396  Caustic  Soda  Sold  as  Caustic  Potash.  {Axi^SS^SSm- 
a  precipitate.  Heating  is  then  continued  below  boiling,  until  the 
precipitate  settles,  leaving  the  supernatant  liquid  clear.  A  drop  of 
the  solution  is  then  added,  the  precipitate  allowed  to  settle  again, 
and  this  continued  until  no  further  precipitation  takes  place.  If 
haste  is  necessary,  a  little  of  the  supernatant  solution,  still  containing 
some  of  the  fine  precipitate  in  suspension,  may  be  poured  out  in 
two  test  tubes,  a  drop  of  barium  solution  added  to  one,  and  the 
difference  in  turbidity  noted.  If  the  precipitation  is  not  complete, 
the  contents  of  the  tubes  are  returned  and  the  procedure  repeated 
as  often  as  necessary.  With  a  little  practice,  fairly  correct  results 
can  in  this  way  be  obtained  very  quickly.  Each  cubic  centimeter 
of  normal  barium  chloride  solution  corresponds  to  0-087  gramme  of 
potassium  sulphate. 
To  test  potash  for  silica,  the  U.S  P.  of  1890  gives  the  following 
directions:  "  If  I  gramme  be  dissolved  in  2  c.c.  of  water  and  added 
to  10  c.c.  of  alcohol,  not  more  than  a  slight  colorless  precipitate 
should  occur  within  ten  minutes  (limit  of  silicate)."  The  same  solu- 
tion is  then  directed  to  be  tested  for  carbonate,  as  follows :  "  After 
boiling  this  alcoholic  solution  with  5  c.c.  of  lime  water  and  filtering, 
not  the  slightest  effervescence  should  take  place  on  adding  the  fil- 
trate to  an  excess  of  dilute  hydrochloric  acid."  Any  sample  con- 
taining an  appreciable  amount  of  carbonate  will  give  a  precipitate 
when  tested  as  above  for  silica,  although  it  may  be  free  from  the 
latter,  since  alkali  carbonates  are  insoluble  in  alcohol.  Potassium 
carbonate  sometimes  precipitates  in  alcohol  in  a  flocculent  form, 
and  may  then  be  mistaken  for  silica.  The  test  for  silicate  and 
that  for  carbonate  are  thus  both  vitiated,  as  the  greater  part  of 
the  carbonate  is  removed  from  the  solution  before  testing  for  it,  and 
the  sample  can  hardly  fail  to  stand  this  second  test.  It  would  be 
preferable  to  perform  the  limewater  test  for  carbonate  with  an 
aqueous  solution  of  the  sample,  or  to  test  for  it  in  the  manner  indi- 
cated in  the  preceding  pages.  The  most  reliable  test  for  silica  is 
probably  the  one  based  on  the  fact  that  it  is  rendered  insol.uble 
by  evaporating  to  dryness  with  an  excess  of  hydrochloric  acid  in  a 
porcelain  or  platinum  dish,  and  heating  for  a  while  at  an  elevated 
temperature.  Upon  cooling,  then  moistening  the  residue  with 
strong  hydrochloric  acid,  and,  after  letting  stand  a  while  dissolving 
in  hot  water,  the  silica  remains  undissolved.  For  quantitative  esti- 
mations it  is  filtered  out,  washed,  ignited  and  weighed. 
