114 MR. W. BATESON OX [Feb, 2, 



that any one of these teeth should he considered supernumerary- 

 more than any other ; and in the absence of such evidence it 

 -would, as I think, be best to regard the four premolars in this 

 specimen as collectively representing the three premolars of the 

 normal. For just as a stick may be broken into three pieces or into 

 four, so would it seem to be with Multiple Parts. The epithehum 

 -which normally gives off three enamel-germs has here given off four 

 such germs, and I believe that it is as impossible to analyze the four 

 teeth and apportion them out among the three teeth as it would be 

 to homologize the sides of an equilateral triangle with the sides of a 

 square of the same peripheral measurement, or to homologize the seg- 

 ments of a 4-rayed Sarsia ^ with segments of its 6-rayed " sport." 

 To make such an attempt would be to disregard the plain and ob- 

 vious indications of the true nature of the phenomenon, and any 

 theory of Homology which recognizes this class of problem as profit- 

 able or legitimate is, I beHeve, founded upon a wrong conception 

 of the physical process of Division. 



For, after all, it is with a process of phj^siological Division 

 that we have here to do, and the Division which results in the 

 formation of a series of Multiple Parts is probably a manifesta- 

 tion of the same physical process as the Division of a cell or the 

 segmentation of an ovum. Whoever will discover by what physical 

 process an ovum segments will give us the key to the problem of the 

 segmentation of tissues into Series of Multiple Parts; and though 

 ■we are far enough from having any such knowledge, we should at 

 least recognize that this is the problem to be dealt with, and any 

 working hypothesis of the nature of Homology should be, at 

 all events, in harmony with what is known of the processes of 

 Division and should be founded upon them. Now the ordinary 

 conception of the relationship of Homology as defined above, though 

 it has been a useful instrument as a basis of nomenclature and so 

 forth, is nevertheless inconsistent with the facts of Division and is 

 founded on assumptions which are not justified, suggesting a view of 

 the physics of Division which is wrong. 



In order to appreciate this, let the reader consider, for example, 

 the case of Omrnatophoca rossii given above. Judged by the ordi- 

 nary rules of morphological criticism, this specimen shows one or 

 both of two things : — 



(1) The first premolar of OmmatopJioca may in itself represent 

 two premolars of an ancestor. 



Or (2) In the descendants o\' Omrnatophoca the single first pre- 

 molar may be represented by two distinct and several pre- 

 molars. 



^ Sarsia is the Medusa of a Gymuoblastic ITydroicl {Syncoryne). The 

 normal Ibrm has 4 radial canals, 4 ocelli, and 4 tentacles. Three specimens 

 having SIX of each of these parts are recorded, two (American) by L. Agassiz, 

 Mem. Amer. Acad. Sci. iv. 248, pi. v. fig. 5, and one (British) by Eomaues, 

 Journ. Linn. Soc. xii. p. 527. All of these were radially symmetrical. 



