Short spark gaps (order of one mm) in air at atmospheric pressure, unless specially radiated, show large spark lag when tested at little above the sparking potential. In engineering literature this property is measured by the impulse ratio, i.e., the ratio of the voltage at which breakdown occurs under a voltage surge rising at a specified rate (about 50 kilovolts per microsecond), to the voltage at which breakdown occurs with slowly increasing impressed 60‐cycle voltage. The spark lag shows statistical variations, and its mean has been shown to be nearly equal to the mean time for emission of an electron from the cathode, prior to the spark. The electron current from the cathode prior to the spark is greatly increased by the presence of a pointed projection on the cathode, so that the mean spark lag at a given voltage is greatly reduced. However, if a single pointed projection is used, of sufficient size to reduce the spark lag at a given voltage, it also reduces the 60‐cycle sparking potential, so that the impulse ratio, which is the quantity important in engineering applications, is not reduced. By using a very large number of extremely small pointed projections on the cathode, the spark lag can be reduced without appreciable lowering of the 60‐cycle sparking potentials. Thus impulse ratios little greater than unity can be obtained. Small particles of carborundum, rutile, alumina and porcelain, of various sizes were attached to the cathodes of spark gaps, and the impulse ratios tested. Minimum impulse ratios were obtained when the particle linear dimensions were between 2×10−3 and 15×10−3 cm.