626 CAUDAL GILLS OF ZYGOPTEtilD LARV.t, 



might just as well have been a gap, or an overlapping, in the 

 two subfamilies. 



Within the subfamily Agrionhice, we can follow the phylo- 

 genetic sequence of gill-forms from the Nodate, through the 

 Subnodate, to the Denodate, which reaches its final form in the 

 tough, thickened, opaque gills of Austrocnemis and some species 

 of Agrion. From such a type as this, a retrogression due to loss 

 of f«nction, (the gills no longer serving as breathing-organs) will 

 explain the presence of the Reduced (Non-functional) Type in 

 such a highl}^ specialised case as the larva of Agrion asietiff, 

 which dwells in the water collected at the bases of the leaves of 

 certain plants, in a region of high rainfall. 



Text-fig.46 shows diagrammatical ly the stages in the develop- 

 ment of the Agrionid form of Vertical Lamellar Gill from its 

 Saccoid ancestor. 



There remains now onh^ the Horizontal Lamellar Type of gill 

 to deal with, as we see it developed in the genus Argiohstes. It 

 seems to me that the development of the two types of Lamellar 

 Gills, vertical and horizontal, is obviously a case of the accumu- 

 lated result of habit on the form of an organ.* If the original 

 JSaccoid Gills were held out separately in the water, as in the 

 case of JVeosticta, then, in course of time, it is inevitable that 

 the Vertical Lamellar Type must be developed. But if, on the 

 other hand, the ancestral form, possessing a Saccoid Gill-system, 

 remains a rock-dwellei', and adopts the habit of clinging closelv 

 to the rock-surface to escape detection, then the natural result 

 must be, in course of time, that the gills will become horizontally 

 flattened. They will, therefore, pass through a triquetro-quad- 

 rate stage, represented in Text-fig.47, b, c, and this stage must 

 he observable in a ditference between the form of the lateral and 

 median gills. The gills of Argiolestes do show this difference, 

 though slightly; for the lateral gills are completely triquetral in 

 section, while the median gill still preserves the quadrate shape, 



* tSince the organs with whose evolution we are dealing are purely larval 

 structures, not present in the imago, this appears to be a very interesting 

 problem for the consideiation of Lamarckians and Darwinians. 



