SYNOPSIS Not enough attention has been paid in the past to locating crops subject to damage by frost on the more frost-free hillsides; and at the present day the phenomenon of nocturnal temperature inversion is not well understood by most fruit growers. Orchards set out 20 years ago in some of the coldest sections in several fruit districts on the Pacific coast are still being operated at a lose, while others have been removed only during the last two or three years. Detailed records of nocturnal temperature differences on slopes, covering entire frost seasons, are scarce. Observations of nocturnal temperature inversions, made at Pomona, Calif., and Medford, Oreg., during the frost seasons of 1918, 1919, and 1920, are given in detail and discussed in this paper. Inversions at Pomona during the winter are compared with those at Medford during the spring. Differences in minimum temperature as great as 28° F. were observed between stations at the base and 225 feet above the base, on a hillside at Pomona. The greatest inversions occur on clear, calm nights, following warm days. The duration of the minimum temperature on the hillside is usually much shorter than on the valley floor below, on account of large fluctuations in temperature during the night on the hillside. On every hill where observations were made, the data indicate that on clear, calm nights the top of the hill is colder than points on the hillside some distance below.