for  Rectifying  High- Frequency  Electric  Currents.      661 
Thus,  for  instance,  we  form  an  oscillatory  circuit  (see 
fig.  2)  by  connecting  a  Leyden  jar  in  series  with  a  square 
coil  of  wire  of  a  few  turns  P,  and  join  the  condenser  and 
inductance  across  a  spark -ball  discharger  connected  to  the 
secondary  terminals  of  an  induction-coil.  At  a  certain 
distance  we  place  another  square  coil  of  wire  S  in  series  with 
a  galvanometer  G  and  oscillation  valve  V.  We  then  find 
that  when  oscillations  are  set  up  in  the  primary  circuit,  we 
obtain  a  steady  deflexion  of  the  galvanometer  indicating  that 
its  coils  are  being  traversed  by  a  series  of  discharges  in  the 
same  direction,  all  those  in  the  opposite  direction  being 
practically  stopped. 
The  author  has  already  described  the  methods  by  which 
the  amount  of  rectification  produced  by  the  valve  can  be 
ascertained  (see  Proc.  Poy.  Soc.  vol.  Ixxiv.  p.  484,  1905). 
Perfect  rectification  does  not  exist,  but,  as  shown,  the  number 
expressing  the  percentage  which  the  actual  unilateral  electric 
flow  is  of  that  which  would  flow  if  the  unilateral  conductivity 
were  perfect,  can  be  ascertained  by  sending  the  current  which 
passes  through  the  vacuous  space  of  the  valve  through  a 
calibrated  galvanometer  and  electrodynamometer  placed  in 
series  with  each  other.  In  valves  as  described  this  rectifica- 
tion may  amount  to  90  per  cent. 
We  may  use  the  above  arrangements  to  investigate 
the  effect  of  different  kinds  of  discharge-balls  and  different 
lengths  of  spark.  If  we  employ  a  fairly  loug  spark  in  the 
primary  condenser-circuit  we  may  find  that  we  obtain  a  very 
small  effect  on  the  galvanometer  in  the  secondary  circuit,  but 
if  we  shorten  the  spark-gap  until  the  spark  at  the  balls  is 
hardly  visible,  the  galvanometer  deflexion  is  generally  in- 
creased. The  reason  for  this  is  partly  because  the  oscillations 
are  damped  out  much  more  by  the  long  spark  than  by  the 
short  one,  and  partly  because  with  a  short  spark  the"  con- 
denser discharges  occur  more  frequently.  Hence,  although 
in  the  latter  case  the  condenser  is  charged  to  a  less  voltao-e 
owing  to  the  lower  discharge  potential,  the  decreased  damping 
and  greater  charge  frequency  causes  the  galvanometer  to 
be  travel  sed  by  a  larger  quantity  of  electricity  per  second 
and  therefore  to  give  a  greater  deflexion. 
In  the  same  manner,  we  call  exhibit  the  difference  in  the 
damping  due  to  variations  in  the  material  of  the  spark-balls. 
Thus,  using  iron,  brass,  and  zinc  spark-balls  of  the  same 
diameter  and  a  spark-length  of  0-1  mm.,  the  galvanometer 
deflexions  in  one  case  were  respectively  40,  57,  and  70  scale- 
divisions,  thus  showing  the  smaller  damping  of  zinc  spark- 
balls.  The  writer  has  found  by  this  means  that  carbon  in 
the  state  used  for  arc-lamp  carbons  presents  many  advantages 
