Although the coefficient of the reciprocal of height was very significantly different from 

 zero, dropping this variable only reduced the coefficient of determination to = 0.741. The 

 simpler prediction equation was: 



S^^ = 0.5592 -0.09022 (l/SC^^ 



with a standard error of estimate of ±0.0153. 



Zobel et al. (1960) reported one of the few studies in which the outer sheath of wood was 

 treated separately. For the regression of outer bole on outer breast-height specific gravity, 

 he found standard errors of ±0.016 and ±0.011 for slash and loblolly pines, respectively. His 

 breast-height samples consisted of 1-inch disks. 



ESTIMATING BOLE SPECIFIC GRAVITY FROM BREAST-HEIGHT CORE DATA 



The average specific gravity of the bole to 3.6" i.b. (Sent) was the dependent variable 

 of a regression on 15 independent variables derived from the same factors studied in relation 

 to increment specific gravity. In addition, the specific gravity of the entire core from cambium 

 to pith (SCent) and its reciprocal were added. Of the 17 independent variables, two variables, 

 the reciprocals of SC^^^ and SCent were sufficient to explain most of the explainable variance . 

 The prediction equations were: 



Sent = 0.5579 -0.05614 (l/SC^^) -0.03299 (l/SCgnt) 

 with R^ = 0.696 and standard error of estimate of ±0.0145, and 



Sent = 0.5175 -0.0738 (1/SC^^) 



with R^ = 0.666 and standard error of estimate of ±0.0151. 



The ratio of crown length to total height explained an additional 1 percent of the total 

 variation^ Although statistically significant, the contribution of this additional variable is too 

 small to be of value for prediction purposes . 



It is interesting to note that the correlation of the outer 20 years of the core with average 

 bole specific gravity was 0.78 while the correlation of the entire core with average bole was 

 0.71. Apparently the central portion of the bole follows a less well-defined trend of specific 

 gravity with height in the tree than does the sheath of increment. Zobel et al. (1960) found a 

 similarly small change in correlation when the juvenile wood was ignored. 



Wahlgren and Fassnacht (1959) obtained standard errors of from ±0.021 to ±0.0285 for 

 four species of southern pine when the independent variable was the reciprocal of specific 

 gravity of a single breast-height core. Gilmore et al. (1961), also working with southern pines, 

 found standard errors of ±0.017 when predicting tree specific gravity from a single breast- 

 height core could be reduced to ±0.015 when the prediction was based on the product of two 

 core values, breast height and stump height. Judging from the within-tree variance of specific 

 gravity at breast height found in the present study, the advantage of Gilmore 's second core may 

 be chiefly a matter of obtaining more samples, rather than a virtue of position. Whether the 

 same argument holds for adding SCent predictions for grand fir is not clear, because in 



this case the SCgnt sample includes two of the four SC^^ cores. 



9 



