EXPERIMENTS IN REARING. 75 



By the foregoing table it will be seen that the species 

 most likely to be of serv'ice are Anax juniiis and ^^shna 

 constricta and heros, which appear in July and August and 

 may occur as late as the middle of September, a i)eriod 

 comprehending some two months, when their usefulness as 

 destroyers of other insects ceases. The pairing of the 

 sexes may be readily observed during the months above 

 named, occurring just over the surface of some fresh water 

 stream or pond, during rapid flight, which, broken to 

 make connection, continues uninterruptedly except for rest 

 in some adjoining thicket or the deposition of eggs. These 

 eggs sinking, lodge among water plants and debris at the 

 bottom of the stream and hatch within a short time into 

 insect devouring larvre, which upon attaining full growth 

 enter into a semipupal stage, in which they are still active. 

 What the combined duration of the larval and semipupal 

 stages is I am unable to state, but it is certainlv not less 

 than one year, and it is quite possible that it may be 

 longer. (Dij)lax berenice, from specimens taken by me in 

 both June and August, may be an exception, and produce 

 two broods. It is, however, of rather infrequent occur- 

 rence, closely confined to its birthplace, and rather feeble 

 in motion.) The knowledge we already have upon this 

 point is sufficiently definite for our ])urposes. ^ 



II, Experiments in Rearing in Tmitation of 

 Natural Conditions. 



An insufficient period to satisfactorily test the question 

 of breeding in all its phases was allowed by the terms of 

 the circular — for the reason stated above, that dragon flies 

 require, with few exceptions, at least a year in whicii to 

 complete their transformations, and, like their cousins, the 



