THE ART OF SCIENTIFIC INVESTIGATION 



be representative. Bacon warned against being led into error 

 by relying on impressions. 



" The human understanding is most excited by that which 

 strikes and enters the mind at once and suddenly, and by which 

 the imagination is immediately filled and inflated. It then begins 

 almost imperceptibly to conceive and suppose that everything is 

 similar to the few objects which have taken impression on the 

 mind." 



A very common way in which mistakes arise is by making 

 unjustified assumptions on incomplete evidence. To cite a 

 classic example, in the lecture in which he enunciated his famous 

 postulates, Robert Koch described how he had been led into 

 error by making what appeared to be a reasonable assumption. 

 In his pioneer work on the tubercle bacillus he obtained strains 

 from a large variety of animal species and after having subjected 

 them to a series of tests he concluded that all tubercle bacilli 

 are similar. Only in the case of the fowl did he omit to do 

 pathogenicity and cultural examinations because he could not 

 at the time obtain fresh material. However, since the morphology 

 was the same, he assumed that the organism from the fowl was 

 the same as those from the other animals. Later he was sent 

 several atypical strains of the tubercle bacillus which, despite 

 a protracted investigation, remained a complete puzzle. He said : 



" When every attempt to discover the explanation of the dis- 

 crepancy had failed, at length an accident cleared up the 

 question." 



He happened to get some fowls with tuberculosis and when 

 he cultured the organisms from these : 



" I saw to my astonishment that they had the appearance and 

 all the other characters of the mysterious cultures." 



Thus it was he found that avian and mammalian tubercle 

 bacteria are different. ^^ Incidentally, this reference, which I 

 found when looking for something else, seems to have been 

 " lost ", for some current text-books state that there is no evidence 

 that Koch ever put forward the well-known postulates contained 

 in this lecture. 



One can easily be led astray when attempting to isolate an 

 infective agent by inoculation and passage in experimental 



ii8 



