378 THE BELL SYSTEM TECHNICAL JOURNAL, MARCH 1954 



Tilt Study 



Routine field practice for the installation of a card translator calls 

 for leveling the table before positioning of the translator. After a trans- 

 lator was placed it had been found that the level was not maintained, and 

 the question of final level reciuirements was raised. Accordingly the test 

 machine was run at in., % in., 1 in., and 1%6 i^i. tilt, Table II. Two 

 test cards were placed in each of five bins. Since the two end card posi- 

 tions on the low side of each bin were suspected as being critical, the two 

 test cards were placed in these positions in the five bins. The experiment 

 is shown schematically below: 



Bins 



where a is the end position, b is the next-but-end position, and the sub- 

 scripts identify the actual card number. Considering any bin, the two 

 card positions and the four values of tilt may be considered as a factorial 

 design. Similarly, the five bins combine with the four values of tilt as a 

 factorial design. 



Balanced Loading 



At the onset of general usage many translators will not be fully loaded. 

 It was desirable to investigate the effects of various patterns of loading 

 at three representative low loads of 200, 400 and (300 cards respectively. 

 Four logical patterns were studied (see Tables III and 1\) all of which 

 are shown below for a load of approximately 400 cards. (Similar patterns 

 follow with loads of 200 and 600 cards.): 



Bin 1 2 3 4 5 6 7 8 9 10 11 12 



W XX X X 



Y X X XX 



Z 000000000000 



X = 100 cards, = K2 load (approximately) 



The ten test cards were all placed in Bin One. 



This experiment involves only two factors; the four loading patterns, 



