4o6 
J. ARTHUR HARRIS 
dividual case ; only 3 of the values of Vps have the negative sign while 
10 of the partial correlation coefficients, oYps, fall below the zero bar: 
Tps =. 1264 ± .0109, oTps = .0507 d= .0090. 
Notwithstanding the material reduction in rps when correction is 
made for the influence of fpo, ns there seems to be a residuum of 
interdependence which cannot be thus accounted for. Evidence for 
this conclusion is seen in four facts, (a) The mean partial correlation 
has a positive and perhaps significantly positive value: ofps = .0507 
rt .0090, oTps/E ofps = 5.64. {h) Of the individual values of or pa 
only 10 of the 32 are negative in sign, {c) The mean value of the nega- 
tive coefficients is only — .0347 while that of the positive constants is 
+ .0895. {d) Of the 10 negative coefficients 5 may perhaps be con- 
sidered statistically significant with regard to their probable error, 
although this number would be reduced if n were taken as plants in- 
stead of pods in the calculation of the probable error. Of the positive 
