Night-sky radiative cooling

The foundational physical principle behind the entire archive. Baer’s “Night and Day” essay opens with the reversal that makes it all work:

Does cold air cool the ground, the trees, and the grass at night? No, it is the other way around; the air is cooled by contact with the ground, trees, and grass, which first cool themselves by shining heat into space, and then pull the clear atmosphere’s temperature down behind them.

Air temperature, Baer argues, is “vastly over honored and over measured.” We accept it as the index of all heating and cooling, but surfaces do the real work — radiating directly to space. Night-sky cooling is not mediated by air; it is a direct exchange between surface and cosmos.

The infrared argument

Baer frames the problem as one of perception. Our eyes work by reflected sunlight — faithful to shape but telling “nothing of all important temperature, for colors are only surface decor, not temperature readings.” An infrared camera reveals that objects are already “red hot” at room temperature. If we could see in the infrared:

We would find the night a busy but gentle scene, with softly glowing clouds crossing an icy blue sky, with every object telling its own temperature story in color. The colors would be the indicators of the energy gained during the day as it is shined away at night. Gusts of wind would brighten blades of grass, or leaves of trees, as heat passed from the atmosphere to them. Massive rocks would glow all night, according to their thickness and previous exposure.

This is the archive’s case for why night-sky cooling has been overlooked: we literally can’t see it happening. The sun “directs all our attention to incoming energy” and “overwhelms the media we might hope would tell us the rest of the story, that of departing energy.”

How the archive builds on it

The physical fact — surfaces with a clear sky view can cool below ambient air temperature — is the foundation for three systems:

System	Application
Cool Cell	Water reservoirs in ceiling + roof radiator/absorbers. Thermosiphon circulation at night; no electricity for cooling.
Di-thermal roofs	Unglazed roof panels that collect solar heat by day (winter) and radiate to sky at night (summer). Same surface, dual function.
Pool-heater cooling	Retrofit: run existing swimming pool solar collectors at night to cool buildings.

The heated vs unheated plates experiments provide the empirical grounding: on clear, dry New Mexico nights, a horizontal plate with 9.5 W/ft² of electric heat cannot reach ambient air temperature. The sky is pulling heat away faster than the heater can supply it.

Prior art: Yakhchal (by 400 BC)

The earliest known precedent for this physics in the archive isn’t 20th-century science — it’s ancient engineering. Persian yakhchals used qanat-fed water frozen overnight in shaded, wind-protected channels, then stored as ice in insulated domed buildings (walls up to two meters thick), sustaining ice through the desert summer with no instruments and no electricity. The same core elements recur nearly 2,400 years later at Zomeworks: water as thermal storage, an arid clear-sky climate, and a surface exposed to radiate heat away at night.

Prior art: Daniels synthesis (1964) and Bliss experiment (1961)

Before Hay’s Phoenix prototype, the physics was already documented in scientific literature. Farrington Daniels, Direct Use of the Sun’s Energy (Yale, 1964, §13.3):

“Important cooling of houses can be achieved in hot clear climates by radiation to the sky during the night. This may amount to 10 to 35 BTU ft⁻² hr⁻¹, and more. In the desert of north Chile it is possible to freeze water by this means when the air temperature is far above freezing.”

“Bliss has made use of this nocturnal radiation in a house in Arizona, with a black cloth radiator 280 ft² in area. The ‘cold’ amounting to 120,000 BTU was stored in a 10-ton pile of rocks buried in the ground. The cooling was equivalent to about 2 tons of refrigeration.”

This is 1961 experimental data (Bliss presented at the 1961 UN Conference on New Sources of Energy, Rome) — six years before Hay’s Phoenix prototype, four decades before the Cool Cell. The 10–35 BTU/ft²/hr figure (2.9–10.3 W/ft²) is directly consistent with Baer’s heated-vs-unheated-plates finding of “night radiation exceeds 9.5 W/ft²” (2002).

Daniels also notes explicitly: “Solar collectors for heating should be covered with glass plates… but the collectors should be uncovered when they are to be used for cooling by nocturnal radiation.” This is the di-thermal surface concept, published in 1964.

The 1958 Yanagimachi paper (cited by Daniels) proposed combining “Solar Energy, Nocturnal Radiation Cooling, Radiant Panel System of Heating and Cooling, and Heat Pump” as a complete year-round air-conditioning system — the Cool Cell concept, 44 years early.

Prior art: Harold Hay’s Skytherm (1967)

The principle was demonstrated at scale before Baer’s work. Harold Hay built a 120 sq ft Phoenix prototype in 1967 and a 1,100 sq ft Atascadero, California house in 1973 — both using water bags on flat roofs exposed to the night sky, covered by movable insulating panels during the day. The Atascadero house maintained 68–72°F indoors year-round without electricity. Hay’s formulation: “You don’t need electricity to cool! You don’t need an air conditioner! You do it with the sky.”

Baer acknowledged the connection directly: he wrote the chapter “Harold Hay’s influence and the Zomeworks Corporation” in Activism in Architecture (Routledge, 2018). The key engineering difference between the two approaches: Hay’s water sits on the roof and doubles as the radiator; Baer’s Cool Cell keeps water in the ceiling and connects to a separate roof radiator via thermosiphon, allowing the system to work in conventionally roofed buildings.

Climate dependency — does this work outside arid New Mexico?

Every measurement in this archive — the 9.5 W/ft² heated-vs-unheated-plates result, Hay’s Phoenix and Atascadero prototypes, the Cool Cell field data — was taken under clear, dry New Mexico skies. A 2026 research round investigating whether this physics still holds in a humid subtropical climate (prompted by a proposal to build a miniature Double Play-style cooler for an outdoor cat in South Texas) found the dependency is severe, not marginal:

The mechanism: radiative cooling works because a surface’s heat escapes through the atmosphere’s “window” to space. Water vapor closes that window — Luciuk’s physics primer shows net outgoing radiation dropping from 98.9 W/m² (clear) to just 15.1 W/m² (overcast) at the same temperature, a 6.5x difference. Aili, Yin & Yang (2021) measured radiative cooling weakening as humidity rose from 13% to 32% RH, but even their “humid” test case is far drier than a Gulf Coast summer night.
The alarming data point: a 2019 Solar Energy study found a radiative cooling surface running 6–7°C ABOVE ambient in humid subtropical Hong Kong versus 4.9°C below ambient in dry Stanford, CA. South Texas summer humidity commonly exceeds the Hong Kong test conditions. This specific comparison was surfaced from a search snippet (the full paper is paywalled) and should be independently re-verified before being treated as settled, but if it holds, a radiative-cooling surface in South Texas summer could become a net heat source, not a cooling mechanism — the exact opposite of what this archive’s New Mexico-derived physics assumes.
The regional data: NWS Brownsville’s Rio Grande Valley climatology shows summer heat index at 108°F+ on two-thirds of days, with remarkably low day-to-day variance — oppressive heat is the norm, not a spike. WeatherSpark’s RGV normals show “muggy” days jumping from 4/month in January to 30.8/month in July, and even the clearest month (May) is only 71% clear/partly-cloudy — nothing close to high-desert sky clarity.

Correction — the actual prototype site is San Antonio, not the Rio Grande Valley. A hyperlocal weather station near the build site placed it at 29.41°N, 98.48°W — roughly 220 miles from the RGV data above, in a different NWS zone with a meaningfully different (though still humid subtropical) climate: average August high ~97°F, relative humidity swinging from ~83% at 6am to ~48% at 3pm, and actual winter freezes most years (unlike the RGV’s ~5-events-per-century record). Afternoon humidity is notably lower than the RGV’s, but the ~83% early-morning/overnight figure matters most here, since that’s when radiative cooling would need to work — so the climate-mismatch concern above likely still applies, just somewhat less severely than the RGV data alone would suggest.

Working conclusion: night-sky radiative cooling should be treated as a real but secondary, supplemental effect outside arid, low-humidity climates — not the primary cooling mechanism the archive’s New Mexico work treats it as. In a humid subtropical climate it may be neutralized or reversed during exactly the months cooling is most needed. See Designing a miniature Double Play experiment for how this reshapes a small-scale build outside New Mexico.

The asymmetry between heating and cooling

The Cool Cell brochure makes an important point about reliability: “Solar gain is not as reliable as radiant cooling; the winter sun can disappear for days, so heat storage must be greater for winter comfort.” Radiant cooling is the more dependable side — night is never more than 12 hours away, so coolth can be replenished on a reliable 24-hour cycle. This asymmetry shapes the entire system design: cooling is passive and automatic; heating needs pumps, louvers, and more storage.