of  the  recent  attack  on  the  Atomic  Theory."  507 
cusses  and  refers  to  in  his  paper  is  something  entirely  different 
from  the  ideas  embodied  and  principles  laid  down  in  what  he 
professes  to  criticise,  the  reply  to  each  of  his  remarks  becomes 
very  simple. 
Thus,  first,  Mr.  Atkinson  considers  that  either  the  experi- 
mental numbers  obtained  in  the  quantitative  analysis  of  a  series 
of  bodies  composed  of  the  same  elements  in  different  propor- 
tions (as,  for  instance,  the  tungsten  chlorides)  must  be  em- 
ployed, or  else  the  atomic  theory  is,  "  unconsciously,  no  doubt/' 
assumed.  It  hence  appears  that  Mr.  Atkinson  has  missed  the 
essence  of  the  paper  he  attempts  to  criticise,  viz.  the  distin- 
guishing between  the  use  of  symbols  to  express  with  more  or 
less  accuracy  observed  facts,  and  the  use  of  an  hypothesis  to  ex- 
plain these' facts:  in  assigning  a  formula  to  a  body  from  the 
results  of  quantitative  analysis,  Dalton's  generalization  only  is 
involved,  his  hypothesis  having  nothing  whatever  to  do  with  the 
matter.  If  what  has  been  already  said  is  not  sufficient  to  con- 
vince the  reader  that  Dalton's  hypothesis  (or,  rather,  modifica- 
tion of  Epicurus' s  hypothesis)  as  to  the  constitution  of  matter  is 
the  result  and  not  the  cause  of  the  generalization  ordinarily 
known  as  the  law  of  multiple  proportions,  no  further  remarks 
could  have  that  effect. 
If  Mr.  Atkinson's  line  of  argument  were  correct,  it  would  be 
impossible  to  assign  any  formula  to  any  thing;  for  on  repeti- 
tion of  an  analysis,  slightly  varying  numbers  are  obtained ;  the 
formula  which  expresses  one  analysis  accurately  would  not  ex- 
press another ;  and  there  is  no  reason  why  one  should  be  taken 
rather  than  the  other.  Again,  does  each  so-called  "  experi- 
mental law"  (generalization),  w7hich  is  not  proved  with  mathe- 
matical accuracy  by  experiment,  involve  the  conception  of  some 
theory  whereby  the  experimental  differences  are  rendered  negli- 
gible? "What  theory,  for  example,  is  it  without  which  the  law 
of  illumination  being  inversely  as  the  square  of  the  distance,  has 
no  real  existence  ?  or  what  theory  is  involved  in  the  experi- 
mental generalization  that  an  electric  current  is  equal  to  the 
electromotive  force  divided  by  the  resistance,  or  in  the  generali- 
zation that  the  volume  of  a  gas  is  inversely  as  the  pressure  to 
which  the  gas  is  subjected  ? 
Precisely  the  same  argument  is  applicable  to  Mr.  Atkinson's 
objection,  that  experiment  does  not  prove  that  two  volumes  of 
each  of  the  vapours  of  several  compounds  containing  the  same 
element  contain  exact  multiples  of  one  volume  of  the  vapour  of 
that  element ;  no  experimental  law,  i.  e.  generalization,  ever  is 
exactly  proved;  the  generalization  simply  records  approximately 
the  result  of  any  experiment,  the  approximation  being  closer 
the  more  carefully  disturbing  causes  are  eliminated. 
