4.1 This practice covers procedures of sampling for obtaining small, clear wood specimens which, when tested in accordance with Test Methods D143 and, in conjunction with full-size product tests, will provide mechanical properties for use in deriving design properties for lumber, panels, poles, house logs, and other products.

4.2 Data obtained by testing specimens sampled in accordance with these methods also provide information on the influence on mechanical properties of such factors as density, locality of growth, position in cross section, height in the tree, and moisture content.

4.3 Cruciform sampling is of principal value when information is desired on the influence on mechanical properties of height in the tree; of age or radial position in the tree; of rate of growth; the change from sapwood to heartwood; the relationships between mechanical properties and factors such as specific gravity; and making general comparisons between species for purposes of rating or selecting species for specific end-use products. Cruciform sampling does not provide unbiased estimates of mean values, percentile or other descriptive statistics, or a means of associating statistical confidence with estimates of descriptive statistics.

4.4 Double sampling is used when it is desired to improve or update existing estimates of mechanical property values that are the basis for establishing allowable design stresses for stress-graded lumber, plywood, poles and piling, and other wood products. The method involves predicting one property by carefully observing a well-correlated auxiliary property that is presumably easier or cheaper to measure. A sample estimate of the auxiliary property is obtained with a high degree of precision by representatively sampling the population. A smaller independent sample or a subsample of the large sample is used to establish a relationship between the auxiliary property and the property for which an estimate is desired. As applied to sampling a forest, double sampling has employed specific gravity to predict mechanical properties. The double-sampling method provides unbiased estimates of mean mechanical property values and an approximation method for estimating percentile values. Statistical confidence may be associated with the estimates of the means but not the percentile values.

4.5 Random sampling is used when probability estimates of descriptive statistics and property distributional characteristics are desired as the basis for establishing allowable design stresses for lumber and other stress-rated products. It is applicable when data for a species do not exist or when existing estimates are believed no longer applicable because of a changing forest character. Random sampling provides better probability estimates than double sampling and is less expensive and quicker if sampling and testing must be completed to establish mechanical property-specific gravity regressions for the double-sampling method.