368 



BELL SYSTEM TECHNICAL JOURNAL 



is iisuall}' found to be slightly larger tiian /' although not markedly so. We 

 may use this information to test the homogeneity of the holding time uni- 

 verse should all hours of the days be grouped indiscriminately. 



TABLE I 



Hourly Holding Time Data, Newark 



(Figures in body of table are average holding times in seconds) 



