APPENDIX 



57 



The sampling error in the data on change in forest 

 acreage and timber volume cannot be estimated. In 

 the 1932-35 Survey, sample plots were spaced one- 

 eighth mile apart on lines 10 miles apart; an estimate 

 of sampling error was not made. Sampling error in 

 the State forest acreage estimate of the first Survey 

 was probably very small, as it is for the second Survey ; 

 and the indicated change in total forest acreage may be 

 considered essentially correct. Indicated changes for 

 Survey regions and other portions of the total acreage 

 should be valued in proportion to the magnitude of 

 the item and of the change. Changes in timber vol- 

 ume, because of the possible effect of nonsampling as 

 well as sampling errors, are shown only for major 

 groupings of the data. 



Growth 



Gross growth estimates are based on radial-growth 

 measurements of mechanically sampled trees on For- 

 est Survey plots. Measured growth over the past 10 

 years was used for projecting growth 10 years ahead. 

 The difference between present and projected volume 

 of the sample trees was reduced to an average annual 

 percentage increase and applied directly to the cur- 

 rent inventory volume. Net growth was calculated by 

 subtracting from gross growth an allowance for mor- 

 tality based on the plot inventory of trees that had died 

 in the previous 4 years. No attempt was made to 

 calculate sampling error in the growth estimates. 



Production and Drain 



Location of forest-products plants and cutting ope- 

 rations, movement of timber products to plants, and 

 production except for items of domestic use were 

 determined from a 100-percent canvass of wood-using 

 industries. Output of fuel wood, fence posts, and 

 other items of domestic use was estimated from an area 

 sample of producers. 



Data on production of each commodity were con- 

 verted to terms of drain upon timber growing stock 

 by using drain-to-production ratios derived from 

 measurements taken in 1947 on 104 quarter-acre 

 sample plots on logging operations throughout the 

 State. The plots were selected by products roughly 

 in proportion to 1946 drain. Their location was con- 

 trolled to insure geographic coverage of each product, 

 but within each locality plots were chosen at random. 

 With few exceptions, plots were taken on going opera- 

 tions so that products could be measured and their 

 volume assigned to the particular trees from which 

 they were cut. 



The sampling error to which the State total cubic- 

 foot drain estimates are liable, on a probability of two 

 chances out of three, is as follows for each commodity. 



Sampling error 

 in cubic-foot 

 Commodity: drain (percent) 



Sawlogs 3. 2 



Fuel wood 12. 



Pulpwood 3. 1 



Hewn ties 2. 5 



Cooperage bolts 6. 9 



Veneer logs 5. 9 



Poles 1 



Fence posts 15. 8 



Piling 1 



Miscellaneous products 13. 



Miscellaneous domestic use 49. 8 



All commodities 2. 6 



Forest-land Ownership and Management 



Forest-land acreage held by public agencies was 

 determined by correspondence with the agencies 

 concerned. 



To determine how much forest land is held by dif- 

 ferent classes of private owners, three sample units of 

 four sections of four square miles each were chosen at 

 random in each county. The name and occupation 

 of the owner and the acreage and boundaries of every 

 parcel of land over two acres in size in the sample units 

 were obtained from county courthouse records and 

 transferred to aerial photographs. Forest area for 

 each owner was calculated from the photographs and 

 expanded from a sample unit to a county-area basis. 

 Number of owners by owner-occupation class was 

 determined by counting all owners in each class whose 

 northeastcrnmost property corner fell within sample- 

 unit boundaries and by expanding the count for each 

 class from a sample unit to a county-area basis. 



Statistical tests of the acreage estimates indicate a 

 sampling error, on a probability of two chances out of 

 three, of 2 percent for the largest class of owners and 

 no more than 12 percent for the smallest class of 

 owners. Estimates of numbers of owners are subject 

 to the probability of greater error: 4 percent for the 

 class with |he most numerous owners, 57 percent for 

 the class with the fewest owners. 



To inquire into management practices, the factors 

 influencing them, and ways to achieve improvements, 

 some 1,000 owners were selected for sampling from 

 among the owners with forest land in the sample units. 



These owners were selected, first, by grouping own- 

 ers by the more important occupation classes (using 



