Minimum limits for sampling 



Stratifying makes sampling applicable on relatively small sales, but 

 the exact size limit depends upon the highest sampling intensity that 

 is considered practicable o Preliminary trials indicate that a sampling 

 rate as high as one in tliree v;ill prove to be a saving over complete 

 tally. If this is true, than a sale of less than 2,000 trees can be 

 sampled vjith an accuracy of 2o5 percent without stratification. With 

 stratified sampling, the lower limit of sale size applicable could be 

 less than 1,000 trees o Of course, as size of sale increases, the 

 relative size of the sample decreases, as shown in Figure lo On large 

 sales, the sampling rate can be light « Figure 1 also shovjs the marked 

 decrease in sample size obtained by lovjering the standard of accuracy o 

 Sales of low value species need not be measured with high accuracy to 

 stay vjithin a reasonable value accuracy o Sampling rates can be governed 

 accordingly* 



FIELD PROCEDURE FOR COLLECTING FIRST SMPLE - 



Sampling must be practicable cr it will net be adopted by timber sales 

 administrators. Although based on rather complex theories 5 the actual 

 application of sampling, fortunately, is simple, and the estimator does 

 not require specialized statistical training, lifter a sampling rate 

 has been determined for a given sale, the marking is carried out much 

 as it is on a regular tree measured sale. Random selection of sample 

 trees is greatly simplified by the use of a crew of three men; two 

 marking and measuring and one recording is an efficient organization. 

 Special forms for field use on vjhich the randomization scheme can be 

 set up in advance of field work are shown in the appendix. Provision 

 must be made, of course, for a separate tally of each species and of 

 each stratum if sampling is stratified, Appendix I gives detailed, 

 step-by-step instructions for sample tree measurement training. 

 Selecting sample trees with the aid of random num.bers is described 

 in Appendix Xc 



One man alone can sam-ple tree measure, but with more than one species 

 cr size strata involved 5 the random selection of sample trees is 

 difficult. A system such as the drawing of colored marbles must be 

 used so that the man does not knov; in advance vjhich tree vjill be 

 measured. If the sampling rate is one in ten 3 then ten marbles, 

 nine white and one black, for example, are neededo After a tree is 

 selected for marking, a marble is dravm from one pocket and placed 

 in the other. The color of the marble indicates -whether the marker 

 measures the tree, or merely counts it. The "two pocket" marble 

 system, besides being cumbersome - especially in cold weather when 

 the marker must wear heavy gloves - introduces a certain bias. For 



-15. 



