of  the  recent  attack  on  the  Atomic  Theory"  505 
nowever,  employ  the  term  law  in  the  sense  in  which  generaliza- 
tion is  above  defined  :  thus  Kepler's  empirical  generalizations 
are  known  as  his  laws ;  while  the  word  theory  is  used  by  some 
in  a  sense  more  nearly  accordant  with  its  derivation,  as  a 
principle  or  system  which  affords  a  point  of  view  of  a  large 
number  of  facts  and  generalizations  at  once.  If  I  understand 
Dr.  Williamson's  essay  aright,  he  usually  employs  the  term  in 
this  sense,  and  as  not  necessarily  entailing  any  hypothesis  at  all, 
although  he  sometimes  uses  it  in  the  sense  of  "  generalization  " 
as  above  defined. 
The  doctrines  propounded  by  Dalton  and  adopted  and  extended 
by  Berzelius  include  the  three  kinds  of  proposition,  viz.  gene- 
ralization, convention,  and  hypothesis,  and  may  be  thus  ex- 
pressed : — 
Generalization. — The  quantitative  composition  of  any  homo- 
geneous body  may  be  represented  with  a  considerable  degree  of 
accuracy  (usually  involving  less  error  than  that  due  to  imperfection 
of  instruments,  impurity  of  specimen,  &c.)  by  taking  and  compa- 
ring simple  multiples  of  certain  fixed  numbers  attached  respect- 
ively to  the  name  of  each  elementary  substance  occurring  in  (i.  e. 
obtainable  from)  the  body  in  question. 
Convention. — This  quantitative  composition  may  be  repre- 
sented more  briefly  still  by  a  system  of  symbols  and  suffixes, 
where  the  symbols  indicate  the  names  of  the  elements  present, 
and  the  suffix  applied  to  any  symbol  indicates  the  multiple  of 
the  fixed  number  mentally  associated  with  that  symbol  which  is 
to  be  taken  in  the  particular  compound  under  consideration. 
This  amounts  to  the  convention  that  all  compounds  are  expres- 
sible by  the  general  formula 
A..B4.CC.D,  .  ..  X,.Yy.Z„ 
where  A,  B,  C,  D  .  .  .  represent  the  symbols  of  the  known  ele- 
ments, and  a,  b,  c,  d .  . .  any  integer  or  zero. 
Hypothesis. — The  observed  facts  summed  up  in  the  generali- 
zation and  expressed  symbolically  by  the  convention  may  be 
accounted  for  by  supposing  that  matter  is  made  up  of  indivi- 
sible portions  of  different  kinds  (atoms),  and  that  the  atom  of 
each  kind  (elementary  atom)  has  a  weight  proportionate  to  the 
fixed  number  attached  to  that  element — a  compound  body  being 
made  up  of  a  number  of  clusters  of  atoms  (molecules),  where  the 
nature  and  number  of  atoms  of  each  given  kind  in  a  cluster  is 
denoted  by  the  symbols  and  attached  suffixes  occurring  in  the 
formula  of  the  body  in  question. 
The  term  atomic  hypothesis,  therefore,  as  I  understand  it  and 
have  employed  it,  includes  two  ideas,  viz.  the  expression  of  some 
proposition  incapable  of  direct  proof,  and,  secondly,  a  particular 
