194 
use and reduced the usable sample size. As the 
reliability of the analysis was in part determined 
by the sizes of samples, unusable specimens 
were both a statistical and an economic loss. 
Whether asymmetrical development, as shown 
in the branchiostegals of Pacific salmon, extends 
in some degree to other paired structures is un- 
known. With knowledge of the presence and 
extent of asymmetries in the characters consid- 
ered, the opposite member of a lost character in 
fish specimens many times could be reliably 
substituted; if it were known that such charac- 
ters essentially developed symmetrically, substi- 
tution would not contribute to the observed 
variability. 
This paper presents an analysis of differ- 
ences between the numbers of elements in left 
and right side structures of four paired meristic 
characters of sockeye ( Oncorhynchus nerka ), 
chum ( 0 . keta ) and pink ( 0 . gorbuscha ) sal- 
mon. In conjunction with this study, the re- 
liability of the criteria utilized in obtaining the 
morphological characters will be discussed. The 
objectives of the analysis were to determine: 
( 1 ) if bilateral asymmetry occurs in these char- 
acters; (2) the extent of asymmetries; (3) the 
direction of the asymmetrical development, as 
shown by one side tending to have a larger 
number of elements; and (4) the usefulness of 
the information in taxonomic and racial studies 
of the three species. 
SAMPLES 
Paired meristic characters were randomly ob- 
tained from specimens collected during the 
years 1956-1959 in the North Pacific Ocean, 
the Bering Sea, and from several North Amer- 
ican and Kamchatkan rivers for racial studies 
conducted at the Seattle Biological Laboratory, 
Bureau of Commercial Fisheries. The characters 
examined included: (1) lateral line scales, (2) 
total gill rakers, (3) branchiostegal rays, and 
(4) pectoral fin rays. Lateral line scales of pink 
salmon were not examined because of extreme 
difficulty in obtaining reliable counts of the 
characteristically very small scales. 
PACIFIC SCIENCE, Vol. XX, April 1966 
COLLECTION OF CHARACTERS 
The paired bony structures were dissected 
from the specimens and left and right sides 
were radiographed separately. When the radio- 
graphs did not provide adequate resolution, the 
structure was stained and cleared, and the ele- 
ments were counted manually. The data were 
obtained using criteria developed by Fukuhara 
et al. (1962: 27-28) for enumeration of char- 
acters. The definitions of criteria, elaborated 
upon below, include discussion of some struc- 
tural characteristics and aberrancies which must 
be taken into account during enumeration of 
the characters. Application of these criteria 
minimized differences between individual inter- 
pretations and the time required for enumera- 
tion of the characters. As measured by periodic 
tests of between-reader variability, disagree- 
ments on independently obtained counts from 
radiographs averaged less than 3 % for the bony 
characters. All scale counts were obtained by 
one biologist. Within-counter variability aver- 
aged less than 2%. 
Lateral Line Scales 
This character consisted of scales or scale 
pockets distinguishable by a tube which pene- 
trates the scale and extends into the underlying 
lateral line canal. The count included the first 
scale posterior to the pectoral girdle and suc- 
ceeding scales terminating at a position on the 
lateral line corresponding to the posterior mar- 
gin of the hypural plate. This margin was de- 
termined by an incision made in the crease that 
resulted from upward flexion of the caudal fin. 
A severed scale or pocket was counted if more 
than half lay anterior to the incision. In im- 
mature specimens, acute scale imbrication re- 
quired scale removal for accurate enumeration. 
Occasional irregularity of typically diagonal 
scale rows was caused by a supernumerary scale 
lying on the lateral line between two scales in 
the proper rows (Fig. 1). Some of these small 
scales lacked a tube and therefore were not 
counted. 
Total Gill Rakers 
This character consisted of rakers on the dor- 
sal and ventral arms of the first branchial arch, 
including the most anterior rakers, which were 
