On quantitative statistical methods in taxonomy; subdivision of a Polymorphous species: Planchonella sandwicensis (Gray) Pierre

An attempt has been made to subdivide a very polymorphous plant species by means of a quantitative statistical method. This method has been based upon the following working hypothesis: 1° as some morphological characters of the material, concerning e. g. the shape of the leaves, the length of the pedicels etc., show an extremely great variation, each of these characters in every specimen at hand may be stated to be in one of three (one intermediate and two extreme) conditions; 2° if a character happens to be in an intermediate condition in a relatively great number of specimens the difference between the extreme conditions of that character may be considered insignificant from a taxonomical point of view; 3° the fewer characters of two or more specimens are differing significantly (in the way mentioned), the more reason there is to consider those specimens to belong together; and, on the contrary, the more numerous the significant differences are, the more reason there is to distribute the specimens to two (or more) different groups. On the basis of this working hypothesis the material at hand, consisting of 143 specimens (all considered to belong to the Sapotaceous Planchonella sandwicensis, which was discussed in a paper by Lam), could be subdivided into two different groups. Five characters were chosen, each allowing to state one intermediate and two opposite extreme conditions. Of these the shape of the leaf proved to be a most important criterion for a subdivision. After this had been stated, a purely quantitative check was made which largely endorsed the result.