The Manufactured Bell

There are two ways to make light, and each leaves a different shape. A hot filament sits at equilibrium with the field around it and spreads its output across every scale at once, leaving a power law with a thermal cutoff and putting five per cent of itself in the visible band. A light-emitting diode is held away from equilibrium and forced through one transition, leaving a bell under two per cent wide and putting ninety per cent of itself where it is wanted. This paper measures that difference and follows what it implies. The dial that sets a diode's colour is a length — the spacing of its lattice — rather than a temperature, and the light it emits corresponds to a body at six thousand degrees while the chip stays at room temperature. What tips each electron over is the zero-point field, which holds half a photon's worth of energy at the emission wavelength and can be reshaped by the crystal to change the emission rate by thousands. The device's emitters are alike as channels into the same field modes, so nothing has to be selected or addressed, and the whole remaining loss sits in getting the light out of the crystal. One construction — something rising crossing something falling — produces a peak in nine separate places, from fusion in a star to a watched double slit; that much is calculus, and what differs between them is what fixes the width. Two open problems in the subject are stated with their numbers, and one measured device that exceeds unity by running as a heat pump. This record holds one paper on its own and is refreshed every couple of months. The complete deposit, 10.5281/zenodo.21788037, holds every paper with its code and data and is updated first; for the newest version of this paper, go there.