the  a,  Rays  emitted  by  Radium.  121 
It  will  be  noticed  that  within  the  limits  of  each  table,  i.  e. 
within  a  length  of  about  2  cm.,  the  radius  of  curvature  varies 
but  little,  or  at  least  that  the  variation,  if  it  exists,  is  within 
the  limits  of  experimental  error.  The  results  may  be 
summarized  thus  : — 
Value  of  product 
EH=^2HX600 
No.  of  experiment.        Field  intensity  a  (cms.).  _  « 
jj.  in  air. 
1.  A.  17/11/1905 12.148  1-165  3'71xl05 
2.  A.  18/  6/1903 9,955  2-040  3-43xl05 
3.  0.  19/10/1905 10,809  2-515  3"53xl05 
4.  A.  16/11/1905 12,143  2-515  3'56xl05 
The  two  experiments  carried  out  with  the  same  apparatus 
but  with  different  field  intensities  have  given  concordant 
results.  Such,  however,  is  not  the  case  if  we  compare  two 
experiments  corresponding  to  different  arrangements  of 
apparatus.  It  would  appear  that  the  principal  cause  of  the 
divergence  is  the  imperfect  knowledge  of  the  coefficient  k 
which  enters  in  the  second  power,  and  which  in  these 
experiments  could  not  be  determined  as  accurately  as  the 
other  data.  With  this  restriction,  a  comparison  of  experi- 
ments 1  and  4  of  the  preceding  table  would  indicate  a 
decrease  of  curvature  along  the  trajectory. 
I  would  more  particularly  call  attention  to  the  table 
relating  to  the  old  plate  (A.  18/6/1903),  in  connexion  with 
which  radium  rays  were  used  and  the  results  obtained  with 
which  have  already  been  published.  It  is  seen  that,  taking 
into  account  the  distance  e,  the  numbers  correspond  very 
closely  to  a  circular  trajectory.  It  becomes  therefore 
necessary  to  reject  definitely  the  explanation  which  I  had 
previously  advanced  and  the  hypothesis  of  an  increase  in  the 
radius  of  curvature  along  the  trajectory. 
A  comparison  of  the  trajectories  of  rays  which  have 
traversed  a  sheet  of  aluminium  0  015  mm.  thick  and  of  rays 
which  have  only  traversed  air  leads  to  conclusions  similar  to 
those  which  have  been  explained  at  the  beginning  of  this 
paper,  as  is  shown  by  the  following  summary : — 
Values  of  8. 
No.  of  experiment.  ( A ^  Ratio. 
In  Air.     Through  Aluminium. 
A.  16/11/1905 17318  1-9037  1-099 
A.  17/11/1905 1-7248  1-9334  1-121 
I  may  finally  add  that  measurements  made  with  respect  to 
the  interior  and  with  respect  to  the  exterior  edges  of  the 
deflected  images,  with  a  view  to  detecting  dispersion,  have 
not  yielded  differences  exceeding  errors  of  observation. 
