NOTES ON VARIATION IN THE TUNICATA. 355 



some cases they belong to Polycarpa, a member of the 

 family Cynthiidae, and they may even be Ascidiidse {e. g,, 

 Ascidia involuta, Heller). It is even possible that such 

 forms might be Compound Ascidians, as Polyclinum sahu- 

 losum and various species of Psammaplidium are incrusted 

 with sand, and, in external appearance, mimic the Molgu- 

 lidae. 



This instance is sufficient to shew how rash it is to 

 attempt to identify a series of Ascidians without a thorough 

 anatomical examination; and it is of the greatest importance 

 that new species should be sufficiently described, not only in 

 their external appearance, but in their internal structure. 

 Most of the older descriptions of Ascidians are, on this ■ 

 account, of little or no value. With a very few exceptions, 

 they consist merely of more or less minute accounts of the 

 external appearance of the animal, and frequently give no 

 clue to even the genus to which the species belonged. Con- 

 sequently, many of the British Ascidians, both Simple and 

 Compound, require to be carefully re-examined and fully 

 described before they can be referred to their proper genera, 

 and before they can be compared with the species described 

 from other European seas by such careful investigators as 

 Traustedt, Lacaze-Duthiers, von Drasche, and others. 



One of the first matters to be determined is, which ana- 

 tomical characters are of most importance in distinguishing 

 allied species ; and, with the view of settling this important 

 point, I have taken advantage of every opportunity afi'orded 

 me during the last six years, of examining the structural 

 details of large numbers of the commoner British species. 

 While conducting this investigation, I have been greatly 

 struck by the large amount of individual variation present 

 within the limits of a species, even in the case of important 

 organs such as the branchial sac and the tentacles. I have 

 already discussed such individual variations in the case of 



