Seven paired sand samples were collected from beach and nearby dunes along the Oregon and California coastlines. Sand was also sampled from the Nebraska Sandhills and some other inland dune locations for comparison. The samples were examined under the petrographic microscope after cementing the loose sand with epoxy, cutting into thin sections and staining for K-feldspar. Size and roundness statistics were generated for the 14 beach-dune samples and three Nebraska Sandhills samples. The quartz and especially K-feldspar showed statistically significant rounding from the beach to the dune over distances as small as 100 m. When comparing the rounding of quartz and K-feldspar from the coastal dunes and the Nebraska Sandhills, K-feldspar was significantly (p ≤ 0.05) more rounded than quartz in all dune locations except for one of the Nebraska Sandhills sites. It is difficult to understand how K-feldspar sand could survive in a large erg without becoming well-rounded, unless there was a nearby beach or stream acting as the source for the angular sand. A predominance of angular K-feldspar grains in supposed ancient eolian deposits (large fossil ergs), without direct sedimentological evidence of nearby streams or beaches (like the Coconino Sandstone), may be cause to question the eolian origin of such deposits.