EXPERIMENTS ON THE THREAD CAPSULES. 151 



stand on. He relates that he once saw a small fish in the 

 convulsions of agony, with one of these filaments in its mouth ; 

 it shortly exphccl. It is a matter of surprise and regret that 

 Mr Gosse, having once made such an observation, did not 

 feel the imperative necessity of repeating and varying it, 

 so as to be sure that the death was not a mere coincidence. 

 If the filament had the power which this single observation 

 fairly seemed to suggest, nothing could be easier than to 

 establish the fact by experiment. But, I repeat, no one has 

 seen the necessity for the verification of an hypothesis so 

 plaasible ; and Mr Gosse, like all his predecessors, was con- 

 tent with recording his observation, as if it carried the point.* 

 Not being so content, I tested it thus : After irritating a 

 Dianthus till it sent out a great many filaments, I dropped 

 a very tiny Annelid among them, and entangled it completely 

 in then- meshes. Yet lo ! these filaments, which are said to 

 possess so powerful a faculty of urtication that even verte- 

 brate animals are killed by them, had no other efi'ect upon 

 a soft Annelid than that of detaining it in their meshes, from 

 which it shortly freed itself and wriggled away mihurt. Nor 



* How little reliance is to be placed on such an observation may be gathered 

 from the following : One evening, at the Scilly Isles, I was startled from my 

 reading by a commotion in the pan on the work-table, which contained a 

 couple of fish, a Doris, and a few Polypes, but no Sagartia. On investigation, 

 it turned out that the commotion was produced by a fish {Ophidium) six or 

 seven inches long, in the convulsions of death. In. its mouth there was a sliijU 

 strip of Viva. As the weed dangled from its mouth, I was forcibly reminded 

 of Mr Gesso's dying fish, with its filament of the Actinia dangUng from its 

 mouth. Should I have been justified in attributing to the sea-weed the power 

 Mr Gosse attributed to the filament ? Clearly not. In both cases the relation 

 observed was one oi coincidence, not of causality. 



