318  VOLUMETRIC  ESTIMATION  OF  TANNIC  AND  GALLIC  ACIDS,  &C. 
hot-filtered  decoction  is  used.  The  results  with  the  decoction 
always  differ  from  those  with  the  substance  itself,  the  latter 
being  higher,  which  seems  to  show  that  a  portion  of  the  tannic 
acid  withstands  the  solvent  action  of  the  water. 
Gall  nuts,  catechu  in  powder  may  be  estimated  in  the  same 
way  as  pure  tannic  acid. 
II.  Iron  Compounds. — These  must  be  reduced  to  the  state 
of  protoxide  by  means  of  zinc,  and  the  excess  of  acid  neu- 
tralized with  caustic  potash  or  soda.  (Ammonia  and  the  car- 
bonated alkalies  must  be  avoided.)  The  solution  is  then  poured 
into  the  absorption-flask  and  pieces  of  potash  wrapped  in  paper 
are  then  dropped  in.  The  absorption  is  complete  in  a  very 
short  time.  For  accuracy  this  process  is  second  to  none,  and 
may  be  recommended  in  preference  to  that  of  Margueritte  and 
Fuchs,  since  it  requires  fewer  precautions.  50  c.c.  of  a  solution 
of  protoxide  which  contained  1  395  Fe  absorbed  in  three  ex- 
periments 148-0  c.c.  148-44  c.c,  and  148  4  c.c.  of  oxygen  at 
19  0.;  the  mean  =  14d-28  c.c,  which  at  this  temperature 
weigh  0-1987  grammes,  answering  to  1-391  grammes  of  iron. 
IU.  Manganese  Compounds. — As  these  are  easily  reduced 
to  the  state  of  protoxide,  the  estimation  is  made  in  the  same 
way  as  in  the  case  of  protoxide  of  iron,  bnt  it  is  necessary  to 
know  to  what  extent  the  protoxide  of  manganese  unites  with 
iron.  Further  experiments  are  required  to  determine  this 
point,  and  the  following  proportional  numbers  are  only  the  re- 
sults of  some  few  experiments  that  have  led  to  the  conclusion 
that  one  part  by  weight  of  absorbed  oxygen  corresponds  to 
4-34  parts  by  weight  of  manganese,  which  may  be  calculated 
as  the  oxide  Mn5  09  In  the  absence  of  more  complete  re- 
searches it  is  superfluous  to  discuss  the  rational  formula  of  the 
resulting  oxide  ;  but  one  thing  is  certain— namely,  that  the  ox- 
idation does  not  proceed  so  far  as  the  state  of  binoxi  le. 
The  estimation  of  manganese  in  the  presence  of  iron  is  per- 
formed iu  two  operations — one  with  the  iron  in  the  state  of 
protoxide,  and  the  absorption  observed  in  the  two  cases.  The 
combination  of  the  two  results  will  give  the  desired  explanation. 
Active  oxygen  may  be  estimated  in  a  similar  way,  by  means 
of  a  solution  of  protoxide  of  iron. 
