Water vapor transfer from a model leaf set in a wind tunnel and that from citrus leaves formulating a canopy in the field were measured by using the leaf-like surface evaporimeter method due to HASESA and TAKECHI (1973), in order to make clear the wind dependence of forced convection water-vapor transfer coefficients over leaves within plant canopies. Local evaporation rates (w) at the point of 2cm from the leading edge of a leaf-like surface evaporimeter were determined respectively in the model and citrus canopies.The following relation was used to separate the effect of buoyancy term from the measured evaporation rates and to determine the local transfer coefficient due to forced convection(Df)Df, EXP=w/ΔC-Dn, CALCwere, ΔC is the departure of water-vapor concentration at the wet surface of the model leaf from that in the general air flow and Dn, CALC the water vapor transfer coefficient due to free convection. Because the leaf-like surface evaporimeter is flat, smooth and rigid, it is possible to assum that the influence of three factors such as leaf-fluttering, surface roughness and leaf curvature upon the transfer coefficient can be disregarded. The comparison between measured water-vapor transfer coefficients and standard coefficients (Df, STD) calculated from theoretical relations for a flat plate (in longitudinal flow) similar to the evporimeter in the size and surface temperature distribution enabled us to make clear the influence of canopy air-flow on the transfer coefficient over leaves.The results obtained for the surface evporimeter in longitudinal air flow are as follows:1. The water vapor transfer coefficient due to forced convection over the model leaf set in the front of model canopy agreed satisfactorily with those calculaled from the laminar boundary layer theory, because the air flowing on the model leaf was free from the air disturbance by the model canopy.2. Although the wind dependence of transfer coefficients over the model leaf set within and behind the model canopy was expressed by the 1/2 power law, the absolute values were to some extent larger than theoretical values. The difference between measured and calculated transfer coefficients was gradually large with the increment of leaf area density of the model canopy. When the leaf area density was retained constant throughout experiments, the values of transfer coefficient were independent of the tion of the model leaf within the model canopy along the direction of air-flow.3. Under the conditions that the model leaves were oriented so as to shelter behind windward leaves, values of exponent of the wind-transfer coefficient ranged betweeh 0.6 and 0.7. In this case, the fetch-dependence of the watervapor concentration over the surface of the model leaf was quite similar to that over a flat plate on which turbulent boundary-layer was built up.4. When the model leaves were set at different distances behind the model canopy, the influence of air-flow disturbance due to the canopy upon the water vapor transfer coefficient diminished gradually with increasing the distance from the canopy.In order to study the effect of the angle of attack of air flow on the water vapor trasfer, experiments with the model leaves inclined against the horizontal plane at different angles were carried out in the model canopy. As can be seen in Fig. 4, in the range of inclination angle between 10° and 20° the values of transfer coefficient were somewhat larger in comparison with those over a flat plate in longitudinal air flow. However, the effect of inclination angle on the transfer coefficient over the model plate decreased rapidly with departure from that inclination range and the values of transfer coefficient became nearly equal to or slightly smaller than the standard transfer coefficient.