ON  PREPARATION  OF  ALUMINIUM  FROM  CRYOLITE. 
239 
The  melted  mass  softens  very  slowly,  owing  to  the  slight  solu- 
bility of  fluoride  of  sodium.  This  is  somewhat  lessened  by  the 
chloride  of  potassium.  After  about  twelve  hours  the  mass  was 
crushed  in  a  porcelain  mortar,  and  the  globules  of  aluminium 
separated.     Their  weight  was  from  0.3  to  0.4  or  0.5  grra. 
The  smaller  globules  could  not  easily  be  separated  by  leviga- 
tion  from  the  undecomposed  cryolite  and  the  alumina,  which  was 
always  formed,  on  account  of  the  greater  hardness  of  the  sub- 
stances. This  portion  of  the  mass  was  treated  with  cold  dilute 
nitric  acid,  which,  though  it  did  not  dissolve  the  alumina,  left 
the  globules  of  the  aluminium  with  a  brilliant  surface.  After 
drying,  the  globules  were  separated  from  the  alumina  and  cryolite 
powder  by  rubbing  upon  muslin. 
Owing  to  the  formation  of  an  almost  imperceptible  film  of 
oxide,  which  is  formed  upon  them  when  heated,  the  small  glo- 
bules of  aluminium  require  to  be  melted  with  a  flux,  in  order  to 
obtain  a  mass.  There  is  always  some  loss  in  melting  aluminium 
with  chloride  of  potassium.  A  globule  of  the  metal,  weighing 
3.85  grim,  lost  in  this  way  0.05  grm.  The  chloride  of  potas- 
sium did  not  contain  any  aluminium  when  dissolved  in  water, 
but  a  small  quantity  remained  undissolved.  There  is  no  doubt 
that  a  portion  of  the  aluminium  decomposed  chloride  of  potas- 
sium, and  that  potassium  and  chloride  of  aluminium  were  vola- 
tilized, as  is  the  case  when  copper  and  even  silver  are  melted  with 
chloride  of  potassium. 
When  the  globules  of  aluminium  were  melted  under  the  double 
chloride  of  aluminium  and  sodium,  as  recommended  by  Deville, 
the  loss,  if  there  is  any,  is  very  slight. 
When  the  metal  is  melted  under  chloride  of  potassium,  the 
surface  is  not  perfectly  even,  but  presents  small  indentations, 
which  is  not  the  case  when  it  is  melted  under  the  double  chloride 
of  aluminium  and  sodium. 
The  best  method  of  preparing  double  chloride  of  aluminium 
and  sodium  is  to  place  a  mixture  of  alumina  and  carbon  in  a 
glass  tube  of  considerable  diameter,  within  which  is  a  narrower 
tube  containing  chloride  of  sodium,  and,  while  passing  a  stream 
of  chlorine  through  the  largest  tube,  to  apply  a  very  strong  heat 
to  that  part  containing  the  alumina  and  carbon,  and  a  less  heat 
to  the  part  containing  chloride  of  sodium.  The  vapor  of  chloride 
