350 



Part III. — Eighteenth Annual Report 



shows a tendency to degenerate in several species and genera, e.g. 

 Scophthalmus, so that this character cannot always be relied upon with 

 certainty. 



Similar difficulties arise with the other characters which have been 

 described. Each by itself is able to give some sort of classification, but 

 the genera are then arranged merely according to their resemblance 

 with respect to these characters and not necessarily according to their 

 natural affinities. Thus we should obtain different combinations of the 

 Heterosomata if we arranged the various genera according to the size 

 and structure of the mouth, or according to the structure of the 

 abdominal or of the last caudal vertebrae, and so on, just as the classifiers 

 of the Heterosomata hitherto have done, and one combination would be 

 as good as the others. Each of these characters would divide the 

 Heterosomata into two or three large groups, but the lines of division 

 would differ in the different combinations. They are too general, 

 making no allowance for the results of convergency, and do not permit 

 of fine transitional stages being displayed. When, however, they are 

 taken along with the two characters which have been mentioned above, 

 the position of the ventral fins, and the structure of the olfactory organ, 

 they furnish important evidence for the further division of the various 

 groups. 



Apart from the particular problems which arise in endeavouring to 

 fix the details of each group, and which will be considered later, there 

 are still a few general questions to be answered. It has been shown 

 repeatedly that what is best for the purposes of classification is not 

 necessarily the best for showing natural affinities, and the question 

 thence arises— Can any classification truly represent the natural affinities 

 of the species classified ? In theory the answer is simple, because the 

 classification that will receive the most respect is the one which displays 

 in a simple and natural manner the groupings of the various species as 

 we find them in nature. In practice, however, there are certain 

 unwritten laws according to which a classification should be formed. It 

 should be clear, not cumbered with too many details, and such that any 

 specimen examined may be readily located. Such have been the pre- 

 valent notions hitherto, and a conflict thence arises between practical 

 expediency and theoretical correctness. Recent opinion, however, as 

 shown in the work of the American writers, has tended strongly in 

 favour of the latter. The work of Jordan and his collaborators has 

 shown clearly in its various developmental phases that, after all, several 

 different classifications would suit the demands of practical expediency, 

 and the chief value of their last great work (33), for the Heteroso- 

 mata at any rate, is that it has passed beyond this stage and really 

 endeavours to display the natural affinities of the various species, genera, 

 and groups. 



The second and more serious question that then arises is whether 

 the characters presented to us — such as have been described here —are 

 really capable of displaying affinities. In these days of the study of 

 variations, when the older notions of the value and fixity of heredity 

 are being rudely shaken, we are not inclined to believe in the per- 

 manence, or even comparative permanence, of any character whatsoever. 

 May not, then, the complex condition of things in the environment with 

 the flexible nature of habits have so reacted on the plastic structures that 

 all the lines of genetic relationship are woven round and round inex- 

 tricably into the form of a net or huge tangle, and any chance of finding 

 a " phylogenetic tree" thereby rendered hopeless? In the case of a 

 large group of animals, such as the Fishes as a whole, or of the Teleostomi, 

 one might assent to this, and agree also with Sagemehl that even for a 



