In many groups of organisms comparison of chromosome morphology is, or should be, an important tool of taxonomists and students of evolution. Objectivity of comparison is attained under three conditions. Firstly, where hybridization between taxa is possible, studies of the chromosome pairing at meiosis may reveal homologies. Secondly, comparisons may be based on patterns of longitudinal differentiation of chromosomes. Because minute detail is observable in only a few tissues like dipteran salivary glands or Trillium root tips, this basis of comparison is of restricted use. The third condition, which is the only one possible in most groups of organisms, is comparison of lengths of chromosome arms. The object of this communication is to show how this can be made quantitative and subject to normal statistical methods. The most useful measure of length is the mean per cent length, i.e., the mean value, obtained from a number of nuclei, of the length of a chromosome arm expressed as a percentage of the total length of all chromosome arms in the same nucleus. This metric has been used by Rothfels and Siminovitch (1958) and by several other authors. The difficulties of karyotype analysis using mean per cent length have been ably discussed by Patau (1960) and by Moore and Gregory (1963). Where these difficulties can be overcome, i.e., where chromosome morphology is quite distinct so that visual pairing of homologues within a cell is most unlikely to be wrong, mean per cent length could be used for comparing the karyotypes of two species. This comparison could be direct only if the two species had the same total amounts of chromosomal material, i.e., the same total absolute lengths at the same degree of contraction of their chromosomes. This would not always be so, and comparisons based on actual measurement of length would be unreliable because the degrees of contraction of the chromosomes and of distortion during slide preparation would be likely to vary considerably and not necessarily at random between the species being compared. The basic assumption of the method reported here is that, in an interspecific comparison, the ratio of total lengths will be the same as the ratio of amounts of DNA per chromosome complement. If this is true then LA, the mean per cent length of a certain chromosome arm in species A, can be compared directly with LBDBIDA where LB is the mean per cent length of a supposedly homologous chromosome arm in species B, and DB and DA are, respectively, the amounts of DNA per chromosome complement in species A and B. This assumption implies that no differences in degrees of polyteny occur between the somatic chromosomes being compared. However, such a situation should be detectable. In the examples which follow DBIDA has been measured microspectrophotometrically using Feulgen-stained mammalian leucocytes which had been grown in vitro by a modification of the method of Moorhead et al. (1960). After the usual colchicine and hypotonic citrate treatment the cells were fixed in acetic alcohol. One drop of species A cells in fixative was placed 34 inch from one end of a microscope slide (subbed as for stripping film autoradiography) and one drop of species B cells 34 inch from the other end. With care the two drops did not mingle and were allowed to dry. Slides were hydrolyzed for ten minutes in N HCl at 60? C., precautions being