602 yRANSACTIONS OF SECTION D. 



2. That these peculiarities were constituted by the presence in the lake of the 

 remains of an extinct freshwater fauna. 



3. That they were due to the presence of a halolimnic remnant. 



The author inclined to the latter of these suggestions, and pointed out that he 

 attached no great weight to his own former suggestion — that the shells of the 

 Tanganyika gasteropods were possibly similar to those common to the Jurassic 

 seas. The similarity was indeed very striking, but any determination based upon 

 the structure of shells appeared to be extremely doubtful. 



(b) By Professor Paul Pelseneer. 



1. In his attempt to demonstrate the archaic character of these halolimnic 

 molluscs Moore has made use of two quite different methods, seeking, firstly, to 

 show the resemblances between their shells and those of secondary fossils, and, 

 secondly, some resemblances in structure between them and other gasteropods with 

 different shells. The employment of this second method clearly vitiates the first; 

 and justly so, because the external resemblances of shells are often illusory, and 

 the results to which they conduce are therefore quite uncertain. The second 

 method, therefore — the study and comparison of internal organisation — is the 

 only one which can throw light on the question at issue. 



2. But even from the latter standpoint I find myself unable to agree with 

 Moore and Digby. In the anatomical studies which they have published on the 

 MoUusca of Lake Tanganyika allusion was made to the affinities of these 

 molluscs with diverse marine forms, Chytra having affinities to Hipponyx and 

 Capulus, Spekia to Lamellaria, and Edyaria ( = Nassopsis) to the Arckitcenio- 

 glossa, &c. But there are really no affinities between these forms in the usual 

 sense of the word; and if one examines the resemblances, they are such as are 

 common to all the Taenioglossa, to which group the haloliminic genera belong. 

 There is no strict and special resemblance between these and marine forms ; and 

 if there be some characters which may be called 'archaic,' other freshwater 

 genera, not * halolimnic,' such as Ampullaria and Pa/Mc?tna, present much more 

 archaic characters. I was led to these conclusions in 1886 by a first study of the 

 molluscs of Lake Tanganyika then collected by Captain (now General) Storms. 



3. These results are confirmed by the investigations which I have begun upon 

 the new material collected by Dr. Cunnington, and placed in my hands by the 

 Director of the Natural History Museum. I conclude, therefore, that all the 

 ' halolimnic ' forms belong to the family Melaniidae, or to very closely related 

 tjpes. This family presents, in its different parts, a well-known polymorphism of 

 the shell ; but along with this there is a great uniformity of organisation, which 

 is also found in the halolimnic genera of Lake Tanganyika. The resemblances of 

 these halolimnic forms to the Melaniidae are clearly proved by reference to the 

 anatomical characters which are generally used in systematic work — the radula, 

 otocysts, &c. — and by special details of their biology, their freshwater habitat 

 and viviparity — all features in regard to which the two groups present analogies. 

 Digby recognised the melanoid character of the radula of Limnotroc/ms, but 

 among the Tanganyilca forms this is certainly the one in which the shell presents 

 the most perfectly marine appearance. 



4. The study of two genera, Giraudia and Laviyeria, the organisation of 

 which was unknown up to the present, also supports the above-mentioned results. 

 Both have multiple otoliths within the otocysts, one otolith being much larger 

 than the others in two species of Giraudia. Laviyeria, the only genus of which 

 a female has been examined, is viviparous, and possesses a radula which resembles 

 most closely that of the Melaniid genus Chiara. The radula of Giraudia is 

 clearly similar to that of the Melaniid genus Ancylotus. The two species 

 G. 2^rceclara [Bourguignot] and G. horei [Smith] have radulae so different in 

 structure as to perhaps necessitate their allocation to two distinct genera. 



