CLASSIFICATION. 413 



those- marks, it will then be usually found to include too 

 many forms ; if the definition be made more particular, 

 the result is to produce so-called anomalous genera, which, 

 while they are held to belong to the class, do not in all 

 respects conform to its definition. The practice has hence 

 arisen of allowing considerable latitude in the definition 

 of natural orders. The family of Cruciferae, for instance, 

 forms an exceedingly well marked natural order, and 

 among its characters we find it specified that the fruit 

 is a pod, divided into two cells by a thin partition, 

 from which the valves generally separate at maturity ; 

 but we are also informed that, in a few genera, the pod 

 is one-celled, or indehiscent, or separates transversely into 

 several joint s m . Now this must either mean that the 

 formation of the pod is not an essential point in the 

 definition, or that there are several closely associated 

 families. 



The same holds true of typical classification. The type 

 itself is an individual, not a class, and no other object can 

 be exactly like the type. But so soon as we abstract the 

 individual peculiarities of the type and thus specify a 

 finite number of qualities in which other objects may 

 resemble the type, we immediately constitute a class. 

 If some objects resemble the type in some points and 

 others in other points, then each definite collection of 

 points of resemblance constitutes intensively a separate 

 class. The very notion of classification by types is in 

 fact erroneous in a strictly logical point of view. The 

 naturalist is constantly occupied by endeavouring to mark 

 out definite groups of living forms, where the forms them- 

 selves do not in many cases admit of any such rigorous 

 lines of demarcation. A certain laxity of logical method 

 is thus apt to creep in, the only remedy for which will be 



m Bentham's ' Handbook of the British Flora' (1866), p. 25. 



