476 University of California Publications in Zoology [Vol.12 



isins may be comeidently distributed as regards latitude and longi- 

 tude, and still be completely isolated in their vertical distribution." 



The results of Michael's work indicate that the increase of exact 

 knowledge in at least one group of oceanic organisms has empha- 

 sized the possible significance to these organisms of isolation. 



It is concluded (Michael, 1913, p. 46) that "Jordan's Law" is 

 only partly true, when tested by vertical distribution, for while the 

 most closely related species do not inhabit the same environment, 

 they do inhabit the most remote environments; and that the more 

 closely related species of Chaetognatha are isolated from each other 

 either horizontally, vertically, or by virtue of physiological differ- 

 ences causing fertilization to take place in different strata of water. 



Attention should be called to the following points in connection 

 with the data presented on pages 465 to 472 of the present paper, 

 at which the evidence apparently does not clearly harmonize with 

 Wagner's theory. 



Sorex halicoetes of the San Francisco salt marshes is separated 

 from its closest relative, Sorex vagrans vagrans of the humid strip 

 along the Pacific Ocean, by several miles of upland which is inhab- 

 ited by Sorex calif 'ornicus calif ornicus. According to the strict 

 requirements of Wagner's hypothesis, the ranges of the most closely 

 related forms should occupy adjoining territory. If ovir interpreta- 

 tion of the relationship of these forms be correct, Sorex calif ornicus 

 californicus must be conceived of as having invaded the territory it 

 now occupies at the expense of the vagrans stock, crowding the 

 latter to one side or the other. 



Island distribution offers many cases of broad gaps between the 

 ranges of species most closely related, but in most eases these gaps 

 were formerly nonexistent or else the closely related stocks in ques- 

 tion formerly enjoyed some means of transversing them, or are able 

 at long intervals still to traverse them, though not to the extent that 

 there is a quashing of incipient differential characters through inter- 

 breeding. The same principle should be remembered in connection 

 with distribution along mountain systems like the Sierra Nevada, 

 where, for instance, the family Ochotonidae has a discontinuous dis- 

 tribution, being found only in certain "islands" of high Boreal. 



Two very interesting and instructive cases are furnished by the 

 maniculatus series of mice. The relations between Peromyscus 

 maniculatus gambelii and Peromyscus maniculatus rubidus, as well 

 as those between Peromyscus maniculatus austerus and Peromyscus 



