METALIC  OXIDES  IN  COMMERCIAL  AMMONIA.  165 
of  copper  was  detected,  representing  0-0368  per  cent.  In  the 
filtrate  from  the  sulphide  of  copper  the  iron  was  estimated  as 
before  stated;  it  amounted  to  2-56  grs.  or  0-0853  per  cent. 
Experiment  6  A  sixth  sample  of  ammonia  examined  like 
No.  5,  gave  no  proof  of  the  presence  of  copper,  but  showed  the 
sesquioxide  of  iron  to  equal  0-97  grs.  per  1000  grs.  or  0-097  p.c. 
The  above  cited  experiments  show  that  the  solvent  property 
of  commercial  ammonia  is  not  confined  to,  though  greater  in, 
those  specimens  perceptibly  colored  by  foreign  matter.  I  believe, 
however,  that  the  solubility  of  sesquioxide  of  iron  is  in  all  cases 
due  to  such  extraneous  matter,  which  may  vary  in  quality  as 
well  as  quantity.  That  such  is  the  correct  view  of  the  matter 
is  substantiated  by  the  following  experimental  results  : — 
(a)  Ammonia,  prepared  chemically  pure,  was  poured  to  a 
liberal  excess  into  a  solution  of  sesquichloride  of  iron,  and  the 
mixture  allowed  to  stand  about  twelve  hours.  After  this,  the 
precipitated  oxide  was  removed  by  quick  filtration,  and  1000 
grs.  of  the  solution  treated  for  iron  with  sulphide  of  hydrogen. 
Not  the  slightest  change  was  affected.  Evidently,  therefore, 
absolutely  pure  ammonia  may  be  safely  used  for  the  precipita- 
tion of  iron. 
(b)  The  last  experiment  was  repeated,  and  the  solution  sepa- 
rated and  tested  without  delay,  but  without  detecting  a  trace  of 
dissolved  iron. 
It  may  therefore  be  inferred,  that  the  iron  dissolved  is  due, 
not  to  the  action  of  ammonia,  but  to  impurities  mixed  with  it, 
and  that  the  latter  are  of  an  organic  nature.  The  influence  of 
organic  substances  in  effecting  solution  and  otherwise  masking 
chemical  reactions  is  by  no  means  new.  That  the  fact  deserves 
prominence  in  connection  with  the  use  of  ammonia  as  a  precipi- 
tant in  analysis  is  borne  out  by  the  following  cases. 
(A)  The  ammonia  employed  in  experiment  1st  was  used  to 
precipitate  sesquichloride  of  iron,  and  the  mixture  set  apart  for 
about  six  hours.  When  the  decanted  solution  was  tested  for 
iron,  the  amount  indicated  was  0-89  per  centage.  The  solution 
of  this  amount  of  sesquioxide  of  iron  must  be  attributed  to  the 
excess  of  organic  matter  present. 
(B)  To  a  solution  of  alumina  in  hydrochloric  acid  was  added 
a  slight  excess  of  the  ammonia  examined  in  experiment  2nd ;  the 
