422 
PREPARATION  OF  ALUMINIUM,  SODIUM,  ETC. 
kept  closed  by  the  fuel  upon  a  shelf  outside  the  furnace.  These 
openings  are  at  such  a  height  above  the  bars  of  the  furnace,  that 
there  would  be  a  space  of  a  foot  between  the  fuel  and  the  tubes. 
The  fuel  used  was  a  mixture  of  equal  parts  coke  and  coal. 
Upon  the  hearth  of  the  furnace,  the  mixture  for  the  prepa- 
ration of  sodium  may  be  calcined  in  crucibles :  and,  when  the  fur- 
nace is  worked  continuously,  the  temperature  becomes  sufficiently 
high  for  working  some  reduction  cylinders  upon  the  hearth. 
Mr.  Deville  did  not  find  the  furnace  which  he  used  very  well 
adapted  for  the  production  of  potassium,  and  considers  that  it 
would  be  better  if  constructed  after  the  manner  of  a  puddling 
furnace  described  by  M.  Dumas.  He  is  of  opinion  that  there  is 
no  question  as  to  the  practicability  of  the  production  of  sodium 
by  a  continuous  operation,  if  the  furnace  is  constructed  in  a 
suitable  manner. 
The  chloride  of  aluminium  was  prepared  by  the  method  recom- 
mended by  Gay  Lusac  and  Thenard.  Alumina,  quite  free  from 
iron,  was  mixed  with  40  per  cent,  carbon,  and  a  small  quantity 
of  coal,  made  into  a  paste,  and  decomposed  at  a  red  heat.  The 
compact  mass  was  broken  into  fragments,  and  introduced  into 
an  earthen  retort.  This  was  placed  in  a  furnace,  and  heated 
in  redness,  while  a  stream  of  dry  chlorine  was  passed  through 
the  contents.  At  first  a  considerable  amount  of  water  was 
evolved  from  the  aluminous  carbon,  which  is  very  hygroscopic. 
When  the  chloride  of  aluminium  began  to  appear,  an  earthen 
funnel  was  fitted  to  the  neck  of  the  retort  by  means  of  asbestos 
and  clay,  and  a  bell-shaped  vessel  fitted  in  the  same  manner  to 
the  rim  of  the  funnel.  The  chloride  of  aluminium  condenses  in 
this  receiver,  and,  however  rapid  the  stream  of  chlorine  may  be 
during  the  chief  part  of  the  operation,  it  is  so  completely  ab- 
sorbed by  the  aluminous  carbon  that  the  carbonic  oxide  which 
escapes  does  not  give  any  indication  of  chlorine.  However,  it 
contains  a  little  chloride  of  silicum,  arising  from  the  action  of 
the  chlorine  and  carbon  upon  the  material  of  the  retort ;  also, 
some  aluminium,  and,  perhaps,  some  chlorocarbonic  acid.  When 
the  receiver  is  full  of  chloride  of  aluminium,  it  is  removed,  and 
replaced  by  another,  and  so  on  till  the  operation  is  finished. 
The  amount  of  chloride  of  aluminium  obtained  from  10  pounds 
of  alumina  was  rather  more  than  20  pounds.    The  carbonaceous 
