Am.  Jour,  fharm. ) 
March,  1909.  f 
Determination  of  Alcohol. 
139 
a  corresponding  increase  in  deviation  from  the  theoretical.  It  is 
significant  that  in  the  case  of  Solution  C  results  close  to  theory 
were  obtained,  while  in  the  case  of  Solution  A  a  deviation  of  more 
than  2.5  per  cent,  below  theoretical  was  observed.  This  is  undoubt- 
edly due  to  the  fact  that  in  each  case  50  c.c.  were  taken  for  the 
experiment,  regardless  of  whether  it  was  approximately  75,  50  or 
25  per  cent,  in  alcohol  strength,  and  it  is  evident  that  when  the 
distillation  from  a  solution  containing  much  extractive  is  continued 
for  a  considerable  time  after  the  alcohol  has  probably  all  been 
recovered,  in  order  to  get  nearly  100  c.c.  of  distillate,  sufficient  non- 
alcoholic material  is  carried  over,  mechanically  or  otherwise,  to  con- 
siderably vitiate  the  result.  The  femedy  for  this  difficulty  is  to  work 
upon  100  c.c.  instead  of  50  c.c.  in  the  case  of  solution  A,  and  upon 
75  c.c.  instead  of  50  c.c.  in  the  case  of  solution  B. 
The  results  on  Experiment  No.  2  are  not  close,  but  withal  are 
closer  than  by  distillation. 
Results  on  solution  A  in  case  of  Experiment  No.  3,  which  were 
subsequently  repeated  and  confirmed,  were  inexplainably  low,  until 
I  discovered  that  on  dilution  to  47.5  per  cent,  alcoholic  strength  a 
considerable  amount  of  precipitation  had  occurred.  This  completely 
accounts  for  the  low  results,  since  a  precipitation  of  extractive  would 
cause  a  lowering  of  specific  gravity,  which  at  that  point  would  lower 
the  final  result. 
The  results  on  solution  B,  Experiment  No.  3,  show  a  wide  varia- 
tion in  the  "  duplicates,"  thus  illustrating  the  enormous  "  experi- 
mental "  or  "  laboratory  "  error  that  it  is  difficult  to  avoid  in  making 
alcohol  determinations. 
The  results  on  the  taraxacum  seeming  to  justify  further  experi- 
ments, it  was  next  attempted  to  apply  the  modified  method,  described 
as  Experiment  No.  3,  to  fluidextracts  of  buchu,  cubeb  and  sandal- 
wood, which  are  among  those  preparations  already  enumerated  as 
causing  difficulty  in  the  method  by  distillation.  It  was  soon  demon- 
strated, however,  that  a  further  modification  of  the  modified  method 
as  already  described  would  be  necessary  owing  to  the  fact  that  it  is 
impossible  to  transfer  the  residue,  remaining  after  evaporating  the 
alcohol  from  such  preparations,  from  a  dish  to  a  measuring  cylinder, 
with  the  aid  of  water  alone. 
After  many  experiments,  the  following  procedure  was  adopted : 
Fifty  cubic  centimetres  of  the  preparation,  measured  at  a  definite  tempera- 
ture, are  transferred  (in  portions  if  necessary)  to  a  test-tube  having  an  inside 
