By pumping and sparking, the resistance of the tube may be gradually raised till the spark would rather jump over 2 inches of air than go through the tube. When this state is attained the Roentgen effect as tested by a screen of calcium tungstate should be very brilliant. No conclusion as to the equivalent resistance of the tube can be arrived at so long as the discharge is kept going continually. When the spark would rather go over an inch of air in the spark gap than through the tube the pumping and sparking may be interrupted and the tube allowed to rest for, say, five minutes. It will generally be found that the equivalent resistance of the tube will be largely increased by this period of quiescence. It may even be found that the spark will now prefer to pass an air gap 3 inches long. In any case the sparking should now be continued, the pump being at rest, and the variations of tube resistance watched by adjusting the spark gap. If the resistance falls below an equivalent of 2 inches of air in the gap the pump must be brought into action again and continued until the resistance as thus estimated remains fairly constant for, say, ten minutes. When this occurs the narrow neck of the exhaust tube may be strongly heated till the blow-pipe flame begins to show traces of sodium light. The flame must then be withdrawn and the discharge again tested. This is necessary because it occasionally happens that gas is given off during the heating of the neck to the neighbourhood of its fusion temperature. If all is right the neck may now be fused entirely off and the tube is finished. Tubes of the focus pattern with large platinum anodes are in general (in my experience) much more difficult to exhaust than tubes of the kind first described. This is possibly to be attributed mainly to the gas given off by the platinum, but is also, no doubt, due to the tubes being much larger and exposing a larger glass surface. The type of tube described first generally takes about two hours to exhaust by a pump made as explained, while a "focus" tube has taken as long as nine hours, eight of which have been consumed after the tube was exhausted to the hammering point.