Point a good enough camera at a mouse in a dark box and the mouse glows. Not metaphorically. It emits a real stream of photons, a few tens to hundreds per square centimeter per second, faint enough that your dark-adapted eye would miss it and a decent EMCCD would not. In a 2025 paper that went around the world, a University of Calgary team imaged that glow, then euthanized the animals and imaged them again. The light went out. This Nature feature by Jo Marchant uses that image as a hook to ask a harder question: is the glow just exhaust, or is it a signal we could read?

Worth separating two things right away, because the press coverage did not. The phenomenon is old, boring, and solid. The clinical dream stapled to it is new, exciting, and mostly unbuilt. This piece is about telling those apart.

The reason the mouse study earned its virality is a clean piece of experimental hygiene. The obvious objection to any "living things glow" claim is that warm objects radiate. A 37 °C mouse is a blackbody; of course it emits. So the team held the dead mice at 37 °C, and kept the imaging stage at 37 °C too. Temperature was pinned; only life was subtracted. The glow still collapsed to a few residual bright spots. That is the control that turns a pretty picture into an actual result: the emission tracks live metabolism, not body heat.

The apparatus was honest work. An Andor iXon 888 EMCCD cooled to about −95 °C, hour-long exposures, single-photon sensitivity. The plant arm reproduced known biology (cut a leaf, watch a burst of light bloom at the wound where reactive oxygen floods in) and turned up one genuinely odd result: a local anesthetic, benzocaine, drove more emission at the injury site than hydrogen peroxide, the standard booster. Interesting. Unexplained. Flagged, to their credit, rather than buried.

Now the sober part. This is n = 4 mice. The live-versus-dead contrast is dramatic but low in information: a dead animal has stopped metabolizing, so the finding confirms that a light powered by metabolism goes dark when metabolism stops. True, tidy, and not far from tautological. The images have essentially no spectral resolution, which matters enormously, because every proposed clinical use depends on the glow carrying a readable fingerprint of which tissue is stressed and how. Showing that the lamp is on is not the same as showing the lamp encodes a message.