1336 THE BELL SYSTEM TECHNICAL JOURNAL, NOVEMBER 1957 



suitable demagnetization data. There are, however, interference effects 

 between cells which are not completely understood at the present time. 



A memory array (16 X 20) has been constructed as a test vehicle. An 

 illustration of this array is shown in Fig. 13. The drive wires have been 

 woven over glass tubes which house the removable magnetic wires. 

 Provision is made for varying the torsion and the tension of the in- 

 dividual magnetic wires. 



As an indication of the performance of the twistor, Fig. 14 is a com- 

 posite photograph showing the minimum and maximum signal over the 

 16 bits of a given column for 3-mil nickel wire. Also included are the 

 noise pulses for these cells, the so-called disturbed zero signals. The 

 write currents were 2.3 ampere-turns on the solenoid and 130 ma through 

 the magnetic wire. The read current was 6.0 ampere-turns. The array 



Fig. 14 — Composite photograph of the 16 output signals from a column of the 

 array of Fig. 13. Average output signal al)out 3.5 millivolts; sweep speed equals 

 2 /isec/cm. 



was transistor driven. A read-write cycle time of 10 microseconds ap- 

 peared to be possible. 



V. DISCUSSION 



The twistor is presented as a logical companion to the coincident- 

 current ferrite core and sheet. ' In many applications it should compete 

 directly with its ferrite equivalents. Perhaps its greatest use will be 

 found in very large ( > 10^) memory arrays. 



From a cost per bit viewpoint the future of the twistor appears quite 

 promising. Fabricating and testing the wire should present no special 

 problems as it is especially suited for rapid, automatic handling. The 

 possibility of applying weaving techniques to the construction of a 

 twistor matrix looks promising. 



It is possible that, for l)oth mode A and C operation of the twistor, an 

 array can be built which consists simply of horizontal copper wires and 



