Experimental  Xegislatton  39
procuring  a  separate  supply  of  pure  drinking
water  for  the  population  of  London,  and  then
finding  that  the  population  would  drink  the  impure ­
  water  ?  Many  other  like  matters  must  be
referred  to  trial,  but  it  is  not  the  purpose  of  this
article  to  present  a  catalogue  of  experimental
reforms,  or  to  follow  the  argument  out  into  all
the  possible  details.
I  am  well  aware  that  social  experiments  must
often  be  subjected  to  various  difficulties,  such  as
the  migration  of  inhabitants,  or  even  the  intentional ­
  frustration  of  the  experiment  by  interested
parties.  I  have  heard  it  said  that  the  prohibition
of  the  liquor  traffic  could  not  be  tried  on  a  small
scale,  because  the  publicans  would  be  sure  to
combine  to  send  liquor  into  the  area.  If  they  did
so,  the  fact  could  readily  be  put  in  evidence,  and
if  they  can  defeat  the  teetotalers  in  detail,  I  am
quite  sure  that  they  will  defeat  them  upon  any
very  great  and  general  measure  like  the  Permissive
Bill.  As  to  migration  of  inhabitants,  it  must  be
provided  against  either  by  suitably  increasing  the
areas  of  experimental  legislation,  or  else  by  collecting ­
  information  as  to  the  amount  and  probable
effects  of  the  migration.  But  the  main  point  of
my  theme  is  to  prove  that  we  cannot  really  plan
out  social  reforms  upon  theoretical  grounds.
General  argument  and  information  of  all  kinds
may  properly  be  employed  in  designing  and  choosing ­
  the  best  experiments,  but  specific  experience