70 THE VARIATION OF ANIMALS IN NATURE 



rats found in different houses in India by Lloyd (191 2) and 

 the statistical differences between communities of ants found 

 in different nests (Alpatov, 1924) and in the ' races ' otPartula 

 found on single trees by Pilsbry, Hyatt and Cook (191 2). 

 For such ' besondere kleine lokal geographische Einheiten ' 

 Semenov-Tian-Shansky (1910) has proposed the name ' natio? 

 We might even include here such groups as are produced by 

 a gregarious instinct and appear as centres of attraction 

 in populations not broken up by topographical obstacles 

 (' schools,' shoals and rookeries). In the majority of cases the 

 groups under discussion represent mere statistical divergences 

 from the mean of the population, such as are seen in the per- 

 centage-difference of colour- and band-classes of land snails 

 and in the different combinations of ear-, tail- and foot-length 

 of Peromyscus. 



How far the groups which we have been discussing are 

 hereditarily stable it is impossible to say. Experimental proof 

 is available to show that the races of ^oarces and Lebistes 

 (Schmidt), Peromyscus (Sumner, 191 5), Cerion (Bartsch, 1920), 

 moths (Goldschmidt, 1922, 1923) and bees (Alpatov, 1929) 

 are stable. We would, however, surmise that a good many 

 alleged racial distinctions are of the nature of ' fluctuations ' 

 (cf. Woltereck on non-inheritable racial characters of the 

 Cladocera, 1928). Much valuable work remains to be done 

 in this field. Crossing experiments have been undertaken 

 by Sumner (191 7), who finds that some subspecies of Pero- 

 myscus maniculatus can be successfully crossed, while others 

 are sterile inter se. 



The fact that populations are divisible into distinct geo- 

 graphical groups such as we have been describing and that 

 some taxonomic species are constellations of geographical forms 

 has led certain students to seek some means of distinguishing 

 such composite groups. They were first called ' Formen- 

 kreise ' by Kleinschmidt ; but Rensch (1929) has recently 

 proposed the term ' Rassenkreise ' for them and has thoroughly 

 examined the subject. He suggests that the term ' species ' 

 should be restricted to groups of mutually fertile and struc- 

 turally similar individuals which exhibit only individual, 

 ecological or seasonal variation, having heritable differences 

 but not divisible into geographical races. Rensch's definition 

 {I.e. p. 15) has to be taken in conjunction with that of his 



