( xcix ) 



into fVmr others liv mie of the organs becoiuinj; reduceil in oacli. anil tliese 

 fonr again into more in the same way, we have the following iliagraiu (for the 

 sake of convenience in printing the middle ]iart of the diagram has not been 

 completed) : — 



V. a'Uiul' a'h'r'il' it'h'c'd' alic'il' alic'iV ii'h'c'il' dHjc'd' ii'h'c'il' dli'i-'il' n'h\-'<V ii'O'c'd' ii'h'c'il' 



I I I I I I I , , I , , , I , I , , , I , „ J , , 



IV. n!Uc'd a'h'cd' abed' a'b'c'd a'b'cd' a'bc'd' a'b'cdl dlicd! a b cd ab cd a be d al) c d 



I I I I I I J I I I I I 



~ '\' "I 1 J' , ' I , 



III. ii'h'cd (I'lic'd ii'hi-il' abed' abed abed 



II. 

 I. 



A : a'bed 



ah'ed 

 I 



alic'd 



abed 



abed' : D. 



The development, starting from I., re.sult.s first in a series of fonr derivations 

 (II.) differing from one another in two organs. The second series of derivations 

 (III.) contains forms deviating from each other in two, three, or fonr organs. 

 In the third series (IV.) the forms differ onl_v in two organs, or are 

 identical. And we arrive finally (V.) at a series of identical forms. There is 

 divergency from I. to III., and convergency from III. to V. Considering now 

 the phylogenetic relation and the similarity of the various modifications, it 

 will be fonnd that the derivations of A in series IV. are more similar to 

 the derivations of D in the same series than to A in series II., and 

 that the derivations of A in series V. are indistinguishable (in organs 

 abed) from the respective derivations of D in the same series, while they 

 are different from the forms in series IV. from which they are derived. It 

 is therefore evident that the classifier who judges from these organs a/icd 

 alone cannot possibly find out the trne phylogenetic connection between the 

 various forms. He will easily mistake the forms which are equally reduced 

 for forms which are closely allied. To him the members of series V. will 

 appear to be very near relations, while they are in fact the end-products of 

 different phylogenetic branches. 



Let us take as illustration the retrogressive development so often observed 

 in SjiliiiK/idae of the tongue, the midtarsal comb, and the claw-segment. If 

 we start from a form with long tongue, with midtarsal comb, and with fully 

 ilcveloped pnlvillus and paronychinm, and assume as before that one of these 

 organs becomes reilnced in each derivation from that ancestral type, we arrive 

 finally at forms which agree with one another in having all the four organs 

 reduced. Therefore the rcdnetion of the tongne and tlie absence of the mid- 

 tarsal comb, of the pnlvillus, and of the paronycliinm in tliose species or genera 

 are not an expression of close relationship. Looking at the j)edigree of the 

 genera of Hphingicae (facing p. 30), it will be notice<l that a reduction and loss 



