Statistical Method. 



31 



periods of 194, 192, and 189 a in order. Similarly on the alcohol day 

 a first period average of 190 a is followed after the taking of the 30 c.c. 

 of alcohol by 223, 214, and 201 a, respectively. These averages, even 

 when so few in number, are not capable of being readily comprehended 

 without further statistical elaboration. It is desirable to relate the 

 two sets of averages and to secure a simplified expression for them. 



Table 2. — Illustrative table showing statistical method em-ployed in -presenting the data of 



these experiments. 



1 The data in this table, taken from table 9, p. 60, give an exaggerated impression of the effect 

 of alcohol, which must be corrected by reference to the fuller data and results in the later table. 



The reaction times in table 2 following the alcohol dose are in every 

 case longer than those which follow the taking of the control mixture. 

 If we average the three values for each condition we should have aver- 

 ages of 192 and 209 <r, the difference between these total averages show- 

 ing 17 a longer reaction after the alcohol. But such a statement would 

 be only partial, as it takes no account of the fact that the so-called alco- 

 ho 1 day began with a normal period which gave reactions considerably 

 shorter (190 a) than those for the similar period (212 a) of the normal 

 day. In the expression for the alcohol effect, allowance must be made 



