Leaf material from 41 herbarium specimens representing 31 Salix taxa was examined for correlation between polyploidy levels and size of petiole epidermal cells and stomatal guard cells. The species included most of the polyploidy levels encountered in contemporary Salix (2n = 38, 76, 114, 152, 190). Petiole epidermal cell size allowed a clear distinction between diploids and tetraploids. The same was true for guard cell size if comparisons were made only within sections or within groups of closely related sections. Polyploidy levels higher than tetraploid could be estimated to +/- 2 - 4x. The data suggest that cell size may depend also, to a minor degree, on climate. Some cultivars that were transplanted from high to lower altitudes showed exceptionally large cell sizes; such specimens should be excluded from reference data. Contrary to chromosome counts where live material is necessary, these cell size methods can be used on herbarium material and on well-preserved fossil specimens (the epidermis of the petiole is often the only organically preserved part of a leaf remain). Cell size methods are also suitable to scan large numbers of living plants for deviations in chromosome numbers, because preparation and measuring procedures are relatively simple. The petiole cell size method was applied to 14 fossil Salix leaves from the Late Miocene Pickett Creek flora (Idaho). Salix succorensis Chaney and Axelrod, S. inquirenda Knowlton, S. wildcatensis Axelrod, S. desatoyana Axelrod, and two undetermined specimens were found to be diploid. The result for S. wildcatensis is of particular interest because its closest modern relative, S. lasiolepis Benth., is tetraploid. The objective of future studies is to estimate the age of some polyploid complexes in Salix by fossil evidence.

Key words: cell size, guard cell, Miocene, polyploidy, Salix, stoma