The intensity of frost varies with differences in microtopography and vegetation. In addition, the grade of frost damage to plants varies considerably with their physiological state and the rates of cooling and forming and melting of frost. Therefore, it is difficult to establish relations between microclimate and frost damage only from the field observations. Many laboratory experiments on the frost damage were done with environmental control apparatuses. Most of them, however, cooled plants by the air of low temperatures, whereas the frost under natural conditions results from the radiative heat loss of plants.<br>The authors constructed a new simple apparatus which was able to lower the leaf temperature below air temperature by the heat of longwave radiation and consequently to form frost on leaf (Figs. 1 and 2). This apparatus, a “frost cabinet”, was set up in an air-conditioned room at 3 to 5°C. Its principle of operation is that a leaf set in the crop chamber at the upper part of the frost cabinet is held in radiative equilibrium with the ceiling of the chamber and with the surface of the cooling plate at the bottom. The temperature of the leaf is controlled by changing the temperature of the cooling plate with the ceiling kept at a given temperature. The amount of water vapor in the crop chamber is regulated through the temperature of the water vessels located near the ceiling.<br>In the experimental setup a plant is hung upside-down from the ceiling. For the preliminary experiment reported in this paper, however, a wet filter paper covered with thin plastic film was used as a model leaf and placed horizontally in the crop chamber. The temperature of the model leaf was found to be several degrees lower than the air temperature (Fig. 4). The relation between beginning time of freezing after onset of cooling and height of leaf above the cooling plate is shown in Fig. 6. The vertical gradient of air temperature in the crop chamber becomes progressively small with height above the cooling plate (Fig. 3). A little variation of the leaf temperature was found with height above the cooling plate (Fig. 7).<br>At a freezing point the leaf temperature rose sharply as a consequence of the latent heat release, as indicated by a fine arrow in Fig. 4. In the course of freezing the temperature reached an extremal value and then decreased in a manner depending on the amount of water contained (Fig. 5). The temperature in the water vessels had a good correlation with beginning time of the leaf freezing, which was affected directly by the amount of the latent heat and the sensible heat transport to the model leaf (Fig. 8).<br>From the results described above, it is recognized that this cabinet can be effectively used for examining frost damages to plants. Its effectiveness will enlarge if developments in the following points are made;<br>1) influence of the upside-down setup of a plant on the physiological functions and 2) improvement of the revolving wiper to remove frost on the cooling plate.