Studies with the ionization gauge reveal that in a positive column discharge in Hg vapor at low pressure there occurs a type of reversible electrical clean‐up in which, under equilibrium conditions, a large and continuous exchange of Hg takes place between the space and the walls. The amount of Hg held temporarily in this way on the walls may be several times that in the gas phase. Thus if the current is suddenly stopped, the pressure may rise momentarily to several times its normal value; while if it is suddenly raised from a low value to a high one the lowering of the pressure may be so extreme as to cause surges across the arc or extinction of the arc (common occurrences during the starting of the Cooper Hewitt lamp). Evaporation from and condensation on free Hg surfaces, if these are at all limited in extent, are not rapid enough to prevent the effects. Electrophoretic and pressure effects are shown to be of great importance in these experiments. Thus in equilibrium, in a 2.1 cm tube, 1 m long, the pressure at the anode end was double that at the cathode end when the current was 2 amp. (mean pressure about 1 micron). This result is shown to be sufficiently accounted for by Langmuir's theory of the positive pressure effect. A method for calculating the drift speeds of ions and of the neutral gas in special cases is described. In the present case, a considerable drift of neutral gas is shown to take place under electron impacts towards the anode which compensates for the motion of the ions in the field. The initiation (due to clean‐up) and phenomena of isolated surge regions in long tubes are discussed. These regions of aggravated low pressure which are usually situated toward the anode end of the tube, are plainly distinguishable by their relatively weak and reddish luminosity. Also the surges are evidenced by a sharp hissing noise which emanates mainly from the inductance. On suddenly stopping an arc in a tube at 185°C, the Hg which came off the wall spontaneously afterward appeared to have a mean life (τ) on the wall of about 4 min. τ did not appear to vary rapidly with the wall temperature nor did the amount of Hg (Q) so coming off. Q did not vary nearly as rapidly as the positive ion current to the walls. Assuming the clean‐up to be due to these ions striking the walls, this and other more direct evidence indicate that Hg is dislodged by ions striking the walls as well as cleaned up thereby and that it is largely the balance between these two effects which determines the equilibrium amount of Hg held reversibly on the wall. The reversible clean‐up here described is probably of considerable importance in all practical low pressure Hg discharge devices wherein rapid changes of current occur. In addition to this so‐called reversible part of the clean‐up, a much larger amount of cleaned up Hg was found which could only be recovered by heating. If the heating were carried out in stages, a quantity of Hg came off with each rise in temperature in such a way as to suggest that the Hg is held with varying degrees of binding. A clean‐up on an iron surface was found similar to that on a glass or quartz surface. A tube containing unsaturated Hg vapor when strongly irradiated with the Hg resonance lines failed to show any clean‐up from which it is inferred that the clean‐up here described is not due to excited or metastable atoms. Careful tests with a McLeod gauge showed that gases other than Hg were not appreciably involved in the present experiments. The observations can be sufficiently accounted for on the tentative theory that ions penetrate into the walls to varying distances or to positions of varying stability, by virtue of their relatively high energies (15 volts or more), where they are held for greater or shorter lengths of time, according to their effective degrees of binding, until they escape as a result of random heat motions occurring either at the original wall temperature or at temperatures artificially elevated either in the oven or locally by ion bombardment.