14G 



THE AMERICAN NATURALIST [Vol.XLVI 



the normal arrangement, occurring in large series of 

 adult specimens. And he has clearly shown that these 

 variations are nearly always significant. There are 32 

 possible arrangements of the plates of the genito-ocular 

 ring and there is no mechanical or structural reason why 

 any one of them should not occur. If variation were 

 perfectly haphazard every one would occur and there is 

 no obvious reason why they might not occur, with equal 

 frequency. Yet in fifty thousand specimens examined by 

 Jackson, representing 137 different species of Mesozoic. 

 and Recent Echini, ten of these possible arrangements 

 never occurred, and of the remaining 22 fourteen are so 

 rare that altogether they aggregated less than H per 

 cent, of the specimens. As a very large proportion of 

 these were individuals abnormal in some other particular, 

 it is fair to say that of 32 possible arrangements of the 

 genital and ocular plates only eight (or at most ten) occur 

 normally. Even more striking are the following facts : 



When only a single ocular plate is insert, it is one of the 

 posterior pair; this is the case in 99^ per cent, of the 

 specimens having one ocular insert. 



When two oculars are insert, they are the posterior 

 pair in more than 99 per cent, of the cases and in every 

 case one of them belongs to that pair. 



When three oculars are insert, they are the two poste- 

 rior and usually the left, but sometimes the right ante- 

 rior; this is demonstrated by almost 99 per cent, of the 



When four oculars are insert, the one exsert is invar- 

 iably either the mid-anterior or right anterior. 



These figures show how surprisingly definite variation 

 is in a character which, so far as we can see, might vary 

 with equal ease in any one of 32 ways. Yet it is only 

 when we examine a particular case that the significance 

 of this definiteness appears. 



Jackson's work is full of such cases, but as most of 

 us are familiar with Strongylocentrotus, we will consider 

 an illustration from that genus, which, in the old, broad 



