Zobel et al. (1960), using disks at breast height rather than cores, arrived at a standard 

 error of ±0.016 for slash pine. They estimated specific gravity at breast height without the 

 sampling error inherent in the increment core methods (but still having some measurement 

 error). From Wahlgren and Fassnacht's data, it appears that their standard errors in pre- 

 diction would have been reduced from about ±0.025 to ±0.015 by replacing the core by the 

 surrounding disk. 



Apparently, several cores to pith are adequate to provide a means of predicting tree 

 specific gravity with a standard error comparable to that obtainable from an entire disk taken 

 at breast height . 



ESTIMATING DRY WEIGHT OF THE TREE BOLE 



Dry weight of standing trees can be estimated by procedures analogous to those used to 

 estimate the cubic contents of such trees . One approach might be to use the equations of the 

 previous section to estimate specific gravity and a conventional volume table to estimate the 

 cubic-foot contents of the bole. Multiplying the product of these two factors by 62.4 Ibs./cu.ft. 

 would give an estimate of tree dry weight. However, such a procedure would make it very 

 difficult to assess the possible effect of crown characters acting through specific gravity and 

 form class on the final estimate of weight. A more direct procedure is to use dry weight of 

 the tree bole in hundreds of pounds (W) as the dependent variable in a multiple regression. 



The variance of dry weight increases with tree size. Hence, a set of weights inversely 

 proportional to the variance was needed to improve the estimates for the smaller tree sizes . 

 In previous analyses, the variance of specific gravity had appeared to be independent of tree 

 size. Thus, most of the change in dry-weight variance must be the result of the increase in 

 variance in cubic feet. Accordingly, a weighting function derived in the course of calculating 

 a cubic-foot volume table for grand fir was used to stabilize the variance of dry weight. 



Independent variables were selected by combining breast-height core specific gravity 

 with the combinations of d.b.h. (D) and height (H), commonly used in volume equations. The 

 specific gravity at breast height was based only on the data for the outer 20 rings of the core . 

 This portion of the core was used because it corresponds to the data collected on the grand fir 

 growth and yield plots . The analysis of the previous section showed that adding the data for 

 the entire core would have resulted in only a small increase in precision of the estimate. 



In addition, crown length (L) and crown class (C) were introduced to account for changes 

 in form class and specific gravity associated with crown characters. 



Four significant variables explained 98 percent of the variance in dry weight , but the 

 combined variable D^'H'SC^q alone accounted for 97 percent'. The most useful additional 

 variable was crown class. Together, these two variables resulted in a prediction equation 

 having a coefficient of determination of 0.978 and a standard error of estimate of ±27.2 lbs. 

 (for an observation of unit weight) . 



The two prediction equations were: 



W(cwt) = -0.2807 +0.001345(D2-H-SC2^) 



s.e. = ±0.307 



and 



10 



