
proportion of abnormal leaves among the whole hundred offspring — the mean 
character of the offspring from which regression must be determined — was certainly 
less abnormal than that of the parents, and did not exhibit an increase in the 
number of abnormal leaves, such as should follow from regression to a new "specific 
mean'' with many leaflets, if the hypothesis put forward by Professor de Vries were 
true. This postulated regression failed to occur in spite of the fact that the 
offspring were reared in a garden, under conditions shown to favour an excessive 
production of abnormal leaves. 
* I have consulted the memoir " Over het omkeeren van Halve-Galton-Curven," Botanisch 
Jaarboek, x., 1898, as well as the account given in the Mutationstheorie. 
39—2 
