1909. } LIFE-HIStORY OF THE AGRIONID DRAGONFLY. 279 
experience it was, with only one or two exceptions, after the 
antenna had become six-segmented that the nymphs showed such 
variation in their rate of progress even when under similar 
conditions. 
RESPIRATION. 
The caudal lamellae have been described as respiratory organs, 
and have been called “ caudal gills,” but Dewitz (1890) pointed 
out that the nymph can live without them, and suggested that 
possibly a rectal respiration, similar to that of the Calopterigide 
and the Anisopterids, existed in the Agrionide also. In those 
forms, however, where rectal respiration is dominant, there is an 
elaborate arrangement of trachee in the walls of the rectum 
which is completely absent in the Agrionide. 
Dewitz has shown that a stream of water does pass in and out 
of the rectum in the Agrionide, but from my observation it is a 
very weak one, nor is there any special apparatus surrounding 
the anus in this group to prevent ingress of foreign particles 
such as is found in the Anisopterids. If rectal respiration exists 
at all, it seems to me that it must be very slight and of but little 
importance, as I could not observe any increased number of 
contractions of the rectum in specimens of Agrion which had 
been deprived of their lamellee. 
In the absence of the lamelle, I thizk the whole of the 
respiration must be carried on through the skin, 
CONCLUSION. 
The occurrence of a pronymph stage in the Odonata * is not 
unique. It has been observed by Pagenstecher (1864, p. 7) in 
Mantis religiosa, and occurs also in other Orthoptera [see Kiinckel 
(1890 (1) p. xxxvii, and 1890 (2) p. 657)], and in several other 
orders (Packard, 1898, pp. 583-585), and is no doubt more 
common than has at present been recorded. 
The stage is apparently for the purpose of freeing a tightly- 
packed larva from the egg, the amnion, instead of breaking and 
remaining within the shell, continuing intact until after the 
larva is clear, and the amnion forms the pronymph skin. 
In the Odonata the larva casts the pronymph skin in exactly the 
same way as it casts its skin in subsequent moults, and it appears 
that, in the other forms in which this stage has been observed, 
the amnion is got rid of in much the same way, except that, as a 
rule, a special ampulla exists in the thorax of the larva which, by 
swelling up, bursts the amnion in that region (Packard, /. ¢. 
p. 984). 
Although the total number of moults may vary somewhat, 
there is an unmistakable connection between certain stages in 
the development of the nymph and certain moults. What I have 
said as to the growth of the antenne, the appearance of the labial 
hairs, and the development of the wings shows this, especially in 
* Exactly the same phenomenon occurs in Sympetrum striolatum, one of the 
Anisopterid dragonflies, where the pronymph stage continues for about 7 minutes. 
