Automatic ^Jercury Vucinun-immp. Ii2l 



resulting- increased efficiency of the water-pump. Probably 

 a reduction in size to 100 c.c. would be attended with similar 

 advantage, but a limit to an advantageous decrease is of course 

 soon reached, when the time taken by the mercury to stream 

 through the capillary tubes becomes an appreciable portion 

 of the whole. 



Comparison of Pump-liead loith Ordinary Form. 

 In the ordinary form of pump-head, of which a diagram is 

 given in PI. XII. fig. 2, it is necessary to have the capillary dd' 

 not only short, but bent so that the free surface of the mercury 

 sealing-column remains permanently in the wider auxiliary 

 chamber D : otherwise on the dowaiward stroke the short 

 column would be entirely sucked back into the pump-bulb. 



This arrangement, however, possesses serious disadvantages. 

 A very obvious one is the violent shock to the apparatus 

 which necessarily accompanies the impact of the mercury 

 from B against the extension into the capillary of the large 

 mass of mercury in D *. Further, the very minute air- 

 bubbles that are trapped at high exhaustions are in a mercury 

 column continuous from B to D, and there is a tendency for 

 them to stick, either onto the side of the capillary, or more 

 particularly at the opening into the wider tube D. In this 

 case, of course, they are liable to be drawn back into the 

 pump-bulb on the downward stroke. It is impossible to ensure 

 their expulsion from the capillary by much increase in the 

 velocity of the mercury from B, both by reason of the increase 

 in the shock which this produces, and of the fact that a high 

 speed of the mercury as it comes up from B into the capillary 

 is probably detrimental to the due escape of air into the latter. 

 Finally, although one does not know, it is quite possible that 

 traces of air are compressed into a film on the surface of the 

 mercury thread by the sudden pressure which accompanies 

 the impact, and if so, this air also will be largely returned to 

 the pump-bulb. 



In the modified form the mercury in the capillary is but a 

 short sealing-column at the most two or three centimetres 

 long (the lower bend of dd^ is about 1 cm. high). The rate 

 of entrance of air into the reservoir is adjusted so that the 

 mercury is rising quietly by the time it has reached the 



* The inertia of the mercury in D made itself evident in an interesting, 

 though at the time very undesirable, way in an earlier pump I endeavoured 

 to construct, in which the electrode X (fig. 2) was in permanent con- 

 nexion with the mercury in D, and an electrical contact was made when 

 the mercury from the pump-bulb reached that in the capillary. At high 

 vacua, on making contact in this way, sparking in the capillary in- 

 vaiiabJy occurred, indicating that the violence of the impact produces a 

 rebound of the two mercury masses from each other. 



Phil. Mag. S. 6. Vol. 6. No. 33. Sept. 1903. Y 



