5*28  Dr.  F.  Horton  on  the  Electrical 
The  conclusion  from  the  present  experiments  is  just  the 
opposite  of  this  view,  and  is  that  the  conductivity  of  the  oxides 
investigated  is  mainly  metallic.  It  may  be  that  the  signs 
of  electrolysis  obtained,  in  some  cases,  by  Nernst  corre- 
spond to  the  signs  of  electrolysis  found  with  some  of  the 
oxides  in  the  present  experiments.  With  mixtures  of  oxides 
it  is  probable  that  signs  of  electrolysis  would  be  more  readilv 
given — for  instance,  the  wandering  of  colour  towards  the 
cathode  in  the  case  of  mixtures  containing  coloured  oxides 
(<?.  g.  oxide  of  iron)  may  be  due,  as  Nernst  suggests,  to  the 
coloured  oxide  being  in  solution  in  the  rest  and  giving  rise  to 
the  ordinary  phenomena  of  electrolysis  ;  but  I  do  not  think  it  is 
justifiable  to  conclude  that  because  such  traces  of  electro- 
lysis can  be  detected  therefore  the  whole  conductivity  is 
electrolytic. 
In  the  present  paper  it  is  shown  that  some  oxides  (e.  g.  CuO) 
show  no  trace  of  electrolysis  ;  and  I  think  that  if  the  conduc- 
tion of  electricity  through  these  oxides  is  carried  on  by  means 
of  electrolytic  ions,  some  evidence  of  the  liberation  of  the 
products  of  electrolysis  wrould  have  been  obtained  in  the 
experiment  in  which  a  current  was  passed  through  oxide  of 
copper  whilst  it  was  heated  in  a  vacuum. 
The  distinction  between  electrolytic  and  metallic  con- 
ductivity is  generally  based  on  the  sign  of  the  temperature 
coefficient  of  the  conductivity  of  the  conductor  in  question. 
If  the  conductivity  decreases  with  rise  of  temperature,  the 
conduction  is  taken  to  be  metallic;  if  it  increases  with  rise  of 
temperature,  the  conduction  is  considered  to  be  electrolytic. 
This  rule  is,  however,  one  with  so  many  exceptions  (some  of 
which  have  already  been  stated  at  the  commencement  of  this 
paper)  that  it  is  useless  for  deciding  the  question. 
The  effect  of  increased  temperature  on  the  conductivity  of 
electrolytes  has  been  studied  by  many  observers.  For  aqueous 
solutions  the  experiments  over  the  largest  range  of  tempe- 
rature are  those  by  Noyes  and  Coolidge  *.  These  observers 
have  studied  the  conductivity  of  solutions  of  sodium  and 
potassium  chloride  up  to  306°  C.  They  find  that  for  very 
dilute  solutions  ('0005  normal)  the  conductivity  increases  at 
a  rate  closely  proportional  to  the  increase  of  temperature. 
With  more  concentrated  solutions  the  conductivity  curves 
rise  more  slowly  and  attain  a  maximum,  after  wdiich  increase 
of  temperature  decreases  the  conductivity.  The  authors  show 
that  this  is  due  to  a  decrease  in  the  number  of  dissociated 
*  Noyes  and  Coolidge,  Proc.  Amer.  Acad,  of  Arts  &  Sci.  yoI.  xxxix. 
p.  160  (1903-94). 
