January 23, 19 13] 



NATURE 



575 



consequence a difference of pressure will be estab- 

 lished between n and ??i which is proportional to 

 the speed of rotation and the internal friction of 

 the gas ; the latter being independent of the pres- 

 sure, the difference of pressure produced should be 

 independent of the pressure. This is true when the 

 pressure is relatively high; if it continued to be 



—■-a 



true down to the lowest pressures we should be 

 able to create an absolute vacuum by exhausting 

 initially with another pump at n to a pressure 

 lower than the (constant) difference of pressure 

 between jn and n. When, however, we come to 

 pressures below o'ooi millimetre of mercury this 

 is no longer the case ; the molecules are then 

 diffusely reflected, and fly from one wall to the 



other without meeting other molecules. If the 

 surface of the cylinder moved with a velocity 

 greater than the molecular velocity we would 

 obtain an absolute vacuum ; such speeds are im- 

 possible in practice. However, at these low pres- 

 sures the ratio of the pressures at in and n remains 

 constant independent of the pressure, and it has 

 been found that attainable speeds of revolution 

 NO. 2256, VOL. 90] 



{8000-12,000 revolutions per minute) are sufficient 

 to give a vacuum better than any hitherto 

 obtained. 



In practice the pump is constructed as indicated 

 in Fig. 2 (a) and (b). Instead of cutting the slot in 

 the case, the cylinder is grooved, and a tongue C 

 from the case projects into the groove ; this is 

 equivalent to a very 

 long slot in the 

 case. For i n- 

 creased efficiency 

 several parallel 

 grooves are cut, 

 and connected with 

 one another so that 

 the low pressure 

 side of one is the 

 high pressure side 

 of the next (Fig. 

 2, V). The com- 

 plete pump is shown 

 in Fig. 3. A pre- 

 liminary pump is 

 needed to reduce 

 the pressure to a 

 few millimetres of mercury initially. 



A great advantage of this form of pump is that 

 it deals with vapours as well as gases, as the 

 low pressure part of the pump remains at low 

 pressure. In other forms of pump the gases are 

 compressed while being removed, and in conse- 

 quence vapours condense which are afterwards 

 brought back into the vacuum again. Without 

 drying agents the new pump has produced a 

 vacuum lower than any hitherto measured, 

 o'oooooo2 millimetre of mercurv ; this pressure 

 was calculated by observing the ratio of the 

 pressures in different grooves. 



Very interesting are the measurements made by 

 Gaede of the kinetic heat effect. Owing to the 

 increased velocity of the molecules the temperature 

 of the gas should be higher near the upper surface 

 of the tongue C (Fig. 2) than near the lower sur- 

 face, and by arranging a thermocouple in place 

 of the tongue C Gaede has detected such an effect 

 as soon as the pressure is low enough to allow the 

 mean free path of the molecules to be larger than 

 the dimensions of the groove. 



A table of the exhaustion attainable with 

 various selected pumps is appended. 



Water pump ... ... ... ... 10 



Ordinary piston pump ... ... ... i 



*Older Geissler pump ... ... ... o'l 



*Ne\ver Geissler pump ... ... ... 001 



*Sprenge1 pump o'ooi 



*Modified Toepler pump ... ... ... o.ooooi 



*Kahlbaum's automatic mercury pump o'ooooo2 



Geryk oil-filled pump ... ... ... o'ooo2 



Gaede rotary mercury pump ... ... o'ooooi 



Gaede molecular pump o'ooooooa 



* Taken from Winkelmar 

 must only be takeii as very i 

 Kahlbaum's pump can give 

 (ihe figure for which is given 



n's " Handbuch der Physik," I. The numbers 

 oueh ; for instanc, it is very doubtful whether 

 a belter vacuum than Gaede's mercury pump 

 by the Physikalisch-Techniscbe Reichsanstalt). 



E. N. DA C. Andrade. 



