Dielectric  Strength  of  Air.  239 
and  alternating  pressures,  that  the  limiting  value  to  which 
the  numbers  approach  is  38  kilovolts  per  centimetre. 
When  the  electrodes  are  small,  or  when  the  disruptive 
voltages  are  only  a  few  kilovolts,  the  numbers  obtained  in 
this  way  differ  largely  from  38  kilovolts  per  centimetre.  It 
is  therefore  necessary  to  explain  why  this  is  the  case. 
It  will  be  shown  in  what  follows  that  when  the  electrodes  are 
small,  the  air  surrounding  them  may  have  broken  down  and 
become  a  conductor  at  voltages  which  are  only  a  fraction  of 
the  disruptive  voltage.  In  this  case  luminous  effects  are 
generally  observed  at  the  electrodes.  When  a  high  alter- 
nating pressure,  less  than  the  disruptive  voltage,  is  maintained 
between  small  electrodes  a  few  inches  apart,  each  electrode, 
when  the  P.D.  is  sufficiently  high,  is  seen  surrounded  by  a 
faintly  luminous  enveloping  cloud  of  a  bluish  colour,  which 
apparently  does  not  touch  the  conductor  it  envelopes.  We 
shall  call  this  cloud  the  corona.  As  the  pressure  is  increased, 
short  violet  streamers  are  seen  issuing  outwards  from  the 
corona,  the  space  immediately  outside  it  being  the  seat  of 
great  electrical  activity.  At  higher  pressures  the  streamers 
are  longer,  and  a  hissing  noise  is  heard.  When  the  potential- 
difference  between  the  electrodes  approaches  the  disruptive 
value,  sparks  take  place  between  them,  and  finally,  when  all 
the  air  is  broken  down,  an  arc  is  suddenly  established. 
Now,  when  luminous  effects  make  their  appearance,  it  is 
obvious  that  the  boundaries  of  the  Faraday-tubes  are  altered, 
and,  consequently,  that  the  electric  field  is  different  from 
what  it  is  at  low  voltages.  We  cannot  apply  formulae,  there- 
fore, which  have  been  obtained  on  the  assumption  that  the 
distribution  of  the  tubes  is  the  same  as  that  for  low  pressures. 
We  have  not  attempted  to  deduce  formulas  which  will  give 
the  dielectric  strength  of  air  from  the  disruptive  voltage 
between  two  electrodes  surrounded  with  coronae,  as  the  space 
occupied  by  the  brush  discharges  is  not  clearly  defined. 
There  are  many  cases,  however,  when  there  are  no  luminous 
effects  and  where  a  disruptive  discharge  ensues  the  moment 
that  the  dielectric  stress  attains  the  breaking-down  value. 
We  have  deduced  the  dielectric  strength  of  air  from  the 
experimental  results  obtained  in  these  cases. 
Many  electricians  consider  that  a  disruptive  discharge 
always  occurs  the  moment  the  electric  stress  at  any  point  of 
the  dielectric  between  the  two  electrodes  attains  a  certain 
maximum  value.  In  what  follows,  however,  we  show  that  in 
many  cases,  when  some  of  the  air  round  an  electrode  breaks 
down,  the  new  value  of  the  "  maximum  electric  intensity  " 
at  the  boundary  of  the  broken-down  air  is  less  than  the  old 
