1898-1902. No. 33.] UPPER DEVONIAN FISH REMAINS. 41 



Other fragments have a coarser sculpture and show evidences of nodose 

 ribs; they therefore belong most probably to another species. 



Observations. The length between the proximal portion of 

 the arm and the anterior ventro lateral plate, which can be determined 

 in specimen No.. 1, shows that not only are we dealing with a Bothriolepis, 

 but is also of essential importance in determining the species, a task 

 that is naturally difficult with such incomplete remains. 



We involuntarily think at first of Bothriolepis canadensis WHIT., 

 which is so common at Scaumenac Bay in East Canada, in strata of 

 something like the same age. 



Closer comparison, however shows me that the form from Ellesmere 

 Land cannot be identified with the former. The latter differs from 

 typical specimens of B. Canadensis by its sculpture, which consists of 

 a network of fine ribs without nodes, and by the short form of the 

 anterior ventro and dorso lateral plates. 



In a small collection made by Dr. 0. HOLTEDAHL, Lecturer at the 

 University of Kristiania, during one of the excursions of the International 

 Geological Congress in Canada 1913, to Scaumenac Bay, and which he 

 subsequently presented to the Palaeontological Museum of Kristiania Uni- 

 versity, we find, however, small specimens of B. Canadensis, shorter and 

 broader in form than the typical examples. The sculpture approaches 

 very closely to that which is characteristic of the remains described here. 



At the same time these small and presumably young examples 

 agree in other respects with the typical form, and can scarcely be 

 separated as a special species. The said characters do not seem, there- 

 fore, to be as constant as assumed earlier, and thus cannot be employed 

 with certainty in distinguishing it from other forms. 



There is, however, another character that can be employed in the 

 determination of these fragmentary remains, viz. the relation between 

 the length of the proximal part of the arm, and that of the anterior 

 ventro lateral plate. This relation must be fairly constant and may 

 therefore be assumed to have great systematic importance. Unfortunately 

 it is not possible to determine the whole length of ihe anterior ventro 

 lateral plate in the specimen in question, (No. 1). It seems, however, 

 to be sufficient that we can determine the length from the posterior 

 point of the plate to the cross rib that runs inwards from the brachial 

 joint on the inside of the plate. 



The following table shows this relation in the forms before us from 

 Galgeodden, B. canadensis, WHIT, and a third form A. hydrophila, AG. 

 from Scotland. 



