October 1958. A section of steel arch swung out beyond the rim of Glen Canyon, seven hundred feet above the Colorado River. Far below, the water looked like a thin ribbon. The section hung from a cable strung across the gorge, a temporary track that from this distance appeared no thicker than a man’s forearm.

Workers on the rim guided it with hand lines toward the growing tip of the arch. The arm extended dozens of feet into open air with nothing beneath it. Nothing at all. It was held in place only by cables running back to anchor points drilled into the sandstone.

If one of those moved, the arm moved with it. Today, traffic on US Route 89 crosses that canyon without slowing down. The bridge is so woven into the desert landscape that most drivers barely notice it. They pass over the Colorado River on a steel arch assembled piece by piece across empty space, with no supports beneath it, no scaffolding in the gorge, and no tower rising from the riverbed.

What made that possible was not widely understood at the time, and it is now almost entirely invisible. The arch looks as though it has always been there. It was not. Closing it at the center demanded a precision that depended on something no engineer could fully control: the temperature of steel on a desert night.

Glen Canyon is not a gentle place. Its walls drop almost vertically toward the river. The gorge was so deep and so sheer that when engineers began planning the crossing needed for the dam’s construction, the obvious solution—erecting timber or steel falsework from the riverbed and building the arch on top of it—was not feasible. A support tower tall enough to reach even halfway up the walls would have required more material than the bridge itself, would have been dangerously unstable in the canyon winds, and would have needed a road to reach it at river level, a road that did not yet exist.

The traditional approach was ruled out before it was ever seriously considered. The alternative was a cantilever arch. Build both halves outward from the rims, grow each arm into the void, and join them in the middle. It sounds straightforward.

It is the opposite of straightforward. An arch carries its load through compression, each section pressing against the next, the whole structure supporting itself through its own geometry. But that geometry only works when the arch is complete. Before closure, each half behaves like a diving board, a projecting arm that wants to rotate outward and fall.

The farther it reaches into the canyon, the stronger that rotation becomes. Something had to hold each arm in place until another arm existed to push against. The engineers solved the first problem with backstay cables. These were heavy steel cables anchored into the canyon walls above the piers, running at sharp angles down to fittings at the tip of each growing arm.

They pinned the tip back against its tendency to rotate forward and fall. Think of them as taut ropes holding the arm up. Every new section added to the arm increased the load on those cables. The anchor points in the sandstone had to bear forces that grew heavier with each lift, and the anchor bolts and bearing plates carrying that weight were part of a temporary system, not the permanent structure.

Moving the sections out to the tips was a separate problem entirely. There was no road into the canyon, no crane large enough to reach the work site, and no way to build anything up from below. The solution was a cableway system: a network of wire ropes stretched between the two rims carrying a traveling carriage. The sections were attached to the carriage, lifted from the rim, and swung out over the gorge.

Each section weighed several tons. The cableway ropes sagged under the load, which meant every section arrived at the tip slightly lower than it should have been. The workers had to account for the geometry of the loaded rope before the section could be maneuvered into place. Guide lines held by workers at the tip kept each section from spinning in the wind as it approached.

The ironworkers receiving those sections at the tip had no room for hesitation. The working surface was the top chord of the growing arm, a steel shape so narrow that two men walking in opposite directions had to squeeze past each other. Beneath them, nothing for hundreds of feet. The canyon winds, accelerated by the gap between the walls, were unpredictable.

Any section arriving at even a slight angle could swing backward as it was lowered. The men holding the guide lines had to absorb that movement without being pulled over the edge. They coordinated through hand signals and by watching each other, because the noise of the cableway machinery made shouted commands unreliable at that distance. When a section was in position, the ironworkers drove temporary alignment pins through the bolt holes to hold it in place, then replaced them with permanent high-strength bolts.

After that, the backstay cables were re-tensioned to distribute the new weight across the anchor system. Then surveyors on both rims took fresh readings of the tip’s elevation and position. Then the entire cycle began again. Another section was rigged, another lift was made, another controlled swing over the gorge.

After each section was bolted into place, surveyors on both rims measured angles and elevations of each arm’s tip. The two arms were growing toward each other from opposite walls across a gap of roughly five hundred feet. Each arm had to reach the midpoint at the correct elevation, the correct horizontal position, and the correct angle—all at once—or the crown section between them would not fit. Any small misalignment early in either arm would multiply across the entire span.

The engineers had to be precise all the way along, with no possibility of adjustment once the sections were bolted. So they measured constantly. Elevation readings after every section, horizontal displacement checks, temperature corrections applied to the survey calculations because steel tapes and survey rods changed length with heat, and desert temperatures swung violently between day and night. As summer gave way to cooler months, the arms themselves moved subtly with each thermal cycle.

The surveyors corrected for that. It was precise, repetitive work, demanding the same exactness as the fitting and bolting, but carried out a thousand feet away on the opposite rim. What the surveyors could not control was the steel itself. The crown section, the final piece that would fill the gap between the two tips, was fabricated to precise dimensions.

But those dimensions matched the gap at one specific temperature. Steel expands when heated and contracts when cooled. Over several hundred feet, a swing of forty degrees Fahrenheit could change the gap between the tips by more than an inch. Glen Canyon is desert.

At the height of summer, the steel face of the arch could reach temperatures far above the surrounding air. By night, that same steel could cool to near the air temperature within hours. The gap the crown section had to fill was not stable. It breathed with the air temperature and the angle of the sun.

The engineers calculated a target closure window: a range of temperatures at which the arch expanded or contracted to produce exactly the gap the crown section required. That window tended to open in the early morning hours, before the desert heated up and the steel began to expand. The workers would prepare the crown section on the cableway, rigged and ready for transport. Thermometers were read on both sides of the arch.

When the numbers fell within the target range, the crew moved. The thermal window proved, in practice, narrower than the engineers initially calculated. Desert conditions compounded the problem. Wind, cloud cover, and the orientation of the canyon could change the arch’s temperature faster than expected.

What this meant on site was that the crew had to be ready to stop mid-operation, return the crown section to the cableway, and wait for another morning when conditions aligned. The most dramatic moment in the arch’s construction was not a dropped bolt, a fallen section, or a snapped cable. It was the waiting for thermometer readings to stabilize on a quiet morning in the canyon. When the crown finally settled into place, the pins were driven, and the spliced connections were made, the bridge changed in an instant.

The backstay cables, which had held each arm against the sky throughout construction, began to shed their load as the now-complete arch took over. The workers slackened them in sequence. The canyon walls that had served as temporary cable anchors became simply abutments: two masses of rock receiving the outward thrust of a compression arch, doing the job they had been designed for from the start. Construction traffic began crossing almost immediately.

Glen Canyon Dam could not have been built without a dependable way to move heavy equipment, aggregate concrete, plant components, and hundreds of workers to the site. The dam required years of construction, millions of cubic yards of concrete, and a workforce that peaked in the thousands. None of that could have reached the other side of the canyon efficiently without the crossing the arch bridge provided. Later arch bridges built by the same general method used precisely controlled cable cranes with remote tensioning, laser instruments to verify alignment in real time, and tighter fabrication tolerances that narrowed the acceptable temperature range and made closure faster and more predictable.

But the essential sequence did not change: extend the arms, hold them with cables, watch constantly, and close at the exact moment when the gap is right. The bridge still carries Highway 89 across the Colorado River at Glen Canyon. Below the deck, inside the painted steel ribs curving down toward the canyon walls, the bolted connections made by ironworkers on a narrow ledge above empty space still transmit the loads. Near the arch’s midpoint, where the two arms met on a cold morning when the temperature held steady just long enough, there is a crown section that fit because the desert cooperated for a few hours.

It carries everything that crosses above it.