cadet — the IBM 1620 had no adder

morning, {{name}}. the 1620 model 1 (1959–1962) didn't have arithmetic circuits. it looked up every digit-pair operation in a 100-entry add table and a 200-entry multiply table loaded into core memory at boot. click any cell to edit it. the machine will not complain. the result is whatever the table says it is.
IBM 1620 · Model 1 · S/N 0042 power busy ready tables edited
addition table core 00100–00199
multiplication table core 00300–00499
nominal current lookup edited by operator
A and B are decimal integers, up to four digits each. results are computed digit-by-digit via table lookup, with carry, exactly as the 1620 Model 1 did it. subtraction uses the nines' complement method — the same add table, the same potential for error.

console output

machine is ready. type two operands and press A to add, M to multiply.

operator log

Click a table cell to edit it. The machine does not check the tables — they are just bytes in core memory, sitting in the address space alongside your data. A single off-by-one in an indexing loop could overwrite one cell and the machine would carry on, telling you confidently that 3 + 4 was 8, printed in formal output on official IBM paper. The Endicott support office's advice when results looked suspicious was to re-bootstrap. The informal advice, recorded in more than one lab notebook, was to print the tables yourself at the start of every session. — C