THE QUESTION OP FATIGUE PROM THE ECONOMTC STANDPOINT. 309 



another, each for a day. The total number of lines they set when 

 added together is distributed as follows : first hour 205, second 255, third 

 222, fourth 211 ; then after lunch and a rest of two hours, 241, 206, and 

 160. Only in its first-hour maximum in the afternoon does this differ 

 from our usual distribution. 



Of Dr. Marsh's three records one refers to workers on a time- 

 wage, and thus introduces a novel element ; but in the other processes, 

 in three spells out of the five the total output of all the workers shows 

 the usual short increase or stationariness followed by a long and gradual 

 decrease; the two exceptions — magazine wire-stitching morning and 

 evening spell — have no decrease, their total output remains on the 

 whole stationary. 



It should be noted that in all the output tables we have collected 

 the wages were wholly or partly paid by the piece, the gradual decreases 

 in the output towards the end of the spell is presumably, therefore, 

 not due to deliberate ' slacking ' : every slackening of effort would 

 mean a loss of the workers' earnings. 



In the case of accidents there is in every one of our twenty 

 odd tables and graphs, except at Cadbury's and H. Renolds' (Tables 

 XVII., XVIII.), a constant increase of accidents in the course of the 

 first three hours of each spell, morning, afternoon, and before breakfast, 

 unless of course there are pauses. This increase is particularly marked 

 in cotton spinning, where the third hour has 665 accidents against 

 the first hour's 316 in the morning spell, and 536 against 222 in the 

 afternoon spell. In some of the American and foreign tables for 

 whole States the increase of accidents apparent between the first and 

 second hour in the morning, though not between the second and third 

 of both spells nor the first and second in the afternoon, is attributable 

 to a certain extent to the fact that the first row of figures given does 

 not always represent a full working hour. However, if from the 

 ' usual working-times ' given at the head of each table we make some 

 rough correction, the general increase of accidents seems to hold good 

 of the early morning hours, and if we look at the "Wisconsin and 

 German accidents classified according to the number of hours' work 

 already done, and thus free from uncertainty, we find this increase as 

 much as 200 per cent, between the first and second hours of work 

 (Table XVI.). 



After the third hour, accidents usually increase still further, but 

 this depends rather on the total length of the spell, for we find that 

 in whichever is the last hour of the spell (fourth or fifth) the accidents 

 decrease, so that the accident distribution or curve may most accur- 

 ately be said to reach its maximum on the penultimate hour of the 

 spell. This increase up to the penultimate hour holds good so long 

 as no mid-morning or mid-afternoon pauses occur, and where the 

 last division of the spell is really a whole hour and not merely as 

 in cotton (Table IX. a) and in Massachusetts (Table XL) a half hour 

 rated to the hour. But exceptions are found in the afternoon spells 

 at Hans Eenolds' (Table XVII.) and at the Cadillac Works (Table 

 XX. ) in some of the Tables dealing with smaller numbers (Tables 

 XVIII., XXIII., and XXIV.) and therefore the more liable to ' error,' 



