Generic Characters in the Coccaceae 147 



In many fields of science the statistical method, in its strict sense, 

 is not applicable. Where laboratory experiments may be made, as 

 in most fields of physics and chemistry, a comparatively small 

 array of data obtained under perfectly controlled conditions may 

 permit the derivation of laws of relationship without extensive statis- 

 tical analysis. The same thing is true in certain fields of 

 biological research. As soon, however, as we proceed to the subtler 

 problems of evolution, it becomes necessary to accumulate a large 

 number of observations and to analyze them by recognized statis- 

 tical methods. These methods alone have brought order out of 

 chaos in anthropology (Ripley, 1899). They have laid the first 

 foundation for a real science of mental and social phenomena (Thorn- 

 dike, 1904; Woods, 1906). They offer the most promising clue for 

 tracing the true relationships among the lower forms of plant and 

 animal life. 



As we have elsewhere pointed out, the classification of the bac- 

 teria presents peculiar difficulties. 



Morphological distinctions are so slight that physiological characters must ne- 

 cessarily be invoked in order to separate and classify the various organisms, and these 

 physiological characters are often variable. Pathogenicity may be taken as a type 

 of those powers of the organism which are easily and profoundly modified by external 

 conditions. On the other hand, there are numerous characters which appear to be 

 extremely constant. Such minute differences as occur in the resistance of different 

 races to unfavorable conditions often remain unchanged through long periods of 

 cultivation. In using these constant characters for classification we are met by another 

 difficulty. Though constant, the differences are very minute, and in studying a number 

 of organisms a perfect gradation is often found between the widest e.xtremcs. This 

 is exactly what should be expected from organisms which reproduce only by asexual 

 methods, since it is the fusion of independent cells which swamps minor differences 

 producing the uniformity of species among higher plants. W'hh asexual reproduc- 

 tion every minute variation which is inheritable must persist unchanged until some 

 Other chance variation occurs. Each such variation means a new and different type of 

 bacterium. 



The immense number of generations which may succeed each other in a short 

 space of time makes boundary lines as shifting as they would become among the higher 

 plants if a dozen geological epochs were considered all at once. 



Since ^"ith unicellular organisms acquired characters may probably be inherited 

 in a higher degree than with other forms, existing races of bacteria will be markedly 

 influenced by the selective effect of environmental conditions, and must bc^ar the impress 

 of their recent history. 



There are, therefore, no species among the bacteria in quite the sense in which we 

 ordinarily use the word — as indicating a group of individuals bound together by a num- 



