C xcix ) 



iuto four others Ly one of the organs becoming reduced in eacli, and these 

 four again into more in the same way, we have tlie following dingram (for the 

 sake of convenience in printing the middle part of the diagram has not been 

 completed) : — 



V. al/c'd' a'b'c'd' a'b'c'J' <,7/cd' aUcd' alicd' 



IV. a'b c'd ah cd' a he'd' a'b'c d a l/cd' a lie d 



I I I I I I 



III. 



II. 

 I. 



n'h'cd 



(I'lic'd 



A : abed 



I 

 „:bt;l' 



ah'cd 



a'b'c'd' a'b'c'd' a'b'c'd' a'b'c'd' a'b'c'd' a'b'c'd' 



. I , , I , , . I , I , , . I , . I , , 



a b cd a'bc d a b cd ab c d a be d' ab'c'd' 



I I I I I I 



,1 , 

 abed' 



abed 



abed 



ab'cd' 



I , 

 abed' : D. 



abed' 



The development, starting from I., results first in a series of four 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 four organs. 

 In the third series (IV.) tlie forms differ only 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 found 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 

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

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

 is therefore evident that tlie classifier who judges from these organs abed 

 alone cannot possibly find out the true phylogenetic connection between the 

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

 for forms wiiich 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 jihylogenetic branches. 



Let us take as illustration the retrogressive development so often observed 

 in Sphinyidae 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 

 developed pulvillus and paronychium, and assume as Ijefore that one of these 

 organs becomes reduced in each derivation from tliat ancestral tyi^e, we arrive 

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

 reduced. Therefore the reduction of the tongue and tlie absence of the mid- 

 tarsal comb, of the pulvillus, and of the paronychium in those species or genera 

 are not an expression of close relationship. Looking at the pedigree of the 

 genera of Sphingicae (facing p. 30), it will be noticed tliat a reduction and loss 



