2i8 Forestry Quarterly. 



A number of points of interest are brought out by a study of 

 these tables. First, the irregularity of the values, suggesting, 

 since it is a reasonable assumption that the frustrum form factors 

 follow some law, that even the 1,394 trees used in the construc- 

 tion of the larch table were an insufficient basis. Secondly, the 

 far greater irregularity of the values in the second table, em- 

 phasizing the decided decrease in accuracy resulting from the 

 employment of the less complete data. Thirdly, the compara- 

 tively small range of the values in the more accurate table, a 

 point which would be much emphasized if one or two more 

 doubtful figures at the extremes of the table were eliminated. 

 Fourthly, that the variation throughout each diameter class and 

 average of all diameter classes is far less in degree or regularity 

 than through the height classes ; in fact, if we take into con- 

 sideration the uncertainty of the figures on the edges of the 

 table, we have forcibly suggested to us that a single value for each 

 diameter class might prove essentially accurate. This is in line 

 with the opinion of many European authorities that stem form 

 factors may be based on diameters only; and it must be born in 

 mind in this connection that such a conclusion is much less 

 radical when applied to the frustrum form factor than to the 

 ordinary form factor with the cylinder as a basis. 



Before proceeding to the practical utility of the frustrum form 

 factor, let us consider for a moment the causes of error in the 

 usual volume table. Assuming that the values from which such 

 a table is computed are averaged by means of a set of harmonized 

 curves, we see that errors may lie in : 



a. minor irregularities of shape of the curves ; 



b. improper general shape or direction of the curves ; 



c. wrong position or location of the curves. 



The first error is of minor importance, and can be made neg- 

 ligible by careful workmanship. Of the other two the first is 

 by far the more dangerous. This is for two reasons. If the 

 general shape and direction of any curve were known, all values 

 along that curve would have their just influence in determining its 

 location, while with the direction unfixed, an abnormally high 

 value at one end of the curve and an abnormally low value at the 

 other end, instead of properly averaging against each other, have 

 a tendency to tilt the curve. The ends of the curves are, more- 



