SUPPLY  AND  DEMAND

87

counteracted.  For  practical  purposes,  therefore, ­
  it  is  probably  safer  to  say  that  supply
prices  are  settled  by  the  cost  per  unit  of
output  of  marginal  firms  ;  remembering,
however,  that  some  theory  of  the  sizes  of
businesses  must  be  brought  in  to  supplement
this  statement.  Cost  per  unit  of  output  in
the  marginal  firm  is,  of  course,  less  likely
than  the  marginal  cost  of  any  business  to
diverge  widely  from  the  results  which  theory
would  lead  us  to  expect  in  a  frictionless
economic  system.
We  must  be  on  our  guard  against  making
our  working  theories  niggling  and  going  to  the
extreme,  so  to  speak,  of  using  a  razor  to  cut
turf.  Nevertheless  it  is  worth  while  taking
notice  of  a  tendency  even  if  it  is  overborne
before  it  attains  its  end.  Its  recognition  may
enable  us  to  account  for  otherwise  puzzling
swells  on  the  surface  of  economic  fact  ;  and
more  important  still,  perhaps,  its  recognition
may  enable  us  to  detect  a  single  law  in  seemingly ­
  disparate  causal  relationships,  a  law
which  is  none  the  less  real  because  in  part  it
is  invariably  counteracted.  These  considerations ­
  apply  to  the  tendency  which  we  have
been  discussing.  It  is  fully  worth  while
knowing  that  a  tendency  exists  which  when
effective  renders  it  possible  to  explain  value