Nitrogen  and  Oxygen  and  of  Argon  and  Oxygen.       645 
percentage  in  the  liquid  to  78*69  per  cent.,  which  is  sufficiently 
near  the  mean  value  78*75  per  cent,  found  by  experiment. 
It  is  certainly  surprising  that  the  amount  of  dead  space  can 
make  so  great  a  difference  in  the  composition  of  the  liquid  ; 
but  it  is  of  course  due  to  the  temperature  of  the  dead  space 
being  so  low  and  to  the  great  difference  in  the  vapour- 
pressures  of  nitrogen  and  oxygen.  These  tests  show  there- 
fore that  the  sample  of  the  liquid  can  be  taken  pretty 
accurately,  the  error  being  certainly  not  more  than  0"3  per 
cent,  and  usually  not  more  than  O'l  per  cent. 
The  sampling  of  the  vapour  offered  no  possibility  of  error, 
but  it  was  necessary  to  ascertain  how  long  the  circulation 
must  be  carried  on  in  order  to  obtain  true  equilibrium  between 
the  vapour  and  the  liquid.  About  5  litres  of  a  mixture  of 
nitrogen  and  oxygen  containing  about  75  per  cent,  of  nitrogen 
were  condensed  in  the  bulb  B,  and  after  ten  minutes  circu- 
lation (the  circulator  pumped  the -gas  at  the  rate  of  about 
50  c.c.  a  minute  j  samples  of  the  liquid  and  vapour  were 
taken.  Circulation  was  then  carried  on  for  a  further  fifteen 
minutes  and  fresh  samples  taken,  and  again  for  a  further 
twenty  minutes  and  a  third  pair  of  samples  taken.  The 
tubes  K,  L,  H  were  then  pumped  out,  and  while  some  more 
of  the  mixture  was  being  taken  into  B  through  the  tap  U  the 
tap  K  was  opened,  thus  filling  the  bulbs  L  with  the  mixture; 
a  second  series  of  experiments  was  then  made  exactly  as  at 
first.     The  results  are  given  in  Table  I. 
Table  1. 
Time  of  circulation 10  miu.    15  min.    20  min.    10  min.    15  min.    20  min. 
Per  cent,  of  Nitrogen!      ^  ^  m  ^  ^ 
in  liquid J 
Per  cent,  of  Nitrogen  |     gg 
m  vapour    J 
930 
These  experiments  show  that  the  liquid  sampling  was  uot 
very  reliable;  but  the  experiments  were  made  before  the 
method  of  liquid  sampling  had  been  perfected.  The  vapour 
samples  show  slight  variations,  and  equilibrium  is  apparently 
reached  after  ten  minutes  circulation  ;  so  a  minimum  of 
twelve  minutes  was  fixed  upon  as  being  sufficient  to  ensure 
equilibrium.  It  should  be  remembered  that  the  percentage 
of  nitrogen  in  the  liquid  sample  must  steadily  fall  ;  for  the 
vapour  is  richer  in  nitrogen  than  the  liquid,  and  some  of  the 
vapour  is  removed  in  taking  each  sample  of  the  vapour.  This 
to  a  certain   extent  explains  the  lower  value  obtained  for  the 
