876 H. H. NEWMAN AND J. THOMAS PATTERSON 
extent the definitive number of scutes is determined at the time 
of fertilization; for presumably, in so far as the four individuals 
of a set are alike, this similarity was determined before they 
became separate individuals, and, in so far as they differ, their 
differences are due to epigenetic factors operating after the sepa- 
ration. Should they prove to be exactly alike in the number and 
arrangement of scutes we would be warranted in claiming that 
hereditary control was absolute; if we do not find them exactly 
alike we may claim that hereditary control is not absolute but only 
partial. As an index of the strength of hereditary control we 
have decided to employ Galton’s coefficient of correlation, a con- 
stant indicating exactly what per cent of complete correlation 
or of absolute hereditary control exists. 
A short method for determining the coefficient of correlation, 
and one which seems especially devised for cases like the one in 
hand, was described by Harris (’09). This method appears to 
have been elaborated by Professor Pearson for use in cases of 
symmetrical correlation tables in which both variates have the 
same mean and standard deviation. The correlation table is 
made symmetrical by using each individual of each set first as 
a ‘subject’ and then as a ‘relative.’ In the case of our twenty 
sets of foetuses such a table requires over one hundred vertical and 
horizontal columns and would not be suitable for reproduction 
here. As a matter of fact it is scarcely necessary to make a cor- 
relation table in order to derive the desired constant. One need 
only determine the positive differences between the subject and 
relative classes and their frequency. In the present case we find 
the positive differences and arrange them as follows: 
Difkerencess. 0) “ly 2 3 45— 5O G6) i 8) 0) 10) a Aa ao 
Frequency 226 39° 10 16-13. 10° 9: 40. 7 34° A ad (OL a Sie 
Multiplying the square of each difference by its frequency and 
adding the products, we obtain the sum of the squares of the 
differences [Sv?]. Now since the negative differences are the 
same as the positive we may-double the above sum and divide 
by the number of cases, 240, and thus obtain the standard devia- 
tion of the difference squared (cv?), which is the only constant 
