﻿Report 
  on 
  the 
  Metabolism 
  of 
  Female 
  Munition 
  Workers. 
  71 
  

  

  of 
  " 
  chance 
  " 
  can 
  he 
  measured 
  by 
  calculating 
  the 
  respective 
  " 
  probable 
  errors 
  " 
  

   of 
  the 
  averages. 
  Not 
  to 
  pause 
  upon 
  the 
  fact 
  that 
  the 
  validity 
  of 
  this 
  process 
  

   in 
  its 
  ordinary 
  application 
  depends 
  upon 
  the 
  assumption 
  that 
  the 
  factors 
  of 
  

   chance 
  variation 
  are 
  numerous, 
  small, 
  and 
  result 
  in 
  a 
  particular 
  type 
  of 
  

   dispersion, 
  we 
  must 
  recollect 
  that 
  in 
  the 
  present 
  case 
  we 
  have 
  special 
  

   difficulties. 
  If, 
  for 
  instance, 
  we 
  find 
  that, 
  judged 
  by 
  this 
  criterion, 
  the 
  

   average 
  energy 
  expenditure 
  in 
  the 
  press-house 
  exceeds 
  that 
  in 
  a 
  lathe 
  

   operation, 
  to 
  what 
  extent 
  is 
  the 
  difference 
  contributed 
  to 
  by 
  (a) 
  a 
  real 
  

   difference 
  of 
  physiological 
  demand 
  made 
  by 
  the 
  two 
  kinds 
  of 
  work 
  ; 
  

   (b) 
  personal 
  differences 
  in 
  the 
  energy 
  output 
  of 
  the 
  workers 
  themselves 
  

   under 
  all 
  circumstances, 
  functions 
  of 
  age, 
  or 
  temperament 
  ; 
  (c) 
  gross 
  

   irregularities 
  due 
  to 
  our 
  having 
  chanced 
  upon 
  keen 
  workers 
  in 
  one 
  operation 
  

   and 
  slack 
  workers 
  in 
  another. 
  It 
  would 
  have 
  been 
  possible 
  to 
  appraise 
  (b) 
  

   had 
  we 
  been 
  able 
  to 
  measure 
  the 
  basal 
  metabolism 
  of 
  our 
  subjects, 
  few 
  

   though 
  these 
  were, 
  but 
  this 
  we 
  could 
  not 
  do 
  owing 
  to 
  the 
  conditions 
  under 
  

   which 
  the 
  research 
  was 
  done. 
  If 
  our 
  numbers 
  had 
  been 
  multiplied 
  tenfold, 
  

   it 
  would 
  have 
  been 
  fair 
  to 
  assume 
  that 
  differences 
  under 
  (c) 
  were 
  smoothed 
  

   and 
  equally 
  distributed 
  through 
  the 
  categories. 
  But, 
  things 
  being 
  as 
  they 
  

   were, 
  the 
  research 
  being 
  brought 
  to 
  an 
  end 
  when 
  only 
  a 
  few 
  hundred 
  

   instead 
  of 
  a 
  few 
  thousand 
  experiments 
  had 
  been 
  made, 
  we 
  have 
  no 
  right 
  to 
  

   assume 
  that 
  (a) 
  alone 
  has 
  been 
  responsible 
  for 
  such 
  " 
  significant 
  " 
  differences 
  

   as 
  are 
  to 
  be 
  discerned. 
  The 
  gross 
  number 
  of 
  observations 
  is 
  so 
  small 
  that 
  

   statistical 
  manipulations 
  and 
  sub-tabulations 
  designed 
  to 
  isolate 
  the 
  several 
  

   factors 
  (in 
  addition 
  to 
  those 
  above 
  enumerated, 
  are 
  considerable 
  variations 
  of 
  

   external 
  temperature 
  and 
  cooling 
  power) 
  can 
  only 
  lead 
  to 
  a 
  spurious 
  air 
  of 
  

   precision. 
  We 
  have 
  accordingly 
  contented 
  ourselves 
  with 
  a 
  summary 
  

   analysis, 
  now 
  to 
  be 
  detailed. 
  The 
  complete 
  protocol 
  of 
  each 
  experiment 
  has 
  

   been 
  deposited 
  with 
  the 
  Metabolism 
  Eesearch 
  Sub-Committee, 
  and 
  the 
  

   detailed 
  particulars 
  will 
  be 
  of 
  use 
  when 
  sufficient 
  material 
  has 
  been 
  accu- 
  

   mulated 
  by 
  other 
  investigators 
  to 
  admit 
  of 
  a 
  numerically 
  adequate 
  tabula- 
  

   tion. 
  The 
  time 
  for 
  this 
  is 
  far 
  distant, 
  and 
  the 
  industrial 
  aspects 
  of 
  

   metabolism 
  cannot 
  yet 
  be 
  fully 
  elucidated. 
  

  

  Section 
  4. 
  

  

  In 
  Tables 
  II-XIII 
  we 
  record 
  the 
  numbers 
  of 
  experiments 
  in 
  each 
  

   operation, 
  the 
  mean 
  liberations 
  of 
  calories 
  per 
  minute 
  and 
  per 
  square 
  metre 
  

   of 
  body 
  surface, 
  the 
  standard 
  deviation, 
  the 
  percentage 
  ratio 
  of 
  the 
  standard 
  

   deviation 
  to 
  the 
  mean 
  (coefficient 
  of 
  variation), 
  and 
  the 
  approximate 
  

   probable 
  error 
  of 
  the 
  mean 
  (taken 
  as 
  - 
  67449 
  multiplied 
  by 
  the 
  standard 
  

   deviation, 
  divided 
  by 
  the 
  square 
  root 
  of 
  the 
  number 
  of 
  experiments). 
  Each 
  

  

  