306 REPORTS ON THE STATE OF SCIENCE. — 1915. 



breaks in the work at all, meals being taken as the work proceeds. 

 Agriculture, lumbei-ing, mining, and iron and steel works will therefore 

 be left for separate consideration, the term manufacture being used to 

 exclude them, though it may be taken to include building. Industry 

 as a whole, when mentioned, will include all these trades, and trans- 

 portation and commerce into the bargain. 



Now, thanks to such standardisation of working times as large- 

 scale industry has introduced, we may know at any given time of the 

 day or the week just how long men in any given factory have been 

 working since the beginning of the week, day, or spell. The output 

 or accidents of that factory at all given times can therefore easily be 

 correlated with the length or duration of work ' previously done.' 

 Few single factories, however, have analysed their output or accidents 

 according to time of occurrence, and to supplement them we have 

 to fall back on figures collected by Governments for a whole State 

 within which the times of work are likely to differ considerably. Such 

 ' Stat/e accidents ' are delimited into hourly periods either by the 

 exact hours (e.g., 7 to 7.59 or 7.01 to 8) or, as in Massachusetts, 

 by the half-hours {e.g., 6.30 to 7.30), but which of these periods 

 corresponds to what hour of the working spell has only been roughly 

 estimated from inquiries on the spot as to the usual working times. 

 These working times are indicated at the head of each of our tables or 

 in a note at the bottom. Accidents or output in hours which are cer- 

 tainly not worked full are not quoted. By thus omitting the hours of 

 meal-breaks in our tables and leaving a space instead, the spells of 

 work are clearly distinguished. 



These tables, then, that follow on pages 323-344 show the output 

 and accidents as distributed over the hours of spell, the spells of a day, 

 and in some cases the days of a week. A yearly distribution could not 

 illustrate the result of (i.e., test) the fatigue of factory workers since 

 they do not usually get a regular yearly holiday," and it is therefore 

 impossible to correlate output accidents to any ' duration of work since 

 start of year.' There is no industrial year. To find the periodic 

 fatigue as contrasted with the accumulated fatigue described in Section 

 I., the only periods we can usefully compare are the days of the week, 

 the spells of a day and the hours of a spell; with the questions in our 

 mind — "What is the weekly fatigue ? the day fatigue ? the spell fatigue ? 



The answer to these primary questions jumps to our eyes. As 

 far as the figures before us show, there is little weekly fatigue, little 

 daily fatigue, but a great deal of spell fatigue. In this nearly all our 

 statistics agree remarkably. 



The output and accidents of the different days that work full time 

 never vary more than 10 per cent, from one another, and even then 

 it is not always the later days of the week that have more accidents or 

 the lower output. 



In Belgium in 1907 (Table XIV.), it is true, Saturday is by far the 

 most fatal day of the week (attributable in some measure to cleaning 

 machinery in motion), and in the case of the English Engineering Trades 

 (Table IX. d), especially locomotive manufacturing and shipbuilding, 



8 There are certain notable exceptions to this, such as the Lancashire wakes-week. 



