318 REPORTS ON THE STATE OF SCIENCE. — 1915. 



comparable with others in the discussion of industry in general 

 (Section V.a). 



But turning now to Table VIII. Col. 2 and Table X., we find that 

 the normal course of the four- or five-hour spell is not reproduced in 

 the twelve-hour spell. There is no gradual increase of accidents in 

 the first three quarters of the time, nor gradual decrease of output in 

 the last three quarters. On the contrary, there seems to be a series 

 di waves. The total accidents by day increase up to a crest in the 

 fourth hour (9-10), decrease to a trough in the seventh (12-1), increase 

 again to the ninth (2-3), and then decrease for the rest of the spell. 

 By night crests occur in the second and ninth hours, troughs in the 

 seventh and last hours. At the beginning of the night, however, there 

 is additional repairing, and this may account for the early crest. 



To calculate how far these accident curves diverge from the normal, 

 we might consider the twelve-hour ' shift ' as a four-hour ' spell ' 

 multiplied threefold. On this basis, the decrease in the last three 

 hours corresponds strictly to the decrease in the last hour of a short 

 spell due to anticipatory excitement, while the increase of accidents 

 in the first four hours of the shift corresponds to the increase between 

 the first and second hours of the spell, attributable to fatigue. The 

 only hours not reconcilable are the fifth to the eighth (10 a.m. -2 p.m.): 

 to correspond to the four-hour spell, accidents should be increasing; 

 as a matter of fact, they form a trough in the curve. This trough may 

 possibly be correlated with the midday or midnight meal, which would 

 first produce a fall in accidents through anticipatory and then 

 positive excitement while being taken ; secondly a sudden rise while its 

 digestion was conflicting with the work in hand. But that all these 

 lesser forces should cloak or overcome the effects of fatigue must be 

 attributed to the main peculiarity of this work, namely, the repeated 

 intervals of rest at the end of each operation, which allow of ' recovery.' 



In studying the separate departments of iron and steel works, it 

 should be noted that the mechanical department and the yards usually 

 have meal breaks, but betweenwhiles work actively all the time. In 

 consequence, the fluctuations of their accident curve are more marked 

 and are actually (not merely proportionately) more similar to the 

 normal. So many different processes are included in these depart- 

 ments, however, that special attention to the figures given is not worth 

 while. 



In the output curve of the Bessemer converter (Table VIII. Col. 1) 

 are seen the same waves up and down and even less regular. Here 

 the effect of 'non-persistence,' especially combined as it is in this 

 process with a relative infrequency in the recurrence of the operation 

 (about three times to the hour), would make the factor practice of 

 more weight than the factor fatigue and likely to last almost to the 

 end of the day. This may account for the very gradual increase 

 of output in each successive two-hour period, if we except 8-10 because 

 of changes and repairs and 12-2 because of lunch and digestion during 

 work. The general drop in the last two hours would on that supposi- 

 tion be due to tlie final overcoming of practice by fatigue. 



(b) The most highly complex processes of those recorded in our 



