VERIFICATORY TECHNIQUES 227 



speaking, methods for the discovery of conditions, as their 

 name implies, but rather methods of verification; they are 

 useless unless the investigator has some conception, how- 

 ever vague, as to what kind of event may be the condition 

 which is being sought after. They are not, therefore, prin- 

 ciples which enable one to infer conditions from consequents, 

 or the reverse; they are rather principles of location. Based, 

 as they are, upon the general kind of connection which 

 condition and consequent bear in nature, they guide ex- 

 perimentation in the sense that they instruct one where and 

 when to look for conditions, given consequents, and where 

 and when to look for consequents, given conditions. This 

 makes them important techniques, especially of infirmation. 

 What should be insisted upon with reference to them is 

 that their use is somewhat more limited than their name 

 implies. 



An important problem in confirmation is that which 

 arises when the predicted consequences fail to receive cor- 

 roboration, i.e., when they are infirmed rather than con- 

 firmed. The usual procedure is to reject the hypothesis 

 from which the deductions were drawn. It is commonly 

 recognized that many confirmations are required to estab- 

 lish an hypothesis, but one single infirmation is sufficient 

 to disestablish it. This is based upon the fact that the 

 hypothesis is the sufficient, but not the necessary, condition 

 for the predictions. Hence the failure of the predictions 

 permits one to deny the hypothesis, but the occurrence 

 of the predictions does not permit one to affirm the hy- 

 pothesis, for there may be alternative hypotheses each of 

 which is sufficient. 



But what has not been so commonly recognized is that 

 there are alternatives to the complete rejection of the hy- 

 pothesis in cases where infirmation occurs. In the first place, 

 the hypothesis need not be rejected but may require only 

 modification. Every hypothesis is merely an aggregate of 

 propositions characterizing the hypothetical entity, and as 

 such suffers no serious shock if some members of the ag- 



