Over a century ago, engineers ran engines, made ice, and pumped rivers using nothing but focused sunlight. The technology worked. We abandoned it for cheap oil. This is the case for switching it back on — and building it into the grandest structures we ever made.
A child's magnifying glass bunches sunlight into a dot hot enough to burn paper. Make the lens big enough and that dot boils water. Boiling water makes steam; steam spins a generator; that's electricity — no fuel, no smoke, no bill. Set one in every bay or hundreds across a district — the count follows the building and the size of the lens. A bigger lens gathers more, but large lenses are hard to build and harder still to cool, so the honest design is many modest spheres rather than one heroic one. Either way the structure itself becomes the power plant.
This isn't science fiction. It's a 250-year-old idea that was demonstrated, commercialized, and then shelved when coal and oil got cheap. The physics never stopped working. We just stopped building it.
Every entry below is documented. Where a number is a period claim and not an independent modern measurement, it's marked — because the honest record is the one skeptics can't break.
Swiss scientist Horace-Bénédict de Saussure builds the first known Western solar oven — an insulated, triple-glazed glass heat trap that reached roughly 110 °C from sunlight alone.
His "heliothermometer" was a simple wooden box with three panes of glass that let sunlight in but trapped the heat. He carried versions up into the Alps to prove the warmth came from the sun itself, not the air. Every greenhouse, solar collector, and "hot box" cooker since descends from this experiment.
Glazing — sunlight passed through three stacked panes of glass on top of a cork-insulated pine box.
Black interior — it struck a soot-blackened inside and was absorbed as heat.
The trap — the panes let visible light in but blocked the outgoing heat and stifled convection, so warmth piled up.
Read-out — a thermometer at the bottom logged the temperature. No lens, no mirror — just trapped sun.
The ~110 °C (230 °F) figure is de Saussure's own 18th-century reading, never re-measured on the original box. It's the direct ancestor of every greenhouse and flat-plate solar collector.
Chemist Antoine Lavoisier and colleagues run a long series of experiments (from 1772) with the great Trudaine "burning lens" — a two-part concentrator that melted metals and even burned diamond. Even so, it couldn't melt platinum — the practical ceiling of 18th-century solar concentration. Platinum finally fell in 1782, to an oxygen blast, not a lens.
The apparatus rode on a big wheeled carriage so it could track the sun. Its primary "lens" was actually a huge sealed cavity of alcohol between two curved glass shells, feeding a smaller solid-glass lens — together bringing the sun to a spot under an inch across, hot enough to melt gold and burn diamond before crowds in Paris. But it never even softened platinum (m.p. 1,768 °C) — the era's ceiling — until Lavoisier cracked that in 1782 with pure oxygen, not light.
Liquid primary — a large lens-shaped cavity filled with alcohol (spirits of wine) gathered the sun.
Glass auxiliary — a smaller ~22 cm solid-glass lens tightened the beam further.
Focus — the rays met in a spot under ~2.7 cm across.
Result — melted metals and burned diamond to gas — but stalled at platinum.
The popular "~89 cm burning glass" is wrong — the primary was a large alcohol-filled liquid lens, not a glass disc. Burning the diamond helped prove it was made of carbon.
Frenchman Augustin Mouchot drives the world's first sun-powered steam engine — a blackened copper boiler under a glass bell, cradled in a silvered truncated-cone reflector (not a parabola) that concentrated the sun onto it.
His design was elegantly simple: a copper cauldron of water under a glass bell, cradled in a polished conical reflector that focused the sun onto it. The trapped sunlight raised steam to run an engine. Napoleon III funded the work, and Mouchot spent years building ever-larger sun-tracking reflectors.
Cone reflector — a silvered truncated cone ("like a lampshade"), ~2.6 m across the mouth, made of twelve removable silver sectors, aimed at the sun.
Boiler — it threw the light onto a blackened double-walled copper boiler holding ~20 L of water on the axis.
Glass bell — a glass "cloche" around the boiler trapped the heat, greenhouse-style.
Engine — the steam ran a small (~½ hp) engine. Origin Tours 1865; shown to Napoleon III in 1866.
The dimensions and Tours performance figures (20 L to 2-atm steam in ~40 min) are period specs for a device that no longer exists, never re-measured. The reflector was deliberately conical, not parabolic — the common "first parabolic trough" label is wrong.
The machine solar had to out-shout: a single 1,400-horsepower Corliss steam engine, 45 ft tall, drove 13 acres of machinery through more than a mile of shafting — the beating heart of Machinery Hall. President Grant and Emperor Dom Pedro II started it together on opening day. World's Fair
Here's the honest detail that kills a popular myth: the 1876 Centennial gave no gold, silver, or bronze grades at all. Every winner — Bell's telephone included — received the same uniform bronze "Medal of Award," and the real honor was the judges' signed written report. So when you read that Bell "won gold for electrical equipment" at Philadelphia, it never happened: there was no gold to win. Steam, not sun, ruled this hall — and that dominance is exactly what solar was up against.
The famous engine made no steam itself. Open fires were banned inside the timber-and-glass hall, so it was fed from a detached boiler house and simply distributed power across the floor:
Boiler house — 20 coal-fired tubular boilers in a separate building on the south side of the hall (no furnaces allowed inside), each rated ~70 nominal hp.
Steam main — steam piped in through a service of pipes; one engineering account gives an 18-inch main ~320 ft long.
The engine — twin vertical cylinders (44-in bore, 10-ft stroke) drove a walking-beam and a 30-ft, 56-ton flywheel at ~36 rpm — about 1,400 hp at 60 psi.
Line shafting — the flywheel turned underground and overhead shafts more than a mile long, belted up to each stand.
The floor — that single engine powered ~13 acres of working machines throughout Machinery Hall.
Period figures vary — steam pressures from ~25 to 80 psi appear across sources, and ~2,500 hp was claimed at 80 psi. Treat the pressures as contemporary claims, not modern measurements.
Mouchot and his assistant Abel Pifre demonstrate the solar engine at the Universal Exhibition — a mirror over 13 ft across driving a boiler to seven atmospheres — and win a Gold Medal in Class 54, most famously for making ice from the heat of the sun. World's Fair
The great ~5-metre reflector stood in the open on the slopes of the Trocadéro, in the annex of the Algerian exhibition — colonial Algiers had put up 5,000 francs to build "the largest mirror in the world." The crowd-stopper was counter-intuitive: solar heat driving an absorption refrigeration cycle (a Carré machine) to produce ice. It so impressed the judges that Mouchot took the Gold Medal in Class 54 and was made a Chevalier de la Légion d'honneur. Four years later, in 1882, Pifre used the same kind of sun-powered engine to run a printing press in the Tuileries Gardens, printing a newspaper aptly named Le Journal du Soleil — "The Sun Newspaper."
Concentrator — the giant ~5 m silvered reflector focused the sun onto a central blackened boiler.
Steam — the boiler raised pressurized steam (period reports cite ~7 atmospheres).
Absorption chiller — that steam drove an Edmond Carré ammonia/water absorption refrigeration machine.
Ice — the absorption cycle chilled hard enough to freeze water — blocks of ice, from nothing but sunshine.
In Bombay, William Adams runs a solar steam plant off a battery of 16 flat-mirror reflectors aimed at a central boiler, driving a hired 2.5-horsepower pump daily for a fortnight — and publishes Solar Heat: A Substitute for Fuel in Tropical Countries (on Internet Archive).
Adams thought Mouchot's single giant dish was fragile and hard to aim, so he arrayed many small flat silvered mirrors on a semicircular rack, each angled at a central boiler — the direct ancestor of every modern "power tower" heliostat field. His 1878 book argued solar could replace coal across the tropics for everything from cotton gins to cremation. You can read the original on Internet Archive.
Mirror battery — 16 concave compound reflectors, each built of many small flat silvered glasses, ringed the boiler on a circular tramway (three-quarters of a circle, a gap left for the engine room).
Central boiler — from ~20 ft away they all threw sun onto one beaten-copper boiler holding 12 gallons, raising steam to ~35 lb.
Pump — the steam ran a hired 2.5-hp steam pump at ~30 lb — "the first steam engine ever worked in India by solar heat."
Tracking — workers shoved the mirrors eastward 4–5 times a day to follow the sun. Manual, not automatic — but the direct ancestor of the heliostat power-tower field.
The widely repeated "36 mirrors / 3 hp" belongs to Adams's experiments and proposals, not this working demo (16 mirrors, a hired 2.5-hp pump). His famous "over 1,140°" focus was an extrapolated supposition — the thermometer only read past 670° for 36 glasses.
The Exposition that raised the Eiffel Tower (18,038 iron pieces) also raised the Galerie des Machines — an iron-and-glass hall vaulting a record ~115 m in a single span with no internal columns, the largest roofed room yet built. The age of iron architecture, on full display. World's Fair
The Galerie borrowed three-hinged arches from bridge engineering to leap 115 metres clear — a structural feat that made the whole hall one uninterrupted volume. Unlike Philadelphia, 1889 Paris used a fully graded award system (903 Grand Prix, 5,153 gold medals, on down). It's the same ambition this project chases: let the structure itself be the machine.
Twenty iron arches — the roof rode on twenty giant trussed arches, a technique lifted straight from bridge engineering.
Three hinges each — every arch was pinned at three points: its two feet and its crown, making it "statically determinate."
Why hinges? — as the iron heated, cooled, or the ground settled, the arches simply rotated a little about their pins instead of cracking under bending stress.
The payoff — the arches sprang from foundation abutments and met overhead with nothing between: a clear span near 115 m and a completely column-free floor, the widest roofed room on Earth at the time.
The span is cited anywhere from 111 to 117 m (115 m is the usual round figure); crown height appears as ~45 m or an oddly exact 48.3 m — treat both as period/derived. Engineer Victor Contamin, architect Ferdinand Dutert.
At the World's Columbian Exposition, Westinghouse won the contract to light the fair with alternating current, underbidding Edison/GE ($399,000 vs $554,000). The "White City" blazed with ~93,000 lamps — the public debut of the power system that would electrify the century. World's Fair
This is the turning point in the honest story: not solar, but cheap, controllable electricity — soon fed by coal and Niagara's water — is what captured the future. (Like Philadelphia, Chicago gave a single uniform bronze medal, no gold grades; the "gold medal to a British locomotive" tale doesn't hold up.) The sun engine wasn't beaten in the lab; it was out-competed on the show floor and at the price of fuel.
Steam first — coal-fired steam engines spun the generators; the "electric" fair still began with fire and boilers.
AC generators — Westinghouse ran a Tesla-licensed polyphase alternating-current system rather than Edison's direct current.
Transformers — AC's trick: step the voltage up to send it efficiently across the grounds, then down again for use. DC at ~110 V could only reach about a mile.
The load — some 93,000+ incandescent lamps lit the "White City," plus motors — the public proof of AC that led straight to the Niagara Falls contract.
Westinghouse won by underbidding GE. Skip the popular "70¢-a-lamp / made no profit" bid story and the exact generator counts — those specifics don't survive checking; and by 1892 Edison's company had already merged into GE, so it wasn't quite "Edison vs Westinghouse" anymore.
At the 1900 Exposition, Valdemar Poulsen's Telegraphone — the first magnetic sound recorder — won a Grand Prix and recorded the voice of Emperor Franz Josef; that ~10-second clip survives as the oldest magnetic recording on Earth. World's Fair
We single this one out because, of all the fair "prizes" in this timeline, it's the most independently documented (Computer History Museum, ITU) — and the artifact itself still exists. The 1900 fair also left Paris the Grand and Petit Palais, the Pont Alexandre III, and the Gare d'Orsay. A reminder to hold every medal claim to this standard: a source you can point to, or an honest hedge.
Microphone — your voice became a varying electric current.
Recording head — that current fed a small electromagnet pressed against a moving steel wire, magnetizing each passing point in proportion to the sound.
The wire remembers — the pattern of magnetization stayed frozen along the wire.
Playback — running the wire back past a pickup head turned that magnetism into current again, heard through a telephone earpiece.
Oberlin Smith published the idea in 1888 but never built it — Poulsen made the first working machine. The 1898 prototype wound piano wire around a brass cylinder; the 1900 Paris exhibits used wire or steel band on spools. The Franz Josef clip survives at the Danish Museum of Science & Technology.
Aubrey Eneas installs a solar motor at the Cawston Ostrich Farm, South Pasadena: a truncated-cone reflector 33 ft across, 8,300 lb, 1,788 pieces of mirrored glass, boiling water at the focus to run a steam pump.
This is a real photograph of it: a truncated cone lined with nearly 1,800 mirror segments, towering over top-hatted visitors, focusing the sun onto a boiler at its mouth to pump irrigation water. Eneas's Solar Motor Company sold several units — until desert hail and windstorms, not the physics, wrecked the machines and the business.
Collector — a sun-tracking truncated-cone reflector ~33 ft across the mouth, lined with ~1,788 mirror-glass segments.
Boiler — the mirrors threw the sun onto a boiler hung at the cone's focus, flashing water to steam.
Engine — that steam ran a small (~10 hp) steam engine.
Output — the engine drove a pump lifting well water for irrigation.
Eneas's company advertised enormous pumping rates (a "1,400 gallons a minute" figure circulated on trade cards) — far beyond a ~10-hp engine at any useful lift — that power moves that much water barely 8 m even at perfect efficiency, and real pumps are not perfect — so read those flow rates as promotion, not measurement.
Portuguese priest-inventor Father Himalaya unveils a parabolic concentrator 13 m high with 6,117 mirrors over 80 m², on clockwork sun-tracking, focusing the sun into a 15 cm spot. Portuguese and period accounts credit it with a Grand Prize plus two gold and one silver medal (Wikipedia lists the two gold and one silver, without the Grand Prize). The museum's own caption claims a focus of 7,000 °F (~3,870 °C) — a contemporary figure, never independently re-measured. World's Fair
This is the actual machine, photographed at the 1904 Fair: it stood outdoors on the high ground southwest of the Palace of Forestry, Fish and Game (entered for judging under the Department of Liberal Arts), tracking the sun on a clockwork mount and bundling the light of 6,117 little mirrors into a spot fierce enough to melt metal — a genuine sensation at the same Fair whose grand architecture lives in the 1800s archive this project grew from.
Mirror array — 6,117 small glass mirrors (each ~123 × 98 mm) tiled an off-axis parabola of ~80 m² aperture — the glass itself totals ~74 m², so the two figures in the period sources don't quite reconcile.
Clockwork tracking — a clock movement turned the whole array to follow the sun across the day.
Focus — the light bundled to a ~15 cm spot about 10 m out, onto a magnesia-lined crucible (~45 × 60 cm).
Result — heat enough to melt iron in under a minute and fire-clay in ~3 minutes. It generated no power; it was built to prove how hot concentrated sunlight could get.
The ~3,800 °C / 7,000 °F figure was a period estimate, never independently measured (sources cluster 3,500–4,000 °C) — and it exceeds what the modern Odeillo furnace reaches, which is itself reason to read it as a claim rather than a measurement. The melting observations — iron, fire-clay, magnesia — are the solid evidence.
American engineer Frank Shuman builds the first commercial-scale solar plant at Maadi, Egypt — five ~60-metre parabolic troughs focusing sunlight onto matte-black boilers to pump Nile water for cotton irrigation. His earlier demo plant made 600 lb of steam an hour and pumped 3,000 gallons a minute.
Shuman was the closest the world came to a solar future before oil. His Sun Power Company's troughs were the direct ancestors of today's parabolic-trough CSP plants, and he openly proposed covering the Sahara with them to power the world. World War I scattered his engineers; he died in 1918; the plant was scrapped for parts.
Troughs — five sun-tracking parabolic-trough mirrors, each ~60 m (204 ft) long and ~13 ft wide, spaced apart so they wouldn't shade each other.
Absorber — each trough focused a line of sun onto a long matte-black boiler tube running down its focus.
Steam — the tubes boiled water into low-pressure steam.
Pump — the steam ran an engine driving a pump that lifted roughly 6,000 gallons (~23,000 L) of Nile water a minute for cotton fields.
The "600 lb of steam an hour / 25 hp / 3,000 gpm" numbers belong to Shuman's earlier 1910 test plant at Tacony, Philadelphia (572 "hot-box" collectors), not to Maadi. All are period figures from Shuman's own reports, never independently re-measured.
San Francisco's Panama-Pacific Exposition crowned itself with a 435-ft Tower of Jewels hung with over 100,000 cut-glass "Novagems" that caught and refracted the sun by day and blazed under 50-plus searchlights by night. A whole building dressed to throw light. World's Fair
It's the closest the fair era came to this project's own image — thousands of little lenses across a structure, turning a building into an instrument for light. (Honesty note: the Tower was an architectural centerpiece, not a prize-winner — and the "102,000 gems / 54 searchlights" counts come from the fair's own promoters, never independently re-counted.) The Palace of Fine Arts from this fair still stands in San Francisco today.
Skeleton — a wood-and-steel armature, 435 ft tall.
Skin — clad in "staff," the plaster-and-fiber material used for every temporary palace at the fair.
The jewels — about 102,000 mirror-backed Austrian cut-glass "Novagems," hung loose so they swung in the breeze, refracting sunlight into color by day.
Night — Walter D'Arcy Ryan aimed 50-plus searchlights at the tower, and the mirror-backed gems threw the beams back as a shimmer of light.
The 102,000 count is a promoter's figure (an older archive says "more than 90,000"), never re-counted; the gems were suspended, not surface-mounted, so the popular "1-7/8-inch jewel covering" detail is wrong. The tower was demolished in 1916, so 435 ft is a design figure.
World War I scatters the engineers; the Maadi plant is dismantled for parts; Shuman dies in 1918. Then coal and oil prices collapse in the postwar recession and again after 1929. Solar wasn't beaten in the lab — it was undercut at the pump. Lost, not stolen
This is the crucial point for credibility: there was no technical defeat. The machines worked. They were simply out-competed by suddenly cheap, abundant fossil fuel — and the few champions died or went to war. For two generations, the economics of cheap coal and oil buried a working technology. That's the honest story, and it's far stronger than any tale of suppression.
It is also a pattern worth naming: a technology can work, be published, and still go missing — no villain required. The same decades produced another line of energy research that went quiet, and I keep it under the same rules as this timeline: cite the source, mark the hypothesis, claim nothing the evidence hasn't earned. Why it went quiet I can't yet show, so I don't claim it went quiet for the same reason this did. The magnetics thread, in Field Notes →
Concentrated Solar Power runs right now: the Odeillo solar furnace reaches ~3,500 °C; California's Ivanpah is a ~392 MW solar-tower plant in its final year; molten-salt plants like Crescent Dunes and Morocco's Noor/Ouarzazate store the sun's heat (~565 °C) to make power after dark. Sun → heat → steam → electricity is mature, deployed, and bankable.
This is Ivanpah in California — 173,500 heliostats aiming sunlight at boilers atop three towers, exactly Adams' 1878 heliostat idea at industrial scale. The principle never changed; only the economics did. The future-cathedral vision is the same machine, brought home to the grandest structures we ever built.
Heliostats — 173,500 sun-tracking mirror units aim sunlight at the tops of three towers.
Receiver boiler — the concentrated light heats water in a tower-top boiler directly to ~550 °C superheated steam.
Turbine — that steam drives conventional steam turbines (Siemens SST-900).
Grid — ~392 MW gross. Ivanpah has no heat store, so it runs only in daylight.
After dark — molten-salt plants (Crescent Dunes; Morocco's Noor) instead heat salt to ~565 °C and store it, running the turbine for hours after sunset.
Exactly the 1876 boiler and the 1878 mirror, scaled up: sun → heat → steam → power. Only the size and the economics changed. (Ivanpah is slated to close in 2026; these are its design specs.)
No exotic science — just optics you can prove with a magnifying glass on a sunny day.
A lens or curved mirror collects sunlight over a wide area and bends it toward one point.
All that light piles onto a tiny focal spot — the "heat point" — driving its temperature up.
The heat point sits on a black receiver full of water (or salt). It boils into high-pressure steam.
Steam spins a turbine/generator — electricity — then condenses and runs the loop again.
Sunlight spread thin is gentle warmth — a black plate left in the open sun tops out around 80 °C — too hot to touch, nowhere near hot enough to drive a machine. Gather that same sunlight from a wide area and pile it onto a small spot, and the temperature climbs with how hard you concentrate it. That is the whole engine: the sphere's only job is to make one spot hot enough to be useful.
A clear solid ball is a real burning lens. It collects sunlight across its whole face and bends it twice — once entering the glass, once leaving — folding it into a tight, blazing focal region just behind itself. A bigger sphere simply gathers more total power. Where that focus lands follows simple, settled optics:
Here is the quietly powerful part. The focus doesn't wander as you change the size of the sphere — it stays at the same fraction of the radius, set only by the glass itself. A marble and a two-metre orb of the same glass focus at the same proportional spot. So you design the receiver plate once and scale the whole building around it: whatever the size of the sphere, the plate that drinks the light belongs in the same place.
And because a sphere never quite focuses to a mathematical dot — the outer rays bend a touch too hard, so the focus is an intense region, not a pinpoint — a flat plate is exactly the right receiver. You want fierce heat spread across a surface you can plumb, not a hole drilled into a single point.
The plate isn't a solid lump — it's a working block with three jobs stacked beneath the focal point. The light makes the heat at the top; everything below decides where that heat goes.
The finned core sits at the focus and turns concentrated light into fierce heat — the hottest point in the whole machine.
Beneath it, a fan pulls the hot air off the fins and pushes it down into the wall and floor channels. The temperature difference can even drive it on its own — no fuel, like a solar chimney.
Copper lines wrap the core. Run them one way and they harvest the heat as hot water or steam; run them the other and flowing water carries heat away to cool the plate on demand.
The plate doesn't only make heat for this instant. It drives that heat down into the building's stone, where it becomes a trapped entity — released slowly over hours, long after the sun has moved on. Boiling water for steam and power is just one way to spend it; the rest is stored in the mass and breathed back into the rooms. The sphere is the furnace; the walls are the radiator.
That this is real isn't theoretical: in January 2024 a decorative crystal ball ignited a pile of towels in a house in Essex, England, simply by focusing window-light. Multiply that by intent, scale, and thousands of units — and you have a power plant.
Put both engines together and the building becomes one machine: the sphere gathers the sun and makes the heat; the receiver hands it to the structure; the outer wall channels carry it down and the mass banks it, then radiates it back from the walls and floors. At the bottom the heat runs under the floor — beneath the pool, which it warms from below, and out under the side patios. Meanwhile cool air is drawn from a lower level and climbs separate interior pillars — the solid floors block it from drifting sideways, so the pillars are its only way up — to be dropped into each room near the top, where it falls, settles, and is then drawn back and recycled. Heat down the outside, cool down from within — a closed loop the heat itself keeps turning. Everything below is just this picture, part by part.
The lens makes power — but the structure itself does something just as old and just as real. It stores the sun's heat in its own stone, moves it through channels in its mass, and breathes cool air up from the earth. Hot from the top, cool from below, distributed through the body of the building.
Pushing heat through channels in a building's mass isn't new — it's the Roman hypocaust: hot air circulated under floors and up through wall flues to warm the whole structure. The idea never died. Medieval engineers built heat-storage hypocausts — granite piled over a furnace vault that radiated warmth for days; some 800–1,000 were built across northern Europe. Korea's ondol and the European masonry stove are the same principle. Modern engineers model the ideal heat channel at just 50–100 mm wide.
Sources: Bansal, Building & Environment 34 (1999); Low-Tech Magazine / Tvauri (Univ. Tartu); Hypocaust & Ondol (Wikipedia)
A rooftop solar collector is the modern furnace. Transpired solar collectors already heat ventilation air at 60–70% efficiency, raising it ~20 °C on a sunny day. Solar-fed radiant floors cut a building's heating fuel by 63–89% in climate studies. And the top-hot / bottom-cool geometry is measured: a passive solar-chimney + earth-tube system moved 252 m³ of air an hour with zero electricity and cut summer indoor temperature by ~9 °C. Every piece of your loop is independently proven.
Sources: ACEEE 2008 proceedings; Renewable Energy v.130 (2018) & v.175 (2021); MDPI Buildings (2026)
Look again at the loop above: the heat doesn't drift through the open rooms — it runs up the corners and pillars. Each of those is built as an active thermal chamber: a single solid block of dense stone with a hollow channel cut into its outer face. That placement is the whole trick. The warm channel sits between the cold outside and the storage mass, so it works as a living shield — it keeps the block's outer face warm, and any heat that tries to escape has to push all the way out through the channel before it's lost, while the heat travelling inward soaks into the solid stone behind and is kept — the warmed stone then radiates gently back into the room. Warm the channel even a little and the whole block stops leaking. A second hollow, cut into the inner face, carries the recycled cool air up through the block and releases it into the room near the top — so one block does both jobs at once: a warm shield outside, a cool supply inside.
This isn't a hunch — it's a measured technique. Buildings with pipe- or channel-embedded envelopes run a thin warm (or cool) layer inside the wall to intercept heat right at the boundary: studies show it cuts heat loss through the envelope by 33–84%, and the very same layer can switch between heating the room and shielding it just by changing the temperature you feed it. Put the channel toward the cold side, the dense mass behind it, and the wall both stores more heat and loses far less.
Sources: Shen, Yu et al. — active / pipe-embedded building envelopes (Energy & Buildings; Applied Thermal Engineering, 2013–2024)
Heat and cool run on separate paths, and the building is plumbed so each moves on its own. The sun's heat travels down the outer wall channels, soaks into the mass, and at the base runs under the floor — warming the pool from beneath and venting out under the side patios. Cool, earth-tempered air is drawn up from a lower level through the interior pillars — the solid floors block it from drifting sideways, so the pillars are its only chimney — and is dropped into each room near the top (about three-quarters up), where, being heavy, it falls and settles. Once it has done its work it's drawn back into the pillars, returned to the lower level, re-cooled, and sent round again — a closed loop, recycled rather than dumped. Warm air rising and cool air falling keep this cool circuit turning with no fan of its own; the heat itself forces the airflow. The one powered part is the small fan at the receiver, driving heat downward against its own buoyancy.
Real buildings already run on exactly this. The Barra–Costantini system threads thermosiphon air channels through concrete floors and ceilings for a 50% solar heating fraction — built into 40 occupied flats in Marostica, Italy. Thermally-activated slabs (TermoDeck) pass ventilation air through hollow cores so the mass itself becomes the heat exchanger, and pull night-cool through those same cores to chill the building by day.
Sources: Barra, Renewable Energy (2004), Marostica flats; Barra system (Wikipedia); TermoDeck / TABS, Applied Thermal Engineering 22 (2002)
Stone is a fine heat store — but water is a better one, and a great structure full of cisterns, basins and pools is a far bigger battery than the walls alone. Run the warm channels past water and the building becomes a water heater as well as a radiator.
And a pool is happiest heated from below. Warm the thermal mass beneath the water and the heat lifts into the pool on its own — warmer water is lighter, so it rises through the cooler water above by natural convection, with no pump and no piping forcing it. The mass holds the heat; the water simply receives what floats up. (Water carries heat through itself slowly — far slower than stone conducts — but this buoyant overturning from beneath does the distributing for free.) Large bodies of water make superb seasonal banks: a salt-gradient solar pond stores the sun's heat at 70–90 °C for months, and the same water that banks the heat can be drawn straight off as hot water.
Sources: GreenBuildingAdvisor & Browning Day (water as thermal mass); Solar pond — Britannica; Journal of Energy Storage review (2023)
A serious vision says what's solid and what's hard. This is the difference between a manifesto and a fantasy.
These aren't carved one-off by hand. The whole cathedral is a kit of stone parts — piers, springers, voussoirs, ribs, vault webs, floor slabs and dome rings — cut off-site and assembled like precast. And the oldest idea in the book does the structural work: the arch.
An arch carries its load as pure compression, running down a curved path called the thrust line — an inverted catenary. Each wedge block (a voussoir) leans on the next; the keystone locks the ring. With masonry, stability, not strength, governs — so the shape itself is the engineer. Get the curve right and the stone simply stands — in pure compression, no steel ties, exactly as the great cathedrals stood for centuries. Today that's not guesswork: the voussoirs are cast and then CNC-milled to the exact taper, and the modern structural-stone revival proves prefab stone is back as a real building system (15 Clerkenwell Close; Webb Yates & the New Stone Age).
Spin that arch around a center and it becomes a dome — the same compression logic in the round. Rings of voussoirs stack into a shell, meridional ribs run crown to base, and an oculus compression ring locks the top — the very place the sphere and receiver sit:
Two more pieces make it durable and quiet. The blocks are cast with lime clasts, so a hairline crack heals itself when water reaches the reactive lime — the mechanism MIT identified in Roman concrete in 2023. And the building sits on a fused "clear-stone" base made with the lens itself (concentrated sunlight vitrifies silica into glass — see the Solar Sinter). Clear stone is tremendously strong in compression, which makes a superb hard foundation; the earthquake trick isn't the stone "absorbing" the shock (glass is brittle) but base isolation — the whole structure rides on low-friction bearings and gently slides, cutting the force that reaches it. Honest, and stronger for it.
And the same logic scales to the whole building. Here is a single bay as a complete order — every block, its cut angle, and how many — from the rooftop sphere down to the isolation bearings:
In the cathedral the airflow isn't bolted on — it runs through the bones. Each pillar is a dual hot-and-cold channel: the same active-thermal-chamber trick from the walls, carried in a single shaft. A warm flue on its outer side runs the sun's heat down and keeps the stone charged and shielded; a separate cool flue on its inner side carries earth-tempered air up from the crypt. One member, two jobs — a warm spine against the cold, a cool riser for the room — so the structure that holds the roof up is the same structure that moves the air.
And the arches do the distributing. Where a pier rises into the springing of an arch, the air turns and follows the curve: the arch and vault rib act as a duct, carrying cool supply across the top of each bay to drop it where it's wanted, and gathering warm return back toward the next pier. The thrust line that makes the arch stand and the air path that makes it breathe trace the same curve — compression overhead, a hollow core within. The vaults become the building's plenum; the colonnade, its lungs.
Sources: Webb Yates Engineers; Architects' Journal & Architectural Record (structural-stone revival); thrust-line & limit analysis (Heyman); MIT/Harvard, Science Advances (2023, self-healing lime clasts); M. Kayser, Solar Sinter (2011); friction-pendulum base isolation (Asian J. Civil Eng., 2024)
The grandest structures we ever built — stone, masonry, peaks reaching for the sky — reimagined as the engines that power their cities. Crystal spheres at the peaks; the building as the machine.
These are the finished pieces. Behind them sit another 140 studies — the sphere worked out over the roofline again and again, the stone that has to carry it, and the craft that raised buildings like these the first time.
Open the concept gallery →It was built. It worked. It was forgotten for cheap oil. The physics never changed — only the price of the alternative did. The future city isn't an invention. It's a reconnection.