corresponding volume as low as 735 board feet, or as high, as 886 

 board feet, assuming the height is in fact 5 logs. The range in 

 volume v;ithin the two-inch diameter class permits a plus or minus 

 error of over nine percent » Now, if the tree actually has only 

 4g- 16-foot logs, an additional error of almost 7 percent is intro- 

 duced, yet the estimator has made no mistakes. On smaller trees 

 the possible statistical error can be as much as 36 percent. Over 

 a large number of measurements, such statistical discrepancies 

 will compensate or "iron themselves out," but in check scaling 

 the correlation of the volume of each scaled tree with the volume 

 estimated for that tree influences the number of check scaled trees 

 needed to meet a certain accuracy. Hence all practicable steps 

 must be taken to improve the accuracy of the first estimate in 

 order to reduce the discrepancy between the estimator's and checker's 

 volumes . 



It is recommended that d.boh, of sample trees be measured and recorded 

 by one-inch classes and heights be estimated by l/2-logSo Also, more 

 care and practice should be devoted to estimating heights. Inaccur- 

 acies of height estimates contribute as much as any other factor to 

 poor correlation between markers' and checkers' volumeso However, 

 with the sample tree measurement system it is reasonable to expect 

 that accuracy of estimating heights will be improved because fevjer 

 trees will be measured and more care can be given themo 



Defect 



Defect Contributes seriously to the discrepancy betv/een estimated 

 and scaled volume. One report states that defect was the greatest 

 single factor that contributed to poor correlation between estimated 

 and scaled volumes. (9_) The present study did not bear this out, 

 possibly because check scale data v;ere not available for many of the 

 sales, and hence a good cross section of defect conditions was not 

 studied. It v;as not possible, from the data available, to set up a 

 table in which a certain defect percentage called for a certain size 

 check scale because too many other sources of error influenced the 

 check scale. 



It has been suggested that if defect were found to be a major source 

 of the error between marker and checker, a random selection of logs 

 at landings might be scaled and the defect percentage calculated 

 from these, rather than to check scale a large number of full-length 

 trees to arrive at the correction factor. Of course, a defect cor- 

 rection factor obtained from logs on landings would have to be applied 

 to the estimator's gross volume, because his net volume has already 

 deducted for part of the defect - a part not distinguishable from the 

 vjhole. A poor representation of defect probably would be obtained 

 if the logger made a practice of leaving defective material in the 



-17» 



