Methods to estimate the length effect on tensile strength parallel to the grain in Japanese larch

To find a desirable method for estimating the length effect on tensile stre ngth (sigma (t)), we used three methods to analyze the sigma (t) data from a Japanese larch (Larix kaempferi) small, clear specimen. These methods inc luded a nonparametric method, the projection method of Hayashi, and a propo sed method. The estimated length effect parameters (g) by the nonparametric method were 0.0237 and 0.0626 for 50th and 5th percentile at distributions , respectively. The projection method requires a standard E-f level (E*: dy namic Young's modulus), arbitrarily chosen for calculating the g value. The g values from the projection method were 0.1122 for low E*, 0.0898 for ave rage E*, and 0.0759 for high E*. The estimated g values by the proposed met hod using selected sigma (t), data were 0.1020 and 0.1838 for the 50th and 5th percentiles, respectively. Among the three methods, the nonparametric m ethod did not consider the different distribution of Young's modulus among specimens, and the estimated length effect parameters (g) by this method we re small. The projection method reduced the influence of Young's modulus, b ut the length effect parameters varied with the E* level. The proposed meth od minimized the dependence on E-f distributions among specimens. We believ e the latter method is desirable for estimating the length effect on tensil e strength.