THE ART OF SCIENTIFIC INVESTIGATION 



of which baffled investigators for years. In Western Australia, 

 H. W. Bennetts for certain reasons suspected that the disease 

 might be due to lead intoxication. To test this hypothesis he 

 treated some sheep with ammonium chloride which is the 

 antidote to lead. The first trial with this gave promising results, 

 which, however, were not borne out by subsequent trials. This 

 suggested that the disease might be due to the deficiency of some 

 mineral which was present in small amounts in the first batch 

 of ammonium chloride. Following up this clue, Bennetts was 

 soon able to show that the disease was due to deficiency of copper, 

 a deficiency never previously known to produce disease in any 

 animal. In Bennetts' own words : 



" The solution of the etiology came in Western Australia from 

 an accidental ' lead ' [clue] resulting from the testing of a false 

 hypothesis."^^ 



Use of hypothesis in research 



Hypothesis is the most important mental technique of the 

 investigator, and its main function is to suggest new experiments 

 or new observations. Indeed, most experiments and many 

 observations are carried out with the deliberate object of 

 testing an hypothesis. Another function is to help one see the 

 significance of an object or event that otherwise would mean 

 nothing. For instance, a mind prepared by the hypothesis of 

 evolution would make many more significant observations on a 

 field excursion than one not so prepared. Hypotheses should be 

 used as tools to uncover new facts rather than as ends in 

 themselves. 



The illustrations given above show some of the ways in which 

 hypotheses lead to discoveries. The first thing that arrests 

 attention is the curious and interesting fact that an hypothesis 

 is sometimes very fruitful without being correct — a point that 

 did not escape the attention of Francis Bacon. Several of the 

 illustrations have been selected as striking demonstrations of 

 this point, and it should not be thought that they are a truly 

 representative sample, for correct guesses are more hkely to be 

 productive than ones that are wrong, and the fact that the 

 latter are sometimes useful does not detract from the importance 

 of striving for correct explanations. The examples are, however, 



46 



