After a brief history of the frost service, the relationship between expected minimum temperature and frost risk is discussed. The appropriate physical and statistical aspects of the problem are considered so that the determining variates used in multiple regression will have a sound theoretical basis. The regressions, using information available at 1800 Central Standard Time explain about 70 per cent of the variation in minimum temperature. This result is superior to that obtained from subjective techniques. The absence of suitable measurements of soil-moisture and of radiation appear to be the chief obstacles to a more complete solution of this problem.
L.G. Veitch (Mon,) studied this question.