650  3Ir,  Ingiis  on  the  Isothermal  Distillation  of 
vapour  as  100  mm. ;  and  the  values  0'8022  and  1*  19-47  respec- 
tively at  the  temperature  corresponding  to  an  oxygen  vapour- 
pressure  o£  200  mm.  According  to  Travers  (loc.  cit.),  Baiy's 
measurements  of  the  temperature  are  erroneous ;  but  the 
thermometer  used  being  the  same  in  the  two  cases  the 
densities  given  are  probably  correct. 
Hence  for  the  results  of  Table  II.  the  volume  of  one  hundred 
gramme-molecules  of  a  mixture  containing  m  molecules  per 
cent,  of  nitrogen  will  be 
28m        32(100  -in) 
0-8225  1-2160       C'C'' 
and  the  molecular  concentrations  will  be  obtained  from  the 
molecular  percentages  by  dividing  each  molecular  percentage 
by  the  corresponding  value  of  this  expression.  Similarly  for 
the  results  given  in  Table  III.  the  factor  is 
28m     ,   32(100 -m) 
i Z — TT — T~^ C.C. 
0-8022  1-1917 
For  simplicity,  however,  the  concentrations  can  be  better 
stated  as  grains  of  the  corresponding  substance  per  100  c.c.  of 
the  mixture;  and  the  results  of  Tables  II.  and  III.  are  given 
in  this  way  in  Tables  IV.  and  V.  If  Henry's  law  holds,  the 
quotient  obtained  by  dividing  the  concentration  by  the  partial 
pressure  should  be  a  constant,  so  the  values  of  this  quotient 
are  also  given  in  the  Tables. 
These  two  tables  show  that  the  solubility  of  nitrogen  in 
the  oxygen  obeys  Henry's  law  quite  rigidly  up  to  a  molecular 
percentage  of  nearly  70  per  cent.,  but  that  oxygen  does  not 
obey  this  law.     The  value  of  the  quotient  for 
J  l  pressure 
oxygen  varies  in  such  a  way  as  to  point  to  association  of 
oxygen  molecules  when  dissolved  in  nitrogen.  Now  the 
surface-tensions  of  liquid  oxygen,  nitrogen,  argon,  and  carbon 
monoxide  were  determined  by  Baly  and  Donnan  {loc.  cit.), 
and  they  concluded  that  the  pure  liquids  showed  no  asso- 
ciation. But  looking  at  their  results  more  closely,  we  see 
that  the  value  of  the  temperature  coefficient  of  the  molecular 
surface-energy  of  oxygen,  viz.  1*917,  is  not  the  same  as 
that  found  for  the  other  three  gases,  wdrich  all  have  a  coeffi- 
cient nearly  equal  to  2'003.  Baly  and  Donnan  concluded 
that  the  probable  value  of  this  coefficient  at  low  temperatures 
was  too  uncertain  for  any  conclusion  to  be  drawn  as  to 
association  ;  but  the  fact  that  arg-on,  nitrogen,  and  carbon 
monoxide  all  have  the  same  coefficient  points  to  the  value 
