Two large passenger ships emerged from the war still afloat. On September 10, 1945, one of them lay anchored at Maizuru on the Sea of Japan, its holds empty and red crosses still painted on its sides. That ship was the Hikawa Maru, built in Yokohama in 1929 for the Seattle route, seized by the navy two weeks before Pearl Harbor, and fitted out to carry patients. She struck mines three times over four years of war and made it home each time thanks to hull plates thicker than anything the shipyard had produced for other vessels.

That morning she took on supplies and put to sea under orders written 11 days earlier. There were six and a half million Japanese outside Japan, and someone had to go and bring them back. Nearly half of them were in uniform, spread across coral islands, the hills of Luzon, along the Yangtze, in Manchuria and Java. The rest were farmers, clerks, teachers, and their children, scattered across an empire that had taken 50 years to build and was lost in eight days.
None of them could walk home. Every one of them had to cross salt water on something, and that something did not exist. The paperwork came before the ships did. Receiving ports were named that autumn, quarantine duties were assigned, and one name on the list was Uraga, a small town on a peninsula at the mouth of Tokyo Bay.
After 17 days at sea, the Hikawa Maru anchored off Mili in the Marshall Islands. The garrison there had been bypassed in 1944 and left to itself, cut off from its supplies, never worth the cost of an assault. Its men spent a year and a half growing what they could eat in coral sand and listening to planes that had stopped attacking them. The ship took on board her passengers and sailed the same day.
On October 7, she landed her passengers at Arawe, where a regiment of Marines stood on the pier watching them come ashore. By the end of the month she was at Wake, and after Wake came Kossol, and after Kossol came Wewak, Rabaul, Morotai, Halmahera, and Manado, island after island, anchorage after anchorage, in a list that read like a table of contents for the entire Pacific campaign. The war had been over for six weeks, and the islands were still full of soldiers. In the southwest Pacific alone, the surrendered garrisons numbered hundreds of thousands of men who had been counted as defeated for two years and had now suddenly become an administrative problem tied to shipping requirements.
Not everyone was pleased about it. American troops waiting to go home complained that ships were being given to the people who had been shooting at them in August. Officials answered that Asia would not settle while its garrisons remained, and that leaving those men where they were would keep more Americans overseas, not fewer. The ship’s file number under the new arrangement was H022, and she carried it printed on her for the next two years.
The arrangement itself had a title so bland it barely counted as one. The Japanese Merchant Marine Shipping Control Authority was established under U. S. Navy command that October.
Its first job was to determine what Japan still had. That proved harder than it sounded, because the Navy Ministry had burned most of its records in the last week of the war. The census had to be carried out by moving from port to port and asking local officials what existed and where. The officers asking the questions noted with surprise that the answers were accurate.
The ministry had destroyed its papers to hide them from the Americans, then had to tell the Americans what was in them. The count found afloat one intact fleet carrier, one light carrier, a pair of light cruisers, some 30 destroyers, and more submarines than anything else. On December 4, it was issued as a printed list of every surviving ship worth fuel. The commercial shipping situation was worse.
What survived was wrecks and small craft: thousands of wooden coasters and harbor boats under 100 tons, almost none of them capable of moving a load of people across an ocean and landing them alive at the other end. The Navy’s answer was to rebuild the fleet. At a conference in Tokyo on December 7, a recommendation was made to hand surplus American shipping to the defeated enemy directly, to be operated by that enemy under American control until the job was done. Nothing like it had ever been attempted, and there were no legal mechanisms ready for it.
What arrived were 106 Liberty ships and 100 landing craft of the flat-bottomed kind with doors in the front. Each one had already been earmarked for scrapping or lay-up. Eighty-five of the landing craft stayed on homeward voyages, and the rest went to supply work in Korea. Japanese yards converted them for passengers, building bunks in the cargo spaces and washing arrangements on the decks.
Instruments and hatch markings were rewritten in kanji so that the new crews could understand what they were looking at. The converted ships entered service at a rate of 25 per week. Six Liberty ships were rebuilt as floating hospitals with more than a thousand beds each. American officers supervised the training of the crews who would take over, teaching a commercial service that had been destroyed beneath them how to operate ships built to another country’s designs.
By the end of March, the authority had enough carrying capacity of its own to release the Seventh Fleet from the task entirely and let it return to being a warship fleet. The crews were the other half of the problem, and the solution there was to stop the clock. The Navy Ministry halted the discharge of thousands of its sailors so that there would be men on board who understood the ships, which meant that the men carrying the enemy home were the same men who had been fighting them two weeks earlier. What remained of the Imperial Japanese Navy went back to sea with its guns removed.
The carriers Hosho and Katsuragi, two of the few that had survived the bombing, worked as ferries. Destroyers worked as ferries too. So did submarines, stripped of their weapons and unable to dive, sent out on short daylight runs on which nobody had to sleep aboard. Something else was moving in the opposite direction.
More than a million Koreans and Taiwanese inside Japan, brought in to work during the war, were waiting to leave, and they left on the same voyages that came in loaded. The American homeward run was given the name Operation Magic Carpet and a place in every account of that period. This operation ran for two years under an abbreviation and received neither. The Rising Sun flag had been abolished, which raised the question of what flag these ships should fly.
The answer came from the International Code of Signals. Ships owned by Japan and crewed by her sailors flew a blue-and-red flag taken from the signal flag Echo. Ships owned by the United States but manned by the same crews flew red-and-green triangles taken from Oscar, so that any Allied ship or aircraft could identify them at a glance. Tokyo handled feeding and supplying the crews.
Fuel had to be purchased through American authorities, which kept an American hand on the controls of the entire operation without the need for issuing orders. Cargo ships were deliberately kept out of the transport pool, because a country receiving no food could not spare the few ships feeding it. When two of those cargo ships docked in Hawaii for repairs, the men aboard were not allowed ashore. Then there was the water itself.
General Order No. 1, issued on the day the surrender was signed, dissolved the Japanese armed forces and at the same time instructed the government to keep its minesweeping organization, because the approaches to the home islands were a killing ground and everyone connected with it knew that. American B-29s and submarines had laid more than 10,000 mines in those waters. Japan had laid several times that number in its own defense and had removed almost none of it.
From November, the sweepers operated under an American task group and worked to a set plan for the first time since the surrender. Ten thousand former sailors and a few hundred small ships served under a mine-warfare specialist named Tamura, who was exempted from the purge of officers because there was nobody else who could do the job. That work went on for years and took men the whole time. Ships kept sinking a few hundred yards from shore in sight of the people waiting for them.
One ship struck a mine near the coast of China with 4,300 people aboard; 20 were lost, which counted as a good day in those conditions. The routes doubled during the winter. Ships came from the Pacific islands, from Shanghai and Canton, landed from Keelung, and by spring the northern routes were opening too. Then in April, a ship arriving from Canton brought cholera into Uraga and entered port before anyone realized it.
Twenty ships were held at anchor offshore with nearly 70,000 people aboard, close enough to shore to see the hills behind it. Some 400 people died there without ever setting foot in Japan. The Public Health and Welfare Section under Crawford Sams, an Army doctor who had spent the previous autumn arguing that Japan’s pre-war quarantine stations had decayed during the war and needed rebuilding before the sailing season, got its stations. Ships coming from infected Chinese ports were routed to Uraga, Sasebo, and later Hakata, the three places where facilities could be operated.
Uraga earned its place because it sat on a small peninsula and could be isolated from everything around it, which was the right answer for disease and the wrong answer for weather. The anchorage was exposed, and the people waiting there took seasickness from the side for weeks. Laboratory equipment capable of processing thousands of bacterial cultures a day was assembled at Uraga within a few weeks, with holding sheds built beside it. On the piers, every passenger and every piece of luggage was sprayed with DDT powder, which is why photographs from those months show people standing in white fog with their arms held away from their bodies.
Formosa presented a particular kind of trouble. Hundreds of thousands of Japanese civilians there had been born on the island and had never seen Japan, and most of them wanted to stay. The Chinese authorities intended to keep the useful ones and charge the rest for their passage and inoculation, both of which the United States was providing for free, producing a series of American protests that came to nothing. The northern half of the map was worse and slower.
Soviet forces had entered Manchuria in the last week of the war and showed no interest in moving anyone out of it. Evacuations there did not begin until the Soviets started withdrawing in May 1946. When they did begin, they depended on American ships through a single port, Huludao, on the Manchurian coast. People reached that port on foot.
Some had walked from near the Soviet border, more than a thousand miles, and had lived through harsh conditions during a Manchurian winter before any ship was available to carry them. On the Japanese side, ships arrived at Hakata, Senzaki, Sasebo, Uraga, Maizuru, Kagoshima, and a set of smaller places, each containing sheds, a disinfection station, and a special pier extending over the water for people to cross on. Who actually ran the Shipping Control Authority is unresolved in the American files. A Pacific Fleet report from February 1946 names Admiral Donald Beary as its administrative head.
A Navy biography of Charles Momsen says he was ordered to assume the same post after the surrender and served in Japan from November 1945 to October 1946. A later account tries to reconcile the two and places Beary in command with Momsen as his chief of staff. The three accounts do not agree, and nobody at the time seems to have thought it worth settling. It was Momsen who put the estimates on paper.
Initial calculations suggested the task would last until July 1947, which everyone considered reasonable for an operation that began with two usable passenger ships and a fleet of old coastal boats. In March 1946, he wrote that the work would end in May of that year. He was close enough. Within a year of that memo, more than six million people had been landed in Japan, and the ships had also returned the remains of thousands of Japanese war dead from islands nobody intended to visit again, something the government in Tokyo had requested and had not expected to receive.
The Hikawa Maru completed her final government voyage in January 1947. She returned to carrying cargo through the lean years, then passengers, reopened the Seattle run in the 1950s, and was retired in 1960 when people stopped crossing the Pacific by ship. She is moored today in Yamashita Park in Yokohama, where schoolchildren walk across her decks, the last great Japanese passenger liner of her generation still standing, with nothing on her indicating to anyone what she was doing in the autumn of 1945. The forecast contained one exception written into it, and that exception was the only thing that outlasted everything else in the file.
The people in Soviet hands were not coming, whatever timetable the Navy wrote for them. Manchuria, Sakhalin, the Kurils, then the camps, and still men were walking off the pier at Maizuru in 1958. A service commendation from that year credited its recipient with an achievement no other American officer in the war could claim: the safe and rapid return of millions of displaced people to their homes. The man who had spent the 1930s devising ways to rescue 33 sailors from a sunken submarine managed to evacuate six million people from a continent in eight months.
The only part of his calculations that failed was the part he himself had identified as the one he could not reach.