KENNELLY. — AN APPROXIMATE LATV OF FATIGUE. 311 



starting plunge made by the swimmer may actually advance him, 

 relatively speaking, on the shortest courses. 



With the exception of the four open- water events, on all of which 

 the speed is distinctly low, and the 8()-yard event, which has an unduly 

 low speed (about the same as in the laO-yard event), the observations 

 cling closely to the straight line. The speeds at both the half-mile and 

 the 1-mile events appear to be distinctly higher than the rest. 



Table VIII indicates that the mean deviation, without regard to 

 sign, of all the events is 3.5 per cent. 



The formulas pertaining to the straight lines in Figure 12 are 



log r = I log L - 0.4196 (20) 



log V = 0.4196 - i log L (21) 



/f 



^=^ seconds (22) 



2 6'^8 

 V — 1 meters per second. (23) 



Men Skating. 



Table IX gives the analysis of 24 yard and mile events and also 5 

 metric events, between the limits of 50 yards and 100 miles. The 

 data appearing in columns I and II of this Table are taken from the 

 records of the Amateur Athletic Union revised in 1905 and published 

 on page 265 of "The World Almanac" for 1906. They represent, 

 therefore, amateur records. 



In Figure 13, log T has been plotted against log L, crossing the 

 sheet twice for economy in space. The straight line drawn in two 

 sections is carried through the 1-mile record point. It also runs very 

 near to the 2-mile, 4-mile, 500-meter, and 5000-meter points. There 

 is a wide deviation of the points from the line between 50 yards (45.72 

 meters) and 440 yards (402.3 meters). This may be attributed to the 

 influence of inertia in acceleration at the start. Column IV of Table 

 IX shows in fact that the maximum speed over a course is not reached 

 until the 440-yard event (402.3 meters), when it attains 11.43 meters 

 per second (25.5 miles per hour). The entries in column IV also 

 reveal considerable relative variation, and do not descend with the 

 same degree of uniformity as is manifested by the corresponding speed 

 entries in Table VIII, or other tables. Perhaps this variability in the 



