Ethel M. Elderton 
71 
probable error is used we can be quite sure they will not be significant if we 
used the true probable error. 
I will now give the inheritance coefficients for each finger for the sibships ; the 
results are summarized in the following table. 
TABLE VIII. 
Eight hand 
Left hand 
Sibships 
Uncorrected 
Corrected 
Uncorrected 
Corrected 
for grouping 
for grouping 
for grouping 
for grouping 
Thumb 
•320 
•379 
•42.5 
•477 
First finger 
•340 
■389 
•361 
•411 
Secoml finger 
•331 
•384 
•443 
•502 
Third finger 
•402 
•477 
•358 
•411 
Fourth finger ... 
•368 
■439 
•366 
•440 
Mean 
■352 
•413 
•391 
•449 
If we compare the values found for the first finger with tlie mean values 
found for the first finger in the longer series we find they are in close agreement, 
but the agreement is less close in the left hand "374 and •411 being the values; 
the difference between them is, however, not significant. It will be interesting 
to see whether there is any significant difference in inheritance in the different 
fingers. Again I have not worked out the p.e. of the contingency but taking 
the p.e. of r as a rough measure we find that all the probable errors in this table 
lie between "020 and •023 and that the probable error of the difference between 
any two coefficients or any one coefficient and the mean is about '03. As this 
probable error is an underestimate of the true probable error it will be wiser to 
take three times the probable error and consider that unless dift'erences in the 
coefficients exceed •09 they are unlikely to be significant. No coefficient in either 
hand differs from its mean by more than '064 when the correction for grouping 
has been made, and if we take all ten coefficients for both hands we find that the 
mean is '431, and we see that no coefficient exceeds this mean by moi'e than 071 
which is certainly not significant. The greatest difference in the coefficients 
found are in the right thumb and in the second left finger where the difference is 
"123. A comparison of all the values for fraternal correlation shows that those 
for finger-prints are lower than we should expect. There is a table of fraternal 
resemblances on p. 25 in Nature and Nurture, The Problem of the Future by 
Karl Pearson* and the lowest fraternal coefficient is that of brother and sister for 
the Cephalic Index ("43). In certain fingers when the whole hand is considered 
one or two values approach to what we should expect, as for instance the third 
right finger, the second left finger and the left thumb, but on the whole the values 
are lower than we now expect to find when we consider fraternal resemblance. 
* Eugenics Laboratory Lecture Series, No. vi. 1910. Cambriilge University Press. 
