Round soda-glass rods were broken in four-point bending. In one series of experiments, the load was increased at a constant rate until fracture occurred, and in another series, the times to fracture under a range of constant loads were determined. Tests in each series were made under three different test conditions-non-rotating, rotating at 14 r.p.m. and rotating at 10,000 r.p.m. It is concluded from the experiments that glass does not fatigue under cyclic loading appreciably faster than it does under static loading. This was in accord with expectations, if the static fatigue of glass is due to the spread of cracks rather than to deterioration of the inherent strength of flawless material. As fracture of glass under static conditions is not preceded by flow and work hardening as it is for metals, the difference in fatigue behaviour of glass and metals is an indirect indication that the fatigue of metals is associated with flow and work hardening, and thus agrees with more direct evidence on the nature of fatigue in metals.