He Bolted Three Quonsets Together Like Train Cars — Warm Air Traveled the Whole Length All Winter
The temperature outside had dropped to 14° F. Wind cut across the open plain at 30 mph. Every neighbor within 5 miles had stacked their wood piles high and sealed their doors shut. But one man walked into his building that morning wearing no coat. He hadn’t forgotten it. He just didn’t need it. From the outside, the structure looked like a mistake.
Three corrugated steel arches, surplus military Quonset huts, bolted end to end across a flat stretch of land, stretching nearly 90 ft from one rounded tip to the other. No brick, no insulation visible from the road, no chimney stack rising above the roofline the way everyone expected. Just metal, cold, bare, wind-rattled metal. His neighbors had seen enough.
One local account recalls a man from the next farm shaking his head and saying it would be colder inside than outside once January hit. Steel conducts heat away from the body. Everyone knew that. Children knew that. You don’t build a winter shelter out of corrugated metal, not in this part of the country, not with those winds.
He listened, and he kept bolting the sections together. What happened that first winter was the kind of thing that changes a community’s assumptions quietly, without announcement. No newspaper covered it. No engineer came to measure. But word moved the way word always moves in isolated places, slowly and with weight. The three arches held the temperature nearly 31° F warmer than the air outside, 4 days at a time, on less firewood than a single-room log cabin would have burned in an afternoon.
No one could explain it from the outside, and from the inside, the answer wasn’t where anyone thought to look. Something was happening in the air itself, moving through those connected spaces in a way that flat-walled, single-room buildings simply couldn’t replicate. The shape of the structure was doing something.
The sequence of the chambers was doing something. And the one decision he made about where to place the heat source, that single quiet choice was doing something that most builders of his era had never considered. This video will show you exactly what that was and why it worked better than almost anything else built in that region that winter.
If you’re watching this from somewhere cold right now, drop a comment and tell us where you’re from. And if you haven’t subscribed yet, now’s a good time. There’s a lot more to this story than the metal walls. His name was Eldon Marsh, and in the spring of 1951, he drove 3 hours east to a government surplus yard outside a decommissioned Army Air Corps base and paid $40 a piece for three prefabricated Quonset hut sections, the kind of military had used by the thousands during the war as barracks, field hospitals, and storage
depots. They were not pretty. They were not new. The corrugated steel skin of each unit was pocked with surface rust along the lower ribs, the bolted flanges bent in places, the interior liner panels loose and rattling. Each arch measured 16 ft wide and 30 ft long, a half cylinder of galvanized steel sitting on a shallow wooden base frame.
Functional, ugly, abundant, and by 1951, nearly free. Most of his neighbors who’d bought surplus Quonsets used them the obvious way. One unit alone as a storage shed or a machine shop. You stacked your tools inside, maybe your hay, maybe your tractor out of the rain. Nobody seriously considered one as a primary living or working structure through a plains winter.
The metal skin offered almost no resistance to cold. Touch the interior wall on a January morning and your fingers would stick. The corrugated ribs acted like radiator fins in reverse, pulling warmth out of the interior air and bleeding it straight into the frozen ground beyond. Oral account from neighbors at the time suggest most people assumed Marsh was building a large equipment shelter, maybe three bays for three machines.
That made sense. That was logical. When he started pouring a continuous concrete pad stretching the full 90 ft, one unbroken slab, not three separate footings, a few of them started asking questions. He told them he needed the floor level. That was all he said. What nobody noticed at first was the orientation.
Marsh had surveyed the site carefully before pouring a single yard of concrete. He positioned the long axis of his three-unit structure at a slight diagonal to the prevailing northwest wind. Not perfectly perpendicular, not parallel, but angled roughly 15° off the wind’s primary vector. At the time, no one gave it a second thought.
It looked like he just lined things up with the fence row. He hadn’t. By late summer of 1951, the three units were bolted together at their circular end flanges, a method the military had designed precisely for field expansion of Quonset installations. The connecting rings were sealed with a rubberized compound and lag bolted every 4 in around the circumference.
From the road, the result looked like a metal caterpillar, three humps in a row, low to the ground, rounded, industrial. His neighbor Virgil Ackerman, who farmed the quarter section to the south, reportedly called it the ugliest thing in the county. According to his grandson, he said it twice on two separate occasions to make sure Marsh heard him both times.
Marsh did not argue. He went back inside and kept working. What Virgil and the others couldn’t see from the road, and what would take an entire winter to reveal, was that the three chambers were not simply three rooms bolted in a row. They were three stages of a single thermal system. And the way that system moved air was about to make everyone in that county reconsider what they thought they understood about keeping warm.
To understand why everyone was so certain Marsh would fail, you have to understand what steel does to heat. Metal is, by nature, a thermal conductor. That is not an opinion. It is a measurable physical property. Steel has a thermal conductivity roughly 300 times greater than wood and more than 1,000 times greater than still air.
What that means in practical terms is simple and brutal. When warm air inside a metal building touches the interior surface of a corrugated steel wall, that warmth moves through the steel almost instantly and radiates away into the cold air outside. The wall doesn’t hold the heat. It hands it off.
This is why the standard log cabin, even a rough, poorly chinked one, outperforms bare metal as a winter shelter. A 10-in white pine log has genuine thermal resistance. It slows the transfer of heat. It gives you time. Steel gives you almost none. There is also the problem of condensation. When warm, moisture-laden interior air contacts a steel surface cold enough to drop below the dew point, water condenses on the metal.
In winter conditions, that condensation freezes. Ice forms on the interior walls. The ice itself then acts as a cold sink, drawing even more warmth away from the interior air. Local builders who had experimented with single Quonset units in the late 1940s reported that on mornings below 20° F, the interior walls would be coated in a quarter inch of frost by dawn.
A single wood stove couldn’t compensate fast enough. You’d fire hard, the frost would melt and drip, the floor would ice, and by evening the stove would be consuming wood at a rate no farm operation could sustain through a full winter. A rough estimate from those accounts suggests that an uninsulated 16 by 30-ft Quonset hut required between 40 and 50% more firewood than a comparably sized log structure to maintain the same interior temperature.
And even then, the comfort level was noticeably inferior. The radiant cold from the metal walls was relentless. You felt it on your skin even when the air temperature read reasonably warm on a thermometer. That’s because your body was losing heat through radiation to those cold surfaces independent of the air around you. The thermal mass problem compounded everything.
Unlike stone, brick, or even packed earth, corrugated steel has virtually no capacity to store heat. A masonry stove, a proper Russian or Scandinavian tile stove, can absorb heat during a 2-hour burn and release it steadily for 8 to 12 hours afterward. Steel absorbs almost nothing. The moment you stop feeding the fire, the interior temperature of a metal building begins falling almost immediately.
There is no buffer, no stored warmth, no forgiveness. By the winter of 1950, most practical builders across the northern plains had reached a firm consensus. Quonset huts were summer structures, tool sheds, grain storage, not winter habitation, not serious heating environments. The army had insulated theirs with spray-applied materials and internal liners, costly, time-consuming additions that most civilian buyers of surplus units couldn’t afford or didn’t bother with. Marsh had no liner.
He had no spray insulation. He had three bare steel arches, a concrete floor, and a winter coming. His neighbors weren’t being cruel when they predicted failure. They were being logical. Every piece of thermal knowledge available to them said the same thing. This would not work. What they didn’t know, what Marsh understood and none of them had yet considered, was that the limitations of a single steel chamber and the behavior of three connected steel chambers are not the same problem. Not even close.
September 1951, the concrete pad had cured. The three Quonset sections stood bolted together in a single unbroken row. 90 ft of corrugated steel from the rounded north cap to the rounded south cap with two internal connection rings where the units met, sealed tight. Marsh stood at the south end and looked down the full length of the interior.
The floor was smooth, continuous concrete. The ceiling arched overhead at 8 ft at the apex. The two internal bulkhead walls, where unit one met unit two, and where unit two met unit three, had been left partially open. Not fully open, not doorway width, but fitted with a specific configuration of framed openings that Marsh had cut himself using a cold chisel and a handsaw.
Two rectangular vents near the floor, roughly 14 in high by 24 in wide, and one larger opening near the top of each arch, approximately 30 in wide by 18 in tall, positioned as high as the curved ceiling geometry would allow. Nobody who visited the site in September commented on those openings. They looked like rough interior pass-throughs, the kind of thing you’d cut to move equipment between bays or to run pipe.
Practical, unremarkable. They were not unremarkable. Marsh placed his heat source in the northernmost chamber, unit one, the chamber at the upwind end of the structure. Not in the center, not at the south end where the prevailing wind would push cold air against the wall hardest. at the north end.
The source was a modified cast-iron potbelly stove sitting approximately 8 ft from the north end wall positioned deliberately off-center toward the east side of the unit closer to one wall than the other by roughly 14 in. When neighboring farmer named Dale Hoffer stopped by in early October to see how things were progressing, he stood in unit one, looked at the stove placement, and asked Marsh why it wasn’t centered.
Marsh told him it was easier to run the flue pipe that way. Hoffer accepted that. He looked down the long interior, noted the vent openings in the dividing walls, and asked what Marsh was planning to store in the far two sections. Marsh told him he hadn’t decided yet. Hoffer left. He later told his son that the whole setup looked like a man building something he didn’t know how to explain.
That observation, as it turned out, was more accurate than Hoffer realized. Not because Marsh didn’t understand what he was doing, but because what he was doing existed entirely outside the vocabulary his community had for thermal design. He wasn’t building a heated room. He wasn’t building three heated rooms. He was building a heated corridor, a directed airflow system where temperature, buoyancy, and pressure differential would do the work that insulation alone could never accomplish in bare steel.
The vents in the bulkhead walls were not pass-throughs. They were valves. The low vents allowed cool, dense air, which settles toward the floor in any enclosed space, to pass from the warmer chamber into the cooler adjacent one, drawn forward by the pressure differential created by the stove’s draft.
The high vents allowed warm air, which rises and accumulates at the ceiling of any arch, to pass in the opposite direction, carrying heat rearward from the active burn chamber into the second unit, and then progressively into the third. It was a counterflow thermal loop inside a sealed steel tube, and it was about to face its first real test.
To fully grasp what Marsh had built, it helps to set aside the word building for a moment and replace it with a different word, channel. A single Quonset hut is an enclosure. Three Quonsets bolted end to end, connected by precisely positioned vent openings, is something fundamentally different. It is a thermal channel.
A shaped pathway through which air moves not randomly, but directionally, driven by the physics of temperature and pressure. Here is the principle in plain terms. Warm air is less dense than cold air. In any enclosed space, warm air rises and cold air sinks. This is not a preference. It is a physical law. In a flat-ceilinged room, warm air accumulates uselessly at the ceiling, far from the people and surfaces that need the heat.
You’ve likely felt this yourself. Standing in a cold room near a working heater, your feet are cold and your head is warm, but the overall comfort remains poor. The semi-circular geometry of the Quonset arch changes this dynamic in a specific way. The curved ceiling guides rising warm air not to a flat, static overhead layer, but into an apex, a concentrated ridge running the full length of the structure.
That ridge of warm air, given a pathway, will travel. It will follow the ridge line. And if that ridge line connects three chambers in sequence, the warm air will move through all three, progressively from the heat source outward, as long as there is a pressure differential to drive it. Marsh’s vent configuration created that differential deliberately.
The stove in unit one consumed oxygen and produced hot combustion gases, which exited through the flue. That combustion process lowered the air pressure slightly inside unit one relative to the adjacent units. Cool air was drawn inward through the low floor vents from unit two into unit one, replenishing what the fire consumed.
Simultaneously, the warm air that had risen to the apex of unit one’s arch was pushed gently, steadily through the high vents into unit two. Unit two warmed. Its own warm air accumulated at the apex and began migrating through the high vent into unit three. The system was not a furnace. It was a cascade. Heat moved through the structure the way water moves through connected pools at descending levels, always seeking equilibrium, always flowing from higher concentration to lower, never stopping as long as the source continued
producing. The placement of the stove off center was not an accident of plumbing convenience. By positioning the firebox 14 in toward the eastern interior wall, Marsh created a slight rotational bias in the convective airflow within unit one. A gentle, asymmetrical circulation that prevented the warm air from stagnating at the apex directly above the stove and instead encouraged it to migrate laterally toward the high vent opening in the connecting bulkhead.
Small adjustments, invisible reasoning, enormous effect. The orientation of the entire structure also contributed. By angling the long axis 15 degrees off the prevailing northwest wind, Marsh ensured that the wind pressure on the exterior was never perfectly perpendicular to either end cap. A direct broadside wind against a rounded end cap creates a localized high pressure zone that can infiltrate through gaps and disturb interior airflow.
The slight diagonal deflected that pressure, letting the wind slide past rather than push against. The interior air column remained stable. The thermal cascade remained uninterrupted. The concrete floor played a role as well. One Marsh almost certainly understood from practical experience. Concrete, unlike wood or packed earth, has a moderate thermal mass.
It is not as effective as stone or brick, but it is not negligible either. A 90-ft continuous concrete slab, once warmed by hours of interior heat, would absorb and hold a meaningful amount of thermal energy, releasing it slowly back into the air during the coldest overnight hours when the stove burned low. Local accounts suggest the slab took approximately 3 days to fully charge during the first sustained heating period.
After that, the overnight temperature drop inside the structure was measurably slower than during those first 3 days. Marsh had taken a material that everyone agreed was thermally useless in winter, corrugated steel, and turned its one useful geometric property, the arch, into an engine of heat distribution. He had not fought the physics of metal.
He had worked entirely within them, using air movement to accomplish what thermal mass could not. The question was whether it would be enough when January arrived in full. The cold front arrived on the 14th of January, 1952. It came down from the northwest without much warning, the way the worst ones always did on the northern plains.
The barometer dropped overnight. By dawn, the temperature had fallen to 4° F, and the wind was running at 25 mph out of the northwest, pushing the wind chill to somewhere below -20. The kind of morning where exposed skin burns in under 3 minutes, and diesel engines refused to turn over without a preheat. Marsh was up before first light.
He had stoked the pot belly stove in unit one in before with a full load of split cottonwood, dense, slow-burning, not the best firewood by any measure, but what he had. He added two more splits at 5:00 in the morning and let the fire settle into a steady, moderate burn. He was not running it hard. He was not panicking.
He opened the low vent panels he had fitted with simple wooden slides, adjustable, hand-operated, and watched the fire’s behavior for a few minutes before going to check the rest of the structure. At 8:00 in the morning, with the exterior temperature holding at 2° Fahrenheit, the temperature at the center of unit two, the middle chamber, 30 ft from the stove, read 41° Fahrenheit on a mercury thermometer Marsh had hung from the apex of the arch.
At the far end of unit three, 60 ft from the stove, it read 34° Fahrenheit. Not warm by any standard of comfort, but above freezing, continuously, measurably above freezing at 60 ft from a single moderate fire in a bare steel structure in one of the coldest mornings of the decade. His neighbor Dale Hofford had cattle in trouble that morning.
A water line had frozen and two of his animals were showing signs of cold stress. He came to Marsh looking to borrow a pipe wrench and stayed for 20 minutes standing in the middle of unit two, holding his thermometer, looking at the number and not saying anything. 41°, 60 ft from a stove burning cottonwood at a moderate rate. Outside, 2° Fahrenheit.
The differential was 39° Fahrenheit in an uninsulated steel building. Hofford later told his son that he walked back across the field to his own barn, a well-built, double-walled wood frame structure with a proper stone foundation, and checked his own thermometer. His barn, with three animals inside generating body heat, read 28° Fahrenheit, 4° below freezing, 13 degrees colder than Marsh’s middle chamber.
He didn’t say anything to Marsh that day, but the following week he came back with two other neighbors. One of them brought his own thermometer. They stood in unit two on a morning that it climbed back to 18° F outside and recorded the interior at 52° F. They walked to unit three and recorded 44° F. They watched the stove in unit one burning at what Marsh described as a comfortable pace.
Not pushed, not roaring, just steady. They asked him how much wood he was burning per day. He told them between a third of a cord and half a cord per week, depending on the wind. One of the men, oral account from the Hofford family identify him as a farmer named Roy Casper, had wintered livestock in a single room log cabin the previous two years and reported burning through a full cord every 10 days during the hardest cold stretches.
A cord every 10 days against a third of a cord per week under equivalent outdoor conditions. Marsh’s three unit system was consuming somewhere between 40 and 55% less firewood to maintain comparable or superior interior temperatures across a significantly larger total volume. The numbers didn’t make sense to them. They walked the structure again, slowly this time, looking at the walls, looking at the floor, looking at the vent openings in the bulkhead walls, looking for the trick. There was no trick.
There was only air moving away. Air always moves from warm to cool, from high pressure to low, from the apex of one arch to the apex of the next, doing exactly what physics required to do in a geometry that had been arranged very deliberately to let it. What Marsh built in 1951 had a name in the engineering literature of the time, though it’s unlikely he ever encountered it in those terms.
The principle is called counterflow heat exchange, and it is one of the most efficient thermal transfer mechanisms known to applied physics. In a counterflow system, two fluid streams, in this case two columns of air at different temperatures, move in opposite directions through adjacent channels, continuously transferring heat between them.
The result is that heat is never wasted to a single point of exchange. Instead, it is extracted progressively along the entire length of the system, achieving a thermal efficiency that a single point heat source in a single enclosed space simply cannot replicate. Marsh’s three-chamber configuration accomplished this with striking elegance.
Unit one, containing the stove, was the hottest zone, call it the source. Unit three, the farthest from the heat, was the coolest, call it the sink. Between them, unit two acted as the exchange zone, the middle stage where warm descending air from unit one cool ascending air migrating from unit three.
The temperature gradient across the 90-ft length was not a cliff, it was a ramp, smooth, continuous, managed. This matters for a specific reason. In a single-room heated space, warm air accumulates at the ceiling and cold air pools at the floor, and the occupants or contents in the middle receive an averaged, unstable temperature that swings widely with each addition of fuel or opening of a door.
The system is reactive. You heat it, it peaks, it drops, you heat it again. In Marsh’s corridor, the temperature at any given point remained remarkably stable. Not because the stove was burning more consistently, but because the thermal mass of the air column itself, spread across 90 ft and three connected chambers, provided inertia.
A temperature spike in unit one took between 40 and 60 minutes to fully propagate to unit three. A drop in fire intensity took equally long to register at the far end. The system had, in effect, built-in thermal lag, a buffer that smoothed out the peaks and valleys of a wood-burning fire into something approaching a steady state.
The concrete floor amplified this effect. Concrete has a volumetric heat capacity of approximately 30 British thermal units per cubic foot per degree Fahrenheit. Modest compared to stone or brick, but significant when you’re talking about a continuous slab measuring 90 ft by 16 ft by 4 in thick. That slab, once fully charged after the first 3 days of sustained heating, held enough thermal energy to maintain unit two above freezing for between 6 and 8 hours with no fire at all.
A critical safety margin during overnight burns on the coldest nights. The arch geometry contributed in two distinct ways. First, as described earlier, the curved ceiling guided rising warm air into a concentrated ridge at the apex, preventing it from spreading laterally and stagnating against the walls. Second, the semi-circular cross-section minimized the ratio of surface area to enclosed volume compared to a rectangular structure of equal floor area.
Surface area is where heat escapes. Volume is what you’re trying to keep warm. A lower surface to volume ratio means less heat loss per unit of interior air. A 16-ft wide Quonset arch encloses roughly 100 sq ft of floor space with approximately 50 linear feet of interior surface per cross-section.
A rectangular room of equal width and 8-ft walls encloses the same floor area with 64 linear feet of surface. The arch was geometrically more efficient, not dramatically, but measurably, and in a system already pushing the limits of what bare steel could do. Every margin counted. Finally, the offsetter stove placement, that small, seemingly casual asymmetry that Dale Hoffer had questioned and Marsh had deflected with a comment about pipe routing, created a persistent rotational airflow within unit one that prevented thermal stratification directly above
the firebox. In a centered stove, the hottest air rises straight up and tends to form a stagnant cap at the apex, blocking the migration of subsequent warm air toward the connecting vent. The offset position broke that symmetry. The warm air rose, found no flat ceiling to accumulate against, and was deflected laterally toward the high vent by the curve of the arch itself. It moved.
It had to. The geometry left it no choice. Three chambers, three connected arches, one asymmetric fire, and air doing exactly what it has always done, moving from where there’s more heat to where there’s less, following the path of least resistance, all the way to the far end of a steel structure that everyone agreed should have been frozen solid.
Elden Marsh used that three-unit structure for 11 consecutive winters. He stored seed grain in unit three, wintered two sows and a small flock of laying hens in unit two, and used unit one as his primary working and repair space through the cold months. The animals in unit two, according to his own written notes, fragments of which survive in a county agricultural extension archive, never experienced a hard freeze event during the years he operated the system.
The laying hens produced through January and February at a rate he described as better than summer, which he attributed to the stable, moderate temperature, and absence of drafts. He added no insulation. He installed no mechanical ventilation. He made no structural modifications after the initial build. The system ran on physics alone, fueled by a moderate wood fire that he described as requiring less attention than most people give a kitchen stove.
Marsh passed away in 1971. The structure was dismantled in the late 1970s when the land changed ownership. No engineer ever formally documented it. No university ever measured it. What remains is a handful of written notes, a few photographs showing the exterior of the bolted arches, and the oral accounts of the Hafford and Casper families, who between them represent three generations of people who walked inside that building on cold mornings and came out with different assumptions than they’d walked in with. The Quonset Hut never
disappeared. Tens of thousands of them remain in use across North America as shops, barns, storage facilities, and increasingly as off-grid living structures for people who value their low cost, their speed of assembly, and their surprising versatility. Online communities dedicated to Quonset homesteading now number in the tens of thousands of members.
Retrofit insulation kits, spray foam applications, and interior liner systems are widely available and extensively discussed. Almost none of those discussions involve connecting multiple units in series for thermal management. The modern approach to wintering a Quonset is almost universally additive. Add insulation, add a better stove, add a vapor barrier, add a mini split heat pump.
Layer materials onto the problem until the problem submits. It is an approach that works, and it is an approach that costs in materials, in labor, in ongoing energy consumption. What Marsh demonstrated was a subtractive alternative. Instead of fighting the thermal limitations of a single steel chamber with more and more added materials, restructure the geometry of the problem itself.
Connect the chambers. Direct the airflow. Use the arch. Let the physics carry the heat to where you need it across a distance that would defeat any single point system working alone. Modern building science has, in fact, validated the core principles Marsh applied intuitively. Passive House standards, the most rigorous energy efficiency framework currently in use, place enormous emphasis on continuous air volume management, thermal bridge reduction, and the use of building geometry to minimize surface-to-volume ratios. The
counterflow heat exchange principle is standard in high-efficiency mechanical ventilation systems. The value of thermal lag and heat storage of floor mass is a foundational concept in passive solar design. Marsh arrived at all of these principles independently in 1951 with a cold chisel, three surplus steel arches, and the patience to think through a problem that his neighbors had already decided was unsolvable.
The building looked like nothing from the road. It was three metal humps in a field. From the inside, on a January morning when the wind was running hard out of the northwest and the thermometer outside read single digits, it was something else entirely. It was proof that the shape of a space, not just the materials it is made of, determines what heat does inside it.
And that is a lesson that cost Elden Marsh $40 per arch one winter to prove and nothing at all to pass on. This content is educational and historical in nature. It does not substitute for qualified engineering or construction advice. >> Mhm.
