2)otassiiU]i In/dwxide hi/ Dicrauura vinula. 289 



a mere trace of a chloride — hardly enough to be sure of ; 

 (3) potassium hydroxide was present in considerable 

 quantity. Tlie presence of potassium was proved by (a) 

 the platinum perchloride test ; from the yellow crystalline 

 precipitate so obtained after treatment with alcohol, the 

 platinum was separated by heat, and potassium chloride 

 recovered ; {h) the sodium and hydrogen tartrate test ; 

 (c) the characteristic colour imparted to the flame of a 

 Bunsen's burner ; (rf) the remainder of the liquid was 

 evaporated to dryness, and yielded a whitish solid, which 

 was strongly alkaline, and deliquesced on exposure to 

 the air, eventually forming potassium carbonate. Having 

 arrived at these results, I confirmed them as far as 

 possible with the few drops of pure liquid in the glass tube, 

 from which also I obtained the same results, including 

 the presence of a very small quantity of a chloride — 

 again almost too little to be quite sure about. 



My next step was to try the relative caustic powers of 

 potassium hydroxide and sodium hydroxide on the 

 cocoons — applying the solutions to the inside. I took 

 solutions of both substances of equal strength, and 

 placed a few drops in four cocoons— two with each 

 solution : the potassium hydroxide reduced the cocoons 

 to which it was applied to a soft pulp in less than three 

 minutes, whereas the sodium hydroxide took more than 

 fifteen minutes to produce the same result. This of 

 course was to be expected, owing to the superior caustic 

 power of potassium hydroxide on most organic substances. 

 It was thus proved that the imago of D. vinula produces 

 caustic potash in order to soften its cocoon, and that 

 this substance is better suited to this end than the 

 other of the two commoner caustic substances. 



II. TJie apparatus employed in perforating the cocoon. 



The imago invariably emerges from the cocoon wearing 

 as a shield a portion of the pupa-case (see fig. 1, j)- 291). 

 This "shield" consists of the median dorsal piece (figs. 

 1 — 3, s) of the head of the pupa, extending as far as the 

 labrum (p. 1.) towards the ventral surface, and also of 

 the two pupal eyes (p. e.), which project laterally further 

 towards the ventral surface than the median labrum. 



At first sight it appears that the "shield" is the tool 

 used to thrust against the softened cocoon, but closer 



