The Ring Head and the Recording Gap

Poulsen magnetised a steel wire in 1898 by dragging an electromagnet's pole along it. It worked and it was terrible: the field spread out over millimetres, so the shortest mark you could write was millimetres long. Eduard Schueller at AEG patented the fix in 1933 and it is still in every drive built today. Bend the magnetic circuit into a RING, and cut one narrow non-magnetic slot in it. Almost all the flux now stays inside the high-permeability core, because that is the easy path — except at the slot, where it has to jump. The fringing field that bulges out of that slot is intense, and it is confined to roughly the slot's own width. The gap does not conduct the flux; it is the one place the flux is forced out into the world, and that is the entire trick. The same ring reads. A magnetised medium moving past the gap pushes flux through the core, and a coil around the core sees a voltage. So the gap length decides both halves at once: the shortest mark you can write, and the shortest wavelength you can read. When the recorded wavelength equals the gap length, the gap averages over exactly one full cycle and the output goes to zero. Every advance in magnetic storage for the next sixty years is that slot getting narrower — 25 micrometres in the 1930s, 2 in a floppy, half a micrometre in a hard disk, a tenth in thin film. This blueprint is fully buildable: wind a coil on a ferrite core, cut a gap with a shim, and measure both the 6 dB per octave rise and the cliff it falls off.