Experiment shows that geologists accept a wide range of pebble shapes for till-fabric analysis. Most of this variation can be accounted for by a single variable—the B:A axial ratio of the clast. The individuals were classified into three groups, “conservatives,” “moderates,” and “liberals,” depending on the pebble elongation they would accept. Data on till pebble fabric were collected sequentially from 19 basal till sites in east-central New Hampshire. Fabric diagrams were produced for 11 different combinations of pebble elongation and quantity from each site, evaluated, and compared to independent indicators of ice-flow direction. The results show that the use of 100 elongate pebbles by conservative standards tends to produce uniform, intense, precise, accurate, and imbricated fabrics; whereas reversed and transverse fabrics are scarce. As shape selection becomes more liberal and as fewer pebbles are used, the fabrics tend to become irregular, weaker, less precise, less accurate, and less imbricated; whereas reversed and transverse fabrics are more frequently generated. At the extreme, where few pebbles are selected by liberal standards, the resulting fabrics approach those generated from random numbers. About 100 pebbles, selected by rather conservative shape standards (B:A ≤0.59) are necessary to obtain dependable diagrams.