NATURAL HISTORY, STATISTICS, AND APPLIED MATHEMATICS 25 1 



basis! It would be as difficult as separating two closely related species 

 statistically. 



The basic data of classical Genetics and of classical Taxonomy are much 

 more alike than has been generally realized. I have myself collected and 

 determined tradescantias from Texas roadsides where two or three species 

 were growing intermingled. I have also scored Primula sinensis crosses seg- 

 regating simultaneously for "oak," "fern," and "tongue," three mutants 

 which affect leaf shape. The two experiences (identifying tradescantias and 

 scoring primrose genes) were all of a piece. One was cataloguing patterns in 

 both cases. The initial judgments were quite as qualitative for the one as for 

 the other. However, after the tradescantias had been identified, one could 

 only catalogue them as Tradescantia gigantca, T. occidentalis, and T. humilis 

 and a host of unclassifiable intermediates. With the primulas one had such a 

 precise understanding of the patterns that the results were set out in neat 

 mathematical symbolism OOTTyy, OottYY, etc. [Parenthetically, it is some 

 satisfaction to realize that if I went back today to that same Texas roadside, 

 I could now, with the method of Pictorialized Scatter Diagrams (1954), born 

 out of my struggles with the Tradescantia problem, catalogue the puzzling 

 intermediate and the parental species, with almost the precision of an experi- 

 ment in Genetics! J 



Though pattern data are more widely used in Natural History than in 

 other disciplines they are by no means confined to it or to Biology. The 

 periodic system (Physics) has already been referred to. Other examples with 

 the fields to which they belong are cold fronts (Meteorology); terminal 

 moraines (Glacial Geology); the equation of a straight line (Mathematics); 

 electron photographs (Physical Chemistry) ; X-ray diffraction patterns 

 (Physics); the bending of Avena coleoptiles (Biology); a continental shelf 

 (Physiography). 



Yet of all the disciplines, Natural History is the most richly endowed with 

 pattern data. For that reason alone, if for no other, it needs to be kept in the 

 science-training curriculum. It is the one method so broadly based that with it 

 you can always discover something you were not looking for. We can nowadays 

 replace many of the functions of the human mind with a properly designed 

 machine. Yet you cannot design a machine, to find something you do not 

 yet know you are looking for, a machine, to take a specific example, which 

 could examine contaminated agar plates, in advance of any knowledge of 

 antibiotics, and discover penicillin. Natural History trains one to think in 

 terms of hypotheses about the phenomena ; to postpone mathematical models 

 until one has at least a rough idea of what kind of a model will be needed. 

 These are healthy attitudes for a scholar in any field. If, however, we are to 

 bring Natural History to its full stature as a science, we need to learn how 

 to use its pattern data with greater precision; need even more to learn how 



