APPLICATION 



In the application of the chart in figure 1, total sale value can be 

 closely estimated from the sale contract or appraisals Scaling costs per 

 unit must be determined from experience for specific conditions. Often 

 scaling costs are high because of lost time in waiting for loads at a 

 landing or because of travel time, and a system of sample scaling may 

 or may not afford a saving. Coefficients of variation are difficulx to 

 estimate without a previous sample of the material » Methods for computing 

 the coefficient of variation have been presented in detail by Lynch <,_/ 



As an example of the use of figures 1 and 2, assiome a timber sale of 

 2,000 M b. m. of mixed timber with an average total stumpage value of $10 

 per M (including all assessments), or a total sale value of $20 ,,000, Be- 

 cause of the overmature nature of the stand, the logs are expected to be 

 rather variable in volume with a coefficient of variation of 0.8. Previous 

 experience has shown that scaling under the particular conditions of this 

 sale costs about 15^ per log. 



On the left-hand graph of figure 1, begin with a 159^ scaling cost, 

 follow vertically to the 0,8 coefficient of variation curve and thence 

 horizontally to a point on axis 1. With a straight edge, connect this 

 point with the value $20,000 on axis 2, and where the line crosses axis 

 3, establish a point. Nextj, connect 1he point on axis 3 and a point on 

 axis 4 opposite the coefficient of variation 0.8 with a straight-edge, 

 and where an extension of this line crosses axis 5 read a value slightly 

 higher than 2 percent,.^/ 



To obtain sample size, turn to figure 2 and the graph for an 0„8 

 coefficient of variation. Assuming th^it the n^amber of logs in this sale 

 is estimated to be 12,000, then for an allowable sam.pling error of 2 per- 

 cent, approximately 1,400 logs will be required in the sample scale. The 

 ratio of 1,400 to 12,000 is about 1 out of every 9 logs. 



3^/ Lynch, Donald Sampling in tree measurement sales on Northern 

 Region National Forests, Northern Rocky Mountain Forest and Range 

 Experiment Station. Station Paper )Nio „ 27. April 1951. 



4/ If a combination of conditions indicates an allowable error of less 

 than 1,5 percent, it is doubtful if sampling is justified unless the 

 number of units in the sale is very large. Sample sizes increase at a 

 rapid rate as allowable errors are reduced below 2 percent. 



-3- 



