PBINCIPLES AND PRACTICE OF CLASSIFICATION. 69 



•not only common to the members of the group, but distinguish it from all others ; and the 

 statement of these constitutes the definition of the group. 



" Thus, among animals with vertebrae, the class 3Iammalia is definable as those which 

 have two occipital condyles, with a well ossified basi-occipital ; which have each ramus of the 

 mandible composed of a single piece of bone and articulated with the squamosal element of the 

 skull ; and which possess mammae and non-nucleated red blood-corpuscles. 



'< But this statement of the characters of the class Mammalia is something more than an 

 arbitrary definition. It does not merely mean that naturalists agree to call such and such 

 animals Mammalia : but it expresses, firstly, a generalization based upon, and constantly 

 verified by, very %vide experience ; and, secondly, a belief arising out of that generalization. 

 Tlie generalization is that, in nature, the structures mentioned are always found associated 

 together ; the behef is that they always have been, and always will be, found so associated. 

 In other words, the definition of the class Mammalia is a statement of a law of correlation, or 

 coexistence, of animal structures, from which the most important conclusions are deducilde." 

 (Introd. to Classif. of Animals, 8vo, London, 1869, pp. 2, 3.) 



But broad as such laws of correlation of structure are, and important as are the conclusions 

 deducible, we must constantly be on our guard against presuming upon the infallibility either 

 of the data or of the deduction, as the author just quoted goes on to show. Such caution is 

 specially required where there is no obvious reason for the particular combination that may be 

 found to exist. In the case of the ostrich-like birds (Batita), for example, we can understand 

 how a flat, imkeeled breast-bone, a particular arrangement of the shoulder-bones, and a rudi- 

 mentary state of the wing-bones, are found in combination, because all these modifications of 

 structure are evidently related to loss of the power of flight ; and, in point of fact, no exception 

 is known to the generalization, that such conditions of the sternal, coraco.-scapular, and 

 humeral bones always coexist. But in all known struthious (ratite) birds, this state of the 

 bones in mention coexists also with a peculiar modification of the bones of the palate, and no 

 necessary connection between these two sets of diverse characters is conceivable. Now, if we 

 only knew struthious birds, and found the combination in mention to hold with them all, we 

 should doubtless declare our belief, that any bird having such palatal characters would also be 

 found to possess such imperfect wing-apparatus. But this would be going too far : in fact, 

 we know that the tinamous (Dromceognathce) Jiave such a palate, yet have a keeled sternum 

 and functionally developed wings. The real use and proper application of such generalizations 

 is to teach the lesson, that creatures exhibiting such modified combinations of characters are 

 genetically related to each other just in the degree to which they possess characters in common, 

 and are genetically remote from each other in the degree to which they do not possess characters 

 in common : i. e., that their similarities and distinctions of structure are sure indexes of their nat- 

 ural affinities. To take another case, derived from consideration of a large number of existing 

 birds : it is an observed fact, that a particular arrangement of the plates upon the back of the 

 tarsus, a peculiar modification of the lower larynx or voice organ, and an undeveloped or abortive 

 condition of the first large feather on the hand, are found associated in a vast series of birds, 

 constituting the gi'oup c)f Passeres called Oscines. What possible connection there can be 

 between these three separate and apparently independent modifications we cannot even sur- 

 mise ; but that they have some natural and necessary connection we cannot doubt, and that 

 the connection is causal, not fortuitous, is a logical inference from the observed fact, that 

 birds which present this particular combination are also closely related in other structural 

 characters; that is, that they have all been subjected to operative influences which have 

 conspired to produce the modifications observed. Given, then, a bird with a known oscine 

 larynx, but unknown as to its feet and wings, it would be a reasonable inference that 

 these members, when discovered, would present the characters observed to occur in like 

 cases. But the first lark {Alaudidce) examined would show the inference to be fallible; 



