Pyrometry  and  Areometry , See.  495 
lengthenings,  the  imperfection  of  the  meafure  might 
become  infeniible  in  the  determination  of  their  relation. 
This  method  is  indeed  very  good  to  compare  with 
each  other  effeCfcs,  the  progreffions  of  which  are  propor- 
tioned; and  one  is  right  to  make  ufe  of  it  whenever  extent 
or  weight  is  concerned.  But  it  is  often  very  deceitful  in 
phylics : for  as  the  co-effeCts  feldom  go  by  proportional 
degrees,  the  more  the  obferved  points  of  their  relations 
are  diftant,  the  more  the  deviations  become  confiderable 
in  the  intermediate  points,  when  they  are  confidered  as 
proportional  to  the  whole.  It  is  thus  that  two  different 
curves,  which  cut  each  other  in  two  points,  deviate  the 
more  from  each  other  in  the  fpace  comprized  between 
the  two  interfeClions,  the  more  diftant  the  points  of  in- 
terfeClion  are  from  each  other.  Now  the  correfpondent 
points,  taken  by  obfervation  in  two  feries  of  phenomena 
\ 
which 
grands  allongemens,  rimperfe&ion  de  la  mefure  devint  infeniible  dans  la  fixation 
de  leur  rapport. 
Cette  methode  feroit  tres  bonne  pour  comparer  entr’eux  des  effets  qui  auroient 
des  marches  proportionnelles ; et  on  l’employe  avec  raifon  qu-and  il  s’agit  de  Feten- 
due  011  des  poids:  mais  elle  eltle  plus  fouvent  fort  trompeufe  en  Phylique.  Car 
des  que  les  co- elfets  marchent  rarement  par  degres  proportionnels,  plus  les  points 
obferves  des  rapports  font  diftans,  plus  les  ecarts  deviennent  grands  dans  les 
points  intermediaires,  en  les  regardant  comme  proportionnels  au  rapport  total. 
C’eft  ainli  que  deux  courbes  difFe: rentes  qui  fe  croifent  en  deux  points,  s’ecartent 
d’autant  plus  1’une  de  l'autre  dans  1’intervalle  des  deux  interfe&ions,  que  ces 
points  de  rencontre  font  plus  eloignes.  Or  les  points  correlpondans  par  obferva- 
7 tion. 
