Recombination  of  Ions  in  Air  and  other  Gases.         475 
than  those  already  described  ;  yet  their  form  is  very  similar. 
The  curve  A  shows  results  with  an  8  mm.  ionization  chamber  ; 
curve  B,  4  mm.  The  ionization  was  due  to  a  thin  layer  o£ 
radium,  surmounted  by  a  set  of  vertical  tubes,  as  described 
by  us  in  the  Philosophical  Magazine  for  September  1905. 
Other  experimental  results  may  be  expressed  in  terms  of  I4, 
the  saturation  current  for  4  turns  of  the  screw  (rather  less 
than  4  mm.),  and  I'4  the  current  for  a  potential  gradient  of 
25  volts  per  cm.,  and  the  same  depth  of  chamber.  On  one 
occasion  it  was  found  that  I4/J/4  =  1*17  :  Is/I's  =  1*21;  and 
on  another  I4/I'4  =  1*18  :  Is/I's  =  1*23.  In  these  experi- 
ments the  radium  was  5*05  cm.  from  the  gauze.  When  the 
distance  was  6'25  it  was  found  that  I4/F4  =  1*27  ;  I8/I'8  =  1*29; 
and  again  I4/I'4  =  1*30,  Is/I's  =  1"30.  In  the  latter  cases 
the  <z  rays  did  not  all  get  across  the  chamber  :  possibly  the 
small  variation  of  the  ratios  with  distance  may,  in  some  way, 
be  due  to  this  fact. 
It  might  be  argued  that  we  ought  not  to  find  much  variation 
in  the  lack  of  saturation  when  the  current  is  increased  by 
shooting  a  greater  number  of  a,  particles  across  the  chamber 
in  one  second,  on  the  following  grounds  : — Each  particle  as 
it  flies  across  makes  something  like  105  ions  in  a  centimetre 
of  its  path.  If  there  are  only  about  10 l  or  103  ions  in  a  c.cm. 
at  any  one  time,  it  is  clear  that  these  must  be  all  the  work  of 
one  particle,  and  that  all  the  ions  it  produces  are  cleared  away 
before  the  next  one  comes.  Thus,  the  ions  made  by  one  a 
particle  have  no  chance  of  combining  with  those  made  by 
another,  and  recombination  cannot  be  proportional  to  the 
square  of  the  number  per  c.c.  But  this  consideration,  though 
no  doubt  true,  cannot  furnish  an  explanation  of  the  fact  that 
the  curves  are  little  altered  when  the  chamber  is  altered  in 
depth.  It  was,  indeed,  in  view  of  this  argument  that  we 
made  the  experiments  with  the  varying  depths  of  the 
chamber. 
It  is  very  instructive  to  compare  these  figures  with  the 
results  obtained  by  Betschinsky,  and  described  by  him  in  a 
paper  contained  in  Drude's  Amialen,  No.  8,  1905.  Very 
careful  measurements  have  been  made  by  this  observer  of  the 
relation  between  current  and  potential  gradient  in  the  case 
when  the  currents  are  of  an  order  100  to  1000  times  greater 
than  those  of  the  experiments  described  above.  Curves  D,  E. 
and  F,  in  fig.  1,  are  plotted  from  the  table  on  page  531  in  his 
paper,  being  reduced  to  a  saturation  value  400,  so  as  to  be 
comparable  with  the  other  curves  in  the  same  figure.  It  will 
be  seen  that  in  this  case  the  curves  for  different  widths  of 
the  ionization  chamber  differ  very  widely  at  low  potential 
2  12 
