14e Loaded direction Evaluation method d(mm) b,(×10 5% 10 5m5wedias Pero the 5aihdine 2608260826 Coefficient of cnfdenc interval:e had a positive with density.The on unpublished data.In the Eurocode 5 standard,embed- ower limit of t the confidence inter ding strength is calculate as follows 0.0821-0.01dp f=e-叭 +1+-we (2) e(1.35 +0.015d)sin a+cosa the regression line. ue or t-a As is shown in the relations between the embedding number of specimens.is the density of wood.is the mean value o the residual variance.The followine owing btai .(2 others.When embedding strength was evaluated by the 5% fe febp+bap+b 3 od,the values tor p)paralle and perpen reement with those reported by Harada t al.and where fefe,and p are the same as in Eq.(2);and bb and econstant value Kawamotoetal This indicates that%embeddings wa woul stimated from dow maximum load up to smm displacement according the 5mm embedding strength and the density was larger EN383,the value sity The variance of the embedding strength can be esti- larger p)perpe h ha(). ma ted by the density from Table 3 by he() show the values of (fp)obtained by Harada et aand Kawamoto et al. num stress according to144 Table 3. Coefficients of regression line and lower limit of 90% confidence interval between embedding strength and density Loaded direction Evaluation method d (mm) al a2 b~ (x 10 3) b2 b3 Parallel to the 5% Embedding 8 0.053 6.15 0.394 -0.308 grain strength 12 0.049 16.04 0.146 0.115 16 0.076 2.15 0.046 -0.037 20 0.073 4.26 0.054 -0.044 5 mm Embedding 8 0.057 11.79 0.181 -0.142 strength 12 0.068 10.21 0.115 -0.091 16 0.082 1.11 0.038 -0.031 20 0.074 4.55 0.054 -0.044 Perpendicular 5% Embedding 8 0.039 -0.53 0.112 -0.087 to the grain strength 12 0.032 1.98 0.048 -0.037 16 0.036 -0.20 0.023 -0.019 20 0.040 -2.80 0.033 -0.027 5 mm Embedding 8 0.049 3.35 0.283 -0~219 strength 12 0.055 -2.60 0.084 -0.065 16 0.047 -0.92 0.026 -0.021 20 0.047 -3.67 0 .048 -0.038 104166 50.34 26.95 20.72 48.16 39.63 22.20 20.72 28.13 16.4t l 3.56 14.86 71.27 28.65 15.19 21.31 Coefficient of regression line: fe = alp + a2 Lower limit of 90% confidence interval: feL = fe - (blp 2 Embedding strength (aCe), in MPa Density: [p(kg/m3)] p, density of the wood; al, a2, bl, b2, b3, constants + b2p + b 3 positivly correlated with the density regardless of the dowel diameter, the evaluation method, or the loading angle to the grain. These results agreed well with reports 3'8 that embed￾ding strength had a positive correlation with density. The regression line and lower limit of the 90% confidence inter￾val between embedding strength and density were calcu￾lated. The lower limit equation of the confidence interval was as follows, z6 sop J (2) where fe L is the lower limit of 90% confidence interval, fe is the regression line, t(~, (o) is the value of t-distribution with the degree of freedom (~0) and significance level (co), n is the number of specimens, p is the density of wood, u 0 is the mean value of the density, Spr is the sum of squares of the density, and Ve is the residual variance. The following equation was obtained by transformation of Eq. (2). fe L = fe - ~/b~p 2 + b2p + b3 (3) wherefeL, fe, and p are the same as in Eq. (2); and bl, b2, and b3 are constant values. The coefficients of regression line and Eq. (3) are shown in Table 3. The inclination of the regression line between the 5ram embedding strength and the density was larger than that between the 5% embedding strength and the den￾sity. The variance of the embedding strength can be esti￾mated by the density from Table 3. The embedding strength divided by the density (re~p) for each dowel diameter is shown in Fig. 10. These figures also show the values of O~e/p) obtained by Harada et al. 9 and Kawamoto et al. 8 and the design values of embedding strength according to Eurocode 57 standard. For the values of (re~p) by Harada et al., the density of sugi (Cryptomeria japonica D. Don) and karamatsu (Larix Ieptolepis Gordon) were assumed to be 430 and 510kg/m ~, respectively based on unpublished data. In the Eurocode 5 standard, embed￾ding strength is calculated from the dowel diameter and the density. The equation in the Eurocode 5 standard is defined as follows. 0.082(1 - 0.01d)p lea = (1.35 + 0.015d)sin2a + cos2a (4) where fe~ is the embedding strength (MPa), d is the dowel diameter (mm), p is the density (kg/m3), and a is the loading angle to the grain. As is shown in the relations between the embedding strength and the dowel diameter (Fig. 5), the slope of the regression line of the 5 mm embedding strength perpendicu￾lar to the grain divided by the density was smaller than others. When embedding strength was evaluated by the 5 % off-set method, the values for (re~p) parallel and perpen￾dicular to the grain obtained by this study showed good agreement with those reported by Harada et al. and Kawamoto et al. This indicates that 5 % embedding strength of softwood would be estimated from dowel diameter and density. When embedding strength was evaluated for the maximum load up to 5ram displacement according to EN383, the values of Oee/p) parallel to the grain were 23%- 29% larger than that derived by Eq. (4). The values of (re/ p) perpendicular to the grain were ctose to that of Eq, (4), Relations between embedding strength and compressive strength Compressive strength parallel to the grain was evaluated with maximum stress according to the Japanese IndUstrial