( 398 ) 



trustworthiness ami the degree of completeness of the observations, and, if this 

 evidence can be relied upon, secondly on the interpretation of the facts. 



Now, in weighing off the evidence sj-stematists follow two very different methods. 

 The first method, a relic from pre-Darwinian times, is by far the dominant one in 

 ordinary systematic work, because it is so very convenient ; it consists in taking as 

 the basis of the primary division of the group (and subseijuently of each subgroup 

 and minor division) certain distinguishing cluiracters which the respective classifier 

 assumes to be of primary importance, withont giving any reason (1) why the forms 

 which possess that character are closer related to each other than to forms which do 

 not possess it, and (2) — and this is the point wliere the grave mistake comes in — 

 why the forms that are devoid of that cliaracter are all phylogeneticnlly closer 

 connected with each other than with forms which have the respective character. It 

 is ob^-ious that the result of this method entirely depends on what characters the 

 classifier selects for gradually dividing up the group of forms, and, as the selection 

 is arbitrary inasmuch as the phyletic significance of the selected character is merely 

 assumed to be warranted, that the classification is based on a string of assumptions. 

 Hence it is self-evident that the classifications drawn up by different authors 

 according to this method must naturally deviate widely from one another, if the 

 classifiers do not make the same assumptions ; and as there is no reason why they 

 should select the same distinguishing characters and use them in the same order, 

 the antagonistic results of different authors are merely antagonistic assumptions. 

 The contradiction between different systems of classification based on different 

 primary assumptions is in the case of Butterflies very obvious. According to 

 Professor (irote, the Butterflies have a diphyletic origin, the forms with a vein on 

 the forewing running into tlie posterior margin of the wing {Fapilionidnii) forming 

 the one, all the other Butterflies, which do not possess such a vein in such a position, 

 the other phylum. However, a diphyletic origin of tlie Butterflies we should also 

 have if we took a character of the claws as the basis of the primary division ; but 

 in this case the Fieridai', wliich have all divided claws, would be the one phylum, 

 and all the other Butterflies (inclusive of Papilionidae) with not-divided claws belong 

 to the other. Or, if the degree of abortion of the forelegs were made the basis 

 of division, we should have Hesperiidae, Papilionidae, and Pieridae as one hexapod 

 phylum, the L>jcaenidae and Kr>/ciniduf (and a few yi/mphalidaei) as a heteropod 

 ])hylum, and the yymphalidai; as a tetrapod phylum. Again, the Si/mphalidae 

 would stand separate from all the rest of the Butterflies if the presence of carinae 

 on the antennae were considered to be of primary importance; while we should have 

 four primary groups if we took into consideration the development of antenna! 

 grooves, the first group containing the forms witliout grooves {Ilesperiidae, 

 Lijcaenidae, many Papilionidae), the second the forms with one groove {^l-lrijcinidae 

 and Pierinae), the third consisting of the forms with two grooves {Nymphalidae and 

 many Papilionidae), the fourth of forms with three grooves ( Diainorphiiiiae). And 

 so on. The great difference in the groujjiug of tlic Butterflies according as we take 

 this or that organ for the basis of division most obviously shows that such classifi- 

 cations are merely a grouping of quantitatively the same degrees in the development 

 ot the respective organ, the grouping being carried out regardless of the individuals 

 exhibiting tlie similar character liaving arrived at that stage of develoi)meut on the 

 same or on different roads, and taking it for granted that the ditt'erence in the 

 character selected for division indicates difierent phyletic origin. Though such 

 a classification is artificial, it is nevertheless striking that, when grouping the 



