STUDIES OF VARIATION IN INSECTS 323 



variation in pattern, appearing, however, in such a way as to 

 demand numerical, /. c, meristic, expression. One speci- 

 men has 9 el3^tral spots, another lo, another ii, and so on; 

 the whole range is indeed from o to i8, with every number 

 between represented, each by various combinations of spots. 

 But it is conceivable, and indeed is really the case among our 

 specimens, that these spots might be either of normal size, or of 

 any lesser size down to the limits of visibility. Some of the 

 spots are of the diameter of pin-points ; some of the pin-shaft 

 and some pin-heads. There is perfect gradation or con- 

 tinuity in this variation. But even in such cases as variations 

 in spines and hairs, this gradation might exist ; and indeed does. 

 Althoucrh in our consideration of the variation in the number of 

 the tibial spines of the locust and the cicada and in the number 

 of the tactile hairs of the bird-lice, we have referred to these 

 variations only numerically, /. e., meristically, as a matter of 

 fact there are obvious differences in the length, /. e., size, of 

 the spines and hairs, so that it would be wholly fair to break 

 down the unit differences and speak of differences by one- 

 quarter, one-third and two-thirds of a spine. For the tibial 

 spines of the locust we actually recorded the conditions in the 

 form of fractions. But in the case of a hook or an antennal or 

 a tarsal segment it is a unit or nothing. To our mind the dis- 

 tinction between substantive and meristic variation is not at all 

 equivalent to a distinction between continuous and discontinuous 

 variation. It is a distinction between two categories of varia- 

 tion only in that one category includes such conditions as per- 

 mit more readily of extremely slight, nearly insensible, prac- 

 tically unmeasurable differences, as those of pattern or shape or 

 extent, while the other category includes particularly conditions 

 in which any variation must of necessity be fairly obvious, and 

 usually capable of numerical expression. 



But we believe, nevertheless, that discontinuous variations 

 occur among insects and that examples of them are presented 

 in the data referring to the species studied by us. For example 

 the occurrence of interpolated, wholly new, and complete cells 

 (determined by the presence of new cross veins or branches of 

 longitudinal veins) in the fore and hind wings of drone honey- 



