196 Jj. LTONELATAVLER [SEPTEMBER 
proportions preserved between the various qualitative 
specialisations. 
(¢) Reproduction on this theory commences when full or nearly 
full development of a structure is reached, when its growth 
capacity is in excess of its demands; from this it will follow 
that the reproductive units will be collected in the repro- 
ductive organs in the order of their evolution. 
(d) A progressively specialising food supply would determine the 
development and the atrophy of the different reproductive 
units. 
(e) The later a specialisation was developed either in the history of 
the species or the individual the less chance of its obtaining 
a foothold in reproduction, and conversely these must be 
the first to be eliminated under stress conditions. It will 
follow from this that the effects of use and disuse, in so far 
as they are of a somatic nature, will be very little if at 
all transmitted to the germinal structures, since develop- 
ment, in so far as the major part of the organism is con- 
cerned, will be completed early. 
The first advantage of a theory like the preceding is that it has 
no need for the supposition of any isolated germ structure, use- 
inheritance being largely negatived by specialisation. The relation 
of germinal to somatic development is on this theory understandable. 
It would account for recapitulation in development, not on the ground 
of a tendency in the organism to repeat certain ancestral characters, 
but simply as the necessary preparatory specialisations out of which 
the later ones are built." It would divide all anomalies into—(1) 
those cases of faulty representation due to the missing of some prior 
stage in development, as in the case of cretins, where the morpho- 
logical element is there but the means of developing it is not, or 
where deficiency of the element itself as possibly happens in the case 
of mongoloid idiots; (2) disproportionate representation (quantita- 
tive anomaly), leading to dichotomy, ete.; (3) under rare conditions 
the reappearance of real ancestral characters. 
If therefore the recapitulation theory has a different meaning from 
that of ancestral repetition, and if most cases of so-called atavism can 
be explained on the assumption of incomplete development, if it is 
further borne in mind that given the power of segmentation then all 
that is chiefly required is a proportionate representation of germs, 
then the complexity of the germ plasm, although very great, need not 
be so inconceivably great as that which involves the representation of 
a large number of ancestral as well as all living characteristics. 
Normal sexual reproduction would on this theory be the right 
1 In a limited sense, however, these stages would represent the history of the individual 
ancestral line. 
