34 MUTATIONS, VARIATIONS, AND RELATIONSHIPS OF THE OENOTHERAS. 



measurements than entry d. The two lots of plants here compared may be 

 contrasted \Adth respect to their origin and treatment as follows, and some 

 one, several, or all of these contrasted factors will probably satisfactorily 

 account for the difference in variability: 



It remains to compare the three mutants here represented — 0. rubrinervis, 

 0. gigas, and 0. lata — with the parent species in regard to mean values of 

 their several characters and the coefficients of their variability. This can 



not be done with entire 

 satisfaction because of the 

 inconsistencies in the meas- 

 urements of plants belong- 

 ing to different pedigrees 

 in the same species. Such 

 comparison would be just 

 only when the lots of mate- 

 rial of all the species had 

 received identical treat- 

 ment, and this, as we have 

 seen, has not been true of 

 these. Wherever several 

 lots of plants of a single 

 species have been used, the 

 average condition of the 

 several lots has been taken 

 for comparison, and as a 

 given amount of deviation 

 in a constant of one value 

 is unfairly compared with 

 a given deviation from a 

 constant of a different value, all the deviations have been reduced to per cents 

 of the equivalent measurement in Oenothera lamarckiana. These values are 

 arranged in the following table in such a way as to allow a comparison 



Fig. 13. — Correlation in the buds of Oenothera, lamarckiana. No. 

 04129. Length of ovary subject, length of bud-cone relative. 

 Coefficient of correlation, 0.5631 ±0.0366. 



