SEVERAL YEARS AGO we published a paper on the permissible roughness to boundary-layer transition, which is referred to by Loftin in connection with his measurement and also by Schlichting in his new book on boundary layers. Our previous idea, based on Schiller's hypothesis, was that kv*/v must be constant when transition occurs at a single roughness element, where k is the height of roughness element, v* the friction velocity, and v the kinematic viscosity. This constant was determined by experiment on transition due to a wire stretched transversely over a flat plate (aluminum, 80 cm. long, 60 cm. wide, and 3 mm. thick), using a' small total-pressure tube in contact with the plate at a fixed point (70 cm. behind the leading edge) for detecting transition. The constant thus determined was 13. We found out later, however, by more careful measurement that transition does not always take place right behind a single roughness but that the boundary layer usually maintains laminar for a certain distance. This fact is also confirmed by Liepmann and Fila. Since our research was published as the Wartime Report in Japan and has been known only within a limited circle, it seems to be worth while to recapitulate its major results. The experiments being described in the present report were also conducted on a flat plate, 2.4 m. long, 2 m. wide, and 3 cm. thick, with leading and trailing edges chamferred, which was held vertically in a wind tunnel of relatively low turbulence. Transition velocity was determined by the total-pressure tube placed in contact with the plate at several points behind the wire, instead of at a fixed position as in the previous experiment. A typical result is shown in Fig. 1, in which the abscissa and ordinate are transition position (location of total-pressure tube) and transition velocity (mean wind-tunnel velocity when total pressure increases suddenly), respectively. The position of the roughness element is indicated by vertical broken line. A summary of all experiments is given in Fig. 2. These results show that the boundary layer maintains laminar for a certain distance behind a roughness element. If we plot the values of transition velocity against the ratio X/XQ, we find that they are approximately independent of the location of roughness element, x0, as shown in Fig. 3, x being the transition position. We were not much successful, however, in getting a single parameter that governs transition satisfactorily. A parameter suggested by Fage and Preston on ordinary transition, which is deduced from Taylor's hypothesis of transition due to incipient separation, is