and  the  Magnetic  Relations  of  the  Heavenly  Bodies.        501 
observations  to  the  north-south  line,  for  which  the  position  of  the 
galvanometer  was  recorded  ten  times  daily  during  a  whole  year. 
The  monthly  means  of  this  series  of  observations  were  then 
compared  with  the  simultaneous  readings  of  the  bifilar  instru- 
ment, and  with  those  of  the  atmospheric  electricity.  The  result 
was,  that  the  turning-points  of  all  three  classes  of  phenomena  cor- 
responded to  each  other. 
These  facts  of  observation  will  be  sufficient  to  show  their  close 
connexion,  first,  with  my  physical  theory  of  the  origin  of  the 
earth's  magnetism,  and,  secondly,  with  the  experiments  to  be  de- 
scribed presently. 
I  took  a  galvanometer  made  by  Sauerwald  for  the  observa- 
tion of  thermo-currents  with  a  reflectiug-mirror,  and  inserted  the 
ends  of  the  copper  wire  into  a  gutta-percha  tube,  into  which  I 
introduced  a  current  of  water  from  the  water-pipes.  The  water, 
after  having  passed  through  the  tube,  was  made  to  flow  into  a 
vessel  partly  rilled  with  water  and  not  insulated.  The  galvano- 
meter always  showed  a  deflection  of  several  degrees,  indicating 
a  current  in  the  direction  of  the  flowing  water. 
The  wider  apart  from  each  other  I  immersed  the  ends  of  the  cop- 
per wire  in  the  water,  the  greater  teas  the  deflection,  so  that  the 
whole  streaming  mass  must  nave  acted  in  all  its  parts  similarly  to  a 
voltaic  battery  if  the  observed  deflection  was  not  due  to  a  secondary 
current. 
After  my  expectations  had  been  fulfilled  in  this  manner  by 
the  above  rough  experiments,  I  changed  the  mode  of  experi- 
mentation. 
The  following  is  a  description  of  such  an  experiment,  with  nu- 
merical results. 
I  used  a  glass  tube  500  miliims.  loug  and  8  millims.  in  dia- 
meter, with  two  holes  in  its  sides  at  a  distance  of  380  millims. 
from  each  other.  The  holes  were  closed  air-tight  by  a  cork,  by 
means  of  which  two  thin  copper  plates  were  introduced  into  the 
tube  in  such  a  way  that  they  were  washed  by  the  flowing  water. 
The  two  ends  of  the  glass  tube  were  joined  to  gutta-percha  tubes 
of  about  its  own  length,  either  of  which  could  be  attached  to  the 
water-pipes,  so  that  the  current  of  the  water  could  be  reversed. 
The  following  are  the  readings  of  the  galvanometer  in  three 
experiments  : — 
"Water  not  flowing.         Water  flowing.  Difference. 
3699  3620  -7-9 
370-0  362-0  -8-0 
3710  3630  -8-0 
Distance  of  the  galvanometer  from  the  scale  2500  millims. 
In  a  second  experiment  the  copper  plates  were  not  at  all  in- 
