A CLASSIFICATION OF GROUPS 313 



a very undesirable goal in investigations like the present one. 

 The habits in the possession of an organism have generally 

 been acquired under every-day conditions and they are there- 

 fore least liable to change under the conditions under which 

 they were developed than under any other conditions. 



Since an observer cannot, in general, be entirely removed from 

 sources of environmental stimulation, it must happen now and 

 then, depending on the one hand on the degree of the con- 

 stancy of the environment or on the "habitualness" of the environ- 

 ment, and on the other on the constitution of the organism 

 itself, that groups will be beat whose nature depends essentially 

 on the environment and not on the organization of the organ- 

 ism, which, if it could be accurately measured and defined, 

 would establish the identity of every group of any observer. 



8. Other analytical methods. It occurred to the writer to ex- 

 tend the methods so generally successful in the consideration of 

 a single series of groups of any given observer to a comparative 

 study of several records of the same observer for the purpose of 

 ascertaining whether the same general relations obtained for the 

 same group in different series. For this purpose, records of the 

 same observer (beat on different days) which contained a com- 

 mon group were selected and compared. The comparisons were 

 effected between the groups that immediately preceded or fol- 

 lowed the group common to the several series. 



In records 3 and 10, observer S beat the following series: 



/ 57 196 47 337 57 354 

 \ 199 18 47 264 73 126 



The common group in the two records or parts of records is the 

 47. If a simple relation obtains between any pair of groups 

 that have about the same temporal positions with respect to the 

 common group 47, or any common group, or any groups that are 

 known to be related, then the significance of the relations that 

 obtain for the groups of any single series is thereby strength- 

 ened because the corresponding probability problem becomes 

 more restricted. Another condition or another set of conditions 



