548 



JOURNAL OF FORESTRY 



The actual differences in two actual tests between these methods of 

 calculating the average deviation are illustrated by the following table : 



The final column is the most accurate, but it will be seen that the 

 next to the last tells about the same story, while the fourth column is 

 misleading. 



In conclusion, then, two tests should be made of every volume table 

 against its fundamental data and the results thereof published with the 

 table itself. The first of these is a comparison of the aggregate true 

 volume of the basic trees with that given therefor by the table, and the 

 second is a computation of the average deviation of the individual tree 

 volumes from it. The former should result in a negligible difference 

 (zero under favorable conditions) but the magnitudes to be expected 

 in the latter cannot be definitely stated. The two tests just given sug- 

 gest that values in the neighborhood of 10 per cent may be normal, but 

 they are a meager basis for any general conclusion. The resulting 

 average deviations would be used for the purpose of comparing dif- 

 ferent volume tables for the same species and (less closely) tables for 

 dift'erent species of similar form. Hence as more and more tables were 

 thus checked the results would become increasingly valuable. It is 

 perhaps all too significant that in spite of the scores of volume tables 

 that have been prepared during the past two decades one cannot yet 

 state with confidence how closely individual trees should (or can) be 

 measured on the average thereby. 



If these two tests are adopted as standard and are hereafter uni- 

 formly and invariably made, not only will we have the satisfaction of 

 knowing that, as it were, we have closed our traverses, but we will 

 gradually build up knowledge which may guide us in deciding between 

 existing tables and determining which of them need the revision. It 

 will give us a quantitative line of attack on the old problem of the 

 relative merits of local and general tables. We will have, too, a new 

 light on testing the applicability of any table to a new region. If 

 the same two tests be repeated for a group of alien trees and if both 

 the aggregate difference is small and the average deviation agrees 

 closely with that published for the table itself, one would have a well- 

 founded confidence that the labor of preparing a new table was un- 

 necessary. 



