In the incessant search for new and simpler techniques of vegetation analysis, truly original ideas are rare. One such idea, the Bitterlich variable plot (or plotless) method, has continued to grow in usefulness for a dozen years. Proposed for estimating tree basal area, it has been adapted to determination of timber volume, and is now widely used in routine forest inventory. Modification of Bitterlich's geometry permits the variable plot technique to be used for direct estimation of shrub cover in open shrub communities. At each of several sampling points, those shrubs whose horizontal crown spread appears larger than the crossarm of a hand-held angle gage are counted, while those covered by the crossarm are ignored. The angle gage is so constructed that the count at each sampling point, divided by an appropriate factor, is an estimate of the percentage of the ground surface occupied by the projected area of shrub crowns. Details of gage dimensions and an outline of the theory were published earlier (Cooper 1957). As commonly used, the gage consists of a wooden base on which a perpendicular crossarm 4 15/64 inches long is mounted 30 inches from the eye. The count obtained with an instrument of these proportions must be divided by 2 to give a point estimate of ground cover. There is no accepted name for this divisor. It seems appropriate to designate it the cm.>er percentage fa.ctor, analogous to the basal area factor used in plotless timber cruising.Tr:STS OF VARIABLE PLOT SAMPLING Comparison of line intercept, variable plot, and loop estimates with true coverage in the open desert shrub vegetation of Nevada showed that line intercept came closest to the value obtained from measurements of every shrub (Kinsinger, Eckert, and Currie 1960). The variable plot method somewhat overestimated shrub cover. primarily because much dead crown material excluded by the other methods was incorporated in the variable plot sample. Even so, the variable plot estimate differed significantly from the measured value on only one of four study areas.There was great variation among individual line intercept samples. Depending on the nature of the vegetation, from 20 to more than 100 separate 100-foot lines, were needed to estimate cover to within 20o/o of the mean with 95o/o confidence. The variance of the angle gage samples was much less, however, so that only three to six were needed for the same precision. The variable plot method took less time per sample than the other methods tested, and the fact that fewer samples were needed aclclecl to its time advantage. There was relatively little difference in variable plot counts made by different observers at the same sample point. Observer variations were considerably greater with the other methods, particularly the loop procedure. Kinsinger et al. (1960) concluded that the variable plot method is faster and more reproducible than the other techniques tested, but that it cannot be used to study minute vegetational changes.Crossarm length significantly affected variable plot estimates in...