A PROPOSED STANDARDIZATION 



oio 



omission of this precaution. The resuh of such a comparison (judging 

 from my personal experience) should disclose an error of not to exceed 

 1 per cent if the table has been even moderately well prepared. If it 

 proves to be more than this the desirability of repeating the work of 

 constructing the table is strongly indicated. If much less it can be 

 disregarded. Intermediate values should be used to correct the table. 

 If the actual total volume of the trees, for instance, is 1 per cent more 

 than that given by the table, all of the tabular values should be raised 

 by that percentage. 



Fio. 1 



This test, however, while necessary, is not sufficient. It is not 

 enough to know that our volume table is accurate in the aggregate and 

 we should demand also that it give as closely as possible a correct value 

 for each individual tree. Let me illustrate the importance of this 

 principle by a simplified example. The line C D in figure 1 will check 

 as perfectly with liie five points marked by the crosses in the aggregate 

 as will the line A B, yet the latter is a far closer approximation to the 

 true position. A comparison of aggregate values fails to bring to 

 light the erroneous direction of C D, and permits absurd plus and minus 

 errors to neutralize each other. The fairer comparison between these 

 two lines consists in calculating the deviations of the individual points 

 from each to <ec whicli gives the smaller total or average. In the 

 following tabic, for instance, the agreement of the totals of the first 



