AmAJp°rlnr;5Srm'}     Contributions  from  H.  M.  Gordin.  165 
OH,  and  shaken  out  with  an  immiscible  solvent.  The  final  estima- 
tion is  carried  out  in  the  same  way  as  in  method  A.  The  method 
Bis  practically  the  same  as  that  given  by  Lyons  in  his  "  Hand- 
book," on  page  43.  It  differs  from  Keller's  method  in  so  far  that 
no  water  is  added  for  the  purpose  of  causing  the  drug  to  ball  to- 
gether, and  that  the  liquids  are  taken  by  volume,  not  by  weight. 
This  addition  of  water  is  certainly  a  source  of  indefinite  error. 
Water  takes  up  some  ether  and  together  with  it  some  alkaloid. 
Besides,  water  itself  dissolves  the  alkaloids  to  a  greater  or  less  ex- 
tent. Codeine,  for  example,  is  soluble  in  eighty  parts  water,  atro- 
pine in  135  parts.  . 
But  even  an  alkaloid  which  is  very  difficultly  soluble  both  in 
water  and  ether  is  taken  up  to  a  considerable  extent  by  water  when 
an  ether-chloroform  solution  of  the  alkaloid  is  shaken  with  water. 
The  following  experiment  proves  this  : 
0-102/  gramme  strychnine  was  dissolved  in  10  c.c.  chloroform; 
to  this  solution  30  c.c.  ether  and  10  c.c.  water  was  added  and  the 
whole  shaken  a  minute  or  two  in  a  separator.  After  a  half  hour's 
standing  the  perfectly  clear  lower  aqueous  layer  was  drawn  off  into 
a  tared  aluminum  evaporating  dish,  the  liquid  completely  removed 
by  evaporating  on  the  water-bath,  and  after  drying  and  cooling  at 
1250  C,  the  vessel  again  weighed.  It  was  found  that  0-0044 
gramme  of  strychnine,  or  about  3  per  cent,  of  the  amount  taken, 
was  taken  up  by  the  water. 
As  in  the  assay  of  drugs,  we  very  often  work  upon  quantities  con- 
taining less  than  01  gramme  alkaloid,  and  the  quantity  taken  up  by 
the  water  depends  only  upon  the  amount  of  the  latter,  the  loss  from 
the  solubility  of  the  alkaloid  in  aqueous  layer  is  liable  to  be  even 
greater. 
On  the  other  hand,  the  water  which,  in  Keller's  method,  is  added 
to  ball  the  drug  together,  taking  up  some  ether,  diminishes  the 
volume  of  the  latter,  so  that  by  drawing  off  half  the  amount  of  the 
ethereal  liquid  originally  added,  either  by  weight  or  by  volume,  we 
actually  get  more  than  half  of  the  amount  originally  taken,  and  the 
results  will  be  too  high.  The  influence  of  this  circumstance  was 
also  proved  by  experiment. 
0-1047  gramme  strychnine  was  dissolved  in  50  c.c.  of  a  mixture 
of  three  parts  of  ether  and  one  part  chloroform;  125  c.c.  water  was 
now  added,  the  mixture  shaken,  and  after  complete  separation  into 
