The best known recent work on rates of evolution is probably that of Simpson (1944), though Small's (1945, 1946) work on diatoms is more extensive. Small is concerned with the origin of species, which in this group seems to be a sudden or almost sudden process. Simpson compares evolutionary rates in different groups mainly by means of data on the origin and extinction of genera. Now among living animals and plants the distinction between genera is much less objective than that between species. On the other hand the paleontologist often finds it hard to be sure of differences less than generic. Moreover the criteria of generic distinction are certainly different in different groups. For example good systematists from Linnaeus onward have assigned the polecat (Mustela putorius) and the ferret (Martes furo) to different genera, while in fact they give quite fertile hybrids and are probably to be regarded as at most subspecifically different. On the other hand I know of no viable, let alone fertile, hybrids between animals assigned to different genera of Diptera or Amphibia. This sugests that animals are rather more readily placed in different genera among mammals than in some other classes. And in view of the incomplete character of the best fossil remains, the systematics of extinct animals are certainly less uniform than that of living ones. On the other hand the dimensions of a solid organ, such as a tooth, a shell, or a bone, are often measureable with great precision; and when we have a series of fossil populations, believed to form a lineage, we can calculate the rate of change of the mean value of any measure. In speci-C-ne o _ t lxr ..a l *.. r+ I:rnilc for% +111 time and for the character. From data on radioactivity we have a fairly precise measure of the duration of the whole tertiary period, and less accurate dates within it; and estimates of ten million years during this period are not likely to be out by much more than 25%o. The precisely dated strata from earlier rocks are not so common. The error is likely to be a good deal greater in paleozoic or mesozoic marine deposits where we have continuous records over some millions of years, as in the English chalk or Lias. Here estimates of time may very well be out by a factor of 2, but are hardly likely to be so by a factor of 10. As biologists we should like to be able to measure time in generations. In a fair number of insects the generation is exactly a year, in others there are two generations per year, and so on. Where generations overlap, as in most vertebrates, we can theoretically define a mean generation length exactly by definite integrals if we have life and fertility tables. In fact we can never do so for fossil forms. But we can be fairly sure that for a small rodent the mean generation length was less than a year, for most ungulates more so. For mammals not larger than a cow we may be fairly sure that it was under ten years. The longest mean initerval between sexual generations is probably to be found in clonally propagated invertebrates such as corals, where it may possibly extend over centuries. If evolutionary rates depend on mutation rates (which is doubtful), the year is as natural a unit as the generation. For it seems possible that the intensity of natural radiation produces a minimum mutation rate per unit of time which is often exceeded (as in Drosophila) but below