270 THE TANGANYIKA PROBLEM. 



of Prosobranchs, which have simply been lumped together 

 through the still prevailing and pernicious habit of using 

 conchological characters to distinguish species, and also 

 because, as Bouvier remarks, a large number of forms, the 

 wide anatomical differences between which are not apparent 

 in their shells, have a fresh- water habit, and have, in con- 

 sequence, been supposed to belong to a single morphological 

 group. 



If we take Melania amarula as a typical Melania, the 

 Ty phobia group in Tanganyika has nothing to do with the 

 Melaniadae at all. All the members of this group are at 

 once and completely dissociated from such a form — (i) by 

 the character of their radula^ ; (2) by the gastric structures 

 which I have described, and which neither M. amarula 

 nor any of the other known members of the so-called 

 Melania group possess ;* and (3) by a number of minor 

 details, including the possession of an introvertible penis 

 in the Tanganyika forms, the peculiar property of the 

 Typhobia group, and of which there is no trace whatever 

 in M. amarula. 



The radula of Typhobia and the radula of Bathanalia 

 are remarkable, and are closely similar to the typical radulse 

 of the genera Xenopkora, Strombus and even Capulas. 

 In the characters of their nervous system the members of 

 the Typhobia group approach both Melania amarula and 

 Cerithium, for they exhibit the shortening of the sub- 

 intestinal cord, and the approximation of the left pleural 

 and sub-intestinal ganglia, which is a characteristic of these, 



* Since the above was written an examination of some of the so-called Melanias from 

 the East Indian Islands has shown that these, at any rate, possess a rudiment of the style 

 sac, an observation which tends to show that divergent and old types have, in different 

 regions, independently taken to the fresh water of the land as members of the secondary 

 fresh-water series to which I have alluded in the second chapter of this work. The 

 structure is also present in Turritella communis. 



