130 
Pedigree  Seed  Corn. 
of  alteration  of  type  through  the  action  of  natural  selection, 
than  with  absolutely  pure  races  started  from  single  individuals 
which  have  not  resulted  from  previous  crossing. 
The  conclusion,  and  this  is  well  confirmed  by  general 
experience,  is  that  whatever  the  origin  of  any  race  of  cereals, 
repeated  artificial  selection  is  necessary  to  maintain  its 
uniformity. 
A further  important  point  to  note  is  that  purely  natural 
selection  does  not,  in  the  case  of  mixed  races  of  cereals, 
necessarily,  or  indeed  probably,  bring  about  increased  rate  of 
grain  production,  or  even  lead  to  a maintenance  of  the  rate. 
The  plants  which  survive  in  every  field  of  mixed  races  of 
corn  are  not  necessarily  or  even  probably  those  of  the  most 
prolific  (that  is,  abundant  seed-bearing)  race.  We  grow  our 
cereal  crops  mainly  with  the  object  of  producing  grain,  rather 
than  straw,  but  the  competition  between  crowded  plants  is 
mainly  in  the  vegetative  stage,  and  it  is  not,  either  wi'h  the 
cereals  or  with  most  other  species  of  plants,  the  good  seed- 
bearers  which  are  necessarily  the  most  vigorous  and  the  most 
likely  to  survive  in  the  vegetative  stages. 
Another  series  of  experiments  in  my  barley  nursery  at 
Warminster  illustrates  this  point. 
In  1906  seed  from  four  pure  races  of  barley  grown  in  the 
nursery  for  several  previous  years  were  taken.  Three  of  these 
were  wide-eared  races  of  the  Goldthorpe  type,  and  the  other  was 
one  of  a Danish  Prentice  {i.e..  Archer)  barley,  called  “ Tystofte.” 
Alternate  rows  of  each  of  the  four  races  were  repeated 
thirty  times.  Rows  were  at  equal  distances  and  plants  at 
equal  distances,  giving  each  plant  equal  soil  space.  Each  of 
the  rows  was  weighed  up  separately  at  harvest.  The  “ wide 
ear  ” race  which  adjoined  “ Tystofte  ” in  each  group  gave  a 
higher  average  grain  produce  per  row  by  nearly  50  per  cent, 
than  “ Tystofte.”  Mr.  W.  S.  Gosset  kindly  examined  statis- 
tically the  figures  for  the  weights  of  the  grain  in  each  row, 
and  found  that  the  probability,  taking  into  account  the 
“ standard  deviation  ” of  individual  rows  from  the  mean,  was 
enormously  in  favour  of  the  “ wide  ear,”  being  more  prolific 
under  the  conditions  than  “ Tystofte,”  assuming  that  there 
was  no  “ interference  ” of  the  one  sort  prejudicial  to  the  other. 
Mr.  Gosset’s  examination  of  the  figures  for  all  the  four  races 
showed,  however,  that  there  was  abundant  evidence  of  “ inter- 
ference,” and  on  his  suggestion  in  the  following  year,  1908, 
these  two  races  were  gruwn,  not  in  alternate  rows,  but  in 
alternate  plots  repeated  twenty  times.  The  field  of  “Tystofte” 
was  10  per  cent,  better  on  the  average  than  of  the  wide  ear, 
and  the  odds,  taking  each  plot  separately  into  account  under 
these  conditions,  was  very  greatly  in  favour  of  “ Tystofte.” 
