BY R. J. TILLYARD. 409 



which has persisted in CorduJephyii alone, of all present-day 

 Libellnlidcr. [As such, they open to our view a most fasci- 

 nating vista of the development of the Ainsopterid labium 

 along its two main lines, viz., the Libellulid and the Aeschtiid 

 forms.] 



Turning now to the Idocordulina, we notice that the larva 

 of i'ordidephya shews not the slightest resemblance to the 

 larva X, which we have taken as typical of that group. In 

 this larva, the general form (large oval abdomen, very flat ; 

 short legs, small head and thorax), and the labial develop- 

 ment (a widely and irregularly torn outer edge, no warts or 

 dots, and few setae), make it almost impossible to believe 

 that Cordulephya is a member of this stock. 



In its habits, the nymph of Cordulephya is altogether 

 E\icord iduin . A denizen of the secluded corners of streams, 

 living quite uncovered on the sandy bottoms, it is able to 

 move with considerable speed, and possesses a peculiar and 

 fascinating mode of swimming, which may be aptly compared 

 to the first regular strokes taken by a frog after diving, — 

 fore-legs outstretched, and hind-legs taking slow and grace- 

 ful strokes through the water. In these habits, and in its 

 colour-pattern, it is so exactly similar to the larva of Ilemi- 

 cordulid (iKsfralid', that I have held a few of each in my 

 hand, and have been quite unable to distinguish them, except 

 by looking at the labium. 



Altogether, therefore, the evidence of the eai'ly stages is 

 t>ery strongly in favour of the inclusion of i^'ordule phyti in the 

 Eucordulina. 



a. The Form of fhr Iinayo. — Under this heading, I take 

 the study of the imago, <(p<irt from winy-venation . As might 

 be expected, the evidence, in most respects, reinforces that of 

 the early stages. The robust head and thorax, and the 

 remarkably long legs of the nymph are repeated in the imago, 

 and shew unmistakably the Eucordidian connection. In 

 t'ordulephya^ however, we do not find the corduliform abdo- 



