716 



that the sacral ribs were at first attached to the nineteenth and then 

 travelled to the twentieth vertebra. While the existence of a myotome 

 between these two positions renders such a passage improbable, the 

 constant presence of a rudimentary rib on the nineteenth vertebra 

 shows very conclusively that such a change has not taken place, for 

 this would imply that originally there were two ribs on the same 

 vertebra or that both had migrated. If, then, the travelling of the 

 sacral ribs, and with them the attached parts of the girdle, is not a 

 real process, but merely a figurative way of explaining the unlike con- 

 ditions, one must admit that the sacral region has the power of de- 

 veloping sacral ribs at several points on both right and left sides. 

 Such an assumption is by no means unnatural, for it offers an easy 

 explanation both for the appearance of symmetrical sacra on the twen- 

 tieth in place of the nineteenth vertebra and for the occurrence 

 of unsymmetrical sacra. It also implies that an animal may possess 

 more than one sacral rib on a side, which might be taken as an ob- 

 jection to this hypothesis, had the actual occurrence of such cases not 

 already been demonstrated by Huxley ('75, p. 752) and by Lucas 

 ('86, p. 562) in Menopoma. This view, then, coincides well with the 

 actual anatomical conditions, but it further implies that, since sacral 

 ribs may be formed in several places, they are not necessarily exact 

 homologues: thus, the right rib attached to the nineteenth vertebra 

 of one animal cannot be the exact homologue of the right rib from 

 the twentieth vertebra of another. Our second general assumption 

 consequently leads to the same difficulty as our first one did, namely, 

 the inability to homologise structures with accuracy. 



This difficulty is due, I believe, not to any defect of observation 

 or reasoning, but to the artificial character of our conception of homo- 

 logy. The general idea of homology is based on the assumption 

 that, in any two given animals not too distantly related, correspond- 

 ing morphological positions can be defined. This is in large measure 

 true, but any position so defined may be subject to variation; and 

 variation, if extensive, may remove all traces of what was once a suf- 

 ficient ground for comparison, thus destroying the possibility of a 

 complete system of homology. The idea of homology, then, though 

 of the utmost practical convenience, is too rigid to respond perfectly 

 to the true morphological relations of organisms and breaks down 

 when extreme pressure is brought to bear on it. This inadequacy has 

 already been clearly pointed out by Bateson ('92, p. 102). 



Cambridge, Mass., Jan. 31, 1896. 



