FLOI'XDERS OF GENUS PARALICHTHYS AND RELATED GENERA 



271 



sppries. tliis diaractcr is constant at all apes, al- 

 ilioiiirli in veiy large fisli the spinules in the species 

 liavinir tliem sometimes are com])aratively less 

 niaikeil. In ver}- large specimens they sometimes 

 change to coarsely granular asperities, but the 

 <listingnishing nature of the scales is still evident. 

 The two exceptional species in this respect, and 

 I lie change of the scales in very large specimens of 

 otiier sj)ecies perliajis explains the inadequate 

 treatment this character has received in the study 

 of the species of Paralichfhys. However, the 

 structure of the scales is of as nnich importance in 

 interpreting the relationship of the species and in 

 the practice of identification, as it is in related 

 genera. Besides the presence or absence of 

 spinules on the scales, another important cliaracter 

 which may be used in generic division is the pres- 

 ence or absence of accessory scales (see p. •284). 



Besides the structure of the scales, their size, 

 which is usually expressed inversely as the num- 

 ber along certain lines of the body, is a valualile aid 

 in distinguishing the species when used in con- 

 nection with the other characters, although it 

 usually shows much variability and considerable 

 intergradation. One serious drawback to a pre- 

 cise use of this chai-acter is the difficulty of de- 

 termining the number of scales with any reason- 

 able degree of accuracy. The ttibes in the lateral 

 line are easiest to count in young fish, but the more 

 or less clear-cut boundaries between the individual 

 tubes disappear to a large extent with growth. 

 Also, with increase in size the normal scales on 

 either side gradually overlap more and more those 

 in the lateral line, while the increasing numbers 

 of accessory scales cover the surface of all the 

 large scales more and more. Consequently, in 

 large or medium-sized fish, it is almost imj)ossible 

 to count the individual scales in the lateral line 

 with any reasonable degree of accuracy. 



After testing diflPerent methods of expressing the 

 scale count, the following procedure was adojited 

 as yielding fairly accurate results with the least 

 amoiuit of labor. The count is made of the num- 

 ber of oblique rows over the straight part of the 

 lateral line, beginning with the row standing di- 

 rectly over that canal in the lateral line which is 

 entirely, or almost entirely, horizontal and end- 

 ing with the i-ow the lowest scale of which is at the 

 end of the hypural as determined by flexing the 



caudal fin. In counting the scales the specimen is 

 held with the back tilted down and away from 

 the observer. When held in this position the re- 

 flection of light is such that the rows of scales 

 ap]iear fairly prominent, and the rows are counted 

 rather than the individual scales. Sometimes the 

 fish has to be turned somewhat at different angles 

 until the rows become prominently visible so that 

 they may be coimted with any fair degree of ac- 

 cuiacy. X check on a number of small specimens 

 shows that the number of scales in the lateral line 

 closely approximates the number of oblique rows 

 placed over it. 



The number of rows along the curved part of 

 the lateral line cannot be determined with as much 

 accuracy as along the straight part, because the 

 rows in the anterior part of the body are more 

 irregular, and because of the greater difficulty of 

 fixing the point to begin the count. Had these 

 rows been included in the count, the small increase 

 in the degi-ee of specific divergence would have 

 been made at the sacrifice of greater accuracy. 

 They were, therefore, omitted and the number of 

 scales stated in the diagnoses in this paper and in 

 table 1 uniformly refers to the numl)er of oblique 

 rows over the straight part of the lateral line 



In current descriptions, the number of scales is 

 usually stilted as so many or ''about'' so many in 

 the lateral line. It seems desirable to have some 

 conversion factor by which current descriptions 

 may be correlated with the present paper, al- 

 though it seems highly probable that counts hither- 

 to recorded by different investigators are not com- 

 parable by a wide margin, because of the use of dif- 

 ferent methods. The number of scales in the 

 curve was determined on a muuber of small speci- 

 mens in which they may be counted with a fair de- 

 gree of accuracy. It was found that, in general, 

 that number closely appioxinuites one-half of the 

 number in the straight part. Therefore, by add- 

 ing one-half to the number given in this paper, 

 counts of scales are obtained which are approxi- 

 mately comparable with those gi\en in current de- 

 scriptions. In the short accounts of established 

 species of which no si)eciniens were examined the 

 number of scales stated is that obtained by using 

 the above conversion factor and subtracting the 

 estimated numbers in the arch from the number in 

 the entire lateral line. 



