Dielectric  Strength  of  Air.  265 
Table  VIII. 
Compagnie  de  ^Industrie  Electrique.     Plate  and  2  cm.  ball. 
X. 
xja. 
fp  (calc.  by  §  10). 
Y  (obs.). 
is" 
Emax. 
46-8 
0-5 
0-5 
1-36 
10 
1-0 
1-77 
26 
44/6 
1-5 
1-5 
2-21 
31 
44-5 
20 
20 
2-68 
35-5 
46-5 
25 
2-5 
315 
39 
48-1 
30 
3-0 
3-63 
425 
50-4 
4-0 
4-0 
4-60 
48-0 
54-2 
50 
5-0 
5-59 
54-0 
59-5 
6-0 
6-0 
6-57 
58-0 
62-5 
(see 
to  increase  rapidly 
e  first  four  values  gives  45*6  kilo- 
is  beginning 
It  will  be  seen  that  Rmax 
9).  The  mean  of  th 
volts  per  centimetre  as  the  dielectric  strength  of  air.  In 
practice  the  plates  used  are  not  large,  and  so  we  are  only 
justified  in  using  our  formula  for  fp  when  the  plate  and  the 
ball  are  close  together.  We  do  not  attach  much  importance 
to  this  test. 
II.  With  Alternating  Pressures. 
(i.)   C.  P.  Steinmetz.     2  inch  spheres. 
In  a  paper  on  the  u  Dielectric  Strength  of  Air,"  published 
in  the  Transactions  of  the  American  Institute  of  Electrical 
Engineers,  vol.  xv.  p.  281,  Professor  C.  P.  Steinmetz  gives 
the  results  of  an  elaborate  and  careful  research  on  the  dis- 
ruptive voltages  between  pointed,  spherical,  and  cylindrical 
electrodes.  Alternating  voltage  was  used  of  frequency  125, 
and  the  shape  of  the  wave  was  practically  identical  with 
a  sine  curve  when  a  particular  smooth-core  alternator  was 
used.  The  ratio  of  the  maximum  to  the  effective  voltage  in 
all  his  experiments  with  this  machine  was  practically  1*42. 
The  spherical  and  cylindrical  electrodes  were  put  in  nitrate 
of  mercury  and  then  rubbed  with  a  clean  cloth.  When  this 
was  done  it  was  found  that  the  disruptive  discharge,  for 
a  given  distance  apart  of  the  electrodes,  always  took  place  at 
the  same  voltage.  If  the  electrodes  were  merely  polished, 
then  at  small  distances  apart  the  results  were  very  erratic. 
The  accuracy  of  the  results  obtained  probably  lies  well  within 
I  per  cent,  in  most  cases.  In  the  experiments  the  barometer 
varied  from  75'2  to  76*2  cms.  This  variation  introduces  an 
uncertainty  of  about  one  per  cent.  The  voltmeter  readings 
may  be  one  per  cent,  out,  and  there  may  be  a   one  per  cent. 
