Hderogynis canalends, n. sp. 75 



does, by preference, on a stem or stalk and not between 

 two or more surfaces. Its structure looks uniform and 

 the thickness of the wall of the cocoon is trifling, on 

 section, however, it is found to consist of an inner and 

 outer layer, closely fitting together, but with less connecting 

 silk than the two separated bat less definite layers of H. 

 paradoxa's cocoon are united by. 



The cocoon of the male of H. canalcnsis is not unlike 

 that of H. penella $, that of the $ is very different. In 

 the first place, it is never laid along a twig or stalk as is 

 the typical position of the $ cocoon of the other species 

 and of all the males. It is sometimes found amongst the 

 thorns of the food plant, but perhaps more frequently 

 under stones adjacent to the feeding place. It is necessary 

 to it to have support on several sides. It is very large, 

 25 to 35 mm. long, 22 mm, across, and more or less in 

 the third dimension according to its situation, against 

 25x13x13 mm. for a large JT. paradoxa, or 16x8x8 

 for a well-sized IT. penella. This outer cocoon often has 

 some external spinning to fix it to its place of attachment, 

 under (or between) stones, one side (or two) is attached 

 to the stone. It is thin and transparent but very closely 

 woven, and probably proof against a good many enemies. 

 Centrally in this outer cocoon is an inner one that looks 

 much like the ordinary cocoon of If. penella, but is fre- 

 quently a good deal larger (17 or 18 mm. long, 16 being 

 large for penella). 



What are the homologies of this cocoon ? When I 

 first met with it, having only memory and no specimen 

 by which to compare it, I thought it must be a variation 

 of the cocoon of paradoxa. This, however, it certainly is 

 not. The manner of spinning is that of penella and not 

 oi paradoxa. The colour does not at all agree with para- 

 doxa, indeed it is paler than that of penella. It is rather 

 a faintly flesh-tinted white than anything of the yellow of 

 penella. 



When one examines the inner cocoon, it is found to 

 consist of two layers like the cocoon of the other species, 

 but they are a little less easily separated. 



The outer cocoon, then, is an entirely new structure 

 not represented in the cocoon of either H. penella or of 

 H. paradoxa. 



This conclusion is fully confirmed when the function 

 it fulfils is considered. It is a continuous envelope with- 



