A sample of 54 specimens of Tetrastemma laminariae was divided into two groups on the basis of one external character. A multivariate analysis of variance, based on 14 quantitative characters, was then performed in order to test the two hypotheses: H(1); the variance is the same in the two groups, H(2); there is no difference in the location of the two group-centroids. Neither of the two hypotheses could be rejected (P=0.002) and the possibility of the sample being impure, i.e. a mixture of at least two taxa, as suggested in an earlier paper on the high intraspecific variation exhibited by the species (Sundberg, P., 1979) is excluded. The variation found in T. laminariae appears to be due entirely to natural biological variability, although some apparent differences may be artifactual and caused, for example, by differential contraction or shrinkage during fixation. This emphasizes the necessity of basing taxonomical descriptions where possible, on many specimens so that an understanding of the degree of variation can be achieved. Additionally it is essential that careful fixation of material is carried out.