On the morning of 14 April 1944, a Royal Navy crew opened the seacocks in a hollow concrete structure the size of a six-storey apartment block and watched it sink into the shallow waters of the Hamble estuary on England’s south coast. The concrete walls were four feet thick. The interior was a series of watertight compartments, each sealed with a steel door, linked by a system of pipes and valves running along the floor. The structure was 180 feet long and weighed 4,500 tons.

It had taken four months to build. It took fourteen minutes to sink. When the water closed over the top and the last air bubbles vanished from the surface, the crew transferred to a waiting boat and returned to the construction site. There, another Phoenix was nearly finished.

There was nothing striking about this moment on the surface. The Southampton Water was crowded with military traffic, and a vessel disappearing beneath the waves was no unusual sight in wartime. What was unusual was that the vessel being sunk was not an enemy ship or a worn-out fishing boat. It was a brand-new structure, built for this exact purpose, deliberately placed on the bottom so that German reconnaissance aircraft, which flew regularly to photograph the English coast, could not see it.

Sinking it was the method of concealment. The plan was to leave it on the bed of the Hamble for six weeks, then pump it out and tow it to France. It would join 145 other structures like it on the floor of the Bay of the Seine, where together they would form a wall. Behind that wall, in the water it protected, the largest amphibious invasion in history would be supplied.

The thing sitting on the bed of the Hamble was called a Phoenix caisson. It was one component of a prefabricated harbour codenamed Mulberry, and Mulberry was the answer to a question Allied military planners had been working on since 1942. How do you supply two million soldiers on an open beach with no port? To understand why the question had to be answered, you have to understand the geography of the French coast and the strategic logic of what the Allies were trying to achieve.

The plan to invade western Europe, carried out on 6 June 1944, required landing Allied forces on the beaches of Normandy and then pushing inland through German defensive lines to liberate France, Belgium, and ultimately Germany itself. The first phase of that operation, getting troops ashore, was a problem of immense complexity that Allied planners had spent two years solving. But the second phase, supplying those troops with ammunition, food, fuel, and replacement equipment for weeks and months of continuous fighting, was a problem that got less attention in public records and was practically just as important. Modern combat forces consume supplies at a rate that is hard to grasp without numbers.

Allied forces in Normandy needed around 15,000 tons of supplies per day once they were fully ashore and engaged in combat. Those supplies came from Britain. They had to cross the English Channel. They had to reach the beach, and there was no port.

The nearest major port to the planned landing beaches was Cherbourg, at the tip of the Cotentin peninsula, sixty miles west of Omaha Beach. Allied planners knew Cherbourg would be heavily fortified. They also knew, from intelligence reports and the example of the 1942 Dieppe raid, which attempted a frontal assault on a fortified port and ended in disaster, that attacking a fortified port would likely produce heavy casualties and destroy the port’s facilities themselves. When the Germans abandoned Cherbourg in late June 1944, they did exactly what the planners expected.

They destroyed the port so thoroughly that it took Allied engineers until late July to restore it to minimal operational capacity. Le Havre was farther and more heavily fortified. Brest was 300 miles from the landing beaches. There was no suitable port between the invasion zone and the front lines that could be expected to survive the initial fighting in working condition.

The alternative was to do without a port. Land supplies across the open beach using landing craft, small flat-bottomed vessels designed to carry troops and vehicles directly to shore from the sea. Landing craft could reach beaches that larger ships could not approach, but their capacity was limited. Each could carry only a few vehicles or a limited load of supplies.

Moving 15,000 tons per day using landing craft alone would require more ships and more round trips than the available fleet could provide. Landing craft were also vulnerable to weather. A storm in the Channel could stop landing craft operations for three days, leaving the invasion force short of ammunition and fuel at a moment when a German counterattack might decide the battle. The planners needed a better answer.

Winston Churchill set the parameters for them in a memorandum dated 30 May 1942. It read, in part: “Ports for use on beaches. They must float up and down with the tide. The problem of mooring must be mastered.

Do not argue the matter. The difficulties will argue for themselves. ” The engineering teams that received that memorandum spent the next eighteen months wrestling with the difficulties. The solution that emerged by the autumn of 1943 was a prefabricated port made of five components, manufactured in Britain and assembled on the French coast in the days immediately following the invasion.

The outermost component was the Bombardon. It was a floating steel breakwater, cross-shaped in section, about 200 feet wide at its broadest dimension, moored in deep water off the beach in a line designed to absorb and dissipate the energy of incoming waves before they reached the harbour anchorage. A wave entering the Bombardon field broke into smaller, less energetic waves. The energy reaching the port was a fraction of the energy in the open sea.

The Bombardons were the first line of defence against Channel weather. Behind the Bombardons stood the Phoenix caissons, the main breakwater wall. A Phoenix was a hollow reinforced concrete box, sealed when its internal seacocks were closed, sinkable when they were opened. Their sizes varied enormously.

The smallest units were about 60 feet long and 25 feet high. The largest were 204 feet long, 55 feet wide, and 60 feet high, weighing 6,044 tons. A 6,044-ton Phoenix was larger than most buildings in London at the time it was built. Placed in position and sunk, each one formed a section of the harbour wall.

Set in sequence, 146 of them formed a breakwater over a mile long, enclosing a harbour basin of several square miles. Inside the Phoenix breakwater, in the protected waters of the artificial harbour, ships could anchor without the wave action that would have made unloading impossible on an open coast. They transferred their cargo to landing craft and to the third component, the Lobnitz pierhead. The Lobnitz pierhead was a floating steel platform about the size of a tennis court, mounted on four hollow steel legs that could be driven into the seabed.

As the tide rose, the legs held their position and the platform rose on them by hydraulic jacks. As the tide fell, the platform lowered. The result was that the pierhead deck remained at a constant height above sea level regardless of the state of the tide, which in the Bay of the Seine could range up to 18 feet. A supply ship moored alongside a Lobnitz pierhead found a stable loading platform at the same height at high tide as at low tide.

From the pierhead to the shore, cargo moved on the fourth component, the floating Whale roadway. The Whale was a system of steel pontoons linked by hinged steel deck sections, so that each section could rotate at its connection points to accommodate the movement of the sea beneath it. When complete, the roadway ran from the pierhead to the shore as a flexible bridge several hundred feet long. Trucks drove across it from the pierhead to the beach as if it were an ordinary road while the ocean moved under their tyres.

The fifth component, the foundation of all of it, was the decision to manufacture every piece in Britain, tow it across the Channel, and assemble it on a hostile coast in the days immediately following the invasion. The director of engineering for the project was Brigadier Bruce White of the Royal Engineers. White was not a famous man. He is not famous now.

He was a civil engineer who had spent his career building things under difficult conditions, and he had been given a problem with no precedent in the history of military engineering and about twelve months to solve it. The Phoenix represented the backbone of the harbour and the single largest manufacturing challenge. 146 reinforced concrete structures, ranging from 60 to 200 feet in length, had to be built and ready by May 1944. There was no facility in Britain designed to build reinforced concrete structures of this size in the water.

Engineers surveyed the available coastline, river estuaries, reservoirs, and dry docks across the country, and requisitioned any site that could be adapted. The sites that entered the programme included a stretch of the Thames near Tilbury, the Clyde estuary at Port Glasgow in Scotland, the dock basin at Cardiff in Wales, the Hamble estuary in Hampshire, a flooded reservoir in central England that was drained and converted into a Phoenix construction facility, a former dry dock at Richborough in Kent, and more than a dozen other locations. At each site, a temporary cofferdam was built in the water, the water was pumped out until it was dry, then the formwork was erected and the concrete pouring began. The workers who arrived at these sites in the autumn of 1943 were told they were building floating platforms for military use.

The supervisors of the workers and the concrete, who were on site longer, were told slightly more. None of them were told the purpose of the structures they were building, or their destination, or their timing. The Official Secrets Act was applied at every level. Workers who asked questions beyond their security clearance were told the information was not available.

They were transferred if they persisted. The concrete workers who built the Phoenixes worked inside cofferdams that reduced the water but did not entirely prevent it. Estuary sediments are saturated with water. You could pump the water from a cofferdam basin down to working level, and still water would seep through the clay layers between the sheet piles, or through gaps in the timber itself, or through the estuary sediments at the base.

The working area inside a Phoenix cofferdam was damp and smelled of salt clay, lit by portable electric lamps on cables strung over the cofferdam walls. In winter, the temperature inside a cofferdam on the Thames was little different from the temperature outside. Concrete is not poured in cold weather the way it is poured in warmth. Below a certain temperature, the hydration reaction that makes concrete set slows down and freezes, and below freezing it stops entirely.

The winter of 1943 to 1944 was cold on the English coast. The construction schedule did not allow for slow concrete curing. The sites working through December and January wrapped the formwork in hessian and straw to retain the heat of curing, and in some cases used portable coal heaters inside the formwork cavities to keep the pour above the minimum temperature required for adequate strength. When frost hit a pour that had not yet reached sufficient strength, engineers documented the failure, and that section was broken out and re-poured.

There were re-pours. The reinforcing bars that gave the concrete its tensile strength were cut, bent, and tied by hand by steel-fixing crews working ahead of the concrete teams. In a Phoenix with four-foot-thick walls, the reinforcement mesh inside those walls was dense. A skilled steel-fixing crew could place and tie a section in the time it took the formwork crew to install the timber shutters, and the two crews had to coordinate precisely because neither could work with the other in its way.

The supervisor’s role at each site was to manage this coordination across simultaneous pours of a structure being built from the bottom up in stages, each stage dependent on the one below it having set, and the concrete had to be poured in continuous lifts. A cold joint, the seam where fresh concrete met set concrete, was a structural weak point. In the design of a box intended to hold back the sea, any structural weakness in the wall was a potential failure point. Once a pour began on one of the large Phoenixes, it did not stop until the stage was complete.

The crews worked in shifts, twelve hours on and twelve hours off, for the entire duration of the pour. In winter, they worked under floodlights. They mixed concrete in conditions no modern site safety officer would permit, in cold and damp, without adequate respiratory protection against the concrete dust that filled the working air inside the cofferdam. At the larger sites on the Clyde and the Thames, multiple Phoenix units were under construction at once.

The Clyde site at Port Glasgow built 12 units. The Thames site at Tilbury built more than 20 units. At these sites, the concrete batching plants were permanent installations, and the supply chains for aggregate, cement, and reinforcing steel were established and reliable. At the smaller sites, supply was more vulnerable.

Any disruption to supply at a single Phoenix site could delay the schedule by a full week. A lost week in the spring of 1944 could not be made up. The total volume of concrete in the 146 Phoenix units was around 55,000 tons. Every ton of that concrete was mixed by hand or in small petrol-driven mixers, poured by workers using buckets and chutes, then compacted in the formwork by hand-held tamping bars to remove air bubbles, and left to set under whatever protection the site provided.

The workers who did this work were paid under wartime wage controls that set rates for the British construction industry. They received no extra pay for secrecy. They were not compensated for the restrictions that prevented them from discussing their work off-site. They were construction workers doing construction work.

The only thing different about their job was that they were not allowed to talk about what they were building. The secrecy requirement added a layer that no previous British construction programme of this scale had faced. By January 1944, the sites on the Thames, the Hamble, and the Clyde contained enormous concrete structures that had no innocent explanation. German aerial reconnaissance over southern England was systematic and skilled.

Luftwaffe photo-interpretation teams had already correctly identified the concentration of landing craft in the southern ports as evidence of an imminent cross-Channel operation. If they also identified the Phoenixes as prefabricated port components, they might conclude that the operation did not depend on capturing a French port, which would narrow the possible landing sites considerably. The British solution was to sink the Phoenix units as soon as they were complete and leave them on the bottom until they were needed. A concrete structure sitting on the bed of a tidal estuary was hard to recognize from 10,000 feet as anything of significance.

From that height, a Phoenix settled into the mud looked like a mass of sediment or a wrecked ship, neither of which was unusual in wartime British waters. The sinking operation required careful execution. When a Phoenix unit was ready for concealment, its internal seacocks were opened in a controlled sequence, flooding the compartments from the bottom up. The pumps that had kept the interior dry during construction were switched off.

The Phoenix settled at a rate determined by the flooding sequences, and if the sequence was wrong, the structure could settle at a tilt, making it more difficult to refloat later. A Phoenix that settled at an angle into the tidal mud had to be freed from the mud at that same angle before its attitude could be corrected, and freeing a 4,000-ton concrete box from estuary mud at an angle was not a simple operation. The crews who carried out the sinking operations were Royal Navy personnel who had trained on one of the completed Phoenix units before it was launched. The training was brief because there was no approved procedure for this particular operation.

The procedures were developed by trial on the first units and refined as problems emerged. When a Phoenix sank with a slight tilt due to a poorly calibrated flooding sequence, the engineers adjusted the sequence for the next unit. When a Phoenix settled into unusually sticky mud and resisted the first refloat attempt, the engineers added a compressed air step to the refloat procedures. The workers who dealt with these problems did not document them in official reports as problems.

They documented them as modifications to working procedures. The Whale floating roadways were manufactured in steel fabrication shops across Britain. The pontoons were welded steel boxes of standard dimensions that could be mass-produced in any facility with the right equipment. The hinged deck sections that connected them required precise tolerances at the pivot points so the sections could flex without jamming as wave action moved adjacent pontoons up and down.

Steel fabrication shops that had been producing hull components for other military programmes were retooled for Whale production over the winter of 1943 to 1944. The Lobnitz pierheads were assembled from structural steel sections manufactured in factories in Scotland and Wales. Each pierhead consisted of a steel deck frame supported by four hollow steel legs, each fitted with a rack-and-pinion lifting system that allowed the deck to be raised or lowered as needed. The legs were designed to be driven into the seabed by a piling attachment mounted on the pierhead itself, after which the deck could be raised to the height the tide required.

The design was tested at a prototype facility off the Scottish coast in the autumn of 1943. The test confirmed that the lifting mechanism worked as designed under tidal conditions, and that the leg structure was adequate for the loads imposed by a supply ship moored alongside. The Bombardons were built in the steel yards on the Clyde, where the working capacity and available water for final assembly were the greatest. They were the last component to have its design finished, and production began later than for the Phoenix units, which meant they were completed in April and May of 1944 with less margin than any other element of the programme.

On 6 June 1944, Allied forces landed on five beaches along the Normandy coast and established a foothold. The operational plan for the Mulberry harbours required the towing of the Phoenix units to begin within 48 hours of the landing, before the beaches were fully secured and while German defensive positions were still active in some sectors. The tow fleet assembled in the southern English ports on 7 June. It consisted of more than 150 deep-sea tugs from Royal Navy resources, commercial towing companies operating under the Ministry of War Transport, and vessels from the merchant fleets of several Allied nations.

The tugs were matched to the caissons based on towing capacity. A 6,000-ton Phoenix required at least two large ocean-going tugs working in tandem. The smaller units could be handled by a single tug. The Phoenix units that had been sunk in their assigned estuaries were refloated in the days just before D-Day.

The refloat crews moved from unit to unit along the estuaries, connecting compressed air lines to the hull fittings, operating the pumps, and waiting for the slow upward movement that indicated the Phoenix had broken free of the mud. The mud was the primary obstacle. A Phoenix that had settled into soft estuary sediments for six weeks had sunk into them across its entire underside, held by the suction force against the hull. Compressed air was used to break the suction at the interface between the hull and the sediment before the internal pumps could generate enough upward pressure to float the structure.

When the caisson rose, it rose at an angle in most cases, because the mud released unevenly along the hull. The refloat crews corrected the tilt by selectively flooding and pumping individual compartments until the deck was level. Then the tow ropes were attached and the Phoenix moved downstream toward the assembly anchorage. The crossing of the Channel took about 20 hours at a speed of 4 knots.

4 knots is walking speed. A person walking at a normal pace covers roughly 4 knots. The crossing distance from the English assembly point to the Bay of the Seine was about 80 miles. At 4 knots in a rough sea state with prevailing westerly winds pushing waves into the Channel from the Atlantic, the journey was not comfortable.

A Phoenix had very little freeboard. The deck rose only a few feet above the waterline. In any significant waves, the deck was awash. The tow crews who accompanied the Phoenix boxes stayed below in the access hatches for almost the entire crossing, listening to the sound of water hitting the hull and the vibration of the tow ropes transmitted through the structure every time the tugs changed speed or direction.

The tug masters handling the tow were working with loads close to the limits of their practical experience. In cross-currents, a Phoenix did not track steadily behind the tug. It yawed. When it yawed, the tow load came at an angle, and an angled load on a tow rope was a higher stress condition than a direct load.

The masters handled the angle by adjusting the length of the rope. In poor visibility and cross-tidal currents, they navigated by compass and radio coordination with the convoy control vessel. The convoy arrived in the Bay of the Seine in stages over the week from 8 June to 14 June. The components of the American Mulberry harbour headed to the anchorage off Omaha Beach.

The British components headed to Arromanches. At each site, naval construction crews were waiting. The construction operation at Arromanches was directed by naval officers who had spent months memorizing the positioning charts. Each Phoenix had a site number in the breakwater plan, and each number had specific coordinates as compass bearings from survey markers on the shore.

The construction officer on a work platform directed the tug masters by radio to the position, and when the Phoenix was over the mark, the seacocks were opened. It settled into place within 15 minutes. The positioning had to be timed to the tidal window. Tidal currents in the Bay of the Seine ran up to 2 knots.

There was a period of about 4 hours around each slack water when the current was slow enough to allow the tug to hold the Phoenix in position while it sank to the bottom. Outside that window, the current pushed the box out of alignment. The construction crews scheduled each placement within the tidal window and carried out as many placements per day as the tide table and the arrival of Phoenixes allowed. The first Phoenix was placed on 9 June.

By 14 June, a sufficient portion of the breakwater was in position that the harbour anchorage inside the Phoenix line was partially protected. Supply ships entered the anchorage and began unloading. Lobnitz pierheads were placed at two of the three planned locations. One Whale roadway was operational to the shore.

Trucks were driving across it. The harbour was not complete on 14 June. It was working well enough to be useful. The construction crews kept working.

On 19 June 1944, the weather forecast for the English Channel indicated deteriorating conditions. By mid-morning on the 19th, wave height in the open Bay of the Seine reached 5 feet and was rising. By midnight, sustained wind speed in the Channel exceeded 25 knots. By the morning of 20 June, wave height in the bay exceeded 12 feet in exposed areas.

The tidal range, which regularly exceeded 18 feet at Arromanches, amplified the storm’s effect on the coast. The American harbour at Omaha Beach was the hardest hit. The Bombardon barriers in the Omaha sector had been placed later than at Arromanches and were not fully moored when the storm arrived. Several barriers broke free of their moorings in the first night and were blown and drifted out to sea.

A loose Bombardon in a storm was a 200-foot steel structure moving with the current and wind at unpredictable speeds. Three of the Bombardons that broke free at Omaha collided with the Phoenix breakwater line. The impacts cracked two caissons and sank a third. Water flooded a breached Phoenix, it lost buoyancy, and settled to the seabed with its upper part protruding above the surface, breaking the continuity of the breakwater wall.

Through that gap, wave energy leaked into the harbour basin. The Whale floating roadways at Omaha were not fully anchored to the shore. The beach anchors, steel stakes driven into the sand to secure the landward end of the roadway, had not been installed on the three planned roadways when the storm hit. The roadways, floating free at the shore end, were caught by the waves surging through the gap in the Phoenix line and driven toward the pierheads.

The weight of a Whale roadway, consisting of hundreds of feet of linked steel pontoons, striking a Lobnitz pierhead at storm speed, was enough to bend the pierhead’s leg structure. Two pierheads at Omaha were put out of service. The supply ships anchored inside the harbour, which had been considered protected water, were now in open water with wave heights exceeding 6 feet. The moorings of most ships held, but the motion in the anchorage during the storm was violent enough to prevent ships from berthing safely at the damaged pierheads.

The ships remained at anchor, and when the storm subsided, they found that the pierheads they had relied on for unloading had been destroyed. The storm lasted 3 days. It subsided on the evening of 22 June. According to weather records, it was the worst June storm in the English Channel in 40 years.

American engineers walked through the harbour at Omaha and counted what remained. The results were not encouraging. Two of the Whale floating roadways were destroyed. Two pierheads were damaged beyond quick repair.

Three Phoenix caissons were breached or damaged. The inshore area was partially blocked with the hulks of supply ships that had run aground and broken up during the storm. The engineers concluded that rebuilding Mulberry A would take longer than the supply situation allowed. The alternative was to salvage what could be saved from Omaha and transfer it to reinforce the British harbour at Arromanches.

The recommendation was approved. Mulberry B at Arromanches had survived better than its American counterpart. The Phoenix breakwater at Arromanches was further advanced in construction when the storm arrived, and the partial enclosure it provided reduced the wave energy inside the harbour enough to make the damage, though serious, repairable. Three Phoenix units were damaged.

Several sections of Whale roadways broke free. The Bombardon moorings held at most positions. The Royal Engineers assessed the damage and began repairs immediately. The salvage operation at Omaha transferred 23 intact Phoenix units and several hundred feet of Whale roadway to Arromanches during the two weeks after the storm.

The American harbour was closed. The British harbour was repaired and expanded with the salvaged components. Mulberry B operated from late June until November 1944, when the Allies had advanced far enough to use the major French ports. During those five months, the harbour at Arromanches unloaded two and a half million soldiers, 500,000 vehicles, and 4 million tons of supplies.

The daily throughput capacity, designed for 7,000 tons, regularly exceeded that figure. When the harbour was taken out of service, the Whale roadways were dismantled. The Lobnitz pierheads were removed. The Bombardon units were recovered.

The Phoenix caissons were left in place. The number of workers who built the Phoenixes ran into the thousands across more than 20 sites. The workforce included concrete pourers, steel fixers, formwork carpenters, equipment operators for cranes and mixers, and labourers who did tasks that did not fit a specific trade title. They were paid under wartime wage controls.

The standard rate for an unskilled construction worker in Britain in 1943 was about one pound and ten shillings a week, higher than before the war because demand for construction labour exceeded supply. The workers were subject to the Essential Work Order, wartime legislation that prevented workers in designated essential industries from leaving their jobs without government permission. A man assigned to build a Phoenix who decided in the winter of 1943 that this was not the work he wanted could not simply leave. He had to apply for a transfer through the Ministry of Labour, and transfer requests from essential workers were not granted easily.

The practical choice was to stay or find a reason compelling enough to satisfy the ministry. The secrecy requirements imposed a social constraint unprecedented in peacetime. These workers could not discuss their duties with anyone outside the site. They could not tell their wives what they were building.

They could not explain to a friend in a pub what they had done that day. The Official Secrets Act was not a formality for them. Violations were met with prosecution. The atmosphere of a nation at total war made additional enforcement unnecessary in most cases.

Most of these workers went home after the war and returned to careers in construction. Some of them eventually saw pictures of the Normandy harbour in newspapers or documentaries and realized, perhaps for the first time, the purpose of the structures they had poured. The workers at the inland sites, where Phoenix units were built in a converted reservoir and towed to the coast, may never have made that connection at all. The Phoenix unit they built was underwater for most of its existence.

It reached France already sunk, then was refloated, towed across the Channel, and sunk again. They never saw it standing in the water at Arromanches. They never saw it doing the job it was built for. The remaining construction records in the National Archives at Kew describe the engineering of the Phoenix programme in detail.

They specify concrete mix designs, sheet-pile penetration depths, seacock specifications, and internal compartment pressures. They describe structural tests on the initial units. They describe the modifications made to the design after early construction revealed problems. What they do not contain, in any systematic form, is a list of the workers’ names.

Names appear in the records only when something goes wrong that requires documentation. Injuries appear in accident logs when they are serious enough to require medical care or affect the production schedule. Deaths appear in the records when they require an administrative response. The total number of construction-related deaths in the Phoenix programme has not been established from the surviving records.

It is believed to be small relative to the size of the workforce and the duration of the project, because the construction of temporary cofferdams, though hazardous, is not as consistently dangerous as work in open water. But that remains a belief, not a known fact. The record was not kept in a way that allows an accurate count. There is one document worth mentioning.

The Hamble site on the riverbank kept a foreman’s log. The log survives in fragments in a local archive in Southampton. The surviving portions cover about four months of construction, from November 1943 to February 1944. During those four months, the log records concrete pour quantities, formwork setup times, curing delays, supply deliveries, and weather conditions.

It records the dates on which specific sections of two of the Phoenix caissons were completed. It records the date on which the first of those caissons was towed to its mooring position in the river. The log does not record the names of the workers. It records crew designations.

Crew A, Crew B, Crew C. The foreman, whose name appears on the cover of the log, is named. The workers are not. In the February 1944 entries, there is a note that Crew B was short one man for three consecutive shifts following an incident in which a worker was struck by falling formwork and required hospital treatment.

The log notes that a replacement was obtained from the labour pool the morning after the incident, and the crew returned to full strength. The worker who was injured is not named. The hospital he was taken to is not named. Whether he fully recovered, whether the injury affected his life after the war, whether he ever learned the purpose of the Phoenix he was building, is not recorded.

The Phoenix that Crew B was building at the Hamble site in the winter of 1944 was completed, towed to its mooring position, sunk, then refloated, towed across the English Channel at four knots, and sunk in the Bay of the Seine. It has been there for 80 years. The Phoenix caissons at Arromanches are still there. The tides of the English Channel have worked on them for 80 years from every direction.

The intertidal zones, the parts of each caisson’s walls exposed to air at low tide and submerged at high tide, have been colonized by barnacles and green-brown algae that grow on any fixed surface in shallow Channel water. The reinforcing steel inside the concrete has been corroding since 1944, and where the corrosion has expanded enough to crack the surrounding concrete, pieces have spalled off and fallen away. The affected areas are visible as irregular patches of exposed aggregate on the faces of the caissons. The original profiles of the walls at the top of each unit have been reduced by erosion.

Some of the caissons now stand several feet lower than they did in 1944. After 80 years of this, the Phoenixes still stand. You can see them from the beach at Arromanches. At low tide, they are not far from shore.

The grey concrete of the Phoenix walls rises above the waterline, irregular and broken at the top, covered with marine growth at the tidal zone. If you stand on the beach and look toward the northwest, you are looking along the approximate alignment of the breakwater. The structures you see are the main wall of a harbour that no longer works, but has not moved from its original position. The Arromanches 360 museum on the cliff above the beach shows original footage filmed by military cameramen in the summer of 1944.

In that footage, the Phoenix breakwater looks complete and solid, trucks move on the Whale roadways, and ships lie at anchor in the protected water behind the wall. You can watch that footage, then walk down from the hill to the beach and look at what remains. The gap between the two versions of the same thing is something you have to see for yourself to absorb. At Omaha Beach, the remains of the American Mulberry harbour are different in nature.

The heavy seas threw the Phoenix caissons and Whale sections that broke up during the storm of 19 June onto the shore. Over the decades, sand covered those remains. Some are visible at low tide, grey concrete masses and rusted steel protruding from the beach. Others are completely buried and appear only at exceptional tides.

They are unmarked. Nothing explains what they are. Many visitors to Omaha Beach pass them without realizing what they are. They assume they are rock formations, or wreckage from the fighting, or natural features of the beach.

They are the wreckage of a harbour built in secret in Britain by workers no one told the purpose of, towed across the English Channel under fire, and destroyed by a June storm in three days before it could complete its mission. Brigadier Bruce White, who directed the engineering programme, was knighted in 1946 for his contribution to the war effort. He lived until 1976. He was photographed in his later years at Arromanches, standing on the beach, looking at the line of Phoenixes in the water.

In that photograph, held in the Imperial War Museum archive, he looks like an elderly man in a coat, looking at grey concrete in grey water. He does not appear to be thinking about engineering, though engineering is what made him who he was. He appears to be looking at something that is hard to look at directly, and yet you cannot help but look at it. The workers who built the Phoenixes are not in that photograph.

They were not at Arromanches in June 1944, except for the few who joined the construction teams that placed the caissons. They were in places like the Clyde, the Thames, the Hamble, and the Midlands reservoirs. They were pouring concrete in winter conditions inside formwork they were not allowed to describe to anyone outside the cofferdam. They worked in Crews A, B, and C, for men whose names appear in the foreman’s log above their own.

The concrete they poured is now in the Bay of the Seine. The tides have moved across it for 80 years. The barnacles have grown over it. The salt has worked on the reinforcing steel inside it.

None of it has moved them from where the construction crew placed them in the summer of 1944. If you have ever crossed the English Channel by ferry from Dover to Calais, then driven south and west on the French motorways toward Normandy, you know the quality of the light in that part of France in the morning. The sky over the Calvados coast has a particular shade of blue-grey, and the grass in the fields beside the road has a particular shade of green, and the light on the water, when you come over the hill above Arromanches, is different from the light anywhere else you have been. You are looking at a bay where men built a harbour.

The harbour is absent in the sense that it no longer works. But it is not absent in the sense that its structure still stands. The Phoenix caissons that the refloat crews freed from the mud of the Hamble, the mud of the Thames, and the mud of the Clyde, and towed through 20 hours of open Channel, are in the water below you, curving around the bay in the formation the construction crew placed them in a tidal window in the summer of 1944. The men who poured the concrete are not there.

The men who tied the reinforcing steel are not there. The members of Crew B at the Hamble site, who were short one man for three shifts in February 1944, are not there. Their names were never in the records. The record kept what needed to be kept, which was the engineering, and the engineering was in the concrete, and the concrete is still there.

The Channel is still there. The crews are not.