Under most glaze icing conditions a smooth region in the vicinity of the stagnation point is followed by a rough region which forms at a well defined boundary. This smooth/rough boundary is believed to be due to increased heat transfer and the formation and freezing of water beads on the surface. It has been postulated that this may be due to boundary-layer transition occurring at this location. In order to study this phenomena, 1 mm diameter hemispheres were placed on the leading edge of a NACA 0012 airfoil and the effect on the boundary layer studied at low subsonic spesds and Reynolds numbers from 750,000 to 2.25 x 10⁶. Flow visualization and hot-wire anemometry were used to detect transition and determine the boundary-layer characteristics. The most forward location where a 0.5mm high roughness element caused transition was at x/c=0.015 at the lowest Reygolds number and at x/c=0.0075 at Re=2.25x10⁶. Standard methods to predict transition due to roughness based on a critical roughness Reynolds number consistently failed, predicting transition much closer to the stagnation point than that measured. It was postulated that these methods failed since most of the roughness elements were larger than the local boundary-layer thickness.