To complement this valuable asset I record a few details concerning 701 I/O that I remember from the IBM 701 computer.

There were read select and write select commands whose address field nominated an I/O device. The devices were, card reader, card punch, printer, drum (perhaps several), and tape (usually several). There was a sense command, and its twin (minus-sense) that, depending in its address field did one of the following:

The printer could receive information for only 72 print columns per 400 ms print cycle. If more information were needed on one print line two such print cycles were required. Plug board logic was required to determine which of the 120 print positions printed on each cycle.

There were commands to write ‘end of file’ on magnetic tapes and variations on read select to read backwards. There was some way to test for a hardware detected magnetic tape parity error after reading a tape record. I do not remember the details. There was a separate opcode: LDA for ‘load drum address’ which specified, after drum select and before copy instructions, where on the drum the operation was to perform at.

Here are more details on 701 IO logic. There was a prominent button on the operator’s console labeled “CLEAR” which caused the machine to stop and set every memory cell to all 1’s. 1’s were less stressful than 0’s for the Williams tubes. A wild branch would after cause the machine to execute such a cleared location which would cause the machine to stop and a ‘copy-check’ light to come on because there was, typically, no I/O operation in progress. The binary op-code for the copy instruction was all ones.