AmAJu0gU,ri87h7!rin'}      Methods  of  Estimating  Tannins.  413 
that  it  is  extremely  difficult  to  absorb  the  whole  of  the  tannin,  that  the 
first  portion  of  liquor  which  passes  is  invariably  lighter  than  succeeding 
portions,  and  that  the  sources  of  error  are  so  large  in  proportion  to  the 
quantities  to  be  measured,  that  the  results  are  of  little  practical  value. 
In  proof  of  this,  I  may  mention  that  a  series  of  nine  analyses  of  the 
same  sumach,  well  mixed,  and  kept  in  a  tightly-corked  bottle,  gave 
results  varying  from  18  to  28  per  cent.,  and  a  mean  error  for  each 
single  experiment  of  3*15  per  cent.,  or  upwards  of  13  per  cent,  of  the 
total  tannin,  while  the  mean  value — 23*9  per  cent. — was  probably 
itself  too  high.  In  these  analyses  the  utmost  care  was  taken,  and  in 
each  case  the  absence  of  tannin  in  the  filtrate  was  proved  by  gelatin. 
If  we  assume  that  tannin  is  worth  20s.  per  ton  per  cent.,  which  is  not 
far  from  the  truth,  the  chemical  valuation  of  this  sumach  would  vary 
from  <£i8  to  X28  per  ton,  with  an  average  error  of  £3  3s.,  and  obviously 
is  far  more  erroneous  than  the  merest  guess.  If  any  further  proof  of 
the  inaccuracy  of  the  method  is  needed,  I  may  quote  the  results  of  a 
series  of  twelve  analyses  of  a  valonia  by  Mr.  W.  N.  Evans,  who, 
perhaps,  has  had  more  practice  with  the  tan-tester  than  any  other  man 
in  England.  The  average  error  exceeds  ten  per  cent,  of  the  whole 
quantity  of  tannin,  and  the  money  values  vary  from,  say,  £17  5s.  to 
X28  10s.  per  ton. 
The  older  method  of  Hammer,  in  which  absorption  by  hide  raspings 
takes  the  place  of  the  raw  hide-filter,  is  in  my  experience  still  more 
inaccurate  ;  and  I  cannot  say  that  the  slight  modifications  proposed  by 
Nickerson  are  any  improvement. 
Another  method  which  has  long  been  in  use  is  precipitation  by 
volumetric  solution  of  gelatin  and  alum.  With  a  solution  of  five 
grms.  of  gelatin  per  litre  I  found  it  impossible  to  say  whether  tannin 
or  gelatin  was  in  excess,  with  less  differences  than  about  4  per  cent,  of 
the  total  quantity  employed,  and  then  the  reactions  were  somewhat 
doubtful.  These  experiments  were  made  with  pure  tannin  j  with 
catechu  or  gambier  the  uncertainty  would  be  far  wider,  and  with  used 
tan-liquors  it  would  be  worse  still.  Under  favorable  circumstances 
and  with  great  patience  it  is  possible  to  obtain  rough  estimates  by  this 
method  ;  but  this  is  all  I  can  say. 
Concerning  Sir  H.  Davy's  still  older  method  of  precipitating  with 
gelatin,  filtering,  drying,  and  weighing,  and  reckoning  four-tenths  of 
the  whole  as  pure  tannin,  Dr.  J.  Watts  says  ("Phar.  Jour."  viii.,  5 17)  it 
