METHOD OF DISCOVERING CAUSAL LAWS 163 



able cause of the phenomenon. Obviously, the only way to 

 reach certitude on this latter point is by instituting a careful and 

 detailed analysis of the whole phenomenon and its surroundings, 

 and so satisfying ourselves by further and more searching ob 

 servation, under special and modified conditions if necessary 

 that, amongst all the surroundings of the phenomenon, none have 

 a causal influence upon it except those supposed by us to have 

 such influence. 



All observation of facts is selective. That is to say, what we 

 observe is largely determined by the nature and direction of the 

 interest we take in what comes under our notice. Our minds are 

 never purely passive, but are constantly interpreting the data of 

 sense experience, and reasoning more or less unconsciously from 

 those data. Hence arises an initial danger, that our observations 

 may be vitiated by bias. What we observe in any phenomenon 

 depends largely on our previous knowledge : the skilled engineer 

 &quot; sees &quot; more in the locomotive than the uneducated peasant, 

 though the latter may have sight, and all other senses, as sharp as 

 the former. Superior knowledge it is that renders observations 

 more fruitful. The well-stocked mind will perceive analogies 

 where the ordinary mind will not. The example of Malus, above 

 referred to, is a case in point. Yet this very psychological fact 

 exposes the observer to the danger of falling a victim to pre 

 conceived notions. We often think we see what we only imagine. 



Again, the selection and isolation of the phenomenon to be 

 examined, and of the elements to be tested as its likely causes, 

 are more or less arbitrary processes, in the sense that they must 

 be determined and prosecuted by the individual himself: for their 

 proper and fruitful prosecution an extensive knowledge of the 

 matter in hand is the only guarantee. The investigator who is 

 not well acquainted with his subject will not be likely to detect all 

 the conditions that are operative, or to eliminate and disregard all 

 that are inoperative, in conducting his observations arid experi 

 ments. Against this, as against the danger of bias, logic can 

 furnish no safeguard. Both will be referred to at greater length 

 in connexion with Fallacies. There, also, we shall deal with yet 

 another mistake which may easily be made : the error of inferring 

 non-existence, whether of instances or of conditions, from mere non- 

 observation of the latter. It is only when these are of such a kind 

 that they could not have escaped observation had they existed, 

 that such inference is legitimate. 



II * 



