822  Dr.  C.  G.  Barkla  on 
differed  little  from  the  primary  ;  that  with  those  substances  o£ 
lowest  atomic  weight  which  were  the  source  of  a  secondary 
radiation  differing  to  a  greater  extent  from  the  primary, 
polarization  was  shown  to  a  smaller  extent;  aud  all  substances 
experimented  upon  with  atomic  weights  beyond  a  certain 
value  gave  no  evidence  o£  polarization  at  all.  Thus  carbon, 
air,  cardboard,  aluminium,  and  sulphur  emitted  a  secondary 
radiation,  varying  in  intensity  in  the  principal  directions  by 
about  the  same  amount.  Calcium  emitted  radiation  exhibiting 
about  half  this  variation,  while  the  radiations  from  iron, 
copper,  zinc,  tin,  and  lead  exhibited  no  variation  at  all. 
Thus  the  character  of  the  pulses  changed  not  abruptly  with 
an  increase  in  atomic  weight,  but  very  rapidly  between  certain 
atomic  weights. 
It  is  significant  that  the  polarization  effect  disappears  with 
the  similarity  between  the  secondary  and  primary  radiations. 
We  thus  see  that  the  change  in  absorbability  of  the  secondary 
radiation  is  accompanied  by  an  acceleration  of  electrons  in 
directions  not  those  of  electric  displacement  in  the  pulses 
producing  the  radiation.  This  is  important,  because  the 
difference  in  character  between  the  primary  and  secondary 
radiations  might  possibly  be  accounted  for  by  a  change  in  the 
average  distance  apart  of  the  separate  pulses.  Here,  however, 
we  see  that  the  change  in  character  is  accompanied  by  a 
change  in  what  might  be  called  the  pulse  structure.  The 
rather  striking  independence  of  the  absorbability  and  the  dis- 
tances of  the  pulses  apart  is  shown  in  the  case  of  secondary 
radiation  from  a  mass  of  carbon  for  instance  ;  for  on  the 
theory  of  the  production  of  secondary  radiation,  each  electron 
is  the  source  of  a  secondary  pulse,  yet  in  spite  of  this  enormous 
multiplication  of  pulses,  the  absorption  of  the  radiation  differs 
very  little  from  that  of  the  primary. 
The  compounds  ammonium  carbonate,  lime,  calcium  car- 
bonate, and  copper  sulphate  were  tested  in  the  same  manner. 
Ammonium  carbonate  emitted  a  radiation  varying  in  intensity 
in  the  principal  directions  by  about  the  same  amount  as  that 
shown  by  the  elements  of  low  atomic  weight.  Lime  and 
calcium  carbonate  emitted  radiations  showing  a  smaller 
variation,  while  from  copper  sulphate  evidence  of  polarization 
of  the  primary  could  not  be  detected. 
These  results  again  are  what  wrould  be  given  by  mixtures 
of  the  radiations  proceeding  from  the  different  constituents. 
All  the  elements  in  ammonium  carbonate  belong  to  the 
scattering  class,  and  hence  the  full  variation  is  produced.  In 
the  radiations  from  calcium  oxide  and  calcium  cabonate,  the 
rays  from  calcium  produce  most  of  the  ionization  on  account 
