Recombination  of  Ions  in  Air  and  other  Gases.        479 
was  with  this  purpose  that  the  work  described  in  this  paper 
was  undertaken. 
It  now  appears  that  our  suggestion  was  justified,  but  it 
is  also  clear  that  we  should  have  obtained  larger  currents  if 
we  had  used  a  higher  potential  gradient ;  500  volts  per  cm. 
was  insufficient. 
Consider  the  curves  in  fig.  3.  In  A  is  shown  the  relation 
between  current  and  potential  gradient  up  to  3000  volts  per 
cm.  for  ethyl  chloride  at  56  cm.  pressure,  the  saturation 
current  per  sq.cm.  being  about  3x  10-13.  B  shows  the  same 
relation  in  the  case  of  air  at  atmospheric  pressure,  the 
saturation  current  being  rather  smaller.  Comparison  of  these 
two  shows  how  much  more  difficult  it  is  to  obtain  the  full 
current  in  the  more  complex  gas.  Again,  C  shows  the  results 
of  experiments  in  which  the  depth  of  the  ionization  chamber 
wTas  varied.  The  crosses  refer  to  a  2  mm.  chamber,  the  dots 
in  circles  to  a  4  mm.  chamber.  The  currents  wTere  of  the 
10~13  order.  The  two  sets  of  observations  lie  on  practically 
the  same  curve.  This  shows  that  general  recombination  is 
not  responsible  for  the  lack  of  saturation,  and  that  the  cause 
is  probably  similar  to  that  whose  effects  in  the  case  of  air  have 
been  described  above.  Curves  D  and  E  refer  to  experiments 
in  which  the  chamber  was  maintained  at  the  same  depth, 
2  mm.,  but  the  currents  were  altered  by  varying  the  distance 
of  the  radium.  In  the  former  curve  the  saturation  current  is 
about  10~~ 13,  in  the  latter  six  times  as  much.  In  the  case  of 
the  results  shown  in  C,  D,  and  E,  the  gas  contained  a  certain 
proportion  of  air. 
These  results  all  go  to  show  that  the  form  of  the  curve  for 
ethyl  chloride  is  almost  independent,  as  in  the  case  of  air,  of 
strength  of  current  and  depth  of  ionization  chamber,  wher 
the  ionization  is  small.  But  also,  as  in  the  case  of  air,  it 
depends  greatly  on  the  density  of  the  gas.  F  represents  the 
results  of  experiments  at  a  pressure  of  36  cm.,  and  is  to  be 
compared  with  A.  All  the  conditions,  except  as  regards 
pressure,  were  the  same  for  the  two  curves. 
We  have  also  carried  out  experiments,  similar  to  some  of 
those  just  described,  for  a  mixture  of  carbon  tetrachloride 
and  air,  and  obtained  similar  results.  Although  there  was 
only  5  per  cent,  (by  pressure)  of  the  denser  gas  in  the 
mixture,  yet  the  current  at  a  potential  gradient  of  330  volts 
per  cm.  was  only  82  per  cent,  of  the  saturation  value,  whils 
in  air  under  similar  conditions  it  was  93  per  cent. 
It  is  hardly  surprising  that  initial  recombination  should 
be  more  effective  in  a  complex  gas  than  in  air.  For  the 
molecule  contains  many  atoms,  each  one  of  which  is   just  as 
