Grass-tempered bowls made by the forbears of historical Karoo Bushmen between ca. A.D. 1500 and 1800 were decorated with many different stamp-impressed motifs. Five of the seven observed motif groups appear to be distributed randomly across the 2,000 km2 study area in the upper Seacow River valley, South Africa. When their percentage frequencies per site cluster were mapped, however, localized concentrations appeared. The validity of the percentage concentrations is tested here by unconstrained cluster analysis, which is not subject to the closure effect. Also, several other tests, all avoiding the use of percentage frequency maps, are introduced as alternative ways to deal with unevenly distributed sites and scarce artifacts or traits unevenly distributed among the sites themselves. The results demonstrate that percentage-frequency mapping can produce occasionally spurious and misleading patterns. Because stratigraphic evidence is scarce and inconclusive, and direct dating of sherds has not yet been conducted, a test also is proposed for determining contemporaneity of motif groups.