MACHINE  OPERATORS

117

This  apparatus  overcame,  to  a  large  extent,  the  difficulties ­
  inherent  in  the  Bogardus  piece.  In  the  first  place,
lt:  was  so  constructed  as  to  make  possible  a  fairly  accurate ­
  performance.  If  the  steel  ball  missed  the  slot  and
fell  on  the  disk,  it  was  prevented  from  falling  off  by  a
r aised  rim.  If  the  ball  fell  through  the  slot  but  missed
the  funnel,  it  was  caught  by  a  sort  of  apron  which  extended ­
  around  the  motor  box.  This  apron  was  constructed
at  a  pitch  which  quickly  brought  every  steel  ball  to  the
feont  of  the  machine  into  a  little  depression  from  which
the  operator  had  to  pick  it  up.  The  apron  was  made  of
t'n,  and  was  very  noisy  at  the  outset.  However,  it  was
later  lined  with  heavy  felt  which  subdued  the  noise  very
effectively.  In  the  second  place,  this  apparatus  made  it
Possible  to  obtain  an  accurate  score.  Every  successful
at tempt  of  the  operator  to  drop  the  ball  through  the  funn
 el  was  recorded  automatically,  which  meant,  of  course,
that  the  failures  were  also  registered.  The  latter  was
fecilitated  by  a  Veeder  star  counter  attached  to  one  corner
of  the  motor  box  in  such  a  manner  as  to  engage  a  pin  projecting ­
  from  the  bottom  of  the  revolving  metal  disk.  This
counter  registered  the  revolutions  of  the  disk  and,  at  the
Sa rne  time,  the  number  of  opportunities  for  dropping  the
to  the  operator.  The  experimenter  could  therehis
  entire  attention  to  the  operator  or  person
oeing  tested.  A  third  advantage  was  the  ease  with  which
the  revolutions  per  minute  of  the  disk  could  be  governed
fe  contrast  with  the  intricate  and  expensive  control  of  the
®°gardus  apparatus.  Fortunately,  the  range  of  the  motor
^as  exactly  suitable  for  this  purpose.  Moreover,  extensive
tr 'als  showed  that  the  disk  would  revolve  at  a  constant
s P e ed  for  eighteen  minutes  without  the  loss  of  a  single
Solution.  After  the  eighteenth  minute,  the  speed  gradupus

  given
fere  devote