460 MR. A. W. WATERS ON 



examined Levinsen's great work are coming to similar con- 

 clusions, and in this way our starting from Levinsen's standpoint, 

 and using it as a stepping-stone, should ultimately advance the 

 correctness of the classification very much. At present the 

 position remains very difficult, since for a large part we are not 

 sure what will be accepted. 



I do not want to be misunderstood as giving an adverse 

 criticism, and so would add that the task was too great to be 

 final, as Levinsen must have fully felt ; and we must now be 

 alive to the fact that, in the future, classifications will to a 

 large extent be based upon the structure of the soft parts, which 

 furnish a very large number of characters, some of which are of 

 great value, while others will not be found very useful. For 

 every external character there are many internal ones of 

 importance. 



It must be emphasized that, as I have often said *, characters 

 of great value in one group or family are almost useless in the 

 next ; and all attempts at fixing certain characters as being 

 of A 1 importance, others of secondary importance, and so on 

 down the scale lead to no result, but we mixst get together our 

 groups of species based upon as many characters as possible, and 

 gradually build from them larger divisions ; and this process must 

 be slow, but it will be natural, whereas the attempt to work from 

 the larger divisions has led to false results. 



As a case in point, Levinsen makes great use of the rosette- 

 plates, which I have found in certain cases to give most usefval 

 results, in other families none at all; and in tabulating Levinsen's 

 results we find that the family character is often uni- or multi- 

 porous, with or without pore-chambers, that is to say, the 

 character in such cases is of no value for 1;he higher group. 

 In those families in which we should have been most glad of 

 help, namely, in Membraniporidpe, Cribrilinidag, Microporidte, 

 Escharidse, Smittinidse, it is pore-chambers or uni- or multi- 

 poi'ous rosette-plates, and in eight other families uni- or multi- 

 porous rosette-plates ; also in genera we find the same range, so 

 that, while Levinsen's work in this direction is very valuable, 

 care is required lest we attach undue importance to the rosette- 

 plates or any one character. 



Where the opercular aperture is not on a level with the frontal 

 surface of the zocecium there is often a shelf upon which the 

 distal end of the operculum I'ests, and Levinsen has shown that 

 this is a character of some value ; but here, again, we must not 

 expect too much. The value of most of the characters used by 

 Levinsen, even for the main divisions, are still on trial. 



Levinsen has followed f Norman in using names given by the 



* Page 71, Levinsen expresses these facts as follows : — " The same structural 

 feature in different systematic divisions can have a very different systematic 

 importance, so that characters which are constant in one genus or family, in 

 other corresponding divisions are not always constant even within the species." 



t Since I wrote the above, Canu has sharply criticised Norman's suggested 

 alterations of generic names (Rev. Crit. de Paleozool. vol. xvii. p. 49, 1913). 



