440 



BELL SYSTEM TECHNICAL JOURNAL 



namely, between the first and second attempts and between the second and 

 third attempts. This table was developed by allocating the 817 observa- 

 tions where a second attempt occurred into 5 ranges of time intervals be- 

 tween the first and second attempts of about 163 observations each. For 

 each range of time interval the number of calls that were respectively 

 O.K., DA, PR and AB is listed. Where a third attempt occurred, the 

 numbers of calls are tabulated by ranges of time intervals between the 

 second and third attempts. The ranges of time intervals are the same as 



Table III 



Disposition of Second Attempts and Correlation of Time Intervals Between 



Data Concerning 817 Observations Having a Second Attempt 



* The asterisk marks the items that had the same range of time intervals between the 

 first and second attempts and between the second and third attempts. 



those used between the first and second attempts in order to see if a correla- 

 tion exists. The significant facts concerning these data are: — 



1. The degree of success in obtaining an O.K. call was better for those subscribers who 



waited longer before making a subsequent attempt. Only 22% of the subscribers 

 who waited from to 45 seconds were successful as against 57% who waited from 

 227 to 600 seconds. 



2. The number of calls referred to the operator or where don't answers occurred are not 



significant to the problem in hand, 



3. The incidence of abandoned calls appeats to be uniform for the five ranges of time 



intervals. This means that the 90% figure estimated from Fig. 3 can be con- 

 sidered to apply with equal effect to all subscribers without regard to the previous 

 time interval between dialing attempts. 



4. The correlation data indicate a tendency for subscribers to establish a tempo or pace 



which they follow when redialing their calls. If this tempo did not exist the items 

 on Table III that are marked with asterisks would not be larger than the surrounding 

 items. 



It was previously indicated that no observations were taken on P.B.X. 

 and coin lines. An earlier attempt to collect data concerning the behavior 

 of subscribers when encountering busies produced data that showed fewer 

 subsequent attempts than was believed to be the case. The differences 

 between the earlier data, which included a high proportion of observations on 

 P.B.X. and coin lines, and the data developed herein are believed to be fully 



