Notwithstanding the great volume of published investigations on the electric discharge in gases, the nature of the negative glow and the mechanism by which the current is carried across the Crookes dark space at moderately low pressures are still matters of speculation. Some years ago, in describing investigations on the latter phenomenon, I drew attention to the outstanding difficulties in the way of formulating a workable theory. One of these difficulties was an apparently hopeless discrepancy between two values for the velocities of the positive ions at the surface of the cathode, the one calculated from their measured space charge and the total current passing through the tube, the other obtained by extra-polation from the mobility measured at higher pressures. On reconsideration of the problem, I now recognise that the latter value was definitely wrong, and in this paper will show that if instead the velocity is calculated from more reasonable assumptions the discrepancy is almost entirely removed. The mean drift velocity of positive ions in a gas under an electric force varies directly with the electric force and inversely with the pressure of the gas. A linear relation has been shown experimentally to hold from atmospheric pressure down to pressures of the order of 1 mm. of mercury, so long as the electric fields are so small that the mean drift of the ions is insignificant compared with their velocity of thermal agitation. In the particular circumstances now under consideration we are concerned with the movements,of positive ions near the surface of the cathode in the Crookes dark space, where the electric field may be as high as 1000 volts per cm. in gas at pressures of the order of 0·1 mm. of mercury. Under these conditions the velocity of thermal agitation becomes negligible compared with that acquired by the ion between successive collisions, so that a linear extrapolation is quite unjustifiable.