Force  in  the  Contact  of  Metals.  91 
to  itself,  independent  of  the  temperature  at  the  points  of  solder- 
ing of  the  wires.  It  shifted  slowly,  and  pretty  regularly,  from 
one  cylinder  toward  the  other ;  and  this  movement  did  not  di- 
minish or  cease  until  the  current  had  been  in  continual  circu- 
lation during  five  or  six  hours.  If  a  weaker  current  was  made  to 
follow  an  intense  current  which  had  circulated  some  hours  in  the 
wires,  the  displacement  of  the  index  was  effected  in  the  opposite 
direction.  It  was  only  after  several  experiments  that  I  succeeded 
in  discovering  the  cause  of  this  strange  phenomenon.  In  con- 
sequence of  the  action  of  the  water  on  the  zinc,  one  of  the  cylin- 
ders of  this  metal  had  become  covered  on  the  inside  with  a  very 
thick  and  dense  coating  of  hydrated  oxide  of  zinc.  Now  this 
covering,  being  a  very  bad  conductor  of  heat,  prevented  it  from 
passing  from  the  annular  space  round  the  cylinder  of  copper  to 
the  mass  of  water;  consequently  the  temperature  about  the  cop- 
per cylinder  rose,  and  occasioned  a  continual  increase  of  heat  in 
the  cylinder  itself.  The  zinc  jackets  having  been  opened  and 
cleaned,  the  inconvenience  mentioned  was,  to  a  very  great  extent, 
removed.  This  shows  how  necessary  it  is  that  the  jackets  be  in 
every  respect  perfectly  equal  the  one  to  the  other.  The  best 
procedure  is,  to  coat  their  interior  with  a  metal  that  will  remain 
unaltered.  As  to  the  outer  surfaces,  it  is  easy  to  see  to  their 
remaining  always  alike. 
It  was  easy,  however,  to  eliminate  this  displacement  of  the 
index.  I  observed  that  during  several  hours  it  moved  in  one 
direction  continually  and  with  a  tolerably  uniform  velocity.  The 
experiments,  besides,  demonstrated  that  the  displacement  pro- 
duced by  the  difference  of  temperature  of  the  points  of  contact 
ceased  in  the  space  of  about  three  quarters  of  an  hour.  Profiting 
by  the  above-mentioned  circumstances,  I  always  proceeded  in 
the  following  manner  in  my  experiments  : — After  the  current 
had  been  in  circulation  for  three  quarters  of  an  hour,  I  read  on 
the  scale  the  position  of  the  index.  This  done,  I  reversed  the 
current,  and  the  index  began  to  move  in  the  opposite  direction  ; 
at  the  end  of  another  three  quarters  of  an  hour  the  position  of 
the  index  was  again  read.  The  result  of  these  two  readings  was 
a  deviation  a.  1  then  restored  to  the  current  its  first  direction, 
and  three  quarters  of  an  hour  afterward  I  took  a  fresh  reading  j 
from  this  reading  and  the  immediately  preceding  one  I  obtained 
deviation  b.  I  next  reversed  the  current  once  more ;  and  from 
the  last  reading,  treated  in  the  same  manner  as  the  others,  I  ob- 
tained deviation  c,  in  which  the  index  was  displaced  in  the  same 
direction  as  in  the  first.  Now,  if  we  take  the  mean  of  a  and  c, 
and  then  the  mean  of  this  mean  and  b,  we  obtain  the  deviation 
sought.  It  is  evident  that  this  process  furnishes  a  result  entirely 
independent  of  the  proper  motion  of  the  index,  since  this  motion 
