404 THE EVOLUTION OF LIFE. 



development and embryonic development, and expresses his 

 general belief thus : 



&quot;The conclusion here reached is that, whereas larval development 

 must retain traces (it may be very faint) of ancestral stages of struc 

 ture because they are built out of ancestral stages, embryonic develop 

 ment need not necessarily do so, and very often does not; that 

 embryonic development in so far as it is a record at all, is a record of 

 structural features of previous larval stages. Characters which dis 

 appear during free life disappear also in the embryo, but characters 

 which though lost by the adult are retained in the larva may ulti 

 mately be absorbed into the embryonic phase and leave their traces in 

 embryonic development.&quot; * 



To set forth the evidence justifying this view would en 

 cumber too much the general argument. Towards elucidation 

 of such irregularities let me name two factors which should I 

 think be taken into account. 



Abridgment of embryonic stages cannot go on uniformly 

 with all disused organs. Where an organ i& of such size that 

 progressive diminution of it will appreciably profit the young 

 animal, by leaving it a larger surplus of unused material, we 

 may expect progressive diminution to occur. Contrariwise, 

 if the organ is relatively so small that each decrease will not, 

 by sensibly increasing the reserve of nutriment, give the 

 young animal an advantage over others, decrease must not be 

 looked for: there may be a survival of it even though of 

 very ancient origin. 



Again, the reduction of a superfluous part can take place 

 only on condition that the economy resulting from each de 

 scending variation of it, is of greater importance than are 

 the effects of variations simultaneously occurring in other 

 parts. If by increase or decrease of any other parts of the 

 embryo, survival of the animal is furthered in a greater 

 degree than by decrease of this superfluous part, then such 

 decrease is unlikely; since it is illegitimate to count upon 

 the repeated concurrence of favourable variations in two or 

 more parts which are independent. So that if changes of an 



* Stitdi^s from the Morphological Laboratory in tlie University of Cam 

 bridge, vol. vi, p. 89. 



