le pluinules on the ab dom en, who se aver age number can 

 * given as about 20 on eacn side,rarely more than . 

 ),are certainly sometimes undivided,or at least 

 ley share a single double portion between them (e.g. 

 e above named spec imens 1,2, *■»-,& 5,of which 2 & 5 

 ive a single branche has four on one side and 1 on 

 le other),. sometimes they are two-parted almost from 

 le base, or with a single unbranched part "Detween them 

 specimen 7 nas a fourtn part of a two-cleft plumul 

 1 one side and on the other side about a half ,while 

 >ecimen S has single one s on one side and three that 

 •e not single, but all are two-parted witn. equally Ion 

 •anches. On the two-parted ones the nevi branch 

 ways comes fortn at the appointed place,namely close 

 > the base and nearer the under s ide, and gradually 

 lcreases and curves a littla,until it is equal in 

 '»ngth to the original plume,and xftxi« the whole ab- 

 Lnal feather acquires a inuch denser appearance. 

 When then we find on the same fish two Pennellas, 

 LJce specimens 3 & 4,fastened alongside of eacn other 

 id in the same stage of development and age, but still 

 Lth such different plumule relations, it becomes ev- 

 lent that this cannot be used to distinguisU. species. 

 When the branches begln to push forth as little 

 rojections on the dor.sal and ventral sides of the 



