470 



TRANSMISSION 



say, they were not " smooth," and if the curve were platted there 

 would be an unaccountable " hump " at each quarter second. 



The second peculiarity had reference to the last record, 2 : 30. 

 By the principle just stated this number should have been about 

 70 per cent of the number recorded at 2 : 29^. On the contrary, 

 it was very much greater. On the principle running through 

 the rest of the records we should predict the number at 2 : 30 to 

 be about 727, whereas the number actually recorded is 1097, 

 showing conclusively that some 370 horses had been admitted 

 to the 2 : 30 list that really belonged at 2 : 30^ or slower ; all of 

 which shows how statistical studies bring to the surface as noth- 

 ing else will the natural irregularities of observations or whatever 

 abnormal facts connect themselves with our investigations. The 

 table on page 469 is suggestive in many ways. 



RELATION OF SEX AND COLOR TO SPEED AS EXPRESSED IN 

 RATE PER CENT 



From the table above it appears that there is little relation 

 between speed and either sex or color. It is true that, as we 



