310 REPORTS ON THE STATE OF SCIENCE.— 1915. 



and finally in several of the figures originally collected by the U.S. 

 Federal Bureau of Labour, namely those of Indiana manufactures 

 (morning) (Diagram J.) and of the American Cotton Industry (Dia- 

 gram III.), 



The divergence of this latter set of figures from the majority, how- 

 ever, is to a large extent due to the unusual hourly limits adopted by 

 the U.S. Bureau of Labour, e.g., from 10.01 to 11 instead of 10.00 to 

 10.59. This at first sight seems a trivial difference, but common- 

 sense backed by experiment shows that factory, occupiers are inclined 

 to return all accidents occurring between, e.g., 9.55 and 10.10 as 10 — 

 hence the importance of the question in which period to place the 

 exact hour. The U.S. Federal method results in pushing back the 

 maximum accidents. 



The decrease in accidents from the penultimate to the ultimate hour 

 of the spell mentioned already varies in amount very greatly. It is 

 slight in all the Illinois figures, only from 1,485 to 1,438 in the total 

 (Table XIII. Col. 4) in the morning and from 1,382 to 1,327 in the 

 afternoon ; but the decrease is very great in some single factories, notably 

 Hans Renolds', in the morning, from 375 to 252 (Table XVII. a.m.), 

 the National Cash Register 308 to 181 (a.m.) and the Cadillac Motor 

 Company from 152 to 88 in the morning, and in the afternoon (where 

 the decrease sets in earlier) 200 to 116 to 94. 



That this falling-off is not entirely due to a fall in output, and, 

 therefore, in the risk, may be seen by the fact that, even where output 

 is humanly variable as in the processes in Tables II. to VIII., output 

 never falls more than 15 per cent, from the penultimate to the last 

 hour of spell.'' That this falling-off is not due merely to a fall in the 

 number employed owing to lunch or the end of work coming sooner, 

 is shown by the accidents of individual factories where the exact times 

 worked were ascertainable, and are stated in Tables XVII. and follow- 

 ing, but furthermore in all State-collected records most detailed inquiries 

 were made on the spot to ensure that, in all figures of hourly accidents 

 given, the number employed at each hour was the same unless other- 

 wise stated. In short, in the last hour of the spell there is a true 

 decrease in the accident rate per given number of men and per given 

 amount of output, or, at any rate, a check to the increase. 



In the case of the statistics for whole States, the question of which 

 is the last hour of the afternoon spell is rather hard to settle, since 

 individual factories will quit work at different times, and though some 

 indication is given by the heading over each table ' Usual Working 

 Times,' evidence for the afternoon decrease of accidents must be 

 sought mainly from the figures of single factories where the exact 

 times of work are ascertained (Tables XVII., XXIV.). 



The individual investigators of the time-distribution of accidents 

 that have gone before us have failed to observe this decrease of 

 accidents at the end of the spell. Bogardus, because he studied only 



* Account must be taken of the fact that our output tables only give processes 

 under piece wage where there is little likelihood of slacking. However, even under 

 time wage discipline and set standards and the speed of the machine would prevent 

 any very much larger falling-off in output. 



