CLARK: BRITTLE-STARS. 297 



reference is made in the following key, to the shape of the upper arm- 

 plates, it should be understood that the plates on the basal third of 

 the arm of adult specimens is meant. With young or small specimens 

 care is necessary to avoid referring it to the wrong section of the genus, 

 but comparison with other specimens will usually prevent mistakes. 

 Errors are also liable to occur with reference to the oral papillae. 

 Species normally mth two on each side of a jaw, sometimes show only 

 one or may even show three. But almost iuA^ariably these variations 

 occur only on one or two of the jaws, so that the actual species -charac- 

 ter is usually easily determined. In 0. savignyi, however, there is 

 extraordinary diversity, individuals with no papillae occurring in the 

 same locality with those having 1, 2, or 3. But if 3 are present, the 

 most proximal is very small and of little importance. Specimens 

 with no oral papillae are often of fully adult size and the papillae 

 seem to have been lost with age. Species with only one oral papilla 

 on a side are seldom variable but occasionally two papillae occur on 

 one or two of the jaws. The exact form of the oral shields and of the 

 adoral plates is apt to be quite variable but the general form often 

 furnishes a good specific character. 



The number of arms is unreliable as many of the species, if not all, 

 reproduce when young, by division. As a result individuals with 

 from 2 to 7 arms may be found, but 5 seems to be the usual number 

 in a typical adult. Of course the number of jaws, oral shields, and 

 pairs of radial shields corresponds to that of the arms. The length of 

 the arms in proportion to the diameter of the disk may furnish a 

 good specific character, but one must be careful that the disk measured 

 is normal and has not recently divided. The number of arm-spines is 

 commonly 3 on the distal half of the arm but is occasionally 4 or rarely 

 5. Basally the number usually rises to 4, often to 5, not rarely to 6 

 and occasionally to 7 or even 8. Only in a very few instances do the 

 arm-spines furnish a good specific character. As already stated the 

 upper arm-plates furnish such characters but in those species where 

 they are very broadly ellipsoidal and fully in contact, the character 

 of their distal margins is quite variable. The imder arm-plates may 

 also furnish good characters but as a rule they vary a good deal with 

 age, and too much reliance must not be placed on them. 



The character of the scaUng of the disk may furnish a good specific 

 mark but it is always coarser in young than adult. In some species 

 the underside of the disk lacks scales (in the adult) and even the back 

 may be quite soft and leathery in certain cases. As a rule spinules 

 are present, at least along the margin of the disk, and they may be 



