The Disappearing Night Sky

AWhen European municipalities began exchanging their orange sodium lamps for white diodes, the change was presented as a rare case in which thrift and environmental care pointed the same way. The new fixtures drew less current, lasted longer and could be aimed with a precision the old globes never allowed. Several councils reported cutting the cost of street lighting by 45 per cent within three years of conversion, and the sky, it was assumed, would darken as a by-product of that accountancy. What followed did not match the forecast. Radiometers aboard weather satellites found that the lit area of the Earth's surface expanded by roughly 2.2 per cent a year between 2012 and 2016, with the brightness of already-lit places climbing at much the same pace. That is a deterioration, but a gradual one, and it is not what observers on the ground describe. Astronomers at long-established sites, and amateurs who had been counting the faintest stars visible from their gardens for decades, reported a far more rapid loss, so that two sets of trusted instruments appeared to be measuring different worlds.

BThe explanation that now commands most support was proposed by Sofia Bhatnagar, an atmospheric physicist at the Aurelian Institute for Sky Quality, and it concerns the instrument rather than the sky. The radiometers in question were designed to observe clouds rather than cities, and, like most such sensors, they register nothing at all below 500 nanometres. Sodium lamps emitted almost nothing there either, so for three decades the blind spot was harmless; white diodes, by contrast, put a substantial share of their output into precisely the region the sensor cannot register. Bhatnagar maintains that a city can accordingly replace its lighting, increase what it emits and appear from orbit to have changed hardly at all. Between 2019 and 2023 her group set the satellite series beside readings from some 500 volunteer stations equipped with identical hand-held photometers, an arrangement that removes most of the objections about differing equipment. The ground network recorded night-sky brightness rising by about 9.6 per cent a year, a figure she presents without adjustment. The discrepancy, she insists, is not a disagreement about the sky but a property of the sensor. Launching more of the same instruments will not settle it.

CNot everyone accepts that the volunteers are measuring what they believe they are measuring. Martin Escobar, a lighting engineer at the Kestenholz Polytechnic, contends that a hand-held photometer records the entire column of air above it, so that a run of humid or dusty years would inflate the trend without a single extra lamp being installed. He notes, too, that the lens of the human eye yellows with age, which makes long-running star counts by the same observers unreliable in a direction nobody intended. To test this he assembled a panel of thirty-one observers of different ages and re-analysed four years of records from one country, concluding that between a fifth and a quarter of the reported increase was an artefact of the method. The correction is real but narrow, and his argument trims Bhatnagar's figure without touching its sign. Escobar concedes that instruments running without any human operator, at stations that neither age nor grow tired, show the sky brightening in the same direction, and he has no account of why they should. That distinction is easily lost in summaries of the debate, which tend to present him as denying the trend altogether.

DHow a more efficient lamp can leave the sky brighter than before is a question with two answers, one belonging to physics and the other to procurement. The physical answer is not in dispute: scattering by air molecules rises steeply as wavelength falls, so a beam at 450 nanometres is dispersed several times more effectively than the amber light it replaced. A diode and a sodium lamp of the same electrical output therefore do not contribute equally to the glow overhead, even when both are shielded and both point downwards. The economic answer is less tidy: because light became cheap, more of it was bought, car parks that had once managed with four lamps were given nine, and hours of operation lengthened. Bhatnagar calculates that only where no ceiling was placed on installed capacity was roughly 60 per cent of the electricity saved returned to the network as additional illumination; in the minority of towns that did impose such a ceiling, consumption and glow both fell. Escobar accepts this second answer without reservation, observing that it belongs to purchasing policy rather than to optics. The saving was genuine; what was done with it afterwards was a decision rather than a consequence.

EUnusually among environmental problems, this one leaves no residue, and the sky above a town that switches its lamps off is dark again the same night. That fact has made the remedies unglamorous. Phosphors that hold the emission below 2,700 kelvin remove most of the short-wavelength component at a cost of a few per cent in efficiency, and curfews that dim residential streets after midnight shorten the hours during which glow is produced at all. A trial in the Norwegian town of Sørvik, where lamps were dimmed by 60 per cent between one in the morning and dawn, is reported to have cut sky brightness overhead by a third within a single season. Shielding, the measure most often demanded, addresses only the light thrown directly upwards and does nothing about the share reflected back from pavements. The writer's own conclusion is cautious: the technology is neither the villain of this story nor its hero, since a diode is simply a cheaper way of doing whatever a community has already decided to do. Reversibility is a genuine advantage, but only for a society prepared to write the rule that reverses it.