Abstract: The paper deals with the numerical classification of 62 wintersweet [ Chim onanthus praecox (L. ) Link ] cultivars, 24 characters within 62 cultivars of wintersweet were selected for the numerical classification on the basis of investigation and nomenclation. Results obtained through Q cluster analysis show that
the inner-perianth color is served as the first classification criteria, faint-center, speckle-center and red-center
can not be separately served as classification criteria, faint-center should be mergered to plain-center, speckle-center and red-center should be mergered to purple-center. Thus wintersweet cultivars are separated into plain-center, purple-center and green-center according to the inner-perianth color, besides, cluster analysis also
shows that the form of the middle-perianth has an important effect on wintersweet cultivars classification, so it
is served as the second classification criteria. R cluster analysis shows that the 24 characters have independent
influences on the genetic relationship. Other questions of wintersweet cultivars classification were also discussed. The p rincipal component analysis showed that the 24 characterswere intergrated into 9 p rincipal components ( PC) and their accumulative contributor ratio amounted to 77.3% , and 16 major discriminating
characteristics were selected according to the relativity between 9 PC and characters.

Key words: Chimonanthus praecox (L. ) Link, Cultivar, Numerical classification, Cluster analysis, Principal component analysis