Sourcing a working 1962 IBM 1401 core memory board

The ask

The museum had been rebuilding a 1962 IBM 1401 mainframe for the better part of a decade. Power supplies, printer, card reader, tape drives — all working. What remained was the memory. Their existing 4KB core plane had two dead cores in a critical row and a warped frame from a leaky roof in 1997.

They needed an original working 4KB magnetic-core memory board — IBM part number 373917, from a Type 1406 unit — with no substitutions, no reproductions, no modern equivalents. The board had to actually work. It had to hold data. It had to have been made before 1965.

We didn't want it to look right. We wanted it to be right. Anything else and the machine is a museum piece pretending to be a computer.

The parts we needed had been out of production for six decades. Most 1401s were scrapped for their gold-plated contacts in the early 1970s. Surviving cores are typically in the hands of a small circle of collectors, most of whom aren't selling.

The search

We started where anyone would start: the Computer History Museum's contacts, the classiccmp.org mailing list, the small network of 1401 restorers still active in the United States and the Netherlands. Everyone knew of a board. Nobody had one to sell.

Two leads went cold. A collector in Utrecht had a 1406 chassis but no working memory. A retired Univac technician in Minnesota had photos of a board from 1978 and no idea where it had gone. We wrote to seventeen former IBM Customer Engineers listed in the SHARE alumni directory. Eleven responded. Ten had nothing.

The eleventh was a man named Don, retired since 1994, who had spent his career at the IBM Poughkeepsie plant and later servicing 1401s in the field across upstate New York. He mentioned, almost in passing, that he had one on his bench. He'd been meaning to test it.

The find

Don's board had come off a 1401 at a Rochester insurance company in 1984, pulled during decommissioning. He'd taken it home, put it on the bench, and it had sat there for forty years, next to a soldering iron and a coffee cup with the IBM logo worn off one side.

We drove to Poughkeepsie in March. Don had already powered the board up on a bench jig he'd built himself in 1985 for exactly this purpose. Every core held. No dead bits. The frame was straight. The sense wires were intact. The board was, by any reasonable standard, the best surviving example we had ever seen.

He named a price that was fair to both of us. We paid it.

The delivery

The board flew to San Jose in a custom foam-lined Pelican case with a humidity indicator and a handwritten inventory tag from Don. The museum's lead restorer met us at the airport. Installation took an afternoon. The 1401 read a deck of punched cards for the first time in twenty-three years the following Tuesday.

The museum held a small power-on demonstration in June. About sixty people came. The machine printed a payroll register from a 1963 test deck at 600 lines per minute, exactly as it had when it left the Endicott plant.

Don was invited. He didn't come, but he asked for a photograph. We sent him one, framed.

The bit that mattered wasn't the memory. It was that somebody had cared for it for forty years without ever needing to.