Gases at Loiv Pressures. 381 



was fairly rapid, but after the first day the change was very 

 slow. This, no doubt, was due in part to the slow passage of 

 the gas towards the charcoal through the somewhat extended 

 form of the apparatus. It was, also, probably due to the fact,, 

 which was noted later in the investigation, that at a given 

 stage of exhaustion, the raising of the free surface of the 

 liquid air in the Dewar vessel surrounding C invariably pro- 

 duced a very appreciable diminution in the gas pressure in 

 the apparatus, and the lowering of the free surface as the 

 evaporation of the liquid air proceeded resulted in a distinct 

 rise in the gas pressure. It is to be understood that the free 

 surface was never allowed to fall as low as the top of the 

 tube C, so that all of the charcoal was always below the free 

 surface of the liquid air. 



The following results show how the decrement chanced 

 with the time in the final forty-eight hours: — 



May 29, 12 m. to 2.53 a.m Decrement 0*000051 



May 29, 7.15 p.m. to 8.58 0*000037 



May 29, 10.53 p.m. to 1.36 a.m. (May 30) ... 0*000031 



May 30, 1.36 a.m. to 4.48 * 0*000028 



11.11 a.m. to 2.06 p.m 0-000037* 



2.06 p.m. to 5.27 0-000024* 



5.27 p.m. to 8.21 0-000028* 



8.21 p.m. to 11.49 0*000022 



The smallest value of the decrement obtained was '000022, 

 and this could be measured moderately well. Its error 

 cannot, I think, be as much as ten per cent. Of course it is 

 clear that the true value of the decrement due to the friction 

 in the fibre is somewhat less than this, for there is still, 

 doubtless, some gas left to offer resistance to the moving 

 disk, so that the number to be used for /x in the above 

 equation should be somewhat smaller than 0000022. I have 

 ventured to make use of the value 0*000020 as the true value 

 to which the decrement will approach as the exhaustion is 

 pushed higher and higher. It will be seen from Table I. 

 that the calculations are carried through not only with this 

 value, but also with the actually measured value 0'000022. 

 This is done simply to show what effect such a change in the 

 value of jm has on the series of results obtained. 



It may be of interest to state that, at the stage of ex- 

 haustion when ^ = 0*000022 was obtained, the McLeod gauge 

 indicated a pressure certainly less than '000001 mm. It is, 



* These were taken in the afternoon when there is considerable jarrino- 

 of the apparatus, and are probably not so accurate as the others. 



