SOMATIC CHARACTERS AND CHROMATIN ROD-LENGTHS. 117 



Taking as a second example the mediastinal area of the elytra, we must 

 expect to find corresponding chromosomes o£ the same length in *S'. paral- 

 lelus, S. hicolor, and aS'. ciirtipe/inis and of a different length in S. viridulus, 

 for in the three first named the area extends abruptly towards the apex 

 whereas in the last it extends gradually. We again fail to observe such a 

 chromosome ; and the same absence of correlation is noticeable with respect 

 to the other characters on the list with the exception of colour, which is 

 variable and untrustworthy. 



Moreover, we are not more successful if we assume the first hypothesis, 

 which postulates unchanging rod-lengths. Disregarding the arrows in the 

 diagram and considering correspondence to depend entirely upon length, we 

 find that the length of antennae and angle of the lateral carinse may be 

 correlated with Chromosome 15 in S. parallehis, S. viridulus, and S. curti- 

 pennis ; but the fact that no other characteristics appear to correspond 

 with chromosome lengths makes justification for this assumption dodbtful. 

 Furthermore, the genus Stenobothrus has been divided by Bolivar and other 

 systematists into subgenera, and these species are now classified as follows: — 

 ChortMppus parallehis, Clwrtldppus curtipenms, Ornocestus viridulus, and 

 Stauroderus hicolor : it is noteworthy that the two whose complexes show 

 the greatest differences in rod-lengths should thus be classed together. 



Four explanations may be put forward to account for this failure. Firstly, 

 we may assume that my measurements are inaccurate ; this, however, seems 

 unlikely, for great care has been exercised, and the^ lengths of the long- 

 chromosomes are such that relative error should be impossible. Secondly, 

 we may assume that the lengths of the five short chromosomes are 

 not respectively identical in all the species ; in this case the characters 

 mentioned may be correlated with these and not with the three long 

 chromosomes : if, however, the principal somatic differences, upon which 

 systematists have based their classification, are not traceable to obvious 

 differences in long chromosomes, why should they be traceable to imper- 

 ceptible differences in short chromosomes, and, if they are so traceable, to 

 what are the obvious differences in the former due ? Thirdly, the arrows 

 may be misleading : Chromosome 17 of S. p>arallelus, S. viridulus, and 

 S. hicolor may, for example, correspond in S. curtipennis with Chromosome 13 

 and not 15, in which case the last named chromosome corresponds with the, 

 medium instead of the longest member of the three, and is accordingly 

 functionally identical with Chromosome 15 of S. p)araUelus and S. viridulus 

 and Chromosome 13 of S. hicolor. This is undoubtedly possible, if rods are 

 continuously increasing in length, for a long chromosome may in the course 

 of evolution be overtaken and passed by one that was shorter^ and the latter 

 may consequently be mistaken for the former : if this occurs, we must realize 

 that measurements cannot always be a trustworthy index to functional 

 correspondence even in the most closely allied organisms. Lastly, if the 



