That the mortality from cancer shows definite variations in different parts of England is now generally believed. Various attempts have been made to ascertain whether any association exists between this variation in mortality and variations in certain features such as elevation and levelness of site, nature of soil and underlying geological formation, proximity to rivers liable to floods, etc., that characterise the several districts comprising different regions, and evidence both for and against the influence of one or other of these factors in predisposing to cancer has been brought forward at different times. An inquiry (1922) into the incidence of cancer in the registration districts in the Severn Valley for the decade 1901–10 seemed to indicate that no definite connexion existed between an excessive mortality rate from cancer and the foregoing factors. It was considered advisable, however, to continue the investigation in another part of the country to determine if the inferences drawn from the analysis of the Severn Valley data were confirmed or if any new features were brought to light. The region chosen included the districts drained by the Yorkshire Ouse, the Trent, the Cambridgeshire Ouse, and those that intervene.