The sample size needed to investigate the disease incidence of rice sheath blight disease was studied by using the relative height of the uppermost lesion to the plant height and the percentage of hills affected by the rice sheath blight disease, which were obtained in 3 fields with different degrees of incidence and 30 fields sampled randomly from an area of 1, 500 ha at Kubiki-mura, Niigata Prefecture from 1981 to 1983. The sample size (n) required was estimated from fluctuation of the confidence limits for the mean of relative height of the uppermost lesions to plant height, and from the Snedecor & Cochran's formula, n=t2σ2/L2 with allowable error (L) limited within±15% of the mean. The sample size required to estimate the percentage of affected hills and the relative height of the uppermost lesions to plant height in a single field were 10 and 100 hills, respectively. Twenty fields were required to estimate disease incidence of rice sheath blight disease in the area of 1, 500 ha. The sample of single hill was selected at random from A, B, C, D and E rows, and picked every fifteenth hill thereafter. The 20 hills were drawn from each row. This systematic sampling was the most efficient sampling method to investigate the rice sheath blight disease.
