IBM's tape breakthrough involved a vacuum cleaner and baby pants
An experimental membrane helped engineers solve a mechanical bottleneck between fast electronics and heavy storage reels.
IBM’s shopping list for a storage breakthrough included baby pants. The computer business had a buffering problem, and someone went to the drugstore for the buffer.
The company’s historical account describes James A. Weidenhammer’s team trying to make thin magnetic tape start and stop without snapping. Heavy reels were the obstacle. The short length near the read-write mechanism needed to respond faster than all the tape wound around a reel could reasonably accelerate.
A loose loop could absorb the difference, provided the engineers could control it. After other arrangements failed, the team tried a vacuum cleaner. Blowing air into a column worked poorly; sucking from below held the loop as needed. Jack Seely fetched baby pants to provide a thin, flexible membrane for the experiment.
The membrane helped air leave faster than it entered. That partial vacuum made the improvised parts useful together. The domestic detail is funny because it sounds so far removed from data storage, yet its job was quite specific: provide a flexible valve while the engineers tested an idea.
Weidenhammer called for “storage loops or buffers,” according to IBM. The important word is buffers. Storing a little slack between two mechanisms lets them move on different schedules, which sounds much less exotic once you’ve watched a charger cable snag because the object at one end moved before the other did.
The IBM 700-series account helps explain why storage needed attention. It describes the 701 using electrostatic storage, drums and tape, with tape holding millions of digits per reel. The machine could calculate far faster than earlier equipment. Faster computation only helps when the rest of the machine can keep supplying and accepting information.
A larger reel therefore solved one problem while worsening another. It held more tape, but the same mass became a burden during fast starts and stops. Asking for a stronger motor would have left the thin tape dealing with the consequences. The loop offered somewhere for the timing difference to go.
This is also a better question to bring to a slow system than demanding a faster component: which two parts are being forced to move together? A buffer may introduce room to breathe, although it also creates something you must measure and control. The tape loop makes that bargain visible in a way a queue-length chart never quite manages.
The next useful storage trick could have an unglamorous first version. Whether it deserves to survive depends on the mismatch it actually solves, and whether the experiment holds up once the improvised valve becomes a proper part.
Baby pants stayed off the spec sheet.