Abstract: Based on the description of the four species of fossil impressions of ginkgolean leaves, Ginkgo digitata (Brongn.) Heer, Ginkgo huuoni (Sternb.) Heer, Ginkgoites sibiricus (Heer) Seward and Ginkgoites taeniatus (Braun) Harris, the authors attempted to expand the qualitative study of their characteristics to quantitative study by means of computerization. The discriminant analysis was applied to the classification of these fossil. The statistic variates selected were the lobe number, length, and width, and the nerve number and the angle between two outer lobes. The diseriminant equations of the four species established by the analysis were substituted, respectively, with the variates of the type specimens as well as the Chinese or the foreign specimens of the same species. The results correspouded well with species classification. The source programme was designed in the FORTRAN 77 algorithmic language and the calculation was done on the IBM-PC computer.

Key words: Ginkgopsid, Impressions, Multiple discriminant analysis