MEASUREMENT OF VARIABLE PROPERTIES 131 



§ 103.— ONE CAST WITH TWO DICE. EXPERI- 

 MENTAL VERIFICATION OF THE EXAMPLES IN §§ loi- 

 102. I have cast two dice 3600 times. Result : 



Table f (Compare Table P, p. 127) : 3600 Casts with 

 2 Dice : Observed Frequencies 



According to calculation, the figure 100 was expected in each 

 of the thirty-six divisions. In twenty cases the difference 

 between the observed and the calculated value does not exceed 

 0*04 of the latter. Although the number of casts is rather 

 small (no less than thirty-six compound events are compared) 

 the concordance is not unsatisfactory. 



In the horizontal rows and in the vertical columns the 

 deviations from the calculated figure (600) are small ; in the 

 most unfavourable case it reaches hardly 0*062 of the latter. 



Looking upon the 3600 casts as being 7200 casts with one die, 

 taking the figures of each die separately and calculating how 

 many times even and uneven figures were obtained (equality is 

 expected), we find : 



Die I Die II 



even (2, 4, 6 dots) : 1785 1801 

 uneven (i, 3, 5 dots) : 1815 1799 



Both dice together (expected 3600 : 3600) : 



even: 3586 uneven: 3614 



Deviation from the expected figures : o'oo4. The observed 

 ratio is even : uneven = o'gg2 : i. 



Compare this very satisfactory result with the irregularity 

 of the series of 100. 



Compare the series of 100, 500 and 3000 in § 93 (p. 118) and 

 the series of 3000 in § 95 (p. 122). 



When we take in each of the thirty-six compound events in 

 Table P the sum of the facial values of both dice, the thirty-six 



