in E d 
under such conditions, especially for an inexperienced inter- 
preter. 
In table 17 the rows with + and — mean overestimation and 
underestimation respectively of density in % of he number. 
Table 19 shows the estimations for all interpreters and all test 
fields. 
Table 19 
Over and under estimations of densities for all interpreters and all test fields. 
  
1 :16000 1:26000 1:33 000 
  
Over estimations (% of the number) 
St 73 54 59 
Under estimations /2; of the number)......... | 16 26 22 
Table 19 indicates that density is usually overestimated and 
that the overestimation is particularly great for the large scale. 
Except for this and, to some extent, for the interpreter III dif- 
ferent scales do not seem to affect the accuracy in the estimation 
of density. 
+ + 3. 23. Estimation of timber supply. 
From mean height and density it is possible to obtain timber 
supply (in cubic metres per hectare = m3/ha, 1 ha — 2 acres) 
by means of a table, as has already been mentioned. If both 
mean height and density have been incorrectly estimated and 
the errors tend in the same direction, considerable errors may 
result. As a rule, therefore, timber supply is estimated by means 
of sample areas, height and density being used for checking. 
The results of timber supply estimations show in Table 20. 
The errors appear considerable. Such a result is, however, as 
good as may be expected from good interpreters when not fa- 
miliar with local conditions and working with a low number of 
sample areas (4 sample areas). 
To facilitate comparison systematic errors and mean square 
errors are given separately in Table 21.