PRINCIFLES AND PRACTICE OF CLASSIFICATION. 73 



which cannot he decided to belong, or not to belong, to the conventional class of birds, because 

 we have found it ccjnvenient and expedient to consider the presence of feathers a fair criterion, 

 or necessary qualification. But what, when an animal is discovered the covering of whose body is 

 half-way between the scales of a lizard and the plumes of a bird, and whose structure is other- 

 wise as equivocal ? This may happjen any day. A feather is certainly a modified scale ; a 

 feather has dciulitless been developed out of a scale. In the case supposed, we should have to 

 modify mir dcliuition of the " Class of Birds " ; that is, change our ideas upon the subject, and 

 alter the boundary-line we established between the classes of birds and reptiles ; whereas, 

 were a " class" something naturally definite, independent, and fixed, all that we could learn 

 about it would only tend to establish it more surely. The same obscurity and uncertainty of 

 definition attaches to groups of every grade — from the Animal "Kingdom'" itself, wliich 

 cannot be cut clear of the Vegetable " Kingdtjm " — down through classes, orders, families, 

 genera, species, and varieties — yes, to tlie iiidividual itself which, however unud.stakable 

 among higher organisms, cannot always be predicated of the lowermost forms of Life. 

 Such divisions, of whatever grade, as we are able to establish f<jr the purposes of classificatii.pn, 

 depend entu-ely upon the breaks and defects in our knowledge. There is no such tiling as 

 drawing " hard and fast " lines anywhere, for none such exist in Nature. 



Taxononiic Equivalence of Groups. — But, however arbitrary they may be, or however 

 obscure or fluctuating may bo their boundaries, groups we must have in zoology, and groups 

 of different grades, to express different degrees of likeness of the objects examined, and so 

 to "classify" them. It is a great convenience, moreover, to have a recognized sliding-scale 

 of valuation of groups from tlje highest to the lowest, and an accepted valuation. Just as in a 

 thermometric scale, there are " degrees" designated as those of the boiling-point of water, tlie 

 heat (if the blood, the freezing of water, of mercury, etc. ; so there are certain degrees of like- 

 ness conventionally designated as those of class, order, family, genus, and species ; always ac- 

 cepted in the (jrder here given, from higher to lower groups. (There are various others, and 

 especially a number of intermediate groups, generally distinguished by the prefix suh-, as suh- 

 faiitihj ; but those here given are generally adopted by English-speaking naturalists, and 

 suffice to illustrate the point I wish to make.) It may sound like a truism to say, that groups 

 of the same grade bearing the same name, whatever that may be, must be of the same value, 

 — nrust be based upon and distinguished by characters of equal or equivalent importance. 

 Erfiivalence of groups is necessary to the stability and harmony of any classificatory system. 

 It will not do to frame an order upon one set of characters here, and there a family up^ m a 

 similar set of characters ; but order must differ from order, and family from family, by an equal 

 or corresponding amount of difference. Let a group called a family difl'er as much frcnn the 

 other families in its own order as it does from some tither order, and by this very circumst;inee 

 it is ncit a family but an order itself. It seems a very simple proposition, but it is too often 

 ignored, and always with practical iU result. Two points should be remembered here : First, 

 that absolute size or numerical bulk of a group has nothing to do with its taxonomic value: 

 one order nniy coutain a thousand species, and another be represented by a single species, 

 without haviug its ordinal valuation affected thereby. Secondly, any given character may 

 assume chffercnt importance, or be of different value, in its application to different groups. 

 Thus, the number of primaries, whether nine or ten, is a family cliaracter almost throughout 

 Osciiies ; but in one oscine family (FiVeornVte) it has scarcely generic value. It is diificult, 

 liowever, to determine such a iioint as this without long experience. Xor is it possible, in 

 fact, to make our groups corre.si^ond in value mth entire exactitude. The most we can Injpe 

 for is a reasonable approximation. As in tlie thermometric simile above given, "blood heat " 

 and other points fluctuate, so does order not always correspond with order, nor family with 

 family, in actual significance. What degree of difference shall be "ordinal"'? What shall 



