The mercury used for the pump must be scrupulously clean. It does not, however, require to have been distilled in vacuo. It is sufficient to purify it by allowing it to fall in a fine spray into a large or rather tall jar of 25 per cent nitric acid and 75 per cent water. The mercury is then to be washed and dried by heating to, say, 110 deg. C. in a porcelain dish. Exhausting a Roentgen Tube. With a pump such as has been described there is seldom any advantage in fusing an extra connection to the vacuum tube so as to allow of a preliminary exhaustion by means of a water pump. About half an hour's pumping may possibly be saved by making use of a water pump. The tube to be exhausted is washed and dried by careful heating over a Bunsen burner and by the passage of a current of air. The exhausting tube is then drawn down preparatory to sealing off, and the apparatus placed upon the pump. It is best held in position by a wooden clamp supported by a long retort stand. Exhaustion may proceed till the mercury in the fall tubes commences to hammer. At this point the tube must be carefully heated by a Bunsen flame, the temperature being brought up to, say, 400 deg. C. The heating may be continued intermittently till little or no effect due to the heating is discernible at the pump. When this stage is reached, or even before, the electrodes may be connected up to the coil and a discharge sent through the tube. Care must be taken to stop the discharge as soon as a purple glow begins to appear, because when this happens, the resistance of the tube is very low, the electrodes get very hot, and may easily get damaged by a powerful discharge, and the platinum of the anode (if a focus tube is in question) begins to be distilled on to the glass. The heating and sparking are to be continued till the resistance of the tube sharply increases. This is tested by always having a spark gap, conveniently formed by the coil terminals, in parallel with the tube. If the terminals are points, it is convenient to set them at about one quarter of an inch distance apart. As soon as sparks begin to pass between the terminals of the spark gap it becomes necessary to watch the process of exhaustion very carefully. In the first place, stop the pump, but let the coil run, and note whether the sparks continue to flow over the terminals. If the glass and electrodes are getting gas free, the discharge will continue to pass by the spark gap, but if gas is still being freely given off, then in perhaps three minutes the discharge will return to the tube, and pumping must be recommenced. The Roentgen effect only begins to appear when the tube has got to so high a state of exhaustion that the resistance increases rapidly.