226 INTRODUCTION TO PHILOSOPHY OF SCIENCE 



''Sodium is silver in color' can be verified observationally, 

 but the proposition "Sodium burns with a yellow flame" 

 must be verified experimentally. However, if all perception 

 involves control the difference between these two situations 

 cannot be great. Both propositions can be expressed in the 

 following form: "If I place a piece of sodium in a certain 

 physical situation (characterized merely by certain condi- 

 tions of atmosphere and light in the one case, and by the 

 presence of a flame in the other), and if I focus my eyes 

 and attend to what is taking place, I shall perceive a cer- 

 tain event (a silvery luster in the one case, and a yellow 

 flame in the other)." The only difference between the two 

 situations is that the modification introduced into the first 

 situation is not presumed to transform the end-object, while 

 the modification introduced into the second is. If experi- 

 mentation is defined as the introduction into nature of 

 changes which would not have occurred without man's 

 intervention, the distinction between observation and ex- 

 perimentation becomes quite obliterated. For man is himself 

 a natural object, and whenever he observes in a controlled 

 manner he introduces changes into nature. None but the 

 most passive sort of observation, therefore, would be non- 

 experimental in character. As a working distinction between 

 the two, one may say that experimentation involves con- 

 trol over both the observer and the observed, while observa- 

 tion involves control only over the observer. 



The so-called experimental methods of John Stuart Mill x 

 have application in this connection, though their use is 

 somewhat more limited than the prominence given to them 

 by logicians would lead one to believe. In their narrowest 

 scope they have reference to the problem of the ascertain- 

 ment of causes from effects, or of effects from causes. In 

 their broadest scope they may be considered as techniques 

 for the determination of laws, or correlations (of which 

 causal correlations are a species). They are not, properly 



1 Svstem of Logic (7th ed., London: Longmans, Green, 1868), Book III, 

 Chaps. VIII, IX. 



