322 An Automatic MercAirij Vacuum-pump. 



entrance to the capillary. Thus the air from B is pressed 

 quietly into the capillary, and the mercury also enters it 

 without shock. Then, on coming down the 6 or 7 cms. of the 

 capillary, the mercury attains a high velocity by the time it 

 strikes the short sealing-column at d' ; and this, having so 

 small a mass, is shot out without any perceptible shock into 

 D, and opens up as it were a clear path for the trapped air- 

 bubble to follow it — at any rate, the freer character of the 

 motion has an obvious effect in diminishing its tendency to 

 stick. The mercury is, moreover, projected in a thin stream 

 through the fairly highly exhausted space of D, a process 

 which is likely to considerably assist it to give up any air it 

 may have condensed on it. On the downward stroke of the 

 pump, the column in the capillary breaks with certainty at d^ 

 and the mercury in it drops wholly into D with the exception 

 of the short sealing-column left in the lower bend of the tube. 

 And, finally, all the mercury which has passed over into D 

 (which may still have air entangled in it) passes back to the 

 reservoir by the path//^, by the insertion in which of a series 

 of bulbs on a capillary tube the dragging of trapped air back 

 into the pump-bulb can be rendered practically impossible. 



The smoothness of action of this form of pump-head is 

 remarkable. The pump works quite silently with the excep- 

 tion of the mercury hammer effect of the mercury streaming 

 into D. As regards its efficiency, I have not been able to 

 make a direct comparison between the two forms under the 

 same conditions, but the following results obtained with the 

 pump show generally the time and number of strokes required 

 with it to evacuate a space of 250 c.c. to certain stages. 



On starting from a vacuum of about 8 cms. pressure pro- 

 duced by the water-pump, the first 10 strokes were complete, 

 each occupying 33 sees., of which the vacuum-tube was open 

 to the pump-bulb for 5 sees. By this time the dark space 

 round the cathode had become 1'5 cm. long. The next stroke 

 was partial (25 sees.), and the 12th complete. On now 

 altering the position of the wire Y so as to allow about a 20 

 seconds' interval of connexion between the tube and the 

 pump-bulb, the next 10 strokes were all partial, and, 14 

 minutes from starting the pump, sufficed to make the dark 

 space extend the whole length of the tube (16 cms.). The 

 rate of exhaustion beyond this point is largely dependent on 

 the rate at which air comes off from the glass of the tube, 

 as well as on the efficiency of the pump. I have, however, 

 obtained in a comparatively short time (e. g. for a constricted 

 tube in about | hr.) a vacuum through which the induction- 

 coil would not discharge (16 cm. spark). 



