INTRODUCTION y 



and sophistries that ordinarily accompany statements of fact de- 

 rived by inspection, much as they might miss their own clothing. 

 The practice of combining facts with non-descriptive inferences is 

 so prevalent that facts seem not to sit comfortably in some minds 

 without that kind of reasoning. The urge toward rationalization 

 is as though natural phenomena did not subsist in their own 

 strength but needed bolstering before they become acceptable. 

 Statements are said to be tedious and dry; they are supposed to 

 need flavoring with imagination and hypothesis before they are 

 presentable. I propose to depend heavily on correlations for 

 flavors and significances. If a reader were to overlook this explicit 

 procedure (§ 154) he might make the mistake of supposing that 

 some data presented have no connotations, as though they had been 

 set down in useless and thoughtless disorder. Recognition of this 

 purpose rather than of some other purpose is a means of grasping 

 quickly the plan and ends of the investigation. 



(2) Quantitative data are obtained and utilized at every point 

 possible. They make precise whatever phenomena are studied. 

 They allow conclusions to be set down in concrete numbers and 

 dimensions. They avoid certain of the intrigues of words and 

 statements whose generality would require further investigation. 



The data selected in what might seem at first to be a prolix pro- 

 fusion lead by induction to conclusions concerning maintenances 

 and regulations. "Illustrations" have been suggested heretofore 

 in the pursuit of these same general phenomena. To me the factual 

 material would not be sufficient for the present inductions so long 

 as it was qualitative. Even in what is currently termed quantita- 

 tive biology, only qualitative conclusions are regularly drawn. But 

 numerical relations seem to me to be precise parts of conclusions ; 

 phenomena are not just large or small, more or less, yes or no, 

 proved or disproved. Equations and graphs may express more 

 forms of relations than words do (§ 152). 



(3) Generalisations result from the quest for uniformities 

 among the quantitative descriptions of particular kinds of physio- 

 logical phenomena. Instead of inferring that a relation observed 

 will turn out to be a constant one, it seems preferable to count how 

 frequently, with what variability, and under what varieties of con- 

 ditions each uniformity appears. The reiteration of certain quan- 

 tities, of combinations, of correlations, becomes the basis of under- 



