76 JLUSOJS BIOLOGICAL MONOGRAPHS [318 



at least in a vestigial condition, in the newly hatched larvae, unless their 

 loss look place sufficiently early in phylogeny to restrict their appearance 

 in ontogeny to embryonic stages. Since these two pairs of larvapods are 

 generally present in the Catocalinae, their absence being rather exceptional, 

 and since Hampson has reported them present in one- genus of the Phyto- 

 metrinae, this latter possibility must be regarded as highly improbable. 



A study of the postembryology of these appendages reveals the fact 

 that these hj'pothetical changes based on recapitulation alone are not found 

 and that the reverse condition usually presents itself. Instead of becoming 

 relatively smaDer during the course of development, the first two pairs of 

 larvapods typically increase in relative size from the first to the last stadia. 

 Moreover, where they are wanting in the fully grown larva, no trace of them 

 is found in the first instar. On the contrary, in certain genera of Agrotinae, 

 Hadeninae, and Acronyctinae, the first pair is absolutely wanting in the 

 first and second stadia, appearing in the third as a minute vestige and 

 increasing in relative size thereafter. This condition is diametrically 

 opposite to the one which would result from an expression of recapitula- 

 tion. From these facts it may be stated conclusively that the appearance or 

 the increase in the relative size of the first one or two pairs of larvapods in 

 the postembryonic development of noctuid larvae are non-recapitulative 

 changes, the effects of recapitulation having been completely obscured by 

 other factors. 



The reduction or absence in lepidopterous larvae of the cephalic one, 

 two or three pairs of larvapods is correlated with the peculiar biological 

 characteristic of walking with a looping gait. Noctuid larvae with this 

 gait have been referred to as semi-loopers as opposed to the loopers of the 

 Geometridae, where this gait is even more pronounced, since but one pair 

 of median larvapods persists in the larvae of this family. Caterpillars of 

 this type, because of the longer steps which they are able to take, can travel 

 more rapidly for the amount of energy expended than those which walk in 

 the usual manner. It is a matter of simple mechanics that the looping gait 

 is the more efficient from the point of view of rapid progress. Loopers 

 appear to be generally more active than other caterpillars, altho certain 

 arctiids whose aptitude for traveling at a high rate of speed is frequently 

 displayed, furnish an exception to this rule. 



The evolution of this habit in the larvae of the Catocalinae, Phyto- 

 metrinae, and Hypeninae is very probably to be accounted for by the 

 advantage of rapid locomotion which is thereby undoubtedly gained. The 

 looping gait enables these caterpillars to withdraw from undesirable situa- 

 tions with the minimum loss of time and with the minimum amount of 

 exposure to the attack of enemies. When dislodged from the trees or the 

 plants upon which they feed, usually in more or less protected situations, or, 

 in the case of most Catocalinae, from the twigs upon which they rest, in 



