Sparking-Polentials  in  Carbon  dioxide  and  Nitrogen.      547 
traces  of  oxygen,  since  i£  the  gas  was  passed  repeatedly 
through  a  liquid  alloy  o£  sodium  and  potassium,  a  constant 
minimum  potential  o£  251  volts  was  obtained.  The  volume 
of  the  apparatus  used  was  small,  and  before  determining  a 
sparking-potential  the  gas  was  sparked  through  vigorously. 
If  the  apparatus  had  previously  been  used  for  the  experiments 
on  hydrogen,  it  is  possible  that  the  sparking  would  drive  out 
some  hydrogen  occluded  in  them,  which  in  a  small  volume 
might  form  an  appreciable  percentage  of  the  gas  present. 
This  would  tend  to  lower  the  spark-potential  and  make  its 
determination  uncertain.  In  the  determinations  of  spark- 
potentials  described  in  this  paper  it  was  found  that  in  the 
neighbourhood  of  the  minimum  spark-potential,  the  spark- 
potential  was  sometimes  affected  to  the  extent  of  6  or  8  volts 
by  the  previous  passage  of  a  current  through  the  gas,  and 
before  the  potential  could  be  again  determined  some  time 
must  elapse  if  the  results  were  to  be  concordant,  so  that  all 
potentials  were  determined  without  sparking  having  taken 
place  immediately  before. 
It  will  be  seen  by  comparison  of  the  tables  given  for 
nitrogen  prepared  by  the  methods  mentioned,  that  there  is 
not  a  great  difference  between  the  results  obtained  in  both 
cases.  The  tables  of  sparking-potentials  show  also  that  a 
small  percentage  of  oxygen  only  produces  small  differences 
in  the  sparking-potentials,  and  that  5  per  cent,  produces 
more  effect  than  1  per  cent.  These  facts  seem  to  show  that 
small  quantities  of  impurities  do  not  produce  disproportionate 
effects.  If  this  be  true,  the  variations  in  the  spark-potential 
of  nitrogen  are  not  produced  by  minute  traces  of  impurity 
but  are  due  to  other  causes.  Further  experiments  in  this 
direction  are  to  be  attempted  shortly. 
The  accompanying  curves,  figs.  1  and  2,  show  the  relation 
between  the  sparking-potential  and  the  product  of  the  pressure 
of  the  gas  p  and  the  distance  between  the  plates  d,  for 
carbon  dioxide  and  nitrogen.  As  the  spark-potential  deter- 
mined with  ultra-violet  light  falling  on  the  negative  electrod  e 
was  in  most  cases  nearly  the  same  as  that  determined  without 
the  light,  one  curve  only  has  been  drawn  in  each  case,  through 
the  points  representing  the  potentials  obtained  with  the 
light. 
The  theoretical  values  of  the  spark-potential  calculated 
from  the  formula  before  given  are  also  indicated. 
