45^  Note  on,  Assay  of  Formaldehyde.       \  Jom-iph^ini. 
'  •>  \      (»itobt'r,  1911. 
formaldehyde  is  treated  with  sodium  liydroxide  in  pressure  bottles, 
the  hydrogen  peroxide  method  of  Blank  and  Finkenbeiner,  and  the 
iodometric  method  of  Romijn. 
On  the  other  hand,  the  meaning  of  the  objection  advanced  by 
Smith  against  the  use  of  the  potassium  cyanide  method  for  deter- 
mining formaldehyde  in  strong  solutions,  namely,  "  that  in  diluting 
strong  solutions  to  the  range  of  these  methods,  a  small  error  in 
weighing  may  be  considerably  multiplied,"  is  difficult  to  understand. 
For  supposing  that  i  Gm.  of  the  formaldehyde  solution  is  weighed 
out,  diluted  with  distilled  water  to  loo  c.c.  and  25  c.c.  of  the 
resulting  solution  taken  for  the  analysis,  and.  supposing  that  the 
delicacy  of  the  balance  used  is  only  ±  0.001  Gm.,  then  the  actual 
amount  taken  for  the  analysis  would  be  0.250  Gm.  ±  0.00025  Gm., 
involving  therefore  a  possible  error  of  ±  o.i  per  cent.  Now  if  the 
step  of  diluting  and  taking  for  the  analysis  an  aliquot  portion  were 
omitted  and  the  total  amount  weighed  out  were  used  for  the  analysis, 
the  amount  so  taken  would  be  i.  0.001  Gm.,  the  possible  error 
involved  would  therefore  be  exactly  the  same  as  before,  namely, 
±  0.1  per  cent.  It  is  possible  that  "  error  in  weighing"  is  an  over- 
sight, the  intention  having  been — error  in  measuring.  If  this  was 
the  intention,  we  can  readily  see  how  an  error  might  be  introduced 
through  an  error  in  measuring.  For  supposing  that  instead  of  using 
for  the  analysis  the  entire  amount  weighed  out,  it  be  diluted  to  a 
definite  volume  and  5  c.c.  of  this  solution  taken  for  the  analysis. 
In  measuring  out  the  5  c.c.  there  might  be  an  error,  say  of  ±  0.05  c.c. 
This  would  introduce  an  error  of  ^  i  per  cent.,  which  would  have 
been  avoided  if  the  analysis  had  been  carried  out  directly  on  the 
weighed  amount  of  formaldehyde  solution.  But  even  in  this  case 
the  error  would  not  be  "  multiplied,"  unless  more  than  one  dilution 
be  made.  Moreover,  if  instead  of  using  only  5  c.c.  of  the  diluted 
solution  a  larger  volume  be  taken,  say  25  c.c,  an  equal  absolute 
error  in  the  measuring  would,  of  course,  relatively  be  only  one-fifth, 
or  reducing  the  possible  error  from  ±  i  to  ±  0.2  per  cent.  In 
other  words,  a  procedure  which  would  involve  the  dilution  of  a  cer- 
tain weight  of  the  strong  formaldehyde  solution  to  100  c.c.  and  the 
use  of  25  c.c.  of  the  diluted  solution  for  the  analysis,  would  ordi- 
narily involve  no  greater  error  than  the  inherent  errors  of  volumetric 
analysis.  And  since  all  the  methods  compared  are  volumetric,  any 
such  errors  may  be  considered  equally  possible  in  all  of  them.  For 
supposing  that  in  measuring  out  the  25  c.c.  of  the  diluted  formalde- 
