The only difficulty in making such a pump as is described lies in the bending of the heads of the fall tubes. This bending must be done with perfect regularity and neatness, otherwise the drops of mercury will not break regularly, or will break just inside the top of the fall tube, and so obstruct its entrance that at high vacua no air can get into the tube at all. The connections at the head of the fall tubes must also be well put on and the joints blown out so that the mercury in dropping over the head is not interfered with by the upper surface of the tube. However, a glance at the enlarged diagram will show what is to be aimed at better than any amount of description. In preparing the fall tubes it is generally necessary to join at least two "canes" together. The joint must be arranged to occur either in the tube leading the mercury to the head of the fall, or in that part of the fall tube which remains full of mercury when the highest vacuum is attained. On no account must the joint be made at the fall itself (at least not by an amateur), nor in that part of the fall tube where the mercury falls freely, particularly at its lower end, where the drops fall on the head of the column of mercury. When a high vacuum is attained the efficacy of the pump depends chiefly on the way in which the drops fall on the head of the column. If the fall is too long the drops are apt to break up and allow the small bubble of air to escape up the tube, also any irregularity or dirt in the tube at this point makes it more easy for the bubbles of air to escape to the surface of the mercury. Any pump in which the supply of mercury to the fall tube can be regulated nicely will pump well until the lowest available pressures are being attained; a good pump will then continue to hold the air bubbles, while a bad one will allow them to slip back [Footnote: For special methods of avoiding this difficulty see Mr. Ram's book.] ... Though three fall tubes are recommended, it must not be supposed that the pump will produce a Crooke's vacuum three times more rapidly than one fall tube. Until the mercury commences to hammer in the pump the three tubes will pump approximately three times faster than one tube, but as soon as the major portion of the air collected begins to come from the layer condensed on the glass surface of the tube to be exhausted and from the electrodes, the rate at which exhaustion will go on no longer depends entirely on the pump.