460         ON  A  PROCESS  OF  FRACTIONAL  CONDENSATION. 
the  room,  having  been  mainly  condensed  in  the  upper  worm, 
and  conducted  thence  down  the  outside  of  the  retort  into  the 
safety-lamp.  This  process  was  still  going  on,  the  lamp  being 
highly-heated  from  the  excess  of  fuel  thus  added  to  it,  but  no 
ignition  took  place  outside  the  lamp.  Although  this  experiment 
was  rather  injudicious,  it  furnishes  a  valuable  test  of  the  effi- 
ciency of  the  safety-lamp  and  furnace. 
Having  described  the  apparatus,  I  now  proceed  to  give  such 
details  of  the  method  ot  conducting  the  separations  as  have 
been  found,  in  my  experience,  most  efficient  and  economical  of 
time.  In  commencing  with  a  crude  mixture  of  unknown  liquids, 
I  deem  it  advisable  to  operate  at  once  on  a  tolerable  large  quan- 
tity of  material,  especially  if  the  constituents  are  supposed  to 
be  numerous,  and  to  omit  chemical  treatment  till  after  the  sepa- 
rations have  so  far  progressed  as  to  indicate  the  number  and 
species  of  bodies  present,  and,  approximately,  their  several  boil- 
ing-points. 
Notwithstanding  the  precautions  taken  to  avoid  loss  from 
evaporation  and  leakage,  I  have  at  times  been  surprised  at  the 
large  waste  of  material  which  has  been  made  apparent  after  a 
long  series  of  operations.  When  it  is  considered,  however,  that 
the  time  required  to  make  a  complete  separation  of  a  very  com- 
plex mixture  of  liquids  must  necessarily  be  very  protracted, 
during  which  more  or  less  of  evaporation  is  constantly  taking 
place,  it  will  be  a  matter  of  no  surprise  that  the  loss  is  so  con- 
siderable. The  quantity  of  material  required  must  depend  also 
on  the  proportions  in  which  the  various  constituents  are  con- 
tained in  the  crude  mixture,  and  upon  their  degree  of  volatility  ; 
but  as  these  cannot  be  known  a  priori,  it  may  suffice  to  make  a 
single  preliminary  distillation  of  a  portion  of  the  mixture,  from 
a  tubulated  retort,  to  ascertain  the  range  of  temperature  within 
which  it  distills,  noting  at  the  same  time  the  proportions  which 
come  over  between  certain  temperatures ;  as,  for  example,  below 
50°  C. ;  between  50°  and  100°,  etc.  ;  from  these  data  one  may 
judge  pretty  nearly  of  the  quantity  which  it  will  be  advisable  to 
take.  It  is  evident  that,  when  very  volatile  bodies  are  present, 
even  in  considerable  proportion,  a  much  larger  quantity  would 
be  required  than  if  the  material  were  but  slightly  volatile    a  the 
