66S 



THE AMERICAN NATURALIST [Vol. XL VIII 



tant step, and the chief contribution has come from very- 

 remote indirect methods. Generally speaking the inves- 

 tigators who choose a direct method of attack often put 

 themselves somewhat in the position of the chemist who 

 would make chemical analysis of living matter when his 

 first step defeats its own purpose by killing the substance 

 to be analyzed. The failure of exclusively direct methods 

 is often evident. Still the ability to obtain results by the 

 method of direct attack, combined with a far too rare 

 ability to tie with them indirectly obtained data, some- 

 times gives noteworthy contributions. 



It accordingly remains to be seriously considered 

 whether or not biology can afford to apply criteria to the 

 measure of the values of investigation. Their application 

 is of course largely unconscious, but the effects are not 

 thereby modified. Noteworthy results of their applica- 

 tion are (a) concentration of work in certain lines indi- 

 cated by a given criterion, and (b) an actual abandoning 

 to a large degree of remote and indirect methods of 

 attacking the problems which the criterion involves. This 

 means the partial abandoning of the methods for which 

 pure science stands. 



Criteria can be safely used only in a very broad gen- 

 eral way, and in application more often to past progress 

 than to current investigation. They are perhaps most 

 valuable as a guide to individual investigators working 

 on problems remote from these more or less central 

 "pure science" questions. That some guide should be 

 in the hands of such workers is beyond question. In the 

 hands of those attacking the problems directly they often 

 appear detrimental because they soon take on an extreme 

 form and become regarded as fundamental. At this 

 stage they are usually in need of extensive revision. If 

 the investigator is contributing observations and details 

 only, he is doing a great service, for such information is 

 needed everywhere. If he is able to combine his own and 

 others results, he almost invariably draws data from all 

 sources, direct and indirect, far and near. Granted the 



