THE GEEEN CELLS OF CONVOLUTA EOSCOFFENSIS. 173 



iu tlie layer surrounding tlie pyrenoid (the starch sheath) — 

 and also of minute rods or granules scattered irregularly 

 throughout the substance of the chloroplast. 



Two views as to the nature of the green cells have been 

 expressed — one, that they are algfe which have been ingested, 

 and which have developed within the animal body, the other, 

 that they are integral parts of the animal tissue. 



The observations of Georgevitsch (1899) that the just 

 hatched young are colourless lend support to the former of 

 these views, but do not suffice to establish it. For, since the 

 animals raised by Georgevitsch — hatched in filtered water — 

 only survived for three days, it is conceivable that ante- 

 cedents of green cells were present in a colourless form, but 

 that, owing to the short period during which the animals 

 lived, these colourless antecedents failed to give I'ise to the 

 green cells. 



Our own observations (Section V) that the earliest recog- 

 nisable stages of the future green cells may be colourless 

 justify this reservation, lending no more support to the infec- 

 tion-hypothesis than to the suggestion of Haberlandt that the 

 green cells may be derived from colourless plastids, the per- 

 sisting remnants of once independent alga) now transmitted, 

 as are the plastids of green plants, by the egg-cell. 



In order to determine finally the origin of the green cells 

 of Convoluta, it is necessary, in the first place, to maintain 

 newly hatched animals under such conditions that infection 

 — if iufection there be — cannot occur, and, in the second 

 place, if the result of the foregoing is the production of 

 colourless Convolutas, to expose the animals to infection 

 and to determine whether this exposure brings about the 

 development of green cells. 



As we state in the Introduction to this paper, this rigorous 

 proof of the intrusive nature of the green or yellow cells of 

 animals has not hitherto been obtained in any single case, 

 we therefore describe in detail experiments which fulfil these 

 conditions, and which have led to a decisive result in the 

 instance of Convoluta roscoffensis. 



