856  Determination  of  Morphine.        { Amvelnr\g*8arm' 
that  present  in  the  original  6  Gm.  sample,  if  the  aliquot  parts  taken 
were  those  recommended  above. 
Remarks. 
The  weight  of  sample  and  the  particular  relative  magnitude  of 
the  aliquot  parts  chosen  may  be  varied  accordingly  to  the  circum- 
stances of  the  case,  but  the  above-mentioned  quantities  have  proved 
generally  convenient.  It  is  important  to  note,  however,  that  at  the 
outset  more  than  half  the  total  volume  of  liquid  used  should  consist 
of  cold  saturated  barium  hydroxide  solution,  and  the  morphine  should 
be  precipitated,  in  the  later  stage,  from  the  most  concentrated  solu- 
tion possible. 
Even  under  the  most  favorable  conditions  some  morphine  will 
remain  in  solution,  after  precipitation  of  the  main  quantity.  If  it 
is  desired,  a  correction  (+0.26  Cc.  ^  acid)  derived  from  experi- 
ment 11  (Part  I)  may  be  made  to  the  titration  result.  In  such  case 
it  is  necessary  to  adhere  strictly  to  the  conditions  of  precipitation 
and  washing  here  laid  down.  There  are  decided  objections  to 
making  such  corrections,  and  probably  they  are  best  omitted.  The 
relative  importance  of  this  error  may  be  reduced  by  increasing  the 
weight  of  sample  worked  with. 
Comparison  with  Other  Methods. 
Since  the  results  given  by  most  standard  methods  must  be  con- 
sidered uncertain  to  at  least  the  extent  of  0.5  per  cent.,  serious  com- 
parisons are  somewhat  Jacking  in  interest. 
The  opium  used  in  Experiments  31,  32  and  33  was,  however, 
examined  by  the  method  of  Stevens  (Pharm.  Arch.,  V,  41-45)  modi- 
fied by  doubling  the  quantities  specified  by  him.  The  result  thus 
found  was  10.42  per  cent.,  which  is  materially  less  than  that  given 
by  the  experiments  quoted. 
It  must  be  remembered  that  extraction  by  barium  hydroxide  is 
likely  to  be  more  thorough  than  by  lime,  while  the  analytical  losses 
in  the  new  method  are  notably  less. 
Experimental  Data. 
Examination  of  Opium.  Experiment  — The  material  used 
purported  to  be  a  Persian  opium.    A  sample  of  6.0000  Gm.  was 
1  These  experiments  are  numbered  consecutively  with  those  in  Parts 
I  and  II. 
