4 1 4  Methods  of  Estimating  Tannins.      { Am^'i87h7a™' 
has  "  been  shown  to  be  both  tedious  and  incorrect,  as  the  solution 
refuses  to  filter,  and  the  first  portions  precipitated  contain  a  far  larger 
proportion  of  tannin  than  do  those  which  fall  towards  the  end." 
Other  chemists  make  the  same  statements,  so  that  I  had  not  thought 
it  necessary  to  repeat  their  experiments.  Very  recently,  however,  I 
have  learnt  that  Mr.  Stoddart,  of  Bristol,  and  Mr.  Dearden,  of  Bury, 
are  again  employing  the  plan,  using  sufficient  alum  to  make  the  pre- 
cipitate coagulate,  and  washing  by  decantation  with  boiling  water. 
Mr.  Stoddart  informs  me  that  the  results  agree  fairly  with  a  modifica- 
tion of  Allen's  lead  method,  which  he  employs.  I  therefore  purpose 
trying  it  at  a  future  time  ;  but  the  results  can  scarcely  be  very  accurate, 
and  it  is  improbable  that  all  tannins  combine  with  gelatin  in  the  same 
proportions.  This  last  objection  is  technically  of  less  importance, 
however,  since  the  power  of  precipitating  gelatin  is  probably  somewhat 
proportionate  to  that  of  making  leather. 
Some  years  ago  Fleck  announced  a  method  depending  on  the  fact 
that  while  tannin,  gallic  acid,  and  coloring  matter  are  all  precipitated 
by  cupric  acetate  solution,  the  two  latter  are  redissolved  by  ammonic 
carbonate.  He  proposed,  therefore,  to  employ  a  standard  copper 
solution,  and  to  estimate  the  excess  by  potassic  cyanide.  Dr.  Watts 
showed  that  this  was  impracticable  (owing  to  the  fact,  as  I  found,  that 
cupric  ammonio-gallate  is  not  blue  but  brown),  but  that  gravimetrically 
some  tannins  might  be  estimated  with  considerable  accuracy,  while 
others  gave  precipitates  more  or  less  soluble  in  the  ammonic  car- 
bonate. The  precipitate  is  complicated,  and  contains  ammonia. 
SchifF  gives  its  formula  in  the  case  of  digallic  acid  as  C14H4Cu2  (NH4)2 
094-OH2  ("Ann.  Ch.  und  Ph."  clxxvi.  171),  which  would  give  1  gram 
of  tannin  (digallic  acid)— 1*54  gram  precipitate,  and  -494  gram  CuO. 
Watts,  employing  the  number  1*489  (deduced  from  the  assumption, 
now  shown  to  be  incorrect,  that  the  salt  was  a  simple  cupric  tannate), 
obtained  analytical  results  fairly  agreeing  with  those  by  gelatin  for 
valonia,  sumach,  divi,  oak  bark,  galls  and  myrobalans  ;  and  also  for 
mimosa,  by  employing  the  number  '2959  instead  of  '489.  All  tannins 
giving  green  precipitates  with  iron  gave  copper  precipitates  more  or  less 
soluble  in  ammonia. 
No  doubt  this  method,  reckoning  the  tannin  as  two-thirds  the  weight 
of  precipitate,  or  twice  that  of  cupric  oxide  left  on  ignition,  would 
give  fair  technical  results  ;  but  it  is  unlikely  that  all  the  various  tannins 
