BV R. .1. TILLVARt). 621 



dence gathered in the preceding section points also in this direc- 

 tion. Probably also the hairiness is primitive too, since all 

 aquatic groups tend to become smooth in the course of evolution. 

 But the one-jointed condition may not be taken as primitive: 

 because, in the case of other organs, e.g., the antenna?, tarsi, etc., 

 the number of joints increases as the larva grows, and yet it is 

 certain that the last and highest number of joints attained is 

 the most primitive number, or, in some cases, even less than that 

 (<.^., in the tarsi, the fossil Tarsophlehiinm had four joints, 

 whereas no existing Odonate has more than three). Thus we 

 must assume that, in the case of a larva that develops two- 

 jointed gills, such as Neosticta, the two-jointed condition is pro- 

 bably primitive. I5ut, in tlie case of those larvae which always 

 have unjointed gills, it is not necessary to assume that they are 

 derived from two-jointed forms, unless there is evidence for it. 



Jn the case of the gills of Calopte,vygid(P. and Lf.stidce., there is 

 absolutely no evidence that these gills were ever more than one- 

 jointed. But in the Agrionid(e, the existence of two-jointed 

 forms in the FrotoiieurincE and AgriouiucE strongly suggests that 

 the comuion ancestor of these two groups had a larva with two- 

 jointed gills. As no larvie of the most primitive subfamily of 

 the Agrionidce (viz., the Megapodagrionince) are known to pos- 

 sess two-jointed gills, I am inclined to suspect that the original 

 Ayriuiiidce, like the other two families of Zygoptera, possessed 

 only one-jointed gills, and that the two-jointed condition arose 

 in the common ancestor of only those subfamilies still possessing 

 it, and, failing to become of any definite service to the larva, is 

 now again reverting to the one-jointed form, by degradation 

 through the Nodate, Subnodate, and finally Denodate stages. 



The most ancient existing forms of Zygoptera possess gills of 

 the ISaccoid Type; for this type appears to be universal through 

 out the two subfamilies Epallagiiuf and Thorvtue. As such a 

 type could l^e easily produced from an original caudal filament, 

 l)y simple enlargement of its internal ca,vity, we shall probably 

 be right in assuming that the Saccoid Gill is the most primitive 

 type. One might expect- this line of evolution, once begun, to 

 run to an extreme; and this is probably the reason for the exist- 



