382  CRYSTALLIZATION  OF  SUPERSATURATED  SOLUTIONS,  ETC. 
But  does  no  other  substance  possess  this  property?  To  clear 
up  this  synthetic  side  of  the  question,  I  tried  the  action  of  a  great 
many  substances.  Among  two  hundred  and  twenty  I  found 
thirty-nine  possessing  this  property ;  of  these,  eighteen  were 
insoluble.  I  washed  them  in  distilled  water,  and  left  them  on 
Alters  shielded  from  the  dusts  in  the  air ;  when,  after  a  few  days, 
they  were  dry,  I  found  them  without  action  on  the  solution  which 
they  previously  crystallized.  Moreover,  the  lixivium  gave,  with 
chloride  of  barium,  a  precipitate  of  sulphate  of  baryta,  and  con- 
tained soda.  The  twenty-one  soluble  substances  I  purified  by 
recrystallization  with  or  without  the  addition  of  chloride  of 
barium.  Not  one  of  them  preserved  its  action  on  the  supersatu- 
rated solution.  These  results  lead  me  to  believe  that  sulphate 
of  soda  is  the  only  substance  which  acts  on  the  supersaturated 
solution. 
Let  us  now  consider  the  action  of  the  air  in  Gay-Lussac's 
experiment  with  a  tube,  as  it  is  ordinarily  made.  If  the  con- 
clusion at  which  I  have  arrived  be  correct,  it  is  necessary  that  a 
particle  of  sulphate  of  soda  should  penetrate  with  the  air  into  the 
tube,  and  so  determine  the  crystallization.  Now  it  seems  diffi- 
cult to  admit  that  in  the  limited  volume  of  air  entering  into  the 
tube  there  should  always  be  a  particle  of  sulphate  of  soda,  but 
this  experiment  succeeds,  at  the  most,  once  in  six  times,  when 
the  precaution  is  taken  of  washing  the  fine  end  of  the  tube  and 
the  pincers  which  serve  to  break  it  and  to  keep  it  at  a  distance. 
If  it  is  generally  most  successful  in  the  lecture  room,  it  is  because 
the  current  of  air  takes  with  it  particles  of  sulphate  of  soda  which 
have  been  thrown  from  the  tube  during  the  boiling  of  the  liquid, 
and  which  have  become  attached  to  the  outer  surface,  and  there 
crystallized.  However,  this  experiment  is  of  little  use  in  re- 
solving the  question  at  issue,  as  it  brings  into  contact  with  the 
solution  only  a  very  small  volume  of  air.  It  is  better  to  pass  air 
very  rapidly  (one  litre  per  minute)  into  a  supersaturated  solution ; 
then,  while  in  the  laboratory,  a  quarter  of  a  litre  is  sufficient  to 
determine  crystallization ;  in  the  open  country,  sixty  and  even 
eighty  litres  are  often  required.  This  result,  obtained  by  avoid- 
ing all  the  causes  which  might  accidentally  introduce  sulphate  of 
soda,  makes  the  existence  of  sulphate  of  soda  in  the  air  more  and 
more  probable. 
