264  Mr.  A.  Russell  on  the 
experiment  is  quite  satisfactory.  The  three  final  values  for 
Umax,  obtained  in  this  table  closely  agree  with  the  mean  of 
the  values  we  deduced  from  Heydweiller's  test.  Considerable 
weight,  therefore,  must  be  attached  to  the  results  of  these 
experiments  in  determining  the  value  of  Rmax.. 
(iv.)  J.  Joubert  and  G.  Carey  Foster.      1  cm.  and 
2  cm.  spheres  (a  =  0'5  &  1). 
In  Foster  and  Porter's  f Joubert's)  'Electricity  and  Mag- 
netism/ p.  135,  tables  of  the  sparking-distances  between 
1  centimetre,  and  2  centimetre  spheres  are  given.  The 
results  for  the  1  cm.  spheres  are  taken  from  Joubert's  Traite 
elementaire  d'electricite  (2nd  edit.),  and  those  for  the  2  cm. 
spheres  were  obtained  by  G.  Carey  Foster.  An  analysis  of 
the  table  for  the  1  centimetre  spheres  shows  that  if  we 
calculate  Rmax.  for  sparking-distances  of  5,  10,  and  15  cms.. 
on  the  assumption  that  the  air  round  the  electrodes  is  not 
broken  down  to  any  appreciable  depth  before  the  discharge 
occurs,  the  values  are  much  too  large.  This  is  in  accord  with 
the  conclusion  of  §  9.  The  mean  of  the  values  up  to  a 
distance  of  2  cms.  apart  makes  Rmax.  42*8.  The  mean  of  the 
values  for  the  2  cm.  spheres  makes  Rmaz.  42*9. 
(v.)   E.  Hospitalier.     1  cm.  spheres. 
An  analysis  of  the  experimental  results  given  by  E.  Hos- 
pitalier in  the  Formulaire  de  V  Electricien,  21st  year,  1904, 
p.  289,  for  the  sparking-distances  between  two  electrodes,  each 
one  cm.  in  diameter,  shows  that  the  potentials  of  the  spheres 
are  not  +Y/2  and  —  V/2  at  the  instant  of  the  discharge.  The 
values  of  Rmax.  calculated  on  this  assumption  diminish  steadily 
from  the  maximum  value  44'1  when  the  spheres  are  0'6  of  a 
cm.  apart  to  40*0  when  they  are  2  cms.  apart.  The  values 
of  /,  however,  are  little  affected  by  the  absolute  values  of 
the  potentials  of  the  electrodes,  provided  that  xja  is  not 
greater  than  0*3.  Taking,  therefore,  the  mean  of  the  first 
three  results  given,  we  find  that  Rmax.  is  42*2. 
(vi.)  Compagnie  de  V Industrie  Electrique. 
Plate  and  sphere. 
The  Compagnie  de  ^Industrie  Electrique  et  Mecanique 
have  published  tests  *  on  the  disruptive  voltages  between  a 
plate  and  a  ball. 
*  Turner  and  Hobart,  '  Insulation  of  Electric  Machines/  p.  -33  (1905). 
