358 .iyV.Y.1/..s' YEW YOh'K ACAUKMY OF SCIfJXCES 



sections of the liiK-s. 'V\\v rankings thus obtained agreed remarkably well 

 with the estimates of the supervisors, showing a correlation of .()2o0, with 

 a probable error of .14 (by Spearman's Footrule). We have, then, in 

 this form of the motor-coordination test a valual)lp means of detecting 

 the quickness and accuracy of telephone operators, — two of the most im- 

 portant traits which make for success at the'switchboard. 



In Prof. Ruger's papei" the piirpose of the stmly was to determine 

 whether a well-formed rival habit or a poorly formed one had the greater 

 influence on the formation of a given habit. The plan of the experiment 

 included an initial and final test series with a given key and a practise 

 series with keys formed by varying the arrangement of the test key. For 

 the practise series the group representing the well-formed habit practised 

 on a single rival key ; the group representing the poorly formed habit 

 either constantly changed to a new key or practised fewer times on the 

 same rival key. In addition to these two main groups there were three 

 (•oiiliol groups and one group whicb practised on tlie test key. One of 

 the control groups read newspapers during the practise period; another 

 did addition, and the third worked on a different type of substitution. 

 All the groups took the initial and final tests with the test key. All the 

 groups did better in the final than in the initial test. However, the rival- 

 habit groups showed niueli less improvement than the control groups. 

 Consequently there was a dominant interference effect. This interfer- 

 ence effect was greater in the group that formed the one strong rival 

 habit than in the one that formed one or many weak rival habits. The 

 control groups were so planned as to have different degrees of relatedness 

 in their practise series to the test keys. The newspaper group simply 

 read what interested them — spontaneous attention : the addition group 

 worked with voluntary attention and at top speed. The substitution - 

 control group worked on material similar to the test series, but not con- 

 flicting with it. The three groups followed this, the above, order in the 

 extent of the improvement of the final over the hiitial test. Since the 

 difference, however, is less than the probable error, the control groups 

 may be considered as equivalent in this particular case. The gro\ip which 

 practised on the test keys showed two and a half times the improvement 

 of the control groups, while the control groups showed twice the improve- 

 ment of the poorly formed rival ha])it group and three times the im- 

 provement of the well-formed rival habit group. Improvement was 

 measured in terms of substitutions per second. 



Three hundred and fifty subjects took part in the experiment. Wood- 

 worth's and Wells's color-naming and geometrical substitution tests were 

 employed. The symbols forming the keys were five different letters or 

 figures. 



