BY R. .T. TILLYARD. 213 



(Stj) it divides into two parallel, main trunks supplying the head 

 (Fig.20). The outer of these seems to represent the original 

 dorsal trunk, while the inner probably represents an originally 

 separate tracheal system for the head, which became hitched on 

 to the dorsal trunk when its original air-supply was cut oiF by 

 the more or less complete functional atrophy of Stj. (The short 

 branch to Stj is still to be seen). Now (Fig. 20) whereas all these 

 tracheae are either purplish-red or coppery colour, and hence 

 belong to the dorsal system, there can be easily detected, lying- 

 just inside DT, a very short trachea ( Vr) of silvery-white colour, 

 arising from the inner of the two divisions of DT close to the 

 point where it divides into two, and running back only a very 

 short way to enter again the doisal trunk itself, ^\'hat then 

 can this useless rudiment be, but the original spiracular ending 

 of VT, the rest of which is now broken off and hitched on to p^ ^ 

 Surely its silvery colour admits of no other interpretation, since 

 at the present time it is completely attached to the dark coloured 

 dorsal system I 



Let us now turn to Palaeontology for support. If our theory 

 be correct, the ancestors of our Odonata should show^ a stronger 

 development of the trache?e on the costal side of the wing than 

 they do to-day. Also, no ivater -dwelling larvca should be found 

 in those deposits whence we have obtained such ancestors. Now 

 this is exactly the case with our Protodonata. Fossil Ephemerid 

 lar\ £6 are recorded far back, almost into Palaeozoic times; but no 

 fossil larvae referable to an Odonate type have yet been found 

 with them. Moreover, in the Protodonata, 8c runs either to the 

 wing-tip or nearly as far, while the crossing of one or more 

 branches of M over R is not accomplished, nor are M and R 

 fused in the imago. Further than this, the Protodonata exhibit 

 extraordinary diffei'ences, in many cases, between the \'eiiation 

 of fore and hind wings — differences that seem to defy explana- 

 tion, unless we assume that these insects are misnamed, and do 

 not lie anywhere near the direct line of Odonate ancestry. Might 

 not an application of this new theory help to solve some oi these 

 very difficult problems, on the assumption that the tracheal 

 shiftings took place not only gradually hut also unequally in fore 



