344 
Merism and Sex in " Spinax Niger" 
Parental Correlations. 
The parental correlations, like the fraternal, exhibit great diversity in value 
for the live characters chosen, varying after correction from "215 for the half 
vertebrae up to •411 for total segments. This last value, which is the highest 
of the series, is low in comparison with the value of parental heredity in other 
forms in which Pearson and Lee find that "its values lie between '42 and '52 
and cluster round "48 " ('03, p. 379). The lowness of the average value for the 
five characters is yet more marked, being when corrected only '307 (of Table 10), 
an exceedingly low value for parental heredity. Allusion has already been made 
(p. 333) to the possibility of some changes in the meristic series occurring during 
growth from the natal to the adult condition. Even if we suppose that such 
changes occur it is probable that they are fairly uniform, and as all the embryos 
examined were at approximately the same stage it is unlikely that the low values 
of the parental correlations are in any measure due to growth factors. 
TABLE 10. 
Parental Correlations. 
Baw value 
Corrected value (a) 
Anterior spine 
Posterior spine 
•359 + -039 
•378 
•196 + -043 
•264 
Whole vertebrae . . . 
•168 + -044 
•215 
Half vertebi-ae 
•238 + -042 
•270 
Total segments ... 
•416 ±-037 
•411 
Mean 
•275 
■307 
(a) The values given in the second column are corrected for selection of mothers out 
of adult ? population. 
The low values of both the fraternal and parental correlations in Spinax 
are of great interest in connection with a recent contribution of Pearson, " On 
a Generalized Theory of Alternative Inheritance " ('04). Basing his calculations 
on the conception of gametic purity Pearson finds that the value of the parental 
correlation should be about '3, that of the fraternal between '3 and "4. These 
theoretical values are considerably below the values hitherto deduced from actual 
observations ('03, pp. 379 and 387), and for this reason Pearson is inclined to 
pronounce against a theory of inheritance involving the conception of gametic 
purity. " The present investigation," he writes, " shows that in the theory of the 
pure gamete there is nothing in essential opposition to the broad features of 
linear regression, skew distribution, the geometric law of ancestral correlation, 
etc., of the biometric description of inheritance in populations. But it does show 
