Word Reaction. 65 



difference -21 <r for the first normal day, June 29, this difference being 

 the only minus value in the group, the effect of alcohol would still be in 

 the same direction, that is, a lengthened reaction for this period. The 

 probable correctness, 0.989, which applies to the difference +14 is so 

 large as to indicate its trustworthiness. For periods 3, 4, 5, and 6 

 the normal differences are commonly plus and are generally very large, 

 while those for alcohol are usually minus, exhibiting clearly the tendency 

 for the reaction to be faster in the later periods following alcohol. 

 The effect of alcohol in per cent for periods 3 to 6, inclusive, is shown 

 respectively as -9, -7.4, -11, and -21 per cent. The probable 

 correctness values for periods 3, 4, 5, and 6 indicate that behind these 

 differences there is certain cause rather than chance cause. The diff- 

 erences are significant; the question for the following paragraphs to 

 elucidate is what produces them. The mean variations do not show 

 a consistent change in all periods. 



Our subject is not mentioned by Dodge and Benedict as being differ- 

 ent from the other members of the normal group in the word-reaction 

 results, 1 their results showing him to be in line with the majority in so 

 far as the effect of alcohol is indicated by averages. On the other hand, 

 the average differences and percentiles for both Subjects VI and IX 

 show a shorter reaction time after alcohol, with the result that the 

 average percentile effect of alcohol for the whole group with dose A 

 isO. 2 



The Dodge and Benedict minus differences for both doses A and B, 

 indicating an increased reaction time after alcohol, are with all but 

 Subject II quite small. There was a distinct tendency in the data of 

 Dodge and Benedict for the reaction time for isolated words to increase 

 gradually from period to period on normal days. If all the data for 

 these normal days are averaged by periods we obtain, beginning with 

 period 1, the following series of values: 445, 455, 465, 469, 474, and 

 507 a. A similar group of averages for the alcohol data for their normal 

 subjects (dose A) produce the averages for periods 1 to 6 of 437, 445, 

 451, 456, 447, and 435 a. The period averages given in our table 10 for 

 word reactions compare favorably with those obtained from the pre- 

 vious series of measurements. Subject VI is, therefore, not peculiar; 



1 Dodge and Benedict's report, p. 107, table 12. 



2 The following correction should be made for Subject VI in Dodge and Benedict's report, page 

 103, table 10: February 12, 1914, which was an alcohol day (dose A) has through error been 

 printed under the normal data. Furthermore, there were actually three periods on that day 

 instead of two, these having the following average reaction-time values: 454, 472, and 458 a. In 

 the pulse section of Dodge and Benedict's book (p. 218, table 40) and in our table 1 (see p. 13 of 

 this monograph), February 12, 1914, is shown as an alcohol day, dose A. Our table 1 was of 

 course compiled from the original records. To place the data where they belong produces minor 

 changes in results for Subject VI, as shown in tables 11 and 12, pp. 106 and 107 of the Dodge and 

 Benedict report, with the final result for the group that the average effect of alcohol as shown by 

 percentiles in the extreme right-hand column of table 12 is -5.8 per cent instead of -6.2 per cent. 

 The average value for dose A, normal group, remains 0. A complete checking of the entire Dodge 

 and Benedict report reveals no other place where alcohol and normal data have been confused. 



