342 
Merism and Sex in " Spina.r Niger" 
when the sexes are considered separately these differences still obtain (of. Appendix, 
Tables 38 to 42). Thus, in the case of total segments, the uncorrected value for 
(/s is -395, for $ s -410, and for siblings -429, values which are all very close together. 
In one case however, that of the whole vertebrae, the value for the ^/s, i.e. '541, 
is considerably higher than that for the $s, i.e. -371. An examination of Table 40 
TABLE 8*. 
Fraternal Correlations. 
Eaw value 
Corrected (a) 
Corrected (b) 
Anterior spine 
•309 + -014 
•320 
•331 
Posterior spine 
•373 ± •013 
•394 
•414 
Whole vertebrae . . . 
•425 + ^012 
■436 
•447 
Half vertebrae 
•228 ±-014 
•241 
•254 
Total segments 
•429 ±-012 
•425 
•422 
Mean 
•353 
•363 
•374 
(a) Corrected for selection of mothers out of adult $ population. 
(b) Corrected for selection of fathers also, on the assumption that they although unknown 
were equally selected with mothers. 
shows that this is largely due to the exceptionally close correlation between the 
six ^ offspring of $ 159 (Table 13, Nos. 261, 263—5, 267, 268), in all of which 
the number of whole vertebrae is unusually low. In the $ s of the same family 
the number of whole vertebrae approximates more closely to the normal. There 
can be no doubt but that this small group of (/s is largely responsible for the 
relatively high value of the fraternal correlation for whole vertebrae. 
(b) The average value of the five correlated characters works out after 
correction to •375. This value is very much lower than the "5 that the Law of 
Ancestral Heredity would have led us to expect-j-, and is markedly lower than the 
•475 for eye-colour in man — the lowest value hitherto recorded (cf Pearson, '03, 
p. 390). In this connection, however, there is a contingency which must not be 
overlooked, viz. the possibility that a ripe $ may be fertilized by more than 
one cT- If this be so it would obviously lower the value of the fraternal correlation. 
We are entirely ignorant of the breeding habits of Spinax, but the following 
circumstantial evidence, slender though it is, seems to tell against the above 
conjecture. In the virgin % the openings of the oviducts are occluded by the 
so-called hymen which must be ruptured before fertilization can be achieved. 
* Sheppard's correction has been made use of in the three following Tables 8 — 10. 
t [The Law of Ancestral Heredity in no way fixes the value of the fraternal correlation, but only the 
ratios of parental correlation to that of higher ancestors. A tniternal value '5 to -6 is found bij 
observation on lo7ig series iu man, horse, dog, etc. Galton's data gave -375 for stature in brother and 
sister. K.P.] 
