The seeds secured from such mutants should be sown as directed above, and as many individuals as possible secured for comparison with the parental type, which should be continued as before. If now the individuals of the mutant 
Fig. 5. Rosette of Raimannia odorata. progeny are placed in a series with respect to any given quality, statistical observations may show whether it is included within the range of fluctuating variability of the parental type. The question therefore as to whether a plant is a continuous or discontinuous variant is one of simple measurement and estimation of qualities, not a matter of opinion. With that simple question easily disposed of, the investigator may next turn his attention to the correlations, a phase of the question of vivid interest, since the sparse data at hand seem to point to a wider range of variation and less degree of correlation in such mutants than in the parental type. That is to say, instead of mutants being derived from forms showing widely fluctuating variability, we have them appearing in the progenies of species in which the range of variation is small and the correlations close, while the derivatives themselves swing through a wide range of fluctuations. 
Fig. 6. Rosetta of Mutant of Raimannia odorata induced by Chemical Treatment of Ovules.
As a complementary means of investigation of the constancy and independence of the various characters dealt with in selection and mutation as observed above, hybridization forms an invaluable means of analysis, and is to be constantly resorted to in all stages of the various phases of the work, since in the observation of the behavior of unit-characters in combination and after resolution the clearest appreciation of their character may be reached.
Still another phase of variability is that in which the greater portion of an individual will be found to comply with the requirements of a continuous series, but which bears a branch or portion of a branch which differs notably by the loss or acquisition of characters. Although all
