482     Prof.  W.  H.  Bragg  and  Mr.  R.  D.  Kieeman  on  the 
that  both  those  effects  take  place.  With  the  object  of  helping 
to  a  decision  on  this  point  we  have  begun  a  <et  of  experiment-, 
of  which  those  now  described  are  the  first  examples. 
The  method  used  is  to  measure  the  co-ordinates  of  some 
standard  point  on  the  ionization  curve  of  the  gas  investigated, 
under  different  pressures.  The  point  chosen  is  that  where 
the  side  of  the  Ra  C  curve  is  struck  by  the  top  of  the  curve 
which  belongs  to  the  ex.  particles  of  next  velocity  to  those  of 
Ra  C.  This  point  in  air,  at  760  mm.  and  20°  C,  is  at  a 
height  of  nearly  4'83  cm.  It  is  a  convenient  point  to  choose, 
for  the  following  reasons  : — Being  on  a  part  of  the  Ra  C 
curve  where  no  great  change  in  the  ionization  takes  place 
for  a  considerable  alteration  in  range,  the  measurements 
there  are  usually  pretty  consonant  with  each  other,  even 
though  they  are  taken  quickly,  and  if  several  be  taken  on 
the  Ra  C  curve  they  check  each  other.  The  ordinate  of  this 
point  can  also  be  determined  with  great  precision  by  measur- 
ing two  or  three  points  along  the  top  slope  of  the  curve  of 
Ra  A  (or  emanation,  whichever  it  finally  proves  to  be).  Thus, 
a  few  readings  can  be  quickly  taken  in  succession  which  deter- 
mine the  point  accurately,  and  very  little  leakage  of  air  into 
the  apparatus  takes  place  while  the  experiments  go  on.  This 
is  a  desirable  thing,  because  our  apparatus  leaks  slowly  when 
the  pressure  within  is  much  reduced,  on  account  of  the  large 
number  of  connecting  tubes  and  mechanical  arrangements. 
We  find  that  this  method  is  very  satisfactory.  We  may 
mention  also  that  to  save  time  it  is  not  well,  in  the  case  of 
gases  like  ethyl  chloride,  which  are  at  first  in  the  liquid  form, 
to  admit  any  of  the  liquid  into  the  apparatus,  as  it  takes  so 
long  to  evaporate  completely.  It  is  better  to  let  the  liquid 
evaporate  in  another  chamber,  which  can  be  quite  small,  and 
then  to  take  over  gas  only. 
Fig.  4  shows  the  results  of  some  experiments  with  (J2H;,C1. 
The  curves  shown  are  portions  of  the  ionization  curves  in  this 
gas  at  different  pressures.  In  all  cases  the  apparatus  was 
exhausted  of  air  to  about  10  mm.  pressure,  then  partly  filled 
with  gas,  re-exhausted,  and  filled  ngain  to  the  desired  pres- 
sure. The  observations  were  made  at  once,  those  in  the 
neighbourhood  of  the  standard  point  being  made  first,  so  that 
the  gas  might  be  as  pure  as  possible  whilst  the  important 
readings  were  being  taken.  A  potential  of  900  volts  was 
used  for  the  three  greater  pressures,  and  of  300  for  the  low 
pressure.  The  chamber  was  3  mm.  wide,  and  therefore  these 
potential  gradients  were,  respectively,  3000  and  1000  volts 
per  cm. 
The   results  for  ethvl  chloride  and   for  air  are   contained 
