DU. M. w ti;aa*ei;s, MIU D. SENTEU, AXD DU. a. .JADUEKUD 
probably be less than 1 niilliin. of iiierciuy, and consequently a very small quantitv 
(.)f this impurity would be sufficient to vitiate the results. 
As it is obviously impossible to separate neon completely from helium by cooling 
the gas to the temperature of liquid hydrogen, it is necessary to employ .helium 
Avliich is initially as free from that gas as is possible. Dewau {loc. cit., p. 155) bas 
employed the gas from the Bath s})rings as a source of helium, but, as bis own 
investigations show tliat the resulting helium contains 7 "4 per cent, of Jieon, it is 
unsuitable for thermometric purposes, ]>articularly for the measui-ement of the 
temperature, which can be obtained by boiling hydrogen under reduced pressure. 
Some years ago it Avas shoAvn (Hamsay and Ti;aa'p:k.s, ‘ Roy. Soc. Proc.,’ a oL GU, 
p. 206) that l)y subjecting tlie gas from the mineral cleAeite to the ]n-ocess of frac¬ 
tional diffusion, it A\ms jAossible to separate fronr it a small quantity of argon and to 
reduce the density of the lighter fractloji to POH (O = IG). Later it AA'as found by 
one of us, though the experiment has not been recorded, that Avhen the gas from 
cleA’eite Avas passed through liquid oxygen, cooled to — 210° C. in a bulb immersed in 
liquid air boiling under reduced pressure, a small (juantity of argon aaas reniOA'ed 
fi'(_)m the gas, but ap])arently no neon. For this reason cleA’elte gas appeared to be 
the iiKAst suitable source for obtaining ])ure helium. The light helium olhained by 
the diffusion experiments Avas emplojmd in these researches; it Avas puritied by 
])assing it at the normal pressure through a coll Immersed in li(pud liydrogen. 
The apparatus enq)loyed aauII uoaa’ be descrilied. 
The gas, of AAdiich aa'o had about 180 cub. ceiit 1ms., AA’as introduced through the 
syphon (tig. 8), into the gas-liolder A, Avhlch contained mercury. The gas-holder 
was connected Iia’ a ridjbei' tube Avltb a mercui’A’ resei’A’oir not shoAA’n in the fimire. 
t/ t. C 
and in order to eliminate any risk ot air entering the gas-holder through leakage of 
the rublier tube, the connection AA’as made through a trap, I). Tlie gas-holder 
communicated through the t\A o-Avay stop-cock, B, A\’ith the syphon for Introducing the 
gas, and AA'ith the refrigerating coil ( ’, AA’hich led through a second stop-cock, B, to a 
second mis-lmlder similar to the lirst. The coll communicated with the mercui'A' 
pump through the stop-cock E. 
After thoroughly exhausting the coil, the stop-cock E was closed, and the gas 
admitted to it. A A’acuum A’essel containing about 300 cub. ceutims. of liquid 
liydrogen, surrounded Ijy a. larger A’essel containing llcpilil air, A\’as brought beloAv the 
coil and raised till the latter Avas thoroughly Immersed in the liquid. The stop-cocks 
B, and B, AA’ere then carefully opened, and by adjusting the leA’el of the mercury 
I'eservoir connected AA'ith the gas-holders, the gas aaiis passed three times backAA’ard 
and foi’AAaird through the coil. When the AA’hole of the gas had returned to the first 
•gas-holder, the stop-cocks AA’ere closed. The gas in the gas-holder AA'e called fraction 1. 
About 100 cub. ceutims. of the orio-inal nas, measured under m)rmal conditions, 
still remained in the ca>11, for at the boiling-point of liquid liydrogen a gas occupies only 
about 1/15 of the A’olume AA hlch it fills at the normal temperature. On opening the 
