lliE QUESTION OP FATIGUE PROM THE ECONOMIC STANDPOINT. 305 



greater is the exposure to accident, since the more specially dangerous 

 movements will have to be performed so much the oftener. 



A nicer test of fatigue than the absolute distribution of accidents 

 would thus appear to be the distribution over time of accidents per 

 unit of output. An absolute increase of accidents might be due to 

 an increase of output, and that due in turn to spurt or some other 

 check to fatigue, and thus an increase of accidents would measure 

 decrease rather than increase of fatigue. Similarly an absolute de- 

 crease of accidents might be due to a decrease of output, and that due 

 to increase of fatigue. 



Unfortunately accident and output statistics are extremely difficult 

 to correlate, since to secure sufficient accident data large numbers and 

 long periods must be studied, while output data suitable to statistical 

 treatment can only be secured here and there. 



Where output or work is likely to vary greatly during the day, as 

 in the tramway, telephone, or shop service with their rush hours, the 

 accident distribution alone is quite unreliable. Separate reference to 

 these ' industries ' is omitted from all statistical tables, though they 

 are included under Total Industry. 



(ii) The Kind of Process. If at certain times of the day the 

 workers studied are engaged in more dangerous work than at other 

 times, as founders are in casting at 3 p.m. or so, then accidents will 

 vary accordingly, without reference to fatigue. 



Section V. — Characteristics of the Actual Time-Distribution. 



A. Geneeal. 



We have now a sufficient grasp of the different causes of fatigue 

 and of the relation of fatigue to the two tests we are using, output 

 and accident time-distribution, to be able intelligently to consider the 

 figures given in the tables on pages 323-344 with a view to finding the 

 solution to our central problem : What is the exact correlation of 

 fatigue with the previous duration of work performed ? 



First, however, a word in explanation of this phrase ' previous 

 duration of work. ' The factory system that now prevails in manufac- 

 turing industry has standardised among many other conditions of life 

 the workers' hours of labour. Except in a few continuous industries 

 like iron and steel, the average ' employed ' factory hand works 5i or 

 6 days a week, 8 to 12 hours a day, and thus his very hours of work 

 per week are standardised and often even ascertainable in ' Blue Books. ' 

 The distribution of the hours into ' spells ' of continuous work diffei's 

 considerably between countries. In America a five-hour spell, then 

 dinner, then four or five hours more is the rule ; on the Continent of 

 Europe short pauses often occur in the middle of the morning and 

 afternoon ; while in England the day generally consists of two hour.«; 

 work, half-hour breakfast, four hours work, one hour dinner, four 

 hours work, tea. This general standardisation applies also outside the 

 factory to the building trades in full, to agriculture, lumbering, and 

 mining as regards day and week, but not as regards spell, since breaks 

 in the work are made somewhat irregulai'ly according to opportunity 

 and feelings. In iron and steel manufacture, however, there are no 



1915. X 



