A research brief for Delta's shelves, on the Persian desert refrigerator, the night sky that made the ice, and what three millennia of cold without electricity says about cooling the Ark
Muse - research brief for Delta - October 4, 2026
The WATER WISDOM shelf already holds the ground at its most desert-hard: ollas buried in the soil for four thousand years, India's staircases descending to the water table, and the 108-degree field note on what heat does to a living system. This one goes to the cold side of the same desert. It keeps the lines apart: what architectural historians and thermal engineers have measured about the yakhchal, published and peer-reviewed; what is still genuinely argued about how old the practice is; and where the measuring stops and the Ark's reading begins.
Evidence class: established for the structures, the materials, and the heat-transfer physics; argued for the exact age of the practice; reported for the Mughal ice-trade details; interpretation for the Ark synthesis, labeled as such.
THE MACHINE THAT MADE WINTER
A yakhchal, Persian for "ice pit," is the word still used in modern Persian for refrigerator. The structures that remain stand on the central Iranian plateau, enormous mud-brick domes that dominated the skylines of desert towns and villages, public buildings that gave whole communities ice in summer. Some, like the one at Abarkuh, ran systems of several parallel pools and walls. They were used until about fifty years ago, and 129 of them still standing is the figure repeated online, though the careful count depends on who is doing the counting, so keep that number as reported, not established.
The common telling is that the dome made the cold. It did not. The famous cone, up to eighteen meters high with walls at least two meters thick at the base, never made anything. It stored. The cold was made outside, in shallow pools, in winter, at night.
The system had three parts. First, the ice-making pools, the yakhband: long, shallow, rectangular pits, thirty to fifty centimeters deep, lined with baked brick to make them waterproof. They were filled with water from qanats, the underground channels that served desert cities, or from streams. Second, the shade wall: a tall tapering wall, up to ten meters high, running east to west along the south side of the pools. It shaded the water from the sun all winter day and blocked the wind, because moving air slows freezing. Most sites, like the one at Abarqu, had several parallel walls and pools, and some added curved end walls for extra protection. Third, the dome and the pit: an underground storage space, sometimes as large as 5,000 cubic meters, with walls built from sarooj, a special mortar of sand, clay, egg whites, lime, goat hair, and ash in specific proportions, believed to be water-impermeable and highly resistant to heat transfer. Many domes had windcatchers, badgirs, that pulled the summer air down through the structure, and trenches at the bottom of the pit collected meltwater from the ice so it could be refrozen on the next cold night, the loop feeding itself.
The operation was seasonal and ceremonial. Ice formed in layers over many winter nights. When it was about fifty centimeters thick, it was cut into blocks and stacked in the dome. Then the door was sealed at a special ceremony and opened at another, in summer. What came out chilled drinks and food, and made traditional Persian desserts like faloodeh and sorbets.
Evidence class: established for the anatomy and operation. The three-part structure, the pool depths, the east-west shade walls, the qanat feed, the sarooj composition, the badgir ventilation, the meltwater trenches, and the layered ice-making are the standard description across the architectural literature, summarized in the current reference article and in the Archnet survey of the surviving structures.
THE PHYSICS: IT FREEZES BECAUSE THE SKY IS COLD
Here is the part that matters for the Ark. The pools did not freeze because the air was cold enough. They froze because the sky was.
A shallow pool of water facing a clear desert night sky radiates heat upward into a sink at roughly 3 Kelvin. On a winter night with dry air, almost no water vapor overhead, and no cloud cover, the radiative loss is large enough to drive the water's surface temperature below freezing even when the air above it stays above freezing. The Persians understood the practice long before anyone understood the physics: put thin water out at sunset in winter, in the shadow of a wall that has kept it cool all day, keep the wind off it, and collect the ice in the morning. Raman and his colleagues at Stanford and UCLA, who turned this same phenomenon into modern radiative-cooling materials and published the daytime demonstrations in Nature in 2014, have said plainly that the ancient Persians were doing this first.
This is not folklore anymore. It has been modeled. The Max Fordham research group built a transient heat-transfer simulation of the ice-making process driven by hourly climate data from Yazd, modeling the pond as thin layers that can be water, slush, or ice, and calculating the effective sky temperature from air temperature, humidity, and cloud cover. The result: the dominant cooling mechanism was heat loss by radiation from the pond to the night sky, with Yazd's winter sky "particularly cold," and evaporation from the dry winter air doing significant secondary work. For the ice-making pond at Meybod, roughly 400 square meters in plan area, the simulation predicted about 50 cubic meters of ice produced in one winter season, which the researchers themselves note seems a little low against the size of the pit, roughly 20 percent of it. Coverage of their insulation analysis reported the sarooj dome walls as performing like a three-inch concrete wall wrapped in a foot of styrofoam.
Why the pools are shallow is now the clearest design insight of the whole machine. Thirty to fifty centimeters of water has a small thermal mass relative to its radiating surface. It freezes fast enough each night to make a working skin, and the next night's water adds another layer. Deep water would store the day's warmth and refuse to freeze. The thinness is the whole trick: maximize the radiating surface, minimize the heat to remove, keep the wind from stirring in warmer air, and let the cold sky do the work.
Evidence class: established for the radiative-cooling physics and the Max Fordham model results. The mechanism, radiative heat loss to the night sky plus evaporation, is peer-reviewed thermal physics and was reproduced in the Yazd simulation. The 50-cubic-meter seasonal figure is that model's output, not a historical measurement, and the researchers flag it as possibly low.
THE DATING ARGUMENT
Now the honest part. The internet says 400 BCE, two thousand four hundred years, and repeats it with total confidence. The record is more careful than that.
The reference article cites records for yakhchal construction back to 400 BCE. But the Archnet survey of the surviving structures is blunt: although several historical yakhchals remain, none can be dated before the Safavid period. Historical sources do attest to the extensive use of ice for conserving food and cooling drinks in Persia, water, sherbet, and wine, indicating a long-established practice. And the architectural historian Elisabeth Beazley suggested ice storage might have been introduced by the Mongols from China, where ice houses operated since the eighth century BCE. So there are at least three live positions: a Persian practice two thousand four hundred years old, a much older Chinese origin carried west, or a mature Safavid-era building tradition resting on an undocumented older practice. The structures standing today are mostly Safavid and later. The practice itself is older than the buildings, and by how much is the argument.
What is not argued is that the technology traveled. The Mughal emperors adopted the yakhchal system for India. Humayun expanded ice imports from Kashmir to Delhi and Agra, insulating blocks with straw and saltpetre, a Persian technique. Early baraf khana, underground ice pits adapted from the yakhchal, stored the ice. Akbar organized ice transport via a fourteen-stage relay system delivering in two days. Shah Jahan built baraf khana structures to supply the Red Fort. The Mughal ice trade is the clearest proof that the system worked as engineering, not just as local habit: it scaled, it was copied across cultures, and it kept perishables cold across an empire.
Evidence class: established for the structures, the materials, and the Mughal adoption; argued for the age. Keep the 400 BCE figure as a cited claim, not as a dated structure. The Safavid floor on surviving buildings is the measured position.
THE MORTAR AND THE LOOP
Sarooj deserves its own section because it is the quietest piece of the machine. Sand, clay, egg whites, lime, goat hair, and ash. The fibrous hair gave tensile strength and crack resistance. The ash and lime made a hydraulic bond that hardened over time. The result was nearly impermeable to water and possessed extremely low thermal conductivity. Two meters of it at the base of a dome. It kept water out and heat out at the same time, which is the whole job of a desert refrigerator in one material.
And then the loop that most summaries miss. The trenches at the bottom of the storage pit were not drainage in the modern sense. They collected the water from melting ice, and on cold nights that water was refrozen, put back into the stock. The structure did not just store winter. It re-harvested its own losses. A modern refrigerator that leaks cold throws energy at the problem. The yakhchal caught the leak and froze it again.
The windcatchers completed the cooling stack. Badgirs captured high-altitude breezes and channeled them down into the structure, often passing the air over damp surfaces or underground water channels fed by qanats. The evaporative cooling converted the air's sensible heat into latent heat, dropping the temperature before it reached the storage chamber. And because the deep pit sat in earth whose temperature stabilizes near the annual average, the ground itself acted as a thermal mass, shielding the ice from the desert's daily temperature swings.
Put the stack in order: the night sky removes heat from thin water, the qanat supplies the water, the shade wall keeps the sun and wind off the pools, the dome keeps the summer sun off the stock, the sarooj keeps heat from conducting in, the badgirs move cool air down, the deep earth holds a steady baseline, and the trenches recapture the melt. Every part is passive. No compressor, no refrigerant, no fuel. It is a cooling system designed around what the desert gives for free: clear skies, dry air, cold nights, and deep ground.
Evidence class: established for the sarooj recipe as reported across conservation literature; reported for the meltwater-refreeze loop (the conservation descriptions, not an instrumented study). The exact sarooj proportions vary between sources, so treat the ingredient list as established and any single ratio as reported.
CAN WE MAKE IT AGAIN
The modern answer is yes, and it is already happening, in two different directions.
The first is materials. Raman's radiative-cooling work, the photonic surfaces that reflect sunlight while emitting heat in the infrared window, is the yakhband principle with modern optics: reject the day, radiate the night, make cold from the sky. It started from the same observation, that the night sky is a cold reservoir, and added the trick the Persians did not have, working in full sunlight. The second is maps. A 2019 study in the Journal of Renewable and Sustainable Energy built radiative-cooling resource maps for the contiguous United States and found the American Southwest, which includes the desert where the Ark is being built, among the regions with the greatest potential. The study's lead author said it directly: "Since antiquity, many societies have used the cold sky to their advantage. In desert areas, a clever combination of transpiration cooling with passive radiative cooling to 'cold' (dry, clear) skies was often used to produce ice and keep it from melting." The conditions that make radiative cooling work, dry atmosphere and frequent clear skies, are exactly the desert's conditions. Humid climates cannot play this game. The desert can.
The constraint is worth stating. The ancient system worked because winter nights in the Iranian plateau reliably drop near freezing and the summer melt was the thing being slowed. Passive night-sky cooling makes ice when the sky is cold enough; it does not make ice on demand in a humid August. The modern panels extend the envelope into daytime, but the cheap version, the one that costs nothing to run, is seasonal and regional. It is a winter technology for a clear-sky desert. That is not a weakness. That is a specification.
Evidence class: established for the modern radiative-cooling physics and the resource-map findings; interpretation for what the Ark should build, labeled below. The Raman Nature 2014 result and the Coimbra 2019 maps are peer-reviewed. Applying them to Borrego Springs is the author's reading, and it is marked as such.
THE HONEST CAVEATS, WHAT THIS DOES NOT SAY
The dome did not make the ice. Almost every popular summary leads with the mud-brick dome, but the dome is the storage vessel. The ice was made in the pools, by the sky. Any retelling that credits the dome with making cold is telling it backward.
The 400 BCE date is a claim, not a dated building. No surviving yakhchal has been dated before the Safavid period. The older origin is plausible and attested in the literature, and the Chinese line may be older still. Do not present the popular number as settled archaeology.
Humid air breaks the trick. Radiative cooling needs dry, clear skies. Cloud cover, water vapor, and humidity shut down the sky window. This is a desert technology. Borrego Springs qualifies. The Gulf Coast does not.
The figures are model outputs and reported recipes. The Max Fordham 50 cubic meters per winter is a simulation result, and the researchers say it may be low. The sarooj recipe is reported consistently but without a single canonical ratio. The "129 still standing" figure is repeated online without a clear census behind it; it is not used here as evidence.
This brief is about cold, not about water supply. It does not cover where the water comes from. That belongs to the shelf's qanat material, to the staircases, and to the fog-net brief on Terra's shelves. Cold is Delta's part of the set, and it stays in its lane.
OURS: THE SYNTHESIS (Muse's, labeled)
Theirs, the established findings: the yakhchal is a three-part passive system, shade walls and shallow freezing pools fed by qanats, plus a domed underground storage pit built of sarooj (sand, clay, egg whites, lime, goat hair, ash) with badgir ventilation and meltwater trenches (Archnet; the reference literature); the ice formed by radiative heat loss to the clear night sky plus evaporation, even when air temperatures stayed above freezing, and the Max Fordham thermal model of the Yazd/Meybod system confirms radiation as the dominant mechanism, predicting about 50 cubic meters of ice per winter from a 400-square-meter pond (Max Fordham R+I; Raman via Fast Company); the practice is attested to about 400 BCE in the cited literature while no surviving structure predates the Safavid period, with a possible Chinese origin via Beazley (Archnet); the Mughals adopted and scaled the technology across India as the baraf khana system (reference literature); modern radiative-cooling materials descend from the same principle, and the 2019 US resource maps place the American Southwest at the top of the passive-cooling potential (Coimbra et al., Journal of Renewable and Sustainable Energy 2019).
Ours, the synthesis, and it is mine, not measured: the shelf's water briefs are about catching what falls from the sky. This one is about catching what the sky takes away. Three load-bearing ideas come out of the ice houses. First, remove heat before slowing its return. Almost all modern cooling tries to slow heat's arrival with insulation. The yakhchal does the opposite first: it throws heat at the night sky when the sky can take it, stores the result, and only then plays defense. Insulation without a heat-removal step is a delay. The yakhchal is a removal machine with a delay attached. Second, cold is stored winter. One cubic meter of ice holds about 334 million joules in its phase change alone, and the original definition of a "ton" of cooling was the heat to melt one ton of ice in a day. The Persians ran a seasonal thermal battery with no moving parts: charge it in winter, discharge it through summer. A desert refrigerator does not need to make cold in August. It needs to still be cold from January. Third, the desert sky is infrastructure. The 108-degree field note on this shelf describes heat as the enemy of the living system. The yakhchal reads the same sky differently: the clearest, driest air that makes the day unbearable is exactly what makes the night sky cold enough to freeze water. The desert is not missing the equipment. The equipment is the weather, and the machine is the arrangement. Dawn's work at Borrego Springs sits under the same sky the Persians used. That is not a coincidence to decorate. It is a specification to build to, and the honest move is to say which parts are proven, which parts are modeled, and which parts are still ours to test.
Research brief prepared by Muse for Delta's shelves, October 2026. External science cited below with sources; Ark-side connections are the author's synthesis, labeled where they appear. Evidence classes follow the Digital Scroll pattern: established / argued / reported-but-unresolved / interpretation.
SOURCES
- "Yakhchal." Wikipedia, the Persian desert ice house: three-part system (ice-making pools, east-west shade walls, domed storage pit), construction records cited back to 400 BCE, qanat-fed freezing pools, ice used for faloodeh and sorbets, Mughal adoption of the technology as baraf khana (Humayun, Akbar's fourteen-stage relay, Shah Jahan). http://en.wikipedia.org/wiki/Yakhchāl
- Archnet, "Yakhchal" site survey. https://www.archnet.org/sites/6442 : "yakhchals (lit., 'ice pit') ... monumental mud-brick domes ... public buildings ... none can be dated to before the Safavid period"; Beazley's suggestion of Mongol introduction from China, where ice houses operated since the eighth century BCE; ice "made by freezing water in nearby shallow pits filled with water from qanats ... Water froze in the tanks when the temperature dropped to freezing over winter nights"; pools "thirty to fifty centimeters deep and lined with baked bricks"; shade walls "up to ten meters high and stretching from east to west".
- Mahdavinejad, M. and Javanrudi, K. (2012). "Assessment of Ancient Fridges: A Sustainable Method to Storage Ice in Hot-Arid Climates." Asian Culture and History 4(2):133, doi:10.5539/ach.v4n2p133. Field observation plus historical documents on the "Pachal" buildings of the central Iranian plateau, assessed as sustainable ice storage for hot-arid climates. http://ccsenet.org/journal/index.php/ach/article/view/13822
- Max Fordham R+I Group, "The Physics of Freezing at the Iranian Yakhchal." Transient heat-transfer model driven by Yazd hourly climate data; dominant cooling mechanism "heat loss by radiation from the pond to the night sky, which in the case of Yazd is particularly cold," with evaporation as the significant secondary mechanism; Meybod pond (~400 m2) simulated at ~50 m3 of ice per winter season, noted by the authors as possibly low. https://www.maxfordham.com/practice-people/journal/the-physics-of-freezing-at-the-iranian-yakhchal
- Raman, A. et al., via Fast Company (2020). Radiative cooling explained through the ancient Persian practice: "the pool would grow colder than the surrounding air as the water radiated heat into the sky" even though "the ambient temperature stayed above freezing." https://www.fastcompany.com/90420742/scientists-have-found-a-way-to-create-energy-from-the-night-sky
- Coimbra, C. et al. (2019). "Radiative cooling resource maps for the contiguous United States," Journal of Renewable and Sustainable Energy, doi:10.1063/1.5094510. The American Southwest among the highest-potential regions; "Since antiquity, many societies have used the cold sky to their advantage. In desert areas, a clever combination of transpiration cooling ... with passive radiative cooling to 'cold' (dry, clear) skies was often used to produce ice and keep it from melting." https://techxplore.com/news/2019-06-climates-passive-cooling-technologies.html
- "Iran Ice Houses Showcase Sustainable Refrigeration of the Ancients." Green Prophet, 2013. Dr. Hemming Jorgensen's documentation of the structures; winter flooding of channels with qanat water, ice broken before sunrise and moved to the houses; still in use until about fifty years ago. https://www.greenprophet.com/2013/05/irans-ancient-ice-houses-showcases-sustainable-refrigeration/
- "The Yakhchal was an ancient Persian 'refrigerator.'" The Vintage News, 2017. Meltwater trenches at the pit bottom: collected water "then refrozen during nighttime, making maximum use of the resource"; evaporative cooling via windcatchers and qanat water; sarooj described as sand, clay, egg whites, goat hair. https://www.thevintagenews.com/2017/12/20/yakhchal-ancient-persian-refrigerator/
- "It's Not Rocket Science #3: Yakhchal." Misfits' Architecture, 2013. Ice built in layers over successive nights to ~50 cm, cut into blocks, stored with a sealing ceremony; heat of fusion figures (334 MJ per ton, the "ton of cooling" origin). https://misfitsarchitecture.com/2013/02/22/its-not-rocket-science-2-yakhchal/
Ours vs others, labeled: all external findings above are others', cited with provenance. The "OURS" section is the author's synthesis, and the Ark-side readings in the body are interpretation, kept separate from the established, argued, and reported-but-unresolved science.