This report covers Phases I and II, the Preliminary Sampling Effort, and Bias Testing, together with a proposed procedure for the Phase III Extensive Sampling Effort. The Widows Creek sampling system was found to be free of systematic error of any practical importance. The procedures of ASTM D-2234 Appendix Al for determination of random and segregation variances was demonstrated to be applicable to constituents of coal other than ash as well as constituents of lumpy bulk materials other than coal. Stopped belt sampling techniques, however, were found to have only limited practical applicability within normal operating routines of a power plant. TVA operating experience indicates that heterogeneity of coal and limestone would rank only as minor factors in facilities performance, the effects of which would tend to be swamped out by several other operating factors. Considering this and the observed disadvantages of stopped belt sampling techniques, the use of random and segregation variances for an index of variability is not recommended. A time series index of variability generated from a series of increments collected by existing equipment at regular time intervals offers some promise, and is suggested as a possible alternative that would have several advantages. A proposed method for obtaining such an index of variability is outlined.