GEODIA AGASSIZII. 



147 



differences of the commensurate proportional amounts of the maximum averages 

 of the homologous dimensions and then divided the sums by six. In this way 

 the average percentage differences of the nine lots in respect to the dimensions 

 of these internal spicules, least subject to the influence of external forces, were 

 obtained. In the following table, where the thirty-six pairs are arranged in the 

 order of their similarity in respect to these dimensions, the numbers thus arrived 

 at are given. 



Pairs of 

 lots. 



Average percentage differ- 

 ences in the length of the 

 sterrasters, the orthopla- 

 giotriaene-rhabdomes. or- 

 thoplagiotriaene - clades, 

 and large amphioxes. and 

 the thickness of the ortho- 

 plagiotriaene - rhabdomes 

 and large amphioxes. 



GH 



BH 



DF 



BD 



CI 



EG 



EF 



AD 



AF 



CD 



AI 



BC 



Average percentage differ- 

 ences in the length of the 

 sterrasters, the orthopla- 

 giotriaene-rhabdomes, or- 

 thoplagiotriaene - clades, 

 and large amphioxes. and 

 the thicl<ness of the ortho- 

 plagiotriaene - rhabdomes 

 and large amphioxes. 



11.6 

 11.9 

 12.2 

 12.3 

 12.7 

 12.7 

 12.9 

 13.3 

 13 6 

 13.7 

 14.6 

 15.6 



Pairs of 

 lots. 



DI 



FH 



AE 



DH 



GI 



CG 



HI 



CF 



FI 



EH 



AC 



CH 



Average percentage differ- 

 ences in the length of the 

 .sterrasters, the orthopla- 

 giotriaene-rhabdomes, or- 

 thoplagiotriaene - clades, 

 and large amphioxes. and 

 the thickness of the ortho- 

 plagiotriaene - rhabdomes 

 and large amphioxes. 



15.6 



16.2 



16.8 



17. 



17.4 



18.4 



18.8 



19.5 



20.1 



21.1 



22.5 



26.8 



The table shows that these differences are very unequal and vary between 

 6.1 % (AH) and 26.8 % (CH). It would therefore seem that it might be possible, 

 by joining the paire differing only slightly, and by keeping apart the pairs 

 differing greatly, to arrive at some systematic grouping. The regular gradation 

 in the increase of the average percentage difference from the lowest to the highest, 

 however, makes it difficult anywhere to draw a line of demarcation between 

 differences insufficiently and differences sufficiently great for systematic dis- 

 tinction. If we consider, as we surely may, differences in respect to these 

 dimensions up to 10.7 % insufficient for systematic distinction, we must unite 

 all the pairs difTering 10.7 % or less. These pairs are AH, CE, DE, FG, AG, 

 AB, BG, BI, DG, and EI. Now if these pairs are united: A joined to H, G, and 

 B; CtoE; DtoE,andG; FtoG; BtoG,andI; andEtoI; it will be seen that 

 all nine lots are united, however different certain pairs may be. In truth, we 

 can say that biometrically a continuous series of transitional forms connect the 

 most aberrant members of the whole group. 



One of the most remarkable features of the peculiarities of the spicules of 

 the nine lots is the want of correlation between them. So far as their dimensions 



