142 EVOLUTION AND ANIIVIAL LIFE 



normal restoration of injured parts. Others, too, are of a char- 

 acter which, to many people, will not seem to be discontinuous 

 at all, but continuous. For example, differences in number of 

 antennal or tarsal segments in insects are called by Bateson 

 cases of discontinuous variation if the differences are only by 

 one segment. But as the differences cannot well be less than a 

 whole segment, variations in number of segments, if represented 

 by all the successive numbers between the lowest and highest 

 number of segments observed, may fairly be called continuous: 

 that is, strictly gradatory. 



It may be of interest to note, for the purposes of explaining 

 by concrete examples the various phases or categories of varia- 

 tion already named, some specific examples exemplifying each 

 category. The following are taken from a paper on variation in 

 insects, which records a number of statistical studies of varia- 

 bility made by the junior author of this book and Mrs. Bell- 

 Smith. 



To distinguish absolutely between acquired variation and 

 congenital (or blastogenic) variation is a matter which can be 

 done in but comparatively few cases. Whether a variation be 

 congenital or whether it be acquired during the development or 

 life-time of the individual showing it, this variation cannot be 

 recognized until after a considerable part of the development 

 has been undergone; if it is a variation in an adult structure or 

 function, all of the development must have been completed. 

 The variation is apparent only after it is unfolded: only after 

 the part it appears in has reached its definite stage of completed 

 growth and development. 



Nov\^, who is to say whether this variation was or was not 

 imposed on the individual showing it, during this long develop- 

 ment and immature life as a result of some external influence 

 brought to bear on the varying part during the development? 

 We know that such extrinsic influences do modify parts and 

 functions during individual development, and so we must be 

 very careful when we claim that this or that variation is con- 

 genital and not acquired. Yet, how all-important it is to make 

 the distinction is apparent when we recall the fact that most 

 biologists are agreed that acquired characters (variations) can- 

 not be inherited, so that new species can be built up only on 

 the basis of congenital characteristics. 



In the c^se of insect variations, a criterion for distinguishing 



