74 Dr. T. A. Chapman's Notes on 



tubular. In both, the long spines appear to arise rather 

 just outside the upper rim than from its margin ; in H. 

 pcnella the rim often very distinctly passes round inside 

 the long spines, and has short spines along this margin. 

 In canalensis the margin rather folds over to the spine, or 

 the inner surface of the spine opens out to either side into 

 the margin, which inclines to fold inwards a little, and has 

 irregularities rarely amounting to short blunt teeth, whilst 

 the margin in iKiuUa is usually armed with a continuous 

 series of long fine needles. 



The cocoons of the three species of Hderogynis present 

 good differential characters (Plate XI). 



In H. imradoxa % the cocoon is much larger than that of 

 penella, and instead of being lemon-yellow is bright reddish. 

 The larva spins first an outer delicate lace-like layer, 

 beginning at the surface of attachment and spinning out- 

 ward on each side a net-like veil, unites these above when 

 they are large enough to meet. Within this is the true 

 inner cocoon, which is similar to the outer one, but hardly 

 as dense and strong; it is not far within the other, the 

 space between being occupied by a comparatively slight 

 web of connecting silk. The outer cocoon is more net-like 

 than a mere fortuitous disposition of the silk would produce, 

 there being numerous net-like holes, the margins of which 

 consist of numerous strands of silk, giving the impression 

 that the silk of paradoxa is coarser than that of the other 

 species ; it does not, however, appear really to be so. 



The % cocoon of H. imradoxa consists then of a definite 

 separate net-like outer layer and an inner layer less dense 

 and slung within the outer one by somewhat abundant 

 threads, the inner and outer layer being frequently so far 

 separated from each other, that the thickness of the wall 

 of the cocoon may be from 3 to 5 mm. The ^ ixtTadoxa 

 cocoon appears to have the same structure, but it proves 

 practically impossible to separate it into two layers. 



The ^ H. pcnella cocoon is much the same as that of 

 H. paradoxa ^as regards divisibility into layers, but agrees 

 with the ^ If. j^enel la in the silk being fairly uniformly 

 distributed, there being little or no aggregation into strands 

 to form a network. The $ H. p)en€lla cocoon is compara- 

 tively small, looks fluffy, with outside silken threads, 

 instead of smooth as in H.paradoxa, so that the method of 

 spinning is probably different. I have not seen H. pcnella 

 spinning its cocoon. It places it, however, like paradoxa 



