390 



THE BELL SYSTEM TECHNICAL JOURNAL, MARCH 1954 



tiny of the data cast grave doii})t on this assumption and each card pos- 

 tion was analyzed separately as shown below. 



Card Position A 

 Analysis of Variance 



* One observation was lost, and Fisher's Missing Plot Technique was used (5). 

 Card Position B 

 Analysis of Variance 



The two estimates of experimental error being 262 and 14 ms^ re- 

 spectively against a previous estimate (Table XIII) on 24 degrees 

 of freedom of 10.9 ms , we must accept Position A as coming from a dif- 

 ferent universe with respect to position B and the previous seven var- 

 ible experiments. Further Position B on the same evidence has been 

 measured with equivalent precision when compared to the previous 

 study with the card support bars out. 



The individual dropping times of Position B cards over the experi- 

 ment range from 19 to 39.5 ms., well within the predicted dropping 

 times. The mean dropping times for each tilt of card Position B are as 

 follows: 



Tilt 



27.2 ms 



1 



3 

 33.8 ms 



27.5 ms 31.2 ms 



By contrast the mean dropping times for Position A are: 



Tilt 12 3 



28.9 ms 39.2 ms 59.9 ms 73. G ms 



with a range of 21 ms to 109.5 ms on the fourth tilt. 



