BY K. .7. TILLYARD. 607 



instai', there was necessarily a considerable delay in i-aising the 

 larvie required from the egg, and in taking them successfully 

 through all the stages to the seventh. It was for this reason 

 that Parts iii. and iv. were not published with Parts i. and ii. in 

 a single complete paper. 



Even with the successful raising of several batches of larvae 

 to the .seventh instar, it has turned out tliat not sufficient evi- 

 dence has been obtained to throw much light upon one very 

 interesting problem, viz., the origin of the constricted or two- 

 jointed gill. The reason for this appears to be, that the only 

 larvje that I was able to raise were not of the Constricted Type. 

 I had hoped that the gills of Aiistroagrion cyane Selys, which 

 are classed as a Nodate Type, would offer some evidence, in tlieir 

 ontogeny, as to how the node developed. But, up to the seventh 

 instar, this larva shows no sign of any division of the gill at all; 

 so that one must evidently examine several more instars, before 

 the beginning of the node would be made apparent. This mean- 

 ing further delay, and the main objects of the paper having been 

 in every other respect attained, I have judged it best to conclude 

 my observations at this stage; being content to indicate, with 

 regard to this one unsolved point, exactly how the problem 

 stands, so that any student of Odonata, who is fortunate enough 

 to obtain material, can pursue it to the end. 



I'he results of Part iii. were obtained from the stud\^ of three 

 species, viz., Aiistroayrion cyane 8elys, Ischnura heterosticfa 

 Burm., and Neost'icla catiesceiis Tillyard. The larvae of the first 

 two were bred from the egg, and every instar up to and includ- 

 ing the seventh (counting the pronymph as the first) was care- 

 fully studied by means of sections.* The Neosticta larva', 

 belonging probably to tlie fourth and fifth instars, were only a 

 fortunate find; not having been raised from the a^g, their instars 

 caimot be stated with absolute certainty. 



I need add little to the account of the methods employed, as 



* As the series of sections obtained fiom AuMroayrion were better than 

 those oV)tained from Jschnnra, 1 have tij^uied only the former. Ausfro- 

 u'/riou lias nodate gills, IscJinnra subnodate; hut, up to tlie seventli instai-, 

 the two forms show no inipoitant differences. 



