ON  A  PROCESS  OF  FRACTIONAL  CONDENSATION.  463 
generally  obtained  for  the  same  number  of  degrees  of  tem- 
perature throughout  the  operation.  In  other  mixtures,  in 
which  certain  bodies  may  seem  to  be  present  in  much  larger 
proportion  than  others,  or  in  which  there  may  be  a  greater 
difference  between  the  boiling-points  of  the  constituents  than  in 
the  cases  referred  to, — facts  which  would  be  indicated  by  the 
thermometer  of  the  retort,  and  by  the  relative  quantities  of  the 
products  obtained, — there  might  be  something  gained  by  ex- 
ercising discretion  in  taking  off  fractions  according  to  these  in- 
dications. 
In  the  second  series  of  fractioning,  the  first  or  lowest  fraction 
of  the  preceding  series,  which  is  large  enough  to  operate  upon 
by  itself,  is  transferred  to  the  retort,  and  brought  into  ebullition. 
The  temperature  of  the  bath  is  then  adjusted  as  above  described, 
and  the  distillation  continued,  the  fractions  obtained  being  placed 
in  their  appropriate  bottles  until  the  temperature  of  the  retort 
shall  have  risen  to,  or  somewhat  above,  the  point  at  which  the 
second  or  next  succeeding  fraction  of  the  first  series  may  be 
supposed,  or  has  been  found  by  experiment,  to  boil.  This  fraction 
is  then  added  to  the  residue  in  the  retort,  and  the  distilla- 
tion is  continued  as  before.  In  the  same  manner,  I  proceed  with 
the  remaining  fractions  of  the  first  series. 
All  subsequent  fractions  are  similarly  conducted.  As  the 
work  progresses,  however,  the  fractions  are  taken  for  a  gradually 
decreasing  number  of  degrees  of  temperature,  until  finally  it 
becomes  necessary,  for  the  attainment  of  absolute  constancy  of 
boiling-point,  to  take  off  a  fraction  of  every  degree,  centigrade  ; 
and  to  continue  thus  to  operate  on  these  fractions,  each  repre- 
senting one  degree  of  temperature,  until  the  desired  end  is  at- 
tained. 
The  operator  will  observe  that,  in  each  series  of  fractions,  in 
which  each  fraction  has  been  taken  for  the  same  range  of 
temperature,  the  difference  between  the  boiling-points  of  any 
two  contiguous  fractions  is  nearly  the  same  as  the  difference  be- 
tween any  other  two  contiguous  fractions. — in  other  words,  that 
the  difference  referred  to  approximates  to  a  common  difference 
throughout  the  same  series.  Once  ascertained,  this  difference 
serves  as  a,  valuable  guide  in  determining  with  sufficient  ac- 
