386 F. W. GAMBLE AND FREDERICK KEEJiLE. 



they became green in the sterilised as well as in the non- 

 sterilised or filtered cultures. Thus, whilst the occurrence of 

 these colourless cells was such as to preclude their develop- 

 ment being considered a normal one associated with the 

 presence of pre-existing leucoplasts, their presence in all the 

 cultures, and their metamorphosis into green cells, seemed to 

 afford proof of their internal origiu. We concluded that they 

 were intrinsic structures. 



Feeling, however, that such a conclusion could only be 

 adopted after confirmatory experiments made with every 

 precaution, we prepared another series of cultures, and in 

 doing so lighted upon a serious source of error, which showed 

 us that not only is every experiment made with sterilised 

 water, into which the egg-capsule of Convoluta is introduced, 

 worthless, but that extraordinaiy precautions have to be taken 

 to avoid contamination when introducing larval Convoluta 

 into the experimental vessels. 



The source of error lay in the cells that adhere to the egg- 

 capsule. During the act of egg-discharge and capsule- 

 formation the tissues of the parent undergo violent dis- 

 ruption. Frequently the body breaks in two across the 

 opening of the oviduct ; not infrequently the hinder half 

 remains attached to the capsule, and there disintegrates. 

 But invariably a discharge occurs of cells that adhere too 

 firmly to the capsule to be washed away. A few days after 

 ovulation the surface of the capsule is covered with other 

 elements, which have the appearance of the growing and 

 dividing cells of an alga in the Palmella stage. Thus we 

 find masses of very pale greenish cells (fig. 3) and linear 

 colonies of colourless elements, some with thick walls and 

 refringent contents, others with no apparent wall and clear 

 contents (fig. 3, C — G). Again, we find linear bifurcating 

 green filaments, evidently the vegetative thallus of an alga, and 

 recalling the filamentous form of the ]iolymorphic Proto- 

 coccus (see Chodat, 1902). There is ground for believing 

 that these growing and actively dividing organisms are 

 derived in some way from the body of Convoluta roscoff- 



