THE FACTORS OF ORGANIC EVOLUTION. 63 



of the epidermis is wholly determined by the natural selec 

 tion of favourable variations ; what must be the position 

 taken in presence of the fact above named, that when 

 mucous membrane is exposed to the air its cell-structure 

 changes into the cell-structure of skin ? The position taken 

 must be this : Though mucous membrane in a highly- 

 evolved individual organism, thus shows the powerful effect 

 of the medium on its surface ; yet we must not suppose that 

 the medium had the effect of producing such a cell-struc 

 ture on the surfaces of primitive forms, undifferentiated 

 though they were; or, if we suppose that such an effect 

 was produced on them, we must not suppose that it was 

 inheritable. Contrariwise, we must suppose that such effect 

 of the medium either was not wrought at all, or that it 

 was evanescent : though repeated through millions upon 

 millions of generations it left no traces. And we must 

 conclude that this skin-structure arose only in conse 

 quence of spontaneous variations not physically initiated 

 (though like those physically initiated) which natural selec 

 tion laid hold of and increased. Does any one think this a 

 tenable position? 



And now we approach the last and chief series of 

 morphological phenomena which must be ascribed to the 

 direct action of environing matters and forces. These are 

 presented to us when we study the early stages in the 

 development of the embryos of the Metazoa in general. 



We will set out with the fact already noted in passing, 

 that after repeated spontaneous fissions have changed the 

 original fertilized germ-cell into that cluster of cells which 

 forms a gemmule or a primitive ovum, the first contrast which 

 arises is between the peripheral parts and the central parts. 

 Where, as with lower creatures which do not lay up large 

 stores of nutriment with the germs of their offspring, the 

 inner mass is inconsiderable, the outer layer of cells, which 

 are presently made quite small by repeated subdivisions, 



